ITEGAM-JETIA

Compact Spiral Transformer Design for High-Efficiency Interleaved Flyback Converters in Solar Power Systems

Benzidane Mohammed Ridha — 2026-03-24

This paper explores the design, modeling, and optimization of a spiral transformer integrated into an interleaved flyback converter tailored for photovoltaic (PV) applications. The transformer features two square planar spiral coils with an outer diameter of 9000 µm and an inner diameter of 6000 µm, yielding primary and secondary inductances of 0.128 µH and 0.226 µH, respectively, at a 5 MHz operating frequency. The miniaturized design includes 3 primary turns and 4 secondary turns, with conductor widths of 500 µm and 300 µm, respectively. An electrical model accounts for parasitic effects, such as series resistances (0.195 Ω for primary, 0.282 Ω for secondary) and capacitances (inter-turn and oxide layer), which are minimized to boost efficiency. Frequency-dependent behavior is analyzed using MATLAB, identifying a resonance frequency of 20 MHz and a coupling coefficient of 0.87. The interleaved flyback converter, simulated in PSIM, delivers a stable 48 V output voltage and 3.2 A output current from a 162 W input, achieving 95% efficiency. This work validates the potential of compact, high-efficiency transformers for advancing PV energy conversion systems.

Development and Implementation of an Industrial IoT-Based Real-Time Process Monitoring System to Improve Overall Equipment Effectiveness (O.E.E.) of Bearing Manufacturing Process

Manan Bhavesh Raval — 2026-03-24

The research study presents the design and implementation of an IoT-based real-time monitoring system to enhance Overall Equipment Effectiveness (O.E.E) of External grinding process used in precision bearing manufacturing. The developed system continuously monitors critical parameters such as cutting temperature, machine vibration, noise and cycle time using calibrated sensors. These critical parameters are directly linked to Key Performance Indicators (KPIs) of O.E.E in terms of Availability, Quality and Performance. A systematic layered IoT architecture is developed to enable real-time data acquisition, cloud storage, visualization and alert generation through a customized web application. Controlled experiments were conducted on various batches of bearing races under fixed machining conditions with critical parameter thresholds defined based on ISO standards and historical production data. The system successfully identified deviations, enabling timely corrective actions and reducing unexpected breakdowns. The real-time dashboard and alert system provided actionable insights, improving operational decision-making and minimizing manual documentation. As a result, OEE improved from 86% to 94.70%, demonstrating an 8.70% increase. Additionally, reductions in breakdowns, defects, and downtime were observed. This study confirms that integrating IoT in manufacturing significantly enhances equipment performance and product quality through data-driven process control.

Performance-Driven Optimization of CMOS-Based Two-Stage Operational Amplifier Using Metaheuristic Algorithms

Sureshbhai Laxmanbhai Bharvad — 2026-03-24

Design of CMOS based analog circuits becomes increasingly complex as transistor sizing plays a crucial role due to the trade-offs among power consumption, silicon area, unity gain bandwidth, slew rate, and open loop gain. This sizing challenge is makes analog circuit design inherently multi objective, and traditional analytical approaches based on simplified transistor level equations often fail to deliver globally optimal results. Metaheuristic optimization techniques have emerged as an effective alternative to explore nonlinear and multi-dimensional design spaces. In this work, the design of a two stage CMOS operational amplifier in the Predictive Technology Model (PTM) 45 nm technology node is optimized using four algorithms: Particle Swarm Optimization (PSO), RAO algorithm, Teaching Learning Based Optimization (TLBO), and the proposed Modified TLBO (MTLBO). The algorithms were implemented in Python and verified through Ngspice-26 simulator on an AMD Ryzen™ processor with 16 GB RAM, 64 bit Ubuntu environment. The proposed MTLBO achieved 86.15 dB voltage gain, 94.05 dB CMRR, and 185 MHz unity gain bandwidth. Comparative analysis shows that the proposed MTLBO algorithm achieves faster convergence with fewer iterations and consistently outperforms PSO, RAO, and TLBO making it a strong candidate for efficient analog VLSI design automation.

Navigating Agile Implementation in Yemeni Software Firms Challenges, Benefits, and Strategic Insights

Ali Balaid — 2026-03-24

Agile methodologies, emphasizing people, collaboration, and shared values, have gained significant traction globally. A thorough understanding of the challenges and benefits of implementing Agile practices within software development companies is vital for the progression of the field and will be the foundation for its advancement in practice and research. Yet there is a scarcity of empirical studies and a limited understanding of the key challenges and benefits that could be gain from Agile implementation, particularly within Yemeni companies. Therefore, this study aimed to investigate the challenges and benefits of implementing Agile practices within Yemeni software development companies. A mixed approach of analysis was used in this study which, started by a rigorous and systematic analysis process of literature followed by a qualitative and quantitative field study. Overall, the key challenges, identified from this process were organizational culture resistance, insufficient executive support, inadequate employee skill sets, poorly defined project scopes, and technological limitations. Conversely, the primary benefits observed were increased flexibility, accelerated time-to-market, reduced costs, and enhanced team communication. This research provides valuable insights to guide Yemeni software development companies in making informed decisions and developing effective strategies for successful Agile implementation.

Experimental Validation of Vibro separator and ANOVA technique for Amplitude

Pavan Maheshchandra Bhatt — 2026-03-24

Over 60% of India's gross domestic product (GDP) is attributed to the agricultural sector. Advancements in mechanical and instrumentation technologies have significantly improved research and development in agricultural machinery. Vibration analysis plays a critical role in predictive maintenance by allowing engineers to detect potential issues in machinery before they lead to system failures. One important factor in vibration analysis is amplitude, which helps us understand how severe the vibrations are and whether there might be any mechanical issues. Amplitude is usually measured in terms of displacement (like micrometers or mils), velocity (millimeters per second or inches per second), or acceleration (g or millimeters per second squared). It’s directly connected to the amount of mechanical stress the equipment is experiencing. In this study, we used analysis of variance (ANOVA) to design experiments, create a regression model, and run tests that looked at how the vibro motor’s angle, its rotational speed, and the material’s modulus of elasticity all work together to affect amplitude. We then optimized these factors to improve the vibro separator’s flow rate performance

Optimization of Process Parameters for Horizontal Vibro-Separator with Dynamic Motion Analysis

Pavan Maheshchandra Bhatt — 2026-03-24

To optimize operating parameters, this paper conducted a dynamic motion analysis of a reciprocating vibro separator. This work examined the impact of motor speed and angle to determine the vibro separator's flow rate. For the dynamic motion analysis, a CAD model of the device was created using parametric software. Then, it was analyzed using a CAE tool at three distinct motor angles and speeds. Various motor angles (28°, 30°, and 32°) and speeds (1,000 RPM) were used in the dynamic motion investigations. In order to determine if a system is in periodic, quasi-periodic, multi-periodic, or chaotic motion, Poincaré maps, FFT (fast Fourier transform), and TDR (time-displacement response) were employed. A sensitivity-based uniaxial piezoelectric sensor was used to gather experimental data from the agriculture industry. Additional parameter changes were made to the computational model after computational studies were verified using real-time industrial data.The outcome of the present research leads us to conclude that the most suitable motor angle and speed for a horizontal vibro separator are 28° and 1050 RPM, respectively.

Electrochemical Potential and Characterization of NiFe₂O₄/Cu-1 and NiFe₂O₄/Cu-2

Naveen Chandra Joshi — 2026-03-24

Ferrite-based materials have been studied for the electrode materials for supercapacitors because of their low cost and rapid, reversible surface Faradaic reactions. In this study, we have developed two electrode materials named NiFe₂O₄/Cu-1 and NiFe₂O₄/Cu-2. The fabrication method was found to be inexpensive, effective, and efficient. The synthesized materials were analyzed via FTIR, XRD, SEM, and TEM methods. FTIR results indicated the presence of O-H, C-H, C-O, Ni-Fe, Fe-Cu, Cu-O, etc. bonds on the surface of these materials. The XRD patterns indicated the amorphous nature of NiFe₂O₄ and the semi-crystalline nature of NiFe₂O₄/Cu-1 and NiFe₂O₄/Cu-2. The SEM analysis indicated that the particles are semi-spherical, and some are irregular in NiFe₂O₄. The sphere-shaped particles were observed in the case of NiFe₂O₄/Cu-1 and NiFe₂O₄/Cu-2, respectively. The electrochemical study has been conducted using different electrochemical parameters. The NiFe₂O₄/Cu-2 exhibited a power density of 4992 W/kg, an energy density of 5.9 Wh/kg, a retention of 91.3%, and a specific capacitance of 222 F/g. It was additionally found that the NiFe₂O₄/Cu-2 was appropriate for device systems with a 42 F/g specific capacitance (under a 2-electrode system).

Comparative Analysis of IECC-QKD with Several Access Control and Security Approaches for Secure Authentication and Data Sharing in Cloud Environments

Akshay Agrawal — 2026-03-24

Blockchain has evolved a lot in the context of digital transformation, as it is decentralized and accessible. However, it has become extremely difficult to keep sensitive data secure in cloud storage and distributed systems. Traditional methods didn't do a good job of controlling who could access data, and there was a risk of unauthorized access, especially in centralized cloud systems, which raised concerns about data security and privacy. This study suggested an Improved Elliptic Curve Cryptography-based Quantum Key Distribution (IECC-QKD) framework that uses a blockchain system to protect data exchange and access control mechanisms in order to reduce security risks. The Edwards curve and binary randomness are both used in the IECC to make sure that the encryption and sensitive data participated on the cloud are safe from tampering. The QKD-based secure key uses quantum mechanics to keep the data safe from unauthorized access by making it impossible to tamper with. The new Proof of Storage (PoSt) consensus mechanism in the Ethereum blockchain also combines the best parts of Proof of Space (PoS) to store and prove the integrity of the data while using as few resources as possible.

Deep Learning Applications in Agriculture and Rural Development: Toward Smarter, More Sustainable Food Systems

Mohammed Mostefa Selt — 2026-03-24

The rapid evolution of deep learning (DL) has profoundly reshaped various activities such as computer vision, natural language processing, and autonomous driving. Agriculture, a typical labor-intensive and risky industry for traditional countries, is also experiencing an innovation by the incorporation of DL technologies that can process high-dimensional multi-source data to support decision-making. This paper discusses the current status research of DL technologies in agriculture and its potential impacts in the context of rural development. We will study the role of various models such as convolutional neural networks (CNNs), recurrent neural networks (RNNs) and hybrid transformer-based architectures in the context of crop monitoring & field monitoring, disease detection, yield prediction, irrigation management, and agricultural robotics. This paper brings attention to the fact that besides laboratory use, innovations in rural peoples' end-users of Artificial Intelligence (AI) created from DL yield livelihood dividends through increased productivity, lower risk and more sustainable resource management. But a host of challenges, from lack of data and infrastructure to ethical and institutional barriers, still stand in the way of widespread use of these tools in resource-poor settings. The article ends by discussing key research and policy considerations to narrow the divide between tech potential and practical impact, underscoring inclusive, transparent and participatory pathways of digital agriculture.

ANFIS-Based MPPT Strategy for Solar Water Pumping System

Islam Hachemi — 2026-03-24

This paper presents the modeling and control of a standalone solar water pumping system, comprising a PV generator, DC/AC inverter, squirrel-cage motor, centrifugal pump, hydraulic circuit, and MPPT controller. Four machine-learning techniques (ANN, ANFIS, SVM, and GPR) are assessed for predicting the MPP voltage to improve system efficiency. Comparative results show that ANFIS gives the highest prediction accuracy (R² = 0.9992 and RMSE = 0.9552), outperforming GPR and ANN, with SVM providing moderate performance during training and testing phases. The performance of predcitive models was evaluated using training and testing data, assessing accuracy in predicting optimal voltage based on irradiance and temperature. Dynamic simulations under step-change irradiance conditions confirmed that the ML-based MPPT controllers significantly outperformed conventional P&O and INC methods, achieving faster time response. Based on this analysis, an ANFIS-based MPPT strategy with a PI controller is used, leveraging its superior accuracy relatively to the other ML-based MPPT controllers and conventional methods (P&O and INC).

Development and Field Evaluation of an IoT-Based Smart Irrigation System for Vegetable Production in Semi-Arid Senegal

Pape El Hadji Abdoulaye Gueye — 2026-03-24

Agriculture plays a central role in Senegal’s economy, contributing about 8% to GDP and providing nearly 70% of economic activity. However, it relies heavily on seasonal crops and is increasingly affected by rainfall deficits linked to climate change. Irrigated agriculture offers an alternative but faces challenges such as limited water availability, labor-intensive manual watering, and low adoption of modern techniques.To address these issues, this study presents the design and implementation of a smart irrigation system that automatically determines crop water requirements based on field parameters (temperature, air humidity, and soil moisture) and local climate conditions. The system was tested on onions, tomatoes, and potatoes, three major crops in Senegalese market gardening.

By automating irrigation and enabling remote control through ICT tools, the system reduces manual labor, optimizes water use, and minimizes environmental impact. Over a one-month experimental period, the system achieved an average 21% reduction in water use compared to manual irrigation, with mean monthly requirements of 4.2 m³ for onions, 4.6 m³ for tomatoes, and 4.1 m³ for potatoes. These results confirm the efficiency and adaptability of the proposed IoT-based approach for smallholder farmers under semi-arid conditions.

Auditory Library Guide: A Digital Solution for Visually Impaired Readers

V. Thamilarasi — 2026-03-24

This paper's main objective is to use the Auditory Library Guide to assist visually impaired people in finding and accessing library books. People with visual impairments have trouble locating the book in the library. Although library services for these individuals were insufficient up until now, everyone now understands how important it is to make information accessible to those who are visually impaired. In light of this, a method is suggested and created to enable them to utilize library resources. Through the use of text-to-speech conversion and voice recognition technology, they can utilize the system to access the library and hear the contents of the books.The purpose of this study is to improve the social issues faced by visually impaired persons and to help them read library books like everyone else.

Progressive BAT for Neural Tuning (PBNT): A Bio-Inspired Hyperparameter Optimization Framework for Skin Lesion Classification

Shunmuga Priya K. — 2026-03-24

Advanced architectures and effective hyperparameter tuning are necessary to achieve high diagnostic accuracy, despite the fact that deep learning (DL) has become a critical instrument for automated skin cancer detection. This research suggests an optimised deep learning framework for the classification of multi-class skin lesions. The framework integrates state-of-the-art CNNs with a novel bio-inspired optimiser, Progressive Bat for Neural Tuning (PBNT). VGG16, ResNet50, EfficientNetV2, AlexNet, and DenseNet were assessed in experiments conducted on the ISIC 2024 (3D-TBP) and HAM10000 datasets. Bayesian Optimisation, Bat Algorithm, Grey Wolf Optimiser, and Firefly Algorithm were benchmarked against PBNT, and the performance was evaluated using the F1-score, precision, recall, and accuracy. PBNT consistently outperformed existing optimisers, with AlexNet-PBNT achieving a 99.0% F1-score, 99.1% recall, 98.9% precision, and 99.0% accuracy, surpassing all other model-optimizer combinations and recent benchmarks. The automated diagnosis of skin cancer is considerably improved by the integratioSSn of CNN architectures with hyperparameter tuning driven by PBNT. The AlexNet-PBNT model offers a clinically viable and extremely accurate solution for early detection.

HFMSTMC: Improved Binary Grey Wolf Optimization in Hybrid Fuzzy Minimum Spanning Tree Clustering with Manifold Learning

Dhanapriya L. — 2026-03-24

Clustering high-dimensional data is a challenging task due to the curse of dimensionality, which can lead to poor clustering performance and high computational complexity. Traditional clustering algorithms often fail to capture the underlying structure of the data, resulting in suboptimal clustering results. Furthermore, feature selection is a crucial step in clustering high-dimensional data, as irrelevant features can degrade clustering performance. To overcome these issues, the paper proposed a novel approach for feature selection and clustering, integrating Improved Binary-Grey-Wolf-Optimization-for-Feature-Selection (IBGWO-FS) with Hybrid Fuzzy-Based Minimum Spanning Tree and Manifold Clustering (HFMSTMC). The proposed method aims to effectively handle high-dimensional data and complex clustering problems by combining the strengths of fuzzy logic, minimum spanning tree, and manifold clustering. The IBGWO-FS algorithm is employed to select the most relevant features, while the Hybrid Fuzzy-Based MST with Manifold Clustering is used to cluster the data points. Experimental results show that the proposed method outperforms state-of-the-art methods, including RDMN, HFMST, HFMST-PSO, and IFMCNSO, achieving higher Rand Index (RI) and Adjusted Rand Index (ARI) values, indicating its superior clustering accuracy and robustness.

Performance Optimization of EV Motor through Novel Modulation Techniques in SiC Multilevel Inverter

K. S. R. Deepika — 2026-03-24

Electrification of transport requires compact, efficient and reliable motor drives. Multilevel, Silicon Carbide (SiC)-based inverters (MLIs) have become an attractive new technology in the propulsion of Electric Vehicles (EVs) because of their high switching frequency, low conduction loss and high thermal stability. This paper provides an in-depth comparison of new modulation techniques, Optimized Space Vector PWM (OSVPWM), Hybrid Multicarrier PWM (HMCPWM), and Selective Harmonic Elimination PWM (SHEPWM) on SiC MLIs to for EV motor drives. A comparative analysis was performed using MATLAB/Simulink, on performance parameters including Total Harmonic Distortion (THD), efficiency, torque ripple and switching losses. According to the results, the OSVPWM method has the lowest THD (3.5%), and SHEPWM reduces switching losses by 18% than the traditional SPWM. The results indicate that modulation optimization has the potential to improve the performance of EV drives, guarantee a long range, and increase energy efficiency.

Improving Depth Image Quality via Super-Resolution and Artifacts Removal: CNN-Based Approach vs. Traditional Methods

Yagneshkumar Jayantilal Parmar — 2026-03-24

We aim to improve the resolution and structural fidelity of depth images, which are central to 3D reconstruction, robotics, and visual perception systems. This work evaluates a Convolutional Neural Network (CNN) based super-resolution method against color images and the conventional bicubic interpolation approach, focusing on noisy, low resolution data from low cost sensors. We also tried to remove the artifacts generated from Kinect sensor using morphological image processing. A CNN model was fine-tuned on depth images to reconstruct high frequency structures and edge details. Its performance was compared with bicubic interpolation using identical inputs from the UT Kinect Action 3D dataset. Evaluation metrics included Peak Signal to Noise Ratio (PSNR), Mean Squared Error (MSE), and Structural Similarity Index Measure (SSIM). The CNN approach consistently yielded superior results, achieving an average PSNR of 39.82 dB and MSE of 6.43 outperforming bicubic interpolation (35.20 dB PSNR and 19.54 MSE). Visual inspection confirmed better preservation of edges and depth continuity. Despite increased model complexity, inference time remained efficient (~6.2 seconds on GPU) compared to ~30.7 seconds for bicubic on CPU. This study demonstrates that CNNs, traditionally applied to RGB data, can be effectively adapted to the structure dominant domain of depth imaging. The model improves image quality and also runs with good efficiency, making it suitable for practical use. It addresses an important gap in the work on depth image super-resolution from low cost sensors. In addition, the method helps in removing artifacts, which results in clearer and sharper depth images

Performance Evaluation of Solar PV-Battery Hybrid System Powered BLDC Motor Speed Control Using Ultra-Lift Luo-Converter

G Dilli Harsha — 2026-03-24

The increasing use of solar energy poses challenges in meeting energy demands due to its unpredictable weather conditions. With the growing need for electric drives, integrating solar energy with battery storage through advanced power electronic converters is essential to ensure system stability and energy sufficiency. This design proposes a solar photovoltaic (PV) array and battery-supported brushless DC (BLDC) motor drive capable of operating efficiently and reliably under varying solar irradiance and temperature conditions. The solar PV array output voltage is boosted to the level required by the motor using an ultra-lift Luo-converter, while a bidirectional buck–boost converter manages the power flow between the battery and the DC bus. The BLDC motor is driven by a three-phase voltage source inverter (VSI), whose electronic commutation is controlled using Hall sensor feedback. Motor speed control is achieved through a Proportional–Integral–Derivative (PID) controller designed for the ultra-lift Luo-converter. By addressing the effects of varying irradiance and temperature, the proposed system ensures stable operation. Due to its modular architecture and hybrid energy management capability, the system shows strong potential for applications in solar–battery-powered Computer Numerical Control (CNC) machines and standalone renewable energy systems. The performance evaluation of the proposed BLDC motor speed control system is carried out in MATLAB/Simulink.

IoT-Based Remote Vital Signs Monitoring and Temperature Forecasting for Pregnant Women

Bilal Mokhtari — 2026-03-24

The importance of maintaining optimal health during pregnancy for both the mother and fetus has driven the development of numerous artificial intelligence (AI)-based monitoring systems. These systems aim to address the growing need for continuous, reliable health tracking in pregnant women, ensuring early detection of complications and promoting better outcomes. While general-purpose health monitoring platforms exist, there remains a significant gap in solutions explicitly tailored for pregnancy. Addressing this need requires not only real-time monitoring but also predictive capabilities based on vital signs. In this work, we propose an IoT-based pregnancy monitoring system that continuously collects key physiological data, namely body temperature, heart rate, and blood oxygen saturation. The collected data is transmitted in real time and processed using a Long Short-Term Memory (LSTM) neural network to build a model capable of forecasting potential health anomalies. The system provides real-time insights and future predictions. This approach enhances proactive care, enabling timely intervention and improving maternal-fetal health outcomes. This system’s approach shifts between personal and centralized monitoring, a capability particularly valuable where regular prenatal visits are difficult, thereby enhancing the overall effectiveness of prenatal care delivery.

Development of a Real-Time Monitoring and Control System for an Industrial Plant Using PLC and SCADA Technologies

Benarabi B. — 2026-03-24

Automated systems are technological solutions designed to operate with minimal or no human intervention, performing tasks with high precision and efficiency. By integrating sensors, actuators, and advanced control logic, these systems optimize operations across various domains, including residential, commercial, and industrial sectors. Their applications range from smart home devices to complex manufacturing processes. This study presents the design and implementation of a comprehensive automation and control system for an industrial milk processing plant. The system aims to optimize critical production processes, including water heating and cooling, milk storage, pasteurization, bottling, and cleaning operations. Programmable Logic Controllers (PLCs) are employed to control the machinery and equipment, while Human-Machine Interfaces (HMIs) and a Supervisory Control and Data Acquisition (SCADA) system provide centralized monitoring, control, and visualization. The proposed solution integrates both hardware and software components to deliver a user-friendly graphical interface for operational supervision and process management. The primary objective is to improve milk production efficiency, ensure process reliability, enhance product quality, and facilitate a seamless and intuitive operator experience..

A Short Review on Battery Thermal Management for EV Application

B Musthafa — 2026-03-24

Battery Thermal Management (BTM) has emerged as a critical aspect of electric vehicle (EV) technology, ensuring safety, performance, and durability of high-energy lithium-ion batteries. As EV adoption accelerates globally, maintaining optimal temperature ranges is essential to prevent thermal runaway, enhance charge–discharge efficiency, and extend battery life. Modern BTM systems integrate active and passive cooling techniques, including liquid cooling, heat pipes, phase change materials, and advanced insulation, to manage heat generation from electrochemical reactions. Innovations such as nano-enhanced composites, artificial intelligence (AI)-driven predictive models, and Internet of Things (IoT)-enabled monitoring are reshaping the field by enabling real-time thermal control and predictive maintenance. Furthermore, advances in solid-state batteries and energy-dense chemistries like nickel cobalt aluminum oxide (NCA) require robust thermal strategies to maintain safety under extreme conditions and during fast charging. Emerging trends show a transition toward lightweight, modular, and highly efficient thermal systems that minimize energy losses while supporting sustainable EV development. The integration of advanced materials, smart sensors, and machine learning will drive the next generation of BTM, supporting reliable energy storage and long-term environmental goals. This review synthesizes current technologies, challenges, and future opportunities, offering insights into designing efficient thermal systems for EV batteries.

Modelling, Verification, and Optimization of Process Plan Generation in Sustainable Reconfigurable Manufacturing Systems Using Evolutionary-Generated Petri Nets

Fareh mohamed elkabir — 2026-03-24

Various evolutionary algorithms have been developed to address environmentally oriented multi-objective process planning challenges in reconfigurable manufacturing systems, though many prioritize effectiveness over flexibility. Research utilizing Petri nets effectively addresses this flexibility issue, with several extensions proposed for modeling complex systems. By leveraging established Petri net theory, the process plan of reconfigurable manufacturing systems (RMS) can be encoded, enabling the optimization of process plan generation and the identification of potential deadlocks. This paper introduces a novel approach to enhance process plan generation, combining Evolutionary Petri Nets (a variant of Petri nets) with the enhanced genetic algorithm N˜SGA-III. The method optimizes four key objectives within the context of Sustainable Reconfigurable Manufacturing Systems (SRMS): total production cost, total production time, greenhouse gas emissions from energy consumption, and hazardous liquid waste generation. Numerical experiments were conducted to validate the effectiveness of this approach.

Leveraging Tumor Induced Frequency Shifts by A 24 GHz Microstrip Antenna with High Sensitivity for Breast Cancer Diagnosis

Yaqdhan Mahmood Hussein — 2026-03-24

Early detection of breast cancer is critical for patient survival, yet current gold-standard methods like X-ray mammography present significant limitations, including ionizing radiation. a new design of the microstrip patch antenna that is optimized towards early detection of breast cancer at 24 GHz through the benefits of millimeter-wave imaging, such as high resolution and non-ionizing radiation. The cross-slot method has been used with Rogers RT 5880 substrate which in the free space has shown good performance of 10 dB, efficiency of 95% and a wide bandwidth of 1.3 GHz (S11 < -10 dB from 22.8-24.1 GHz). The design was tested in three critical conditions using CST Microwave Studio simulation that included free space operation, multilayer human tissue (skin, fat, fibro) integration, and tumor-embedded conditions. Findings indicate the clear tumor signatures as a frequency shift of 700 MHz, bandwidth reduction to less than 3 MHz and radiation pattern distortions (beam splitting in multiple co-polarized lobes and increased backscatter). It is worth noting that tissue integration decreased gain to 9 dB, but tumor detection sensitivity (90% efficiency) was increased (through an increase in energy coupling) by high permittivity of the tumor. These quantifiable electromagnetic perturbations offer a viable, non-invasive substitute to traditional mammography, and have shown great potential to wearable diagnostic systems in form of the compact design, high sensitivity tumor and real-time monitoring. The paper takes microwave breast imaging one step further and determines definite correlations between changes in the parameters of the antennas and the presence of the malignant tissues.

Smart Classification of Submersible Pumps

Mourad Abed — 2026-03-24

Submersible pumps play a vital role in diverse industrial applications, ranging from water supply to the transfer of hazardous chemicals. Their classification, traditionally based on operational and design characteristics, often faces limitations in precision and adaptability. This study explores the integration of artificial intelligence (AI) and machine learning (ML) to enhance the classification and optimization of submersible pumps. Utilizing datasets of pump parameters, advanced ML algorithms like Random

Forest and Support Vector Machines were applied, achieving significant improvements in prediction accuracy. Experimental results highlight the inverse relationship between flow rate and head, as well as the impact of pump diameter on performance. The research underscores the potential of smart systems in creating adaptive, data-driven classification models that surpass traditional methods. Future work involves refining algorithms, expanding datasets, and incorporating real-time decision-making systems to address dynamic operational challenges. This approach offers a promising direction for improving the efficiency and reliability of submersible pump systems in modern engineering.

Groundedness-Aware Retrieval in Government Document Chatbots: Systematic Literature Review and Semantic Alignment Score (SAS) Formulation

Frendy Rocky Rumambi — 2026-03-24

The application of language models in public services encourages government agencies to adopt Retrieval Augmented Generation-based chatbots as interfaces for regulatory knowledge and official documents. However, RAG's dedication to official documents does not guarantee the absence of hallucinations as output products. RAG also does not reduce public trust and legal confidence. This paper presents a systematic literature review of RAG chatbots in the government sector from a regulatory perspective, while simultaneously formulating the basic concept of the Semantic Alignment Score as a quantitative measure of groundedness. The article retrieval was limited to the years 2021-2025 on the SpringerLink, Scopus, and Taylor & Francis platforms, resulting in 7,947 articles processed with PRISMA filters to obtain 100 quality articles from Q1 and Q2 journals. Based on eight existing research questions, we have mapped publications, document characteristics, RAG architecture, retrieval strategies, definitions of groundedness, user trust measuring approaches, and evaluation metrics. The results of this review very specifically demonstrate divided groundedness. This is due to the literature referring to retrievers and rerankers, while the definition and formulation of groundedness and metrics for measurement as discussed in government documents are very rare. Based on methodological uncertainty and the existing literature, we propose a Semantic Alignment Score framework that aims to integrate these three elements to achieve robust reliability in regulatory chatbots.

Deep Learning-Based Detection of Vernacular Heritage Houses in Sumba Island Using BSTO-VGG16-ALNN and Big Data Analytics

Nandhini P — 2026-03-24

In the conventional settlements of Sumba Island, the nearby local area progressively changes design to meet their consistently moving necessities. The social legacy, exemplified in vernacular houses, holds extensive interest for the travel industry. By grasping and advancing the social meaning of these houses, we can add to manageable the travel industry improvement, supporting the neighborhood economy and encouraging consciousness of the social lavishness of the district. As we dig into the examination of the social legacy inside vernacular houses, it becomes evident that these designs wrestle with difficulties like rot, underlying weaknesses, and natural tensions. Despite this, the ever-changing technological landscape presents promising opportunities for the preservation of these priceless cultural assets. The incorporation of progressively complex PC vision innovation, joined with the openness of high-goal remote detecting pictures, presents an extraordinary way to deal with exactly assess and quantify the many-sided subtleties of Earth's regular and fake conditions for enormous scope. In this undertaking, our work proposes a strategy using profound learning methods and huge information examination for distinguishing the social legacy of vernacular houses in Sumba Island, Indonesia. At first, we utilize the boosted sooty tern optimization (BSTO) calculation for target division, really isolating vernacular houses from the grave remote detecting pictures. Subsequently, we present the pre-prepared VGG-16 engineering to extricate highlights from the fragmented objective picture. Also, we carry out the adaptive learning neural network (ALNN) for the exact programmed discovery of vernacular houses. Utilizing Sumba Island tiles, we validate the efficacy of our BSTO-VGG16-ALNN method and demonstrate impressive results.

An Integrated Recommendation System for Customised E-Learning Using the BERT Model

Jatin Singh — 2026-03-24

Digitization of education has made education more accessible. With the growing accessibility, the primary challenge for e-learning is to customize the learning environment to the needs and preferences of the learners. The learning can be customized by considering features such as persona type, skill level, learning goal, learning style, educational background, past knowledge, and memory span of learners. The tailored learning environments enhance learner engagement and significantly increase the number of learners who successfully achieve their educational goal. This work presents a recommendation system using the Case-Based Reasoning (CBR) and the Rule-Based Reasoning (RBR) with the Bidirectional Encoder Representations from Transformers (BERT) model embedded for sentence classification. The proposed integrated system has an F1 score of 0.74, indicating the balance between making correct and useful recommendations, and a higher normalized Discounted Cumulative Gain (nDGC) of 0.81 shows that the system ranks the most relevant learning modules at the top of the recommendation list. The classification of learning objectives using the BERT model into predefined domains achieved an accuracy of 95%. The system results in a structured learning path, which is more organized and engaging. The system is beneficial to novice learners, as it reduces failure rates and improves completion time of the learning process.

Lifting-Based Block Fractional Wavelet Filter Compression of Hyperspectral Images over Wireless Multimedia Sensor Network Platforms

Purushottam Lal Nagar — 2026-03-24

In the rapidly development of remote sensing technology, the compression of Hyperspectral Images is a pivotal yet formidable task. Hindered by inherent limitations in hyperspectral imaging, enhancing the accuracy and efficiency of compression algorithm remains a critical and much-debated issue. Algorithms using set partition wavelet transforms excel in hyperspectral image compression due to their embedded nature, coding efficiency, and low complexity. Specifically, the Fractional wavelet-based zero memory set partitioned embedded block algorithm achieves high coding efficiency with lower memory demands, though its method of repeatedly comparing coefficients to a threshold is time-intensive. To solve this, a new algorithm has been developed that optimizes both computational and memory complexity. It employs a block-based fractional wavelet filter (BFrWF), which delivers the same accuracy as conventional transforms but requires far less memory.

The Block-based Fractional Wavelet Filter is a low-memory technique for image transformation, but its high computational complexity makes it impractical for resource-constrained devices in IoT and Wireless Sensor Networks. Additionally, it produces blocking artifacts due to improper handling of block boundaries. This paper introduces a new lifting-based version of BFrWF that eliminates these artifacts by correctly overlapping image blocks. This new implementation with low complexity zero memory set partitioned embedded block (LC-ZM-SPECK) achieves higher coding efficiency, making it well-suited for resource constraint visual sensor nodes.

Artificial Intelligence Strategies in Enhancing Cybersecurity: An Analysis of Advanced Applications and Techniques

Ali Chasib Alhasnawy — 2026-03-24

In this research, we aim to detect the important role of Artificial Intelligence (AI) in increasing cyber security measures against rapidly refined cyber threats. Since cyber-attacks develop in complexity and frequency, traditional safety methods are often inadequate, which requires integrating AI-controlled solutions to increase danger, reaction time, and general safety flexibility. The study will be engulfed by various AI applications in the cyber security domain, including detection of nonconformities, automatic danger information analysis, and future analysis. By analyzing existing literature and case studies, research will emphasize how AI technologies such as machine learning and natural language treatment can be utilized to improve the effect of cyber security structure. Besides, there will be challenges and moral ideas related to distributing AI in security contexts, including privacy and concerns related to algorithm bias. Large findings of recent studies indicate that AI can increase cyber safety ability by quickly identifying and more accurate risk assessment of hazards. This research outlines the transformation effect of artificial intelligence on cyber security practices and provides recommendations for organizations to effectively use AI-controlled strategies. By promoting a deep understanding of skills and limitations in the cyber defense structure, the purpose of this study is to contribute valuable insights that can lead to future development in this important field at the present time.

Enhanced DeepFake Image Detection via Swin-B Transformer with Visual Attention Analysis

Vineela Krishna Suri — 2026-03-24

Deepfakes, synthetic media created using advanced machine learning techniques, pose significant societal challenges by spreading misinformation and undermining trust in media. With the increasing sophistication of deepfake technologies, distinguishing between genuine and synthetic media has become increasingly difficult. This paper presents a robust deepfake image detection framework using the Swin-B Transformer, a pre-trained model fine-tuned for our application. By integrating a hybrid dataset that combines real images from the FFHQ dataset and synthetically generated fake images from a publicly available Kaggle dataset, we simulate real-world media scenarios. Our model achieves an impressive accuracy of 97.47% on the test set, demonstrating superior generalization to both real and synthetic visual data. Using Grad-CAM, we visualize the spatial segments of the image that the model focuses on during classification, providing insight into the decision-making process. This work contributes to enhancing content authenticity, controlling fake news, and ensuring digital trust and safety.

Effect of CVD-Coated Multilayer Carbide Insert on Surface Finish of Cryogenically Treated EN31 Hardened Alloy Steel Using Box-Behnken Design

Shivaji Vithal Bhivsane — 2026-03-24

The present study is focusing on cryogenically treated (CT) EN31 hardened alloy steels. The experimental runs were designed by using Box-Behnken approach of RSM using multiple process parameters- cutting speed (v, 160-260 m/min), feed (f, 0.1-0.2 mm/rev), depth of cut (d, 0.05-0.15 mm), tool nose radius (re, 0.4-1.2 mm), and hardness (HR, 45-49 HRC) for optimizing surface finish. The specimens and inserts underwent a 24-hour socking time in nitrogen medium (-196° C) as part of the deep cryogenic treatment and hardness of the materials were measured before and after the treatments. In experimentation wet turning was performed on LMW Make LX20T L5 CNC machine with multilayer TiCN+Al2O3+TiN CVD coated inserts. According to the study, Ra dramatically rises with an increase in d and falls with an increase in v. When comparing Ra to the f, re, and HR, comparable variations are seen. The Box-behnken approach of RSM is more capable of predicting surface finish by 99.96% utilizing a polynomial quadratic equation.

Experimental Investigation of Rotary Friction Welding of Al2017A Alloy Reinforced with Alumina (Al₂O₃) Particles

Raouache elhadj — 2026-03-24

A series of similar joints made from Al2017A aluminum alloy were produced through rotary friction welding, with localized reinforcement provided by alumina (Al₂O₃) particles, a ceramic known for its high hardness and extensive use in metal–matrix composites. These particles were placed at the weld interface by filling drilled holes of different diameters in the fixed component, allowing the influence of varying volume fractions on joint behavior to be examined. Tests of tensile strength and microhardness indicated that adding Al₂O₃ leads to a noticeable improvement in the mechanical performance of the welded joints when compared with those lacking ceramic reinforcement.

Real-Time Path Rerouting and Obstacle Aware Navigation in Autonomous Vehicles: A Simulation and Data Analysis Approach

Lakshmi Narayana I — 2026-03-25

Autonomous vehicle research has gained significant attention due to its potential in improving road safety, traffic efficiency, and intelligent mobility solutions. However, real-world testing remains costly and complex, making simulation-based models an effective approach for validating navigation and obstacle avoidance strategies. In this project, we present a simulation-driven autonomous vehicle framework capable of navigating between user-defined source and destination coordinates while ensuring real-time obstacle detection, dynamic rerouting, and journey visualization. The methodology integrates a virtual GPS for location tracking, an A*-based pathfinding algorithm enhanced with dynamic obstacle avoidance, and a simulation interface that allows users to input coordinates and visualize the entire navigation process. Camera-based or sensor-simulated modules are employed to detect obstacles in real time, triggering the rerouting logic to compute safe and collision-free alternative paths. Live data such as route progress, obstacle events, and estimated time of arrival are continuously displayed through the simulation dashboard. Following several iterations of testing, data logs were collected and analyzed using machine learning techniques to evaluate navigation efficiency, obstacle response time, and rerouting accuracy. Results demonstrate that the system successfully adapts to dynamic environments, offering a cost-effective and scalable platform for smart transportation research, algorithm benchmarking, and assistive mobility applications.

A Development of Fuzzy-PID Controller Applied on an Autonomous Surface Vehicle

Thien M. Tran — 2026-03-25

Precise trajectory tracking remains a significant challenge for Autonomous Surface Vehicles (ASVs) due to their inherent nonlinear and strongly coupled surge-sway-yaw dynamics, which are often exacerbated by environmental disturbances. This study aims to enhance the trajectory tracking performance and robustness of an ASV by developing an adaptive Fuzzy-PID controller and comparing its efficacy against a conventional Proportional-Integral-Derivative (PID) controller. A comprehensive 3 degree of freedom (DoF) mathematical model of the ASV is derived using the Fossen modeling framework, incorporating rigid-body kinetics, hydrodynamics, and environmental disturbances. A novel single-loop Fuzzy-PID control architecture is then proposed, wherein a Mamdani-type fuzzy inference system dynamically tunes the PID gains online based on the tracking error and its derivative. The performance of the proposed controller is rigorously evaluated against a classical PID controller through simulations involving circular and figure-eight trajectories, which are designed to stress the system's dynamic coupling. The simulation results demonstrate the superior performance of the Fuzzy-PID controller. It achieves a substantial reduction in steady-state heading bias up to 46% in Root Mean Square Error (RMSE) and 64% in Integral Absolute Error (IAE) for the circular trajectory compared to the conventional PID. Furthermore, the proposed controller delivers smoother control signals, faster settling times, and improved robustness under strong dynamic coupling and external perturbations. The proposed Fuzzy-PID controller provides a significant improvement in tracking accuracy and adaptability over the conventional PID controller. It offers an efficient and practical solution for the autonomous guidance and control of small-scale surface vehicles, effectively handling nonlinearities and coupled dynamics. Future work will focus on hardware-in-the-loop validation and incorporation of advanced disturbance observers.

Optimization of Energy Consumption by HVAC System in Buildings Using Deep Learning-Based Control Strategies

Rajalakshmi K — 2026-03-25

The building industry, which uses the most electricity, has a significant potential to contribute to energy consumption reduction. Commercial structures use more energy than other types of structures because of their productive and logistic features. In these types of structures, one of the main energy consumers is the HVAC system which comprises of heating, ventilation, and air conditioning, especially in arid conditions. Energy-efficient environment friendly HVAC system conception and execution can significantly lower the use of energy and support ecologically sound growth in business establishments. On the other hand, inadequate implementation of methods for reducing energy use may lead to a decline in the welfare of the environment. Therefore, in order to achieve energy efficiency and maintain the optimum degree of temperature regulation, a comprehensive energy conservation strategy is needed. To accomplish this goal, model predictive control strategy-based methodologies are used in this work. To estimate how much energy will be used in commercial buildings, four deep learning-based methods are utilised: radial basis function networks, multi-layer perceptrons, artificial neural networks, and back propagation neural networks. To further cut down on energy use, four distinct control mechanisms are used. The performance of the suggested solution is examined using performance measures like Mean Absolute Error and Mean Absolute Percentage Error.

Investigation of Fracture Behavior in Mode I and II for Repaired Edge-Inclined Cracks with Trapezoidal Composite Patches

Toufik Achour — 2026-03-25

Numerous studies have shown the effectiveness of composite patch repairs. However, many of these investigations primarily address the enhancement of repaired components' lifespans, focusing mainly on opening mode (Mode I). In real-world applications, cracked components often undergo mixed mode loading that includes both Modes I and II. This article examines the stress intensity factors for Modes I and II in relation to the fracture behavior of a tensile-loaded aluminum plate (Al 7075) featuring a 45° inclined lateral crack repaired on both sides with a unidirectional graphite/epoxy composite trapezoidal patch. A three-dimensional finite element model of the repaired specimen is employed to explore how composite patching affects critical crack tip parameters (KI, KII, and stresses). This approach demonstrates how the properties of the composite and adhesive impact the repaired structure's behavior and the effectiveness of the bonded composite patch. The findings reveal that trapezoidal composite patch can significantly reduce the stress intensity factors KI and KII, thereby extending the service life of cracked structures.

Digital Twin Implementation Strategies for Complex Manufacturing Ecosystems: A Multi-Level Integration Framework

Suchita B. Jadhav — 2026-03-25

Complex manufacturing ecosystems require sophisticated digital transformation strategies that can integrate multiple production systems, processes, and organizational levels while maintaining operational efficiency and data integrity. This paper presents a comprehensive multi-level integration framework for digital twin implementation in complex manufacturing environments, addressing challenges in standardization, interoperability, and scalability. We evaluate implementation strategies across four integration levels: component-level, machine-level, system-level, and enterprise-level digital twins, utilizing International Organization for Standardization (ISO) 23247 standards and edge computing architectures. Our analysis encompasses 15 industrial case studies spanning wire arc additive manufacturing, Computer Numerical Control (CNC) machining, flexible manufacturing cells, and multi-plant operations. The proposed framework demonstrates 34% reduction in implementation time, 28% improvement in data processing efficiency, and 42% enhancement in decision-making capabilities compared to traditional approaches. Results show that standardized digital twin architectures based on ISO 23247 enable seamless integration across manufacturing levels while maintaining scalability for complex ecosystems. The study establishes that edge computing-enhanced digital twins achieve 15-millisecond response times for real-time control applications and support zero-defect manufacturing initiatives through predictive analytics and closed-loop optimization. The framework provides practical guidelines for organizations implementing digital twin strategies in complex manufacturing environments.

Impact of Roughness on the Emptying Time of a Tank: Experimental and Numerical Analysis

Belaid Larbi — 2026-03-25

This study investigates the hydrodynamic influence of surface roughness on the discharge time of free-surface tanks through a combined experimental and numerical approach. The research evaluates the mechanisms by which varying roughness magnitudes alter flow resistance, turbulence modulation, and energy dissipation during the drainage process. Experimental data were obtained from a tank with controlled roughness parameters, providing validation for Computational Fluid Dynamics (CFD) simulations performed using ANSYS FLUENT. The Volume of Fluid (VOF) method was employed to resolve the multiphase flow dynamics. Results indicate a distinct inverse correlation between surface roughness and emptying time. Specifically, the introduction of surface roughness (5–15 mm) reduced discharge duration by up to 4.67%, with the most pronounced efficiency gains observed within the lower roughness interval (0–6 mm). This phenomenon is attributed to the modification of near-wall flow structures and the reduction of laminar flow resistance. The study validates the accuracy of the numerical framework in predicting complex fluid behaviors and underscores the critical role of surface roughness optimization in enhancing the design and operational efficiency of hydraulic systems.

Experimental Validation of a Novel Dual-Slot SIW Bandpass Filter for 5G Ka-Band Applications

Fatiha Louakhche — 2026-03-25

This study details the design, optimization, and experimental validation of an innovative dual slot Substrate Integrated Waveguide (SIW) bandpass filter intended for 5G Ka-band applications. The proposed filter is implemented on a cost-effective Rogers RO5880substrate and exhibits a second-order quasi elliptic response achieved by etching a novel transverse dual-slot topology within the SIW cavity. The filter was fabricated using standard PCB procedures. The measured results show strong agreement with simulations, confirming the robustness of the design. The small device measures14.15×27.86 mm² and functions effectively inside the 27-29 GHz range, attaining a fractional bandwidth of 8.81%,a return loss of -42 dB, and an insertion loss of -3.16 dB. These results demonstrate that the proposed dual-slot SIW filter provides a highly selective, miniaturized, and low-cost solution, making it a promising candidate for integration in next-generation 5G millimeter-wave front-end systems.

AES-Enhanced Blockchain Intrusion Detection System for Secure Networks

Hemanth Uppala — 2026-03-25

Blockchain is a distributed ledger technology that can securely, transparently, and tamper-proofly record transactions across a network. IDS on blockchain monitors nodes and activities in transactions to detect malicious or unusual patterns, thereby improving network integrity. However, blockchain also has disadvantages such as high computational overhead, vulnerability to specific attacks, and limited scalability. As a method to enhance IDS performance, the Z-score is used to normalize feature values, stabilizing the model and facilitating convergence. A combination of Support Vector Machines (SVM) and Decentralized Identification (DID) architecture can effectively and reliably distinguish between normal node behavior and malicious node behavior. Key verification uses consensus-driven transaction authentication, and a Multi-Layer Perceptron Neural Network (MLPNN) can detect complex attack patterns. All sensitive information is protected using the Advanced Encryption Standard (AES) to provide robust encryption for both stored and transmitted data. By combining these methods, blockchain applications can offer a comprehensive, secure, and scalable network intrusion detection system, addressing the limitations of detection accuracy, computational efficiency, and privacy in existing data and traditional blockchain implementations, to achieve an accuracy of 91%.

Deep Learning Approach for Oil Palm Leaf Disease Classification Using VGG16 Enhanced with Adam Optimization

Hafiz Irsyad — 2026-03-25

Early detection of oil palm leaf diseases is essential to minimize economic losses and ensure sustainable plantation management. This study proposes a deep learning approach using the VGG16 architecture optimized with the Adam algorithm to classify oil palm leaves into three categories: healthy, infected, and initial infection. The dataset was obtained from Roboflow and preprocessed through cropping, annotation, and standardization before being split into training, validation, and testing sets. Experimental evaluations were performed across multiple training epochs and compared against two baseline models: a shallow Convolutional Neural Network (CNN) and YOLOv11-CLS (version S). Results show that VGG16 combined with Adam achieved the highest accuracy of 97% at 25 and 50 epochs, with balanced precision and recall across all classes. In contrast, the baseline CNN reached a maximum accuracy of 88%, while YOLOv11-CLS produced fluctuating results with a peak accuracy of 82%. Statistical significance testing confirmed that the performance improvements of VGG16 + Adam were consistent and reliable, validating its suitability for practical implementation in precision agriculture. These findings highlight the potential of combining deep architectures with adaptive optimization to enhance disease diagnosis in oil palm plantations and reduce reliance on manual inspections.

Comprehensive Analysis of Energy Audit Strategies for Enhancing Energy Efficiency in Higher Educational Institutes

Seema V Vachhani — 2026-03-25

Increasing demand of energy and associated cost create considerable significance for energy conservation and energy audit. Energy audit creates opportunities to improve energy efficiency and to reduce energy cost. The authors looked into and peer reviewed energy audit technical articles to find innovative approaches for energy efficiency improvement, including methodologies employed, findings and future work for the energy audit case studies of educational institutes. The review shows that most articles emphasis on the lighting load during energy audits of higher educational institutes. In this paper, the authors have discussed the identified areas with energy conservation potential for the higher educational institute in Saurashtra region of Gujarat, India.

Quality Enhancement Techniques for Breast Carcinoma in Epithelial Tissue Identification Process

Bharathi K — 2026-03-25

Breast Cancer is considered to be the deadliest disease among women due to the carcinoma in epithelial tissue development in breast. The cause of the disease many vary due to many circumstances, but identification procedure followed are mostly similar. The clinical way of identifying the cancer effected tissues in breast are followed in advance stages or pre advanced stages, which is due to the lack of adequate knowledge about breast cancer. The treatment given during the final stages are mostly not feasible solution and eventually ends with negative result. Digital Image Processing (DIP) technique coupled with Data Mining and Machine learning algorithms are most recently used breast cancer identification procedure. The identification procedure followed using those techniques are not only accurate, it also gives very fast analyzing report based on the historical record. This research article proposes pre-processing technique, which is a part of the overall research work of breast cancer identification procedure. The Mammography images collected from the source may contains many irrelevant information as well as missing values. The article gives a clear idea of pre-processing techniques followed as well as filtering techniques implemented to enhance the quality of the collected breast cancer Mammography images.

Automation of the food deep-freezing process using a PLC-controlled freezing tunnel.

Washington Stalin Jácome Bastidas — 2026-03-25

This paper addresses the design, construction, and evaluation of an automated freezing tunnel for the deep-freezing process of food, implemented using a programmable logic controller (PLC). The research was conducted in the context of the food industry, where energy efficiency, quality preservation, and sustainability are critical factors for competitiveness. The system integrates temperature sensors, control modules, and a human-machine interface (HMI), allowing for accurate, real-time monitoring of the process. The experimental results show that the prototype reached temperatures of up to –30 °C, freezing different products in an average of 12 minutes. This performance represents a significant reduction in time and costs compared to traditional methods, ensuring the preservation of organoleptic and nutritional properties. It is concluded that the automation of the deep-freezing process using PLCs is a viable and scalable alternative for the food industry, providing improvements in productivity, operational efficiency, and food quality.

Systematic Review of Coffee and Rice Ash in Physical-Mechanical Properties and Concrete Durability

Michael Whitman Quiñones Mas — 2026-03-25

Concrete is one of the most widely used building materials worldwide, but its production generates a high carbon footprint due to the high clinker content in Portland cement. In response, coffee husk ash (CCC) and rice husk ash (CCA) are being investigated as sustainable alternatives to partially replace cement. This paper presents a systematic review of recent studies in Scopus and ScienceDirect, applying selection criteria related to technical relevance, availability of experimental results and incorporation of both ashes in concrete and mortar mixtures. Likewise, a bibliometric analysis is developed using VOSviewer to identify research trends, countries, outstanding authors and predominant keywords. The compiled studies allow us to compare the effect of CCC and CCA on physical properties (workability, density, absorption), mechanical properties (compressive strength, tensile strength and bending) and durability (permeability and resistance to chlorides and sulfates). Overall, the evidence shows that both ashes can be used as supplementary cementitious materials in moderate replacements, valorizing agro-industrial waste and reducing the environmental impact of the concrete, as long as parameters such as fineness, calcination conditions and dosage are controlled to guarantee optimal and sustainable results

A Novel TwinNet Transformer-Based Deep Learning Model for Accurate and Efficient Paddy Leaf Disease Diagnosis Using Explainable AI

Daniel Raj K — 2026-03-25

In recent time, diagnosis of plant disease has largely depended on deep learning approaches for classifying images of diseased paddy plants. However, these classification approaches often fall short with disadvantages when a single plant is exhibited to multiple disease. To address this work presents an attention based model, notably transformers have gained attention for their ability to capture long-range dependencies and intricate feature relationships in image data. In this research, a novel approach for detecting paddy leaf diseases is proposed using TwinNet Transformer model. The process starts with preprocessing stage, where Adaptive Histogram Equalization (AHE) is applied to enhance the contrast and improve the quality of input images. Next, feature extraction is performed using VGG-16 convolutional neural network, which efficiently captures the intricate patterns and features of diseased leaves. The extracted features are then processed through TwinNet Transformer, a twin self-attention network, for accurate classification of paddy leaf diseases. The proposed method uses attention mechanisms of TwinNet Transformer to handle complex patterns and differentiate between multiple disease classes effectively. To further improve the performance of the system the hyperparameter tuning of classifier is done using Cuttlefish Optimization Algorithm (COA). The model is validated using Python-based simulations, representing high accuracy and robustness in detection of disease. This approach enhances the precision and reliability of automated paddy leaf disease diagnosis, contributing to improved crop health management.

Smart AI-Intensive Software Defect Predictions Based on Optimized Deep Feature Classification Using GWO-Code2Vec CNN Learning

Ruckmani V. S — 2026-03-25

Software defect prediction plays a crucial role in improving software reliability, reducing maintenance costs, and enhancing software quality. However, traditional prediction methods often fail to accurately detect software defects due to non-relational dependencies among features, redundant datasets, and imbalanced data distributions. These limitations result in lower true positive rates and reduced predictive accuracy. To overcome these challenges, this research introduces an optimized Deep Learning–based Defect Prediction Framework that integrates advanced preprocessing, feature optimization, and classification mechanisms. Initially, Z-Score Logarithmic Transformation (ZSLT) is employed for preprocessing to normalize feature scales and eliminate noise in defect logs. To address data imbalance, the Clustered Synthetic Minority Oversampling Technique–Edited Nearest Neighbour (CSMOTE-ENN) method is applied, effectively balancing the dataset while preserving meaningful feature diversity. Subsequently, Grey Wolf Optimization–Deep Neural Network (GWO-DNN) is utilized for optimal feature selection, enabling the identification of highly correlated defect-related attributes. Finally, the Code2Vector–Graph Convolutional Neural Network (Code2Vector-GCNN) model is employed for deep feature classification, capturing both semantic and structural code representations to improve defect detection accuracy. Experimental results demonstrate that the proposed framework significantly outperforms conventional machine learning and deep learning models in terms of precision, recall, and F1-measure, providing an intelligent, adaptive, and high-accuracy solution for software defect prediction.

A YOLO-Based Automatic Bangladeshi Vehicle License Plate Recognition System

Ashraful Islam — 2026-03-25

In this paper, a YOLO based automatic Bangladeshi vehicle license plate recognition system is proposed. The proposed system has four parts: license plate detection, extracting the region of interest (ROI), applying image processing to the ROI, and character segmentation & recognition. In the detection & extraction stage, the system receives a vehicle image and then detects the license plate and extracts the plate region. The recognition part consists of three consecutive stages: city name recognition, vehicle type recognition and vehicle serial number recognition. As the presented method recognizes the license plate characters in three consecutive stages, a data serialization algorithm is proposed to serial the data. The dataset contains 500 license plate images from four major cities of Bangladesh. The images are used to train, test and validate the proposed model. The proposed method has provided very impressive results and outperformed many other existing methods.

Experimental Evaluation of Multi-Sensor AI-Based Fault Diagnosis for Induction Motors in Industrial Applications

Saranya M — 2026-03-25

Induction motors are critical components in industrial systems, and unexpected failures can lead to significant production losses and maintenance costs. This paper presents an experimentally validated multi-sensor fault diagnosis approach for induction motors using vibration and stator current signals under practical operating conditions. A laboratory test bench was developed to simulate common industrial faults, including bearing defects, rotor abnormalities, and stator winding faults, across multiple load levels. Conventional machine learning techniques, namely artificial neural networks and support vector machines, were employed as baseline classifiers, while an adaptive hybrid convolutional neural network–long short-term memory (CNN–LSTM) model was used to improve fault classification robustness. The proposed approach achieved a maximum classification accuracy of 98.4 %, with stable performance across varying load conditions and repeated experimental trials. The results demonstrate that integrating vibration and current measurements enhances diagnostic reliability compared to single-sensor methods. The study highlights the practical applicability of adaptive AI-based diagnostic systems for industrial predictive maintenance, offering improved fault detection capability while maintaining feasibility for real-world deployment.

A Deep Learning Approach to Tomato Disease Classification Using a CNN-LSTM Hybrid Network

Youssef M Laatiri — 2026-03-25

Agriculture is a crucial sector that meets the fundamental nutritional needs of humanity. Plant diseases exacerbate economic and food security issues for nations and hinder their agricultural planning. Conventional techniques for identifying plant diseases necessitate considerable labour and time. As a result, numerous scholars and institutes endeavour to tackle these challenges through sophisticated technical approaches. Deep learning-based plant disease detection shows significant progress and optimism over traditional techniques. When trained using extensive, high-quality datasets, these systems detect diseases on plant leaves in their early stages. This article conducts a systematic evaluation of deep learning methodologies in plant disease detection by analyzing a number of research publications between 2015 and 2025. Our research examines three specific areas: the categorization, detection, and segmentation of diseases on plant leaves, while rigorously evaluating publicly available datasets. This systematic review offers an in-depth evaluation of the existing literature, describing deep learning architectures, the most frequently studied tomato diseases, datasets, challenges encountered, and diverse perspectives. It offers new perspectives for researchers in the field of agriculture. In addition, it addresses the main challenges of identifying agricultural diseases.: datasets, barriers, and diverse perspectives. It offers new perspectives for researchers in the agricultural field. In addition, it addresses the main problems related to agricultural diseases.

Well-Posedness in M-Ultradifferentiable Spaces for Weakly Hyperbolic Cauchy Problems with Hölder Continuous Coefficients

Said Bouaziz — 2026-03-25

In this article, we demonstrate the weakly hyperbolic Cauchy problem under Hölder's regularity of a coefficient depending on time in the context of M-ultradifferentiable well-posedness. We find an equivalent condition for well-posedness within the framework of Gevrey regularity, ensuring well-posedness in a class of M-ultradifferentiable functions, dependent on the associated function of the sequence M. viscoelastic materials.

Firefly-based Optimized Link State Routing Protocol for Energy-Efficient Vehicular Ad Hoc Networks

Mustafa Tareq — 2026-03-25

Vehicular ad hoc networks (VANET), a burgeoning technology within the wireless technology field, hold great promise. As it continues its progress, it stands ready to make increasingly significant contributions to advancing intelligent transportation systems in the foreseeable future. The scale of infrastructure, the number of vehicles, the complexity of scenarios, and the mobility of nodes can collectively contribute to heightened energy consumption during route selection between source and destination nodes, as well as during the transmission of application packets. VANET opts for routes containing a minimal count of intermediary nodes to reach the intended destination efficiently. With the expansion of the distance between nodes, there is a proportional increase in the requirement for broadcast power. The power capacity of the nodes directly influences the ease of establishing a route between two nodes. In this paper, we proposed a Firefly Optimized Link State Routing (F-OLSR) protocol, which utilizes swarm intelligence principles through the firefly algorithm (FFA). The goal is to enhance energy efficiency within an Optimized Link State Routing (OLSR) framework. The route is chosen depending on the amount of power the intermediary nodes between the source and destination nodes. The evaluation performance of the F-OLSR protocol was compared with that of popular protocols. The F-OLSR protocol was demonstrated to be an effective solution for choosing the optimal path with high power from source to destination and improving performance. Regarding energy consumption, the F-OLSR protocol achieved 2.30% and 2.31% concerning the packet size and node speed, which is better results than in terms of packet size and node speed, which is a better result than the other protocols.

Consistency Validation of Transforming UML Statechart Models to Flat State Machine Models with USE Approach

Ali lalouci — 2026-03-25

Model-Driven Engineering (MDE) supports software development through the systematic use of models, meta-models, and model transformations, making transformation validation a critical concern. This paper addresses the consistency validation of a transformation from UML statechart diagrams to Flat State Machines (FSMs). The transformation is formalized as a transformation model using a UML class diagram enriched with Object Constraint Language (OCL) invariants. The UML-based Specification Environment (USE) model validator is used to automatically verify transformation consistency and its implied properties, including weak consistency, class instantiability, and class and association instantiability. A case study based on an Automated Teller Machine (ATM) system demonstrates the effectiveness of the proposed approach in supporting reliable software development within Model-Driven Engineering.

Power Quality Improvement through Harmonic Filters Using Tasmanian Devil Optimization Algorithm

Ashokkumar Parmar — 2026-03-25

To enhance power quality (PQ) through active power filters (APF) in radial distribution systems (RDS), this study investigates the application of the Tasmanian Devil Optimization (TDO) and Osprey Optimization Algorithm (OOA). Despite their benefits, the growing use of solar photovoltaic (PV) systems presents challenges with PQ, including harmonic distortion because of their nonlinear features. Harmonics are the leading cause of low PQ in such systems. Here, the PV injects harmonics into the RDS and is categorized as a nonlinear distribution generator (NLDG). This study examines the effect of the nonlinear loads (NLs) of two end nodes and the incorporation of the NLDG into the RDS on the entire RDS. APFs are positioned carefully to reduce the harmonics and improve the PQ. The suggested method minimizes the APF current while abiding by inequality limitations by utilizing an optimization algorithm. The TDO was used to determine the appropriate APF size. It is inspired by natural processes such as photosynthesis. It has a good balance between exploration and exploitation for effective search. The efficacy of the TDO was demonstrated through simulations on the IEEE-69 bus RDS and was compared with that of the OOA. The outcomes validate that the TDO is stable and efficient in resolving this optimization issue for PQ enhancement in RDS.

Cost-Effectiveness and Efficiency of PV Pumping Systems Utilizing Artificial Intelligence Algorithms

Zoubir Chelli — 2026-03-25

We conduct this research for thirty-six months from 2021 to 2023 in Souk-Ahras (Tiffech), Algeria, to evaluate the effectiveness and costs of a direct-coupled photovoltaic pumping system for agricultural irrigation, considering the region's potential for solar energy. This study presents an analytical method of dimensioning based on water quantity, sunlight data, pump group efficiency, and well and reservoir characteristics to provide insights for better economic installation management. We evaluated the viability of a solar photovoltaic- SPV, water pumping system using the software PVsyst. However, we employed artificial intelligence- AI, methods to optimize irrigation and conserve water resources, thereby enhancing energy efficiency. Four AI methods, Support Vector Machine- SVM, Random Forest- RF, Naive Bayes- NB, and k-Nearest Neighbors- kNN, are utilized. Energy-efficient systems promote a reduction in overall costs, enabling users to maintain financial balance while meeting their pumping needs. According to AI findings, the k-NN model exhibits high accuracy at 0.960 precision. The cost-effectiveness comparison of the SPV system to a diesel generator demonstrated a significant decrease in the cost of pumped water, thereby making the SPV system more viable and profitable. This approach has demonstrated its effectiveness and robustness.

Evaluation of Methylene Blue Removal Using a Food-Industry Byproduct via Simulation and Modeling

Salah Eddine Bencheikh — 2026-03-25

Our work examines the use of peanut shells, a common and cheap food waste, as a natural adsorbent for treating industrial dyes, particularly methylene blue in this study. We also performed a comprehensive analysis of adsorption, which showed that it follows both Langmuir and Freundlich adsorption isotherms, signifying that peanut shells are efficient adsorbents for methylene blue.

To enhance the outcome of the adsorption, a Box-Behnken design was used in our experiment. This design enabled us to examine how various important parameters, such as the amount of adsorbent, concentration of the dye, pH, temperature, rate of stirring, and the ionic strength of dissolved salts, can affect this procedure. Also, a mathematical model and simulation of this adsorption reaction enabled us to find optimal conditions for efficient removal of this dye.

In our experiments, the highest adsorption efficiency of 97% was obtained under optimal conditions of 1.5g of adsorbent, a concentration of 40mg/L of MB, a pH of 11, a temperature of 65°C, 0 mg/L ionic strength, and a stirring speed of 165 rpm. These results reveal that peanut shells can be considered a potential, environmentally sustainable alternative for removing dyes.

Design of a Hybrid Ensemble Feature Selection Framework for Big Data Text Mining

Smah Smari — 2026-03-25

The growing volume of textual data often exceeds the capacity of available computing resources, and conventional machine learning algorithms struggle to scale up. Today, the quality of data is becoming more critical than its raw quantity: it is therefore essential to transform massive data into intelligent data through appropriate pre-processing steps. Feature selection plays a key role in this process. In this work, we propose the design of a hybrid ensemble-based feature selection framework for processing large-scale textual data. The approach is based on the MFD-AFSA algorithm combined with different feature evaluation functions, applied on multiple data subsets. To improve scalability, we also outline a distributed strategy in an Apache Spark environment, based on the Random Sample Partitioning model. Finally, we introduce an automatic approximation mechanism, which we call auto-approximation, enabling selection sets to be built dynamically via an approximation technique. This work is part of a methodological design approach; experimental validation and practical evaluations will be the subject of future work.

Experimental Investigation and Optimization on Abrasive Waterjet Drilling of Inconel -939 Alloy

Praveen Jayapalan — 2026-03-25

Inconel 939, the nickle-based superalloy has exceptional properties that include its high thermal and mechanical behaviours. Owing to this properties Inconel 739 demand in various industries like aerospace, automobile. Machining is unavoidable to use Inconel 739 alloys. Abrasive water jet drilling (AWJD) adopted to analysis and Optimize the MRR (Material Removal Rate) and SR (Surface Roughness) by integrating the RSM (Response Surface Methodology)-CCD (Central Composite Design)-Desirability approach. The L27 Experiments conducted by adopting variation in the Pressure, Traverse Speed, Abrasive Flow Rate and Standoff gap distance. The developed model validated statistically, its suitable for further analysis. Analysis of Variations (ANOVA) illustrations the Pressure was high influential variable in the responses and optimization processes. The developed model of regression shows minimal error between predicted and actual values. Three dimensional plots and used for capture the parameters effect in the responses. Pressure was influencing the 72.7% and 78.8% on SR and MRR Respectively. Desirability approach shows the validation error less than the acceptable limits.

Efficient Extraction of Patterns and Insights from Large-Scale Data from Distributed System Using Deep Learning and LSTM

Syed Kousar Niasi K — 2026-03-25

Extraction of useful insights and patterns from massive datasets is impossible without data mining. The widespread use of Internet services and the subsequent creation of vast quantities of data have ushered in the modern era of big data. Every aspect of human existence generates copious amounts of data. Data analysis and utilization processes need to take into consideration a growing number of factors to deal with the growing volume and complexity of today's data sets. The importance of deep learning in data mining rises in proportion to the problem's complexity. Traditional data mining methods face major hurdles due to the ever-increasing number and complexity of data. To address this issue, we offer a unique distributed data mining strategy that utilizes deep learning and Long Short-Term Memory (LSTM) systems. In this study, we combine deep learning with distributed computing to provide a powerful tool for data mining. In order to capture long-term relationships in sequential data, the proposed model uses recurrent neural networks (RNNs) of the LSTM kind. Since LSTM networks are so effective with time-series and sequential data, they may be used in a wide variability of data mining tasks. We create a parallel computing framework that pools the processing power of a cluster's many nodes to facilitate decentralized operations. The training and inference of the LSTM-based data mining model are sped up by the distributed design, which also facilitates the efficient processing of big datasets. We compare the proposed model to standard data mining techniques and show that it outperforms them on a number of real-world datasets. The outcomes demonstrate the superior accuracy of 99.5% and efficiency of our deep learning-based method for identifying useful patterns and making predictions. The findings validate the model's horizontal scalability and the benefits of distributed computing, guaranteeing its practical application in large data settings.

Red-Tailed Hawk (RTH) Algorithm for Optimal PV Reconfiguration in Irrigation Systems Under Partial Shading Conditions

Abdelouadoud Loukriz — 2026-03-25

Photovoltaic (PV) water pumping systems offer a sustainable solution for agricultural irrigation in regions where conventional energy sources are unreliable or expensive. However, partial shading (PS) significantly degrades PV performance and limits water delivery. This paper proposes a dynamic PV array reconfiguration strategy based on the Red-Tailed Hawk (RTH) optimization algorithm to mitigate PS effects. The reconfigured PV array, combined with maximum power point tracking (MPPT), supplies a brushless DC (BLDC) motor-driven water pumping system to enhance operational efficiency. The proposed method ensures stable steady-state operation while maintaining the PV system near the maximum power point under varying shading conditions. Performance evaluation is carried out in MATLAB using a 4×5 PV array subjected to dynamic shading caused by a tree, with comparisons made against the conventional total cross-tied (TCT) configuration using reported experimental data. The results demonstrate clear improvements in water flow, with average gains of 36 liters per hour in the first case and 111.6 liters per hour in the second case. These outcomes confirm the effectiveness of the proposed approach and highlight its potential for improving the sustainability and reliability of photovoltaic water pumping systems in energy-constrained agricultural regions.

Improvement in BER Performance of GFDM-Based 5G Communication System Using New Hybrid Error-Correcting Code

Divya Jain — 2026-04-24

In 5G wireless communication system, Generalized frequency division multiplexing (GFDM) is a promising substitute to OFDM that offers improved spectral efficiency and lower PAPR. However, performance of GFDM based wireless communication system can still be degraded by channel impairments such as noise and fading. The error correction codes are one of the techniques used for improving the performance. Different error correction codes are used for error correction in 4G and 5G communication system. This paper proposes a hybrid error-correcting code (ECC) scheme that combines polar code and convolution code to improve the bit error rate (BER) performance of GFDM compare to conventional ECCs. Simulation results illustrate that the proposed hybrid ECC scheme outperforms traditional error correcting code, in terms of BER performance. The proposed scheme also achieves near-optimal BER performance under various channel conditions.

A Proposed Model for Designing a Metamaterial Unit Cell 9 - 14GHz Frequency Band Communications

Jamal Mohammed Rasool — 2026-04-27

A compact monopole antenna on a printed circuit board incorporating a single negative (SNG) metamaterial targets application in the X and Ku bands at 9.2 and 14.2 GHz. The design utilizes circular split-ring resonator (SRR) unit cells, structured with periodic arrangements smaller than the guided wavelength. Using CST Studio Suite simulation, the antenna demonstrates a wide 750 MHz bandwidth and very low radiation losses. This efficient design enhances performance and coverage, making it well-suited for radar and satellite communications within the targeted frequency bands.

Forest Fire Trends in Indonesia and Australia: Lessons Learned and Mitigation Strategies

Evizal Abdul Kadir — 2026-04-24

Forest fires present significant environmental, economic, and social challenges in both Indonesia and Australia, with each country exhibiting different causes and effects. In Indonesia, forest and peatland fires are frequently driven by land-use changes, agricultural practices, and extended dry seasons, particularly during El Niño events. In contrast, Australia experiences intense bushfires primarily fueled by extreme heat, drought, and natural ignition sources such as lightning. This research analyses fire trends in both countries, identifying key factors that contribute to their frequency and severity. Lessons learned from past incidents, including the 2019–2020 Australian bushfire crisis and Indonesia’s recurrent peatland fires, emphasize the need for improved fire management strategies. Various mitigation approaches, such as early warning systems, controlled burns, policy regulations, and community-based prevention programs, are discussed. In addition, advancements in remote sensing, artificial intelligence, and IoT based monitoring systems are examined as potential solutions. This comparative analysis highlights the importance of regional cooperation, adaptive policies, and sustainable land management to mitigate future fire risks. Learning from each nation’s experiences, policymakers and stakeholders can develop more effective strategies for combating forest fires and reducing their long-term effects.

Assessing Moral Development and Learning Outcomes in STEM Education: Evidence from an Indonesian Informatics Program

Deby Erdriani — 2026-04-27

This study examines the dual role of mathematics learning in developing moral character and STEM competencies among informatics engineering students. Despite mathematics being a fundamental component of engineering education, its potential as a vehicle for character development remains underexplored. Through a mixed-methods approach involving 55 informatics engineering students from an Indonesian university, significant correlations were found between mathematical ability and moral character development (r = 0.42, p < 0.01) and career readiness (r = 0.38, p < 0.05). The integrated STEM mathematics learning approach, which connected discrete mathematics to real-world informatics problems, proved effective in fostering both computational thinking (demonstrated by 42% improvement in programming abstraction) and professional ethics (28% enhancement in ethical awareness). Thematic analysis revealed emerging themes “including ‘increased awareness of data privacy’ and ‘appreciation for logical consistency in ethical reasoning.’ ” Results indicate that mathematics serves not only as a technical foundation but also as a medium for character development in engineering education. The study recommends an integrated curriculum design that leverages mathematics learning for holistic student development, addressing the growing need for responsible innovation in technology fields.

Safety as an Organizational Dimension: Challenges of Public Management in Risk Prevention and Institutional Responsibility

Magda da Silveira Elkfury — 2026-04-27

The study seeks to reflect on the importance of occupational safety in working at heights during public building maintenance, emphasizing organizational challenges and structural proposals that may contribute to risk mitigation. Methodologically, the research is developed through bibliographic review and documentary analysis, with data interpretation based on categorical content analysis. The investigation is relevant as it highlights the importance of safety in activities performed at heights in public buildings with high circulation, since the results indicate that analyzing recurring failures and adopted preventive measures supports the implementation of management systems that promote a culture of prevention and significantly reduce accident rates. The study contributes to broadening the understanding of height safety in public administration; socially, it reinforces its ethical dimension by valuing life, dignity, and human integrity in public service.

Development of a Microcontroller-Based Automated Pest Control Spraying System

Bless Gordo Ampuan — 2026-04-27

This study aimed to develop and evaluate an automatic pest control spraying system designed to enhance precision in pest management while minimizing human exposure to pesticides and reducing environmental impact. The system was tested on a plant bed to assess its efficiency, effectiveness, sensitivity, and economic viability, using parameters such as payback period, internal rate of return (IRR), and benefit-cost ratio (BCR). Employing a descriptive-experimental design, the prototype automatically detected pests and sprayed pesticides only at the detected location. The system consisted of a 230V AC source, digital time relay switch, Arduino Uno, PIR sensors, relay module, solenoid valve, and pump. Results showed that larger sample sizes improved effectiveness and sensitivity. The pump and solenoid valve efficiencies were 95.49% and 93.46%, respectively, while sensor efficiencies were 75.00%, 66.67%, and 29.17% for caterpillar, cockroach, and grasshopper detection. Economic analysis indicated a payback period of 8 months and 5 days, an IRR of 39.46%, and a BCR of 1.41, signifying strong financial feasibility.

Enhancing Playfair Cipher Security Using Chaotic Maps: A Comparative Analysis of Logistic, Hénon, and Arnold Cat Maps

Radhika Patel — 2026-04-27

Today, classical encryption systems such as the Playfair cipher are easily broken in the current computing environment. Traditional ciphers based on digraph interactions cannot resist attacks using frequency analysis and pattern recognition, making them unsuitable for modern security. This work presents an improved Playfair cipher by integrating three chaotic maps—Logistic, Hénon, and Arnold Cat—to generate encryption keys dynamically and distort ciphertext. This integration creates a hybrid system that is highly secure yet computationally efficient. Experimental results using Shannon entropy and Lyapunov exponent metrics show clear performance advantages. The Hénon Map proves superior in randomness, achieving a Shannon entropy of 4.11257. The Arnold Cat Map, with a Lyapunov exponent of 0.89813, demonstrates strong sensitivity to initial conditions, preventing brute-force attacks. The Logistic Map provides a balanced compromise (entropy: 3.97695, Lyapunov: 0.63663) between security and resource efficiency. This approach augments the traditional Playfair cipher into a robust modern security solution, showing how classical cryptographic techniques combined with chaos theory can effectively meet contemporary digital security demands.

The Use of Machine Learning in Online Course Sentiment Analysis: A Case Study Post COVID-19 Pandemic

Darwan Darwan — 2026-04-27

The number of tweets on Twitter containing information related to online lectures in the post-pandemic Covid-19 period or after the government lifted the pandemic period has drawn many pros and cons. This research discusses the best level of accuracy of the three machine learning methods, namely Naïve Bayes (NB), Decision Tree (DT) and Support Vector Machine (SVM) in analyzing Twitter results related to online lectures. The data used is crawled data from Twitter using the keyword “Online Lecture” and 5,978 tweets were obtained after verification. Three category labels were used in this research, namely Positive, Negative and Neutral. Pre-processing was carried out using the stages of Data Cleansing, Tokenizing, Stopword, Normalization and Stemming. While the feature extraction stage uses the TF-IDF stage. The average accuracy results for the NB method are 78.02%, precision is 78.18%, recall is 78.02% and f1-score is 77.45%. DT obtained an average accuracy of 86%, precision is 89.14%, recall is 81.14% and f1-score is 82.31%. The average accuracy obtained by SVM is 89%, precision 90.97%, recall 87.31%, and the F1 score is 85.99%. Among the three algorithms, it is known that SVM achieves very good performance in online lecture sentiment analysis based on Twitter data.

Impact of Digital HR Practices on Environmental Sustainability in Healthcare Institutions of Andhra Pradesh

Suresh Yamarthi — 2026-04-27

Background: Electronic Human Resource Management (e-HRM) has applied increasingly as a strategic tool to improve HR effectiveness and sustainability of healthcare organizations. Despite this, there is little empirical evidence on utilization and potential sustainability value in the Andhra Pradesh hospitals.

Objectives: The present study aims at reviewing the prevalence, reporting and perceived impact, practices of e-HRM on sustainability of organization operating of in health care sector of Andhra Pradesh.

Methods: This was descriptive research with quantitative approach. Participants HR practitioners, administration, and health workers in private hospital settings situated in urban and semi-urban areas. A sample size of 250 was determined based on stratified random sampling method and a structured questionnaire was used to collect data. Instrument reliability and validity were obtained through Cronbach alpha and content validity, respectively. Analytic tools such as correlation, t-test ANOVA, Factor Analysis and SmartPLS were used to analyze collected data.

Results: We conclude that adoption of e- HRM practice in hospital level is moderate and in HR and administration functions higher than clinical support functions. Respondents believe that e-HRM as a practice makes the organization more efficient and sustainable.

Conclusion: The study highlights the need for e-HRM in the healthcare sector to be employed as a strategic HR performance tool and sustainability enabler. Actualities: Perhaps, there is a call to increase the level of awareness, training, and infrastructural support in so far as they could and can support the optimal benefits of e-HRM implementation.

Intelligent Security Surveillance System Based on Multi-Modal Object Detection and Edge Computing

Shraddha More — 2026-04-28

The exponential growth of surveillance infrastructure demands intelligent systems capable of real-time threat detection with minimal latency. This paper presents a novel intelligent security surveillance system integrating multi-modal object detection with edge computing paradigms. Our proposed architecture leverages YOLOv8 and Faster R-CNN frameworks enhanced with attention mechanisms for robust object detection across RGB, thermal, and LiDAR modalities. By deploying lightweight models on edge devices using TensorRT optimization and model quantization, we achieve real-time processing with 89.7% mean Average Precision (mAP) while reducing inference latency to 47ms. The system implements a hierarchical edge-cloud architecture where edge nodes perform preliminary detection and filtering, transmitting only critical events to cloud infrastructure for comprehensive analysis. Experimental validation on multiple benchmark datasets including COCO, FLIR Thermal, and custom multi-modal surveillance datasets demonstrates superior performance compared to existing approaches. Our system achieves 94.3% detection accuracy for person detection, 91.8% for vehicle detection, and 88.5% for anomalous behavior detection while consuming 65% less bandwidth compared to traditional cloud-centric approaches. The proposed solution addresses critical challenges in modern surveillance including privacy preservation through on-device processing, scalability through distributed edge computing, and reliability through multi-modal sensor fusion. Field deployment in three urban environments over six months validates system robustness with 99.2% uptime and <50ms end-to-end latency. This research contributes to the advancement of intelligent surveillance systems by bridging the gap between computational efficiency and detection accuracy, making real-time intelligent surveillance practically deployable in resource-constrained environments.

Evaluating the Effectiveness of Deep Learning Models for Chest X-Ray Image Classification

I Komang Somawirata — 2026-04-27

A chest X-ray (CXR) examination is one of the radiological examinations used to help a doctor diagnose a disease in patients safely, quickly, and inexpensively. The development of Computer-Aided Diagnosis (CAD) systems has prompted numerous researchers to explore methods for detecting diseases using X-ray imaging. By implementing this research, it is hoped that it will enable medical personnel to accurately and quickly diagnose patients' diseases. This study utilizes three datasets of aortic enlargement, cardiomegaly, and COVID-19. A Convolutional Neural Network (CNN) is one method that researchers widely use to build Computer-Aided Design (CAD) systems. This study aims to compare the performance of seven CNN models with different architectures to determine which one produces the highest accuracy. The seven CNN models used include DenseNet121, DenseNet169, DenseNet201, InceptionV3, MobileNet, ResNet50, and Xception. The test results show that the DenseNet201 model, with an input size of 224 × 224 pixels, achieves the highest accuracy value for all datasets, reaching over 90% accuracy.

Multi-Scale Attention-Guided CNN-BiLSTM Framework for Emotion Recognition in Multimodal Video Data

J. Biju — 2026-04-27

A mental state, emotion is connected to human behaviour, thoughts, and the degree of positive or negative experiences. Human emotion does not yet have a precise definition. By allowing AI systems to precisely comprehend and sympathetically react to human emotions, this discovery has the potential to completely transform human-machine interaction and open the door for increasingly sophisticated and emotionally intelligent computers. The main research problem is creating models that accurately read emotions from multimodal data; this calls for big, diverse datasets for video data to capture complex emotional cues and fine-tuned CNNs for audio data to identify minor speech changes. This study introduces a novel multimodal emotion detection method that seamlessly combines voice and video modalities to correctly infer emotional states. The attention-based CNN-Bi-LSTM model handles the video component and provides deep semantic understanding through its bidirectional layers. An attention-based fusion process is used to blend the results of both modalities, balancing their respective contributions. Here, the suggested methodology is thoroughly tested using two different datasets: the YouTube and Carnegie Mellon University SAVEE datasets.The results show higher efficacy compared to current frameworks. This comprehensive technology enables accurate emotion recognition and contributes to a number of noteworthy developments in the industry.

Resource Allocation Metaheuristic Techniques in Cloud Computing

Sonia Sharma — 2026-04-27

Cloud computing is a computer paradigm that delivers IT resources as services, such as platforms, apps, and infrastructure, via the Internet. The cloud Computing offers the infrastructure needed to process and compute any kind of data resource, and it is used to handle massive volumes of data. Large, well-known businesses have moved their processing and storage to cloud computing in recent years. Businesses and organizations may lower their infrastructure costs by utilizing cloud computing. Businesses can test their apps faster, more effectively, and with less maintenance. Cloud computing enables the IT team to adjust resources to fluctuating and changing requirements. Allocating resources in cloud computing is intrinsically difficult since more and more people are using various cloud apps in some infrastructure. The majority of resource allocation solutions now in use focus on performance, which is impacted by the volume of applications from scientific and business domains. This article presents an analysis of meta-heuristic approaches for resource allocation in cloud computing systems. When allocating resources in the cloud, the examined meta-heuristic algorithms can achieve much better performance, lower costs, shorter turnaround times, better resource usage, and increased energy efficiency. This study compares several scheduling algorithms for cloud and grid systems using three well-known metaheuristic approaches: Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Ant Colony Optimization (ACO).

Enhanced Detection of Student Depression Using an Optimized Machine Learning Model

Saad Adnan Abed — 2026-04-27

Depression is an increasing concern among students adversely affecting academic performance and mental well-being. Early prediction is important for timely intervention. This paper aims to classify students into "Depressed" or "Not Depressed" categories utilizing the Depression Student Dataset. A comparative analysis of several traditional machine learning (ML) approaches, counting Gradient Boosting (GB), Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), XGBoost, LightGBM, AdaBoost (AB), Naïve Bayes (NB), and Decision Tree (DT) was performed to evaluate their predictive abilities. To enhance accuracy of prediction, this paper proposed an optimized LR model fine-tuned utilizing GridSearchCV. The optimized model shows superior performance with an accuracy rate of 98%, outstanding all other algorithms in this study. The findings highlight the efficiency of model tuning in improving depression classification results. This research proposed a robust framework for used ML to classify depression among students, contributing to early pridiction and support strategies.

Experimental Investigation of Orientation-Dependent Tensile Behaviour in FDM-Printed ABS and CF-ABS Components

Kaustubh Pravin Joshi — 2026-04-27

Fused Deposition Modeling (FDM) is widely utilized for manufacturing polymer parts, but attaining optimal mechanical properties continues to be difficult. This research investigates how build orientation and carbon fiber (CF) reinforcement affect the tensile properties of Acrylonitrile Butadiene Styrene (ABS) components made via FDM. Specimens of pure ABS and CF-reinforced ABS (CF-ABS) were produced in X, Y, and Z orientations and underwent tensile testing. The findings showed a significant reliance of mechanical properties on material composition and orientation. The Y-oriented CF-ABS sample demonstrated the greatest ultimate tensile strength (28.90 MPa) and Young’s modulus (2296.65 MPa), whereas Z-oriented specimens displayed considerably reduced strength as a result of inadequate interlayer adhesion. CF reinforcement improved stiffness and strength in the X and Y directions but lowered ductility, as elongation dropped from 1.06 % (ABS) to 0.67 % (CF-ABS). The analysis using Taguchi-based Design of Experiments (DOE) demonstrated that the most significant factors were orientation and material type. In summary, CF reinforcement significantly enhances tensile properties in advantageous orientations, offering essential information for refining FDM process settings in polymer composite production

Web Service Qos Prediction Based on Autoencoder with Mini-Batch Gradient Descent

Le Van Thinh — 2026-04-27

Reliability prediction of Web services has become very important in related research communities. Especially,
predicting the Quality of Service (QoS) for active users has been a hot issue of research and application. On the other hand,
with the rapidly growing in number of service providers and users, resulting in a large number of data sets. It has a big
impact to the QoS such as managing and monitoring for describing functional and non-functional characteristics of Web
services. Therefore, we will certainly struggle at processing large data sets in the future, unless the issues are resolved
quickly before it happens. In that context, QoS prediction on big data set is an urgent problem to be solved. In this paper,
we present a new model for handling this problem based on autoencoder, it is called autominibatch. We use this model to
cope with large data sets by using Backpropagation and Mini-batch gradient descent for predicting QoS values of Web
services. This also is a new method for evaluating prediction of the field of web service quality. Our experiments were
performed on two data sets in the WS-DREAM data set and the experimental results have proved the effectiveness of the
proposed model.

A Machine Learning and Hybrid Feature Selection Framework for Predicting Preterm Birth in Nulliparous Women Using Electronic Medical Records

T Ummal sariba begum — 2026-04-27

Preterm birth (PTB) is a leading cause of neonatal mortality worldwide, underscoring the need for accurate early prediction. This study presents a machine learning framework combined with hybrid feature selection to predict PTB risk in nulliparous women using data from electronic medical records collected during prenatal consultations. Clinical, demographic, and physiological data were obtained from the publicly available nuMoM2b dataset, covering three gestational intervals: 6–13 weeks, 16–21 weeks, and 22–29 weeks. Data preprocessing and standardization were performed using Differential Evolution (DE) to enhance quality and improve model performance. A hybrid feature selection approach, integrating the Dragonfly Optimizer with entropy-based relevance ranking, was employed to identify informative and non-redundant predictors, reducing dimensionality and noise while maintaining clinical interpretability. A Multilayer Perceptron (MLP) classifier trained on the selected features differentiated term and preterm deliveries. The framework achieved 88.41% accuracy, AUC = 0.80, precision = 85.84%, recall = 87.24%, and F1-score = 88.15%. Incorporating ultrasound features such as cervical length and Plasticity Index further improved predictive performance. Notably, at the third prenatal consultation, the model reached 85.62% sensitivity for predicting very preterm infants. These findings highlight the importance of ultrasound measurements and demonstrate that integrating machine learning with evolutionary optimization and entropy-based feature selection can significantly enhance early PTB risk detection. The approach enables timely interventions for high-risk pregnancies, potentially improving maternal and neonatal outcomes. This study underscores the value of computational methods in clinical decision support and emphasizes how machine learning can transform prenatal care for nulliparous women.

Personalized Information Retrieval Using LETOR Machine Learning Re-Ranking Algorithms in MSLR-WEB10K Dataset: A Comprehensive Study

M Poomani — 2026-04-27

Personalized Information Retrieval (PIR) aims to tailor search results to individual user preferences and contexts. With the exponential growth of digital information, traditional retrieval systems often fall short in delivering relevant results to diverse users. This study explores the integration of machine learning re-ranking algorithms into personalized information retrieval systems to enhance search relevance and user satisfaction. The LETOR based model is experimented for relevance re-ranking in personalized retrieval. A comprehensive analysis of LETOR based models are analyzed to find a best hybrid re-ranking framework based on the performance of each model. The findings demonstrate that the LGBMRegressor model demonstrated the most consistent and best performance across the majority of metrics.

SegRivWidth: River Width Measurement Using Deep Learning

Hasmukh P Koringa — 2026-04-27

A precise river width measurement is crucial for various river modeling, habitat evaluation, flood risk analysis, and other hydrological environmental, and technical applications. Conventional techniques is time-consuming and error-prone, such as direct field measurements. Nowdays fast and accurate riven identification and its width measurement is highely needed to save human life during flud and other natural disaster. Recent deep learning technology can greately be applied to identify and measure width of river automatic, fast and accurate from a remote place. The proposed deep learning based method is executed in two steps, identification of river and river width measurement. Deep learning based segmentation is used to identify river from remote sensing image. The accuracy of the semantic segmentation to identify river depends on rich spatial data and the resolution of the remote sensing images. In this work proposed SegRivWidth algorithm for automatic river width measurement from segmented images. The obtained results are compared with the ground truth river width and found better accuracy. The obtained results are also compared with the existing methods in terms of Average Absolute Error (AAE) and Root Mean Square Error (RMSE). The proposed SegRivWidth has an RMSE of 4.76 m and an AAE error of 2.16 m for the river width measurement.

Embryo Selection and Classification in IVF Using Hybrid Deep Learning Approach

J. Deepa — 2026-04-27

Accurately evaluating and choosing viable embryos for implantation is crucial to the success rate of in vitro fertilisation (IVF). Traditional evaluation techniques mostly rely on the subjective and variable eye observation of embryologists. This study suggests a deep learning-based automated embryo classification framework that combines image preprocessing, segmentation, and classification in order to overcome these drawbacks. First, to improve the contrast and clarity of small morphological characteristics in embryo photos, Contrast Limited Adaptive Histogram Equalisation (CLAHE) is used. The enhanced images are then subjected to a sprint semantic segmentation network (SSS-Net) to ensure that only relevant features support classification and accurately differentiate the embryo region from the backdrop. Using a hybrid Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM) model, the last step successfully classifies embryos into different quality categories. This method captures both spatial and sequential feature dependency. The experimental evaluation demonstrates the durability and dependability of the proposed framework with high precision, recall, F1-score, and accuracy, demonstrating strong predictive performance. When compared to conventional techniques, the combination of sophisticated preprocessing, effective segmentation, and hybrid deep learning architecture greatly increases classification consistency. All things considered, the suggested method offers an automated, scalable, and objective tool for evaluating the quality of embryos. It may also help embryologists make clinical decisions and eventually increase the success rates of IVF.

Segmenting Hepatic Blood Vessels and Liver Tumors in CT Images: An Improved Evolutionary Algorithm Approach

Selvakumar Subramanian — 2026-04-27

Among the many forms of cancer, liver tumours are among the most dangerous. Liver neoplasms can be effectively predicted, identified, and managed with the help of computer-aided technology and liver interventional surgery. Accurately understanding the morphological nature of the liver and its blood arteries is a crucial task. An essential part of medical analytic planning is the segmentation of liver tumours in CT scans. The enormous difficulty, however, lies in correctly identifying and segmenting the hepatic blood veins in CT scans. Finding and segmenting hepatic vessels manually in CT scans is an inconvenient and time-consuming process. In order to segment liver tumours, this study employs a variety of techniques to clean up the input images before feeding them into the STDCSL, an algorithm for short-term dense concat segmentation. Three parts make up the STDC-CT network: detail guidance, multi-scale contextual information, and small object attention extractor. Small affected area attention guides the merging of detailed and contextual information branches. This study proposes an improvement to the Barnacle Mating Optimizer (BMO), an evolutionary algorithm that takes its cues from nature, in order to fine-tune the STD-CSL parameters. Levy flight is used to enforce and replace the sperm cast equation, which improves the exploration phase of the original BMO. Next, the enhanced BMO (IBMO) is teamed up with the suggested STDCSL. The technology proves to be reliable and applicable to automatic analysis of liver tumours in everyday clinical practice, proving its generalizability. The method's great accuracy in stroke detection further supports its potential use as a clinical tool for preoperative clinical planning.

Evaluation of the operational performance of an automated deep-freezing tunnel: cycle times, thermal stability, and handling errors.

Franklin Wilfrido Salazar Logroño — 2026-04-27

This study assesses the operational performance of an automated blast-freezing tunnel using a reproducible metric framework. Three scenarios are evaluated: (A) baseline operation, (B) improved sealing, and (C) PLC-optimized sequencing. Primary variables include total cycle time (broken down by stages), thermal stability across zones (standard deviation and inter-zone gradients at inlet–center–outlet), and the manipulation error rate (jams, retries, misalignments). The setup uses thermal probes sampled at 1 Hz, PLC/HMI timing markers, and a structured classification of operational events, while controlling for confounders (product mass, placement, and set-points). Findings indicate that enhanced sealing reduces thermal variability and inter-zone gradients, whereas optimized sequencing significantly shortens total cycle time without increasing errors, meeting non-inferiority criteria for thermal uniformity. Practical implications are discussed regarding traceability, operational availability, and continuous improvement of the freezing process in academic-industrial environments.

A GIS-Based Site Suitability Assessment System for Sustainable and Productive Fish Rearing

Gerry L. Contillo — 2026-04-27

The study focused on the development of a web-based fishpond planner and recommender for Ilocos Norte that fills the gap in sustainable aquaculture site selection. Adopting a descriptive-development approach, the system combined the Analytic Hierarchy Process with Geographic Information System (GIS) for evaluating major environmental factors; water availability, soil type, flood susceptibility and water quality in terms of their weighted contributions to site suitability for fishpond. The system provided user- friendly visualization and decision supporting the built-in capabilities of Geographic Information System (GIS), remote sensing, and interactive web mapping. Major findings indicate that the tool ranks fish species well for appropriate locations and generates trust-worthy, simple, implementable recommendations. Acceptance testing responses were overall positive in usability and reliability, with recommendations on slight improvement regarding interface responsiveness/guidance amenities. The study concludes that the GIS-based, AHP-driven platform created is potentially useful as a decision-support tool for sustainable fish farming in Ilocos Norte. The study provides a step toward promoting efficient, environmentally friendly and technology-based planning in the aquaculture sector.

A Communication-Free PQ-Based Control Strategy for 7.7 kW Bidirectional ‎‎Wireless EV Chargers

Abdelmoumin Hamrani — 2026-04-27

The widespread deployment of electric vehicles (EVs) has significantly heightened the need for more efficient and intelligent charging technologies. Bidirectional wireless power transfer (BD-WPT) systems enable both grid-to-vehicle (G2V) and vehicle-to-grid (V2G) operation, supporting applications such as renewable energy integration, demand peak mitigation, and system frequency stabilization. However, existing control strategies face challenges related to synchronization, communication delays, and robustness under detuning and misalignment. This paper investigates a PQ-based synchronization control technique for a 7.7 kW series–series compensated BD-WPT charger, designed in compliance with SAE J2954 standards. The adopted method relies solely on local measurements of active (P) and reactive (Q) power, eliminating the need for communication links while achieving decoupled control of power direction and magnitude. A voltage-controlled oscillator (VCO) ensures robust phase synchronization between the inverter and rectifier. Simulation results demonstrate that the PQ-based control achieves accurate bidirectional power transfer, seamless G2V/V2G transitions, effective power regulation, and resilience against detuning and coil misalignment. These findings underscore the promise of PQ-based synchronization in improving the efficiency, reliability, and scalability of BD-WPT systems for future smart grid integration.

Numerical Study on the Influence of Surface Modification Enhancing Thermal Performance and Optimization in Microchannel Heat Exchangers

Sorna Latha — 2026-04-27

Numerical investigations on the thermal performance of microchannel heat sinks featuring pin-fin surface modifications have been performed using Computational Fluid Dynamics (CFD) simulations. The simulations have been carried out to assess the flow and heat transfer behavior across 55 distinct microchannel configurations with varying pin fin heights with water as the coolant medium. The performance of pin-fin-enabled microchannels was compared against conventional flat microchannel to quantify the enhancement effects. Two fin heights (40 µm and 60 µm) were examined in laminar flow regime across Reynolds numbers ranging from 500 to 2000. Results indicate that the integration of pin fins significantly improves thermal response by amplifying heat transfer surface area and inducing localized flow disturbances. The channel configuration with 60 µm fin height demonstrated optimal performance, achieving a notable increase in the Nusselt number and overall heat transfer efficiency. These findings suggest that surface modifications through pin fins, even with water as a base fluid, can substantially enhance the thermal capability of microchannel heat exchangers and are suitable for compact, high-performance cooling applications.

Keywords: Microchannel heat sinks; Pin-fin surface modification; CFD simulation; Energy-efficient thermal management; Sustainable cooling technology; Water-based coolant; Heat transfer enhancement

Analytical and Numerical Analysis of Stress Concentration Singularities in Perforated Composite Wing Ribs

Ishak Berkane — 2026-04-27

Aeronautical structures commonly use perforated skin panels and wing ribs to reduce weight while maintaining satisfactory mechanical performance. However, the combined influence of aperture size, material anisotropy, and fiber orientation on global and net stress concentration factors in orthotropic composites remains insufficiently quantified. This work aims to analyze the stress concentration behavior of perforated panels made of isotropic and orthotropic composites. A uniaxial tensile load is modeled using the finite element method (Abaqus) and classical analytical models (Heywood, Lekhnitskii, and Green-Zerna), supplemented by parametric calculations in MATLAB for glass/epoxy and carbon/epoxy plates. The results show a maximum stress concentration factor of K_tg ≈ 6.6 for a geometric ratio d/W = 0.5 under an applied stress of 10 MPa, with a maximum error of 5% between the analytical and numerical models. A reduction of approximately 60% in the concentration factor is observed for a fiber orientation of 0°. Finally, the study proposes global and local stress concentration maps that incorporate fiber geometry/orientation compatibility limits, enabling the optimization of composite wing ribs while preserving the advantages of lightness, stiffness, and control of local stress amplifications, and opening up prospects for the study of damage and fatigue.

Fine-Tuning Strategies for Sentiment Analysis in the Algerian Dialect: A Comparative Study on DziriBERT

Salima Brachemi-Meftah — 2026-04-27

The aim of this work is to explore sentiment analysis in the Algerian dialect through the adaptation of pre-trained linguistic models. We focus on DziriBERT, a model specifically developed for the Algerian dialect and derived from the BERT (Bidirectional Encoder Representations from Transformers) model, recognised for its performance in various natural language processing tasks. Three fine-tuning approaches were studied: full fine-tuning, freeze tuning, and LoRA tuning (Low-Rank Adaptation). Experiments were conducted on two separate corpora, ADArabic and Adouane, to evaluate the robustness and generalisation of DziriBERT model. The results show that the LoRA method achieved the best performances, on the ADArabic corpus, it achieved 83.26% on terms of accuracy and 80.94% for F1-score. On the Adouane corpus, LoRA reached the highest performance, with an accuracy of 85.28% and an F1-score of 82.52%. These results confirm the relevance of using DziriBERT for sentiment analysis in the Algerian dialect and highlight the effectiveness of LoRA tuning as a lightweight and efficient alternative to full fine-tuning, with significantly reducing the number of adjustable parameters.

A Novel Approach of Transfer Learning for Accurate Brain Tumor MRI Classification in Big Data Healthcare Environment

Sunil kumar Agarwal — 2026-04-27

The complex structure of brain tumors and the need for prompt and accurate detection present a major challenge to the medical community. This paper suggests an improved CNN framework for classifying brain cancers in the big data healthcare arena, addressing the shortcomings of current diagnostic techniques. The CNN model was enhanced by integrating transfer learning methods and data augmentation using Magnetic Resonance (MR) images. The model's predictive performance was further improved by adding more training parameters in pre-trained deep learning models like ResNet-50, VGG-16, Inception V3, DenseNet201, Xception and MobileNet. According to experimental results, the suggested model performs better than baseline models and achieved a 99.40% classification accuracy rate. According to this study, a more precise and effective way to diagnose brain tumors could be achieved in clinical settings by using the suggested model. The future direction suggested enhancing the dataset and further refining the model to improve its generalization capabilities in diverse clinical scenarios.

Review of Toward A Circular Economy In The Tire Industry: Integrating Life Cycle Assessments And Recycling Innovations

Siti Nadiah Binti Mohd Saffe — 2026-04-27

The global tire industry generates ~1.5 billion end-of-life tires annually, amplifying waste, emissions, and health risks. This review synthesizes contemporary life-cycle assessment (LCA) evidence and recycling innovations—including pyrolysis, microwave devulcanization, and hybrid thermochemical routes—to map credible pathways toward circularity. We identify four persistent gaps: heterogeneous LCA methods and data silos; misaligned policy incentives; scalability and profitability limits of advanced recycling; and fragmented adoption of sustainable materials. To address these, we propose an integrated framework that couples standardized, sector-specific and dynamic LCAs with modular, upgraded pyrolysis for high-value recovery (carbon black substitutes, oils, steel), design-for-circularity (renewable elastomers, sensor-enabled tires), and enabling instruments (extended producer responsibility, differentiated subsidies, carbon crediting). The framework establishes feedback loops between tire design and end-of-life performance, supports regionally deployable solutions, and prioritizes interoperable data platforms, pilot deployments, and policy realignment. Collectively, these measures can accelerate the tire sector’s transition from linear to circular models while minimizing environmental burdens across production, use, and end-of-life stages.

Spotted Hyena Bio-Inspired Optimization Clustering and Ford Distributing Shortest Routing for Efficient Communication in Wireless Sensor Network

S. Shameema Begum — 2026-04-27

Wireless Sensor Networks (WSN) contain sensor nodes that transmit with each other using only wireless channels. They are deployed in unreachable regions to gather information about the circumstances. Nonetheless, sensor nodes have limited energy, which poses a significant challenge to expanding Network Lifetime (NL). Therefore, the most crucial issue is to reduce nodes' energy consumption and increase the NL. To cope with these problems, this research uses the Ford Distributing Shortest Routing Protocol (FDSRP) with Spotted Hyena Bio-inspired Optimization (SHBiO) technique for efficient data transmission in WSN. Initially, the proposed forms Cluster Members (CMs) using the Reliable Congestion Free Node Selection (RCFNS) method. Based on the CMs, the optimal Cluster Head (CH) picks using the SHBiO method. Subsequently, the proposed FDSRP protocol finds the shortest path between CH and the Base Station (BS). This proposed protocol works based on a proactive process to transmit packets. The proposed simulation experiment is conducted in a Network Simulator version 2 (NS2) environment. Analysis shows the proposed method produces significantly better throughput, packet transmission, and drop rate while consuming less energy than other methods.

A Review on AI based Fault and Anomaly detection in Power Systems

Jenisha K J — 2026-04-27

Power systems are becoming increasingly complex because of renewable integration, distributed generation, electric vehicles, and automation. These changes have created new types of disturbances and faults that are difficult to manage using traditional relay based Protection systems. Artificial intelligence provides data driven techniques that can learn pattern, classify faults, detects anomalies and support intelligent decision making I real time. This review summarizes key AI methods such as machine learning, deep learning, fuzzy logic and reinforcement learning, and explains how different researchers have applied them to power system fault detection and anomaly monitoring. The paper also compares the advantages and limitations of each techniques, and cyber security. Finally, future research directions toward adaptive, explainable, and resilient AI driven protection systems are discussed.

Multi-Objective Pelican Optimization Algorithm for Optimal Placement of EVCSs and DGs in Radial Distribution Networks

Swetha Veligaram — 2026-04-27

The rapid adoption of electric vehicles poses considerable challenges for the operation of radial distribution networks, due to the increased burden and voltage issues at electric vehicle charging stations (EVCS). This paper introduces the Pelican optimization algorithm (POA) for the optimization of distributed generators (DG) to alleviate the negative effects of EVCS loading on the distribution system. Five cases of DG are examined: Type-I (unity power factor), Type-II (reactive power only), Type-IIIA (power factor of 0.85), Type-IIIB (power factor of 0.9), and Type-IIIC (optimized power factor). The proposed methodology is validated by implementing it on two test systems, and the obtained POA results are compared with Particle Swarm Optimization (PSO) results. Type-IIIC DG provides the best overall performance, achieving an impressive 90.39% loss reduction for the IEEE-33-bus system and a remarkable 98.02% reduction for the 69-bus system, and also ensures excellent voltage stability. Type-II DG shows poor performance and is considered unsuitable for practical applications. POA has shown strong effectiveness in achieving the desired objectives, especially with multi-DG configurations

Intelligent Worksite Protective Equipment Compliance Detection Using Deep Neural Networks

Geetha G — 2026-04-27

This research introduces a real-time framework for Personal Protective Equipment (WPE) identification utilizing YOLOv8, an advanced object detection model. The suggested system is intended for the ongoing surveillance of building sites and industrial settings to autonomously confirm worker adherence to obligatory safety equipment, including hardhats and highly visible safety vests. The system analyses input stream from a live camera or pre-existing picture datasets to execute real-time detection of WPE violations in video frames. Upon detecting non-compliance, automated alarm notifications are promptly created and dispatched over the Telegram instant messaging platform, facilitating swift safety intervention. The implementation is constructed with Python, utilizing OpenCV with video processing as well as the YOLOv8 system for object identification and inference. The suggested solution presents an effective scalable, and scalable safety compliance tracking system that can be effortlessly incorporated into current industrial safety overall surveillance infrastructures to improve worker security and workplace safety.

Power Quality Enhancement using a FOPID–Controlled Quadratic Boost Converter Integrated Hybrid Active Filter

Venkata Ramana Rao S — 2026-04-27

This work presents an integrated renewable energy interfacing system that combines a Quadratic Boost Converter (QBC) with a Hybrid Active Filter (HAF) to enhance power quality and voltage regulation in photovoltaic (PV) and battery-supported hybrid energy sources. The coordinated operation of the converter–filter system is supervised by two closed-loop controllers: a classical Proportional–Integral (PI) controller and a Fractional-Order Proportional–Integral–Derivative (FOPID) controller. The controllers are designed to regulate the DC-link voltage of the QBC while ensuring harmonic suppression in the grid-side current. The Simulink-based evaluation demonstrates that the FOPID regulator delivers faster transient response, improved steady-state accuracy, and reduced harmonic distortion compared to the conventional PI scheme, making the topology suitable for power-quality enhancement in distributed renewable systems.

QoS-Aware Internet of Environmental Things Algorithm for IoT-Fog-Enabled Smart Waste Monitoring System

Pooja Rani Agarwal — 2026-04-27

Modern times have seen the rise of smart cities that rely on the Internet of Things. To process data, smart objects require multi-hop communication to transmit it to the sink. Internet of Things (IoT) enabled, data-driven smart cities enhance public utilities, infrastructure, and services for the benefit of residents. Smart towns that rely on the Internet of Things have a tough time collecting clean municipal garbage. Cities produce more waste as they expand. Two big problems with trash management are garbage collection and sorting. Public trash cans overflow before they are cleaned, leading to bacterial growth, unpleasant odours, and potential illness. Urbanization, inefficient technologies to aid solid waste management, and poor trash collection and disposal all contribute to these problems. Trash everywhere makes garbage collection more challenging. Solid waste management systems that aren't scientific, efficient, or good enough lead to a lot of issues. By combining the strengths of the IoT and fog computing, this paper proposed an integrated IoT-Fog-Cloud architecture for smart waste management systems and a QAIoET Algorithm for efficient resource allocation, which can improve service quality while reducing response time by 46.8% and costs by 2.6% compared to cloud solutions.

Temperature Effect on pH and Residence Time as Essential Parameters for Biogas Production: A Review

Córdova-Rodríguez Daniel David — 2026-04-27

This study presents a systematic review of literature on the influence of temperature on methane production as biogas. A total of 337 scientific documents were collected from the Web of Science database and processed using tools such as Excel, Mendeley, and PRISMA approach literature review tool as PICOT method to identify the articles that best match the objectives of the study by eligibility criteria. The systematic review investigated 25 relevant studies, comparing mesophilic (30–40°C) and thermophilic (50–100°C) conditions. It was found that, although temperature is a determining factor in the efficiency of methanogenesis, its effect is conditioned by variables such as substrate type, residence time, pH, and the pretreatment methods applied. The results show that the mesophilic regime tends to offer greater operational stability, while the thermophilic regime can accelerate anaerobic digestion, but with a higher risk of inhibition by compounds such as free ammonia and volatile fatty acids. The synergy between adequate pretreatments and substrate selection is key to optimizing methane production. This work provides an updated overview of scientific knowledge in this area, identifying gaps and opportunities for future research aimed at the efficient design of thermally optimized anaerobic digestion systems.

A Bottom-up k-anonymization approach for big data publishing

Abderrahmane Saidi — 2026-04-27

As governments and other organizations share larger datasets, keeping individual information private has become increasingly difficult to solve. When publishing the data, data anonymization models like k-anonymity and l-diversity are employed to ensure the trade-off between privacy and data utility. This paper presents a method called Bottom-Up k-anonymization (BU-K), implemented on Apache Spark. It improves efficiency by applying the Bottom-Up Generalization (BUG) approach. BU-KC performs better than Top-Down Specialization (TDS) in terms of scalability, and data privacy, while still keeping the data useful. Moreover, using Apache Spark’s distributed computing architecture significantly improves processing time compared to traditional MapReduce approaches. This work fills a gap in distributed anonymization on Spark by offering a new, efficient, and scalable solution

C2DR-VAE: Adaptive Recommendation Framework via Cluster-Conditioned and Dynamically Refined Variational Autoencoders

R. Navin Kumar — 2026-04-27

Recommendation systems are crucial to customer interaction in the digital realm and nevertheless, it continues to face the old classic issues of the cold start problem, scalability, and capacity to keep up with rapidly evolving customer preferences. This study presents a new adaptive, scalable, and privacy-aware recommendation system C2DR-VAE (Cluster-Conditioned and Dynamically Refined Variational Auto-Encoder), which combines clustering and generative modeling dynamically. First, the data of user-item interaction is clustered with the K-means to create behaviourally consistent clusters, which also serve as previous knowledge to the VAE. These clusters are optimized upon training through the proposed structure, as opposed to the traditional means of training that employs a fixed preprocessing, by dynamically updating the centres of the clusters at a specified frequency, with the learned latent embeddings. This dynamically refining process will allow the model to dynamically update dynamically changing-user behaviors and learn the representation of the model efficiently. C2DR-VAE addresses issues of cold start sparsity based on a mixture of robust data-driven clustering and the generative capability of VAEs, albeit with high scalability to large data sets and high-quality personalized recommendations with dynamic environments. In order to be robust, the framework is tested with cross-domain transfer experiments across various domains. Evaluation is conducted through an integrated framework that combines standard accuracy and ranking measures, user-centric metrics and system-level performance indicators to comprehensively assess both the scalability and effectiveness of the developed model. The specified framework fosters the studies of the recommendations that offer user-centered, domain-strength and self-enhancing solution.

Hybrid Feature Selection for COVID-19 Text Classification using Cuckoo Search Optimization and Mutual Information with DeBERTa

Mohamed Goismi — 2026-04-27

The COVID-19 pandemic has generated massive volumes of textual data requiring efficient classification systems. Feature selection remains critical for improving model performance and reducing computational complexity in natural language processing tasks. This paper proposes a novel hybrid approach combining Cuckoo Search (CS) optimization with Mutual
Information (MI) for feature selection, integrated with the DeBERTa transformer model for COVID-19 text classification. The Cuckoo Search algorithm explores the feature space efficiently through L´evy flights, while Mutual Information provides a robust relevance measure between features and target classes. Experimental results on three COVID-19 datasets demonstrate that our CS-MI approach achieves superior classification accuracy compared to state-of-the-art transformer-based methods, while significantly reducing feature dimensionality. The proposed method achieves 94.2% accuracy on Twitter data, 93.5% on news articles, and 95.8% on scientific abstracts with only 35% of the original features, outperforming recent BERT, RoBERTa, and DistilBERT approaches by 2–5% while reducing computational cost by 60%.

HCAVR-PLE: A Dual-Phase Adaptive AI-VR Framework for Equitable, Explainable, and Engaging Physics Education

Premalatha R — 2026-04-27

There is still an issue of equity and involvement in secondary STEM education, especially among diverse students dealing with the concepts of physics in resource-limited environments. Traditional teaching methods, such as traditional virtual reality (VR), tend to be non-personalized, restrictive in inclusivity, and thus ineffectual. The study proposes HCAVR-PLE, a Human-Centered Adaptive Virtual Reality Personalized Learning Environment that combines the Cognitive Affective Model of Immersive Learning (CAMIL), multi-armed bandit algorithms, explainable AI (XAI), and reflective prompts to provide clear, bias-free teaching. HCAVR-PLE had enormous effects, where the AI-VR (Real) had an increase in PAT of 29.2% compared to the non-adaptive groups ( =0.39, p 0.05). CAMIL engagement was greatest in adaptive AI-VR ( =0.52) and gender/SES equity gaps were least in the AI-VR (Real) condition. The framework was validated using the dual-phase method: 300 synthetic learner profiles (Phase I) based on PISA and OULAD data sets, and 221 Grade 8 students (Phase II) in the state of Tamil Nadu, India and compared to using Teacher dashboards, built based on HG-SCM-based explainable AI, made delivery of personalized experiences interpretable and equitable, whereas Phase II had established efficacy of framework without intervention of dashboard. These results reveal the potential of HCAVR-PLE to revolutionise STEM education by delivering greater engagement and equity solutions, potentially providing a scalable, universal classroom model of learning.

Response Surface Methodology-Based Optimization of Injection Timing and Hydrogen Flow in an Mahua Biodiesel–Hydrogen Dual-Fuel Engine

Kumaran P — 2026-04-27

Investigations on alternative diesel engine fuels have intensified due to the growing demand for cleaner energy sources. This study adopts a novel methodology by examining the performance, combustion, and pollutant emissions of a single-cylinder, four-stroke variable compression ratio engine fueled by an M25 biodiesel-hydrogen dual-fuel blend, with optimal injection timing in conjunction with hydrogen enrichment. The subsequent injection timings and hydrogen flow rates evaluated were IT21, IT24, IT27, and IT30 degrees prior to TDC. Under full load conditions, IT30 operating with a hydrogen flow rate of 16 lpm attained a minimum brake-specific fuel consumption (BSFC) of 0.28 kJ/kWh and a maximum brake thermal efficiency (BTE) of 32.7%. The experimental settings elevated the cylinder pressure by 11.7 % and the heat release rate by 8 % relative to usual operation. The emission investigation results indicated a 54% reduction in carbon monoxide (CO), a 24.7% reduction in hydrocarbons (HC), and a 37.2% reduction in smoke opacity; nevertheless, due to elevated combustion temperatures, NOx emissions increased by 7.2% to 820 ppm. The ideal working condition, as determined by Response Surface Methodology optimization is IT30.25˚ and 16 lpm H2. This design forecasts a BTE of 32.52% and NOx emissions of 790 ppm. Statistical validation was accomplished by diagnostic methods, and this study introduces an innovative methodology that combines biodiesel combustion enhancement with meticulous regulation of hydrogen flow and timing adjustments. M25-hydrogen blends demonstrate potential as a viable decarbonized engine technology, with a performance-enhanced approach to satisfy BS-VI and Euro VI standards

Algorithmic-Statistical Model Under a Computational Scheme for Forecasting Insolvency Under Financial Risk

Alexander Fernando Haro Sarango — 2026-04-27

This research focuses on the analysis of insolvency risk in credit unions in Ecuador using a predictive approach based on automatic learning. Based on international and regional precedents on the determinants of financial solvency, the study aims to develop a model capable of classifying credit unions as solvent or not solvent, according to a return on eq-uity (ROE) threshold of 5 %. The methodology adopted was quantitative, explanatory and predictive, using the Random Forest algorithm on a structured database with coded finan-cial variables. The slope variable, called ROE - LOGIC, classifies as “not solvent” (1) those observations with ROE lower than 5 %, and as “solvent” (0) the rest. The model was trained with 80 % of the data and validated with the remaining 20 %. The results show excellent performance, with precision, recall and F1-score metrics above 0.88 in the test set, and an AUC of 0.95, indicating a high discriminative power. The most influential variables were the net interest margin on promised assets, operating expenses relative to net interest margin and the proportion of performing assets. These were used to construct an interpretive formula that estimates the probability of insolvency

Modernizing Electrical Utility Networks: The Role of Geographical Information System in Enhancing Power Distribution and Management

Kanmani Kalaichelvan — 2026-04-27

Electrical utility networks, structured as centralized power systems with one-way flow, are undergoing significant transformation. Geographic Information Systems (GIS) have become pivotal in modernizing these networks, enabling spatial analysis, real-time monitoring and optimization of power distribution. Traditional electrical utility networks, reliant on manual operations, struggle with fault detection, load balancing, and integrating renewable energy sources. This study focuses on a comparative analysis of these traditional and modern networks, highlighting the structural evolution, technological advancements, and the critical role of GIS in creating a scalable, future-ready power infrastructure. To increase operational efficiency, modern systems make use of IoT sensors, smart grid technology, and predictive analytics. GIS supports dynamic decision-making by integrating real-time data with geographic context. The transition marks a paradigm shift toward decentralized, data-driven, and resilient energy networks.

Design and Development of a Self-Stability System for Real-Time Motorcycles

C. Dineshkumar — 2026-04-27

Modern motorcycles are now manufactured with more attention to how they look and how air moves around them to make them more attractive to buyers. However, this focus sometimes ignores how well the rider can stay balanced, especially when moving slowly. Keeping the bike steady in these situations can be tough. This study looks at how to improve motorcycle balance and make the bike more stable by using a gyroscope. The research includes testing how the motorcycle reacts when it starts to fall, and then designing and checking a gyroscope system based on those tests. The main goal of this paper is to check if adding stability using two gyroscopes along with a double flywheel can make motorcycles more balanced without making them less safe. This method helps both skilled riders and new riders who are learning to ride two-wheelers.

Design and Structural Evaluation of a Roll Cage for Baja All-Terrain Vehicles

DineshKumar C — 2026-04-27

The roll cage is a critical structural element of an all-terrain vehicle (ATV), providing essential rigidity and enhancing occupant safety during off-road operation. This study presents the design and structural evaluation of an ATV roll cage, focusing on CAD modeling, finite element preprocessing, and static structural analysis. The roll cage geometry was developed using Creo to satisfy key requirements such as strength, weight efficiency, and compatibility with the vehicle chassis. The completed model was then discretized using Hyper mesh, where appropriate meshing was applied to ensure reliable numerical results. Static structural analysis was carried out using the Opti Struct solver to evaluate stress distribution, deformation behavior, and potential failure regions under representative loading conditions, including gravitational effects and external impact loads encountered during off-road use. The results confirmed that the roll cage design meets safety and performance criteria, with factor of safety values maintained within acceptable limits, supporting its suitability for demanding off-road environments.

Design and Experimental Validation of a Sensor-Based Safety-Enhanced Braking System for Low-Cost Automotive Applications

Dineshkumar c — 2026-04-27

In recent times, automobile depends upon the active and passive safety systems to enhance the safety features due to increased rate of accidents. The causes of the accidents are increasing nowadays and therefore the fatality rate also increases. The automotive safety system had developed widely in costlier vehicles but unfortunately not in low end vehicles. The statistics describe that the safety systems in the low-cost vehicles are as par with high-cost vehicles and hence the fatality rate is also more than the high-cost vehicles. The causes of accidents were investigated by many researchers and they highlighted that the several road accidents were caused by inattention of drivers and in specific 77% fatality were mainly due to drivers’ fault during regular driving. In this work, research was carried out to reduce the fatality rate and damages in emergency condition. The proposed research is to make the vehicle to come to stopping position by activating the braking system at the time of emergency. Ultrasonic sensor senses the obstacles and object present ahead of the vehicle. A new hydraulic actuator is developed and is connected with the brake pedal and actuator, it is actuated when signal from the control unit, press the brake pedal to slow down the vehicle in critical situations. The experimental setup is used for the testing the proposed system and to evaluate the performance of control algorithm with sensor-based technology. The safety enhances braking system is implemented for the proposed system by adding the actuator on the brake pedal without any change in the existing brake system. The concept of this research is to reduce the severity of injury, fatality in emergency condition.

Hybrid Federated Neural–Observer Predictive Control for Robust Electric Vehicle Battery Management

V Rajesh Kumar — 2026-04-27

The rapid growth of electric vehicles (EVs) has intensified the demand for reliable battery management systems (BMS) capable of ensuring safety, longevity, and performance under diverse operating conditions. Conventional observers such as the Extended Kalman Filter (EKF) and data-driven neural networks have shown limitations in scalability, robustness to noise, and interpretability. This paper proposes a Hybrid Federated Neural–Observer Predictive Control (HFNOPC) framework for robust EV battery management. The framework integrates three innovations: (i) a robust super-twisting sliding observer for noise-resilient state estimation of state-of-charge (SOC), state-of-health (SOH), and temperature; (ii) a physics-guided neural residual network that enhances the baseline equivalent circuit and thermal models by capturing nonlinearities due to aging and hysteresis; and (iii) a federated learning strategy that enables distributed EVs to collaboratively train the neural residual component without sharing raw data, thus ensuring scalability and privacy. The enhanced state estimates are coupled with a model predictive control (MPC) scheme, which optimizes charge–discharge trajectories subject to safety and thermal constraints. Simulation studies demonstrate that the proposed HFNOPC reduces SOC estimation error by up to 32% compared with EKF and decreases control cost by 24% compared with conventional PID–based charging strategies. Furthermore, robustness tests under ±10% sensor noise and thermal stress confirm improved stability and accuracy. These results highlight the potential of the proposed framework as a next-generation BMS solution, offering interpretability, robustness, and fleet-wide scalability, thus paving the way for safer and more efficient EV deployment.

Techno-Economic Assessment of Hybrid Renewable Energy Systems for Sustainable Energy Supply

Koppola Vasavi — 2026-04-27

This study investigates the feasibility of a Hybrid Renewable Energy System (HRES) as a sustainable and cost-effective solution for rural electrification in regions of India with limited grid access. The research focuses on identifying the optimal hybrid configuration and conducting a pre-feasibility techno-economic analysis for supplying reliable electricity to an engineering institution located in Avalahalli, Bengaluru, Karnataka (560049). The proposed HRES integrates multiple renewable sources and was modelled and optimized using HOMER Pro software. Simulation results indicate that the most economically viable configuration achieves the lowest Net Present Cost (NPC) and Levelized Cost of Energy (COE), along with a 100% Renewable Energy Fraction (REF). The optimal design comprises 6170 kW of photovoltaic (PV) arrays, a 93 kW G10 wind turbine, 28,780 batteries (83.4 Ah each), and 761 kW power converters, yielding a COE of $0.388/kWh, a minimum NPC of $44.1 million, and only 4.16% unmet load. The system effectively meets the daily energy demands of 22,644 kWh (DC) and 2,508 kWh (AC) while achieving zero carbon emissions. Moreover, the HRES demonstrates strong economic performance with a payback period of 3.61 years, a return on investment of 34.8%, and an interest rate of 39.8%, confirming its viability for rural electrification.

Integrating Bottleneck Attention Module (BAM) into YOLOv8 for Automated Industrial Label Quality Inspection

Habib Ja’far Nuur — 2026-04-27

This study proposes an automated inspection solution to address the vulnerability of manual Quality Control (QC) systems to human error in detecting expired date labeling defects on plastic bottle products used for liquid packaging. The developed system employs the single-stage YOLOv8 architecture for object detection, which offers high inference speed, a crucial aspect for real-time applications. This study enhances model accuracy through the integration of a Bottleneck Attention Module (BAM) as an attention mechanism, strategically placed at the 9th layer of the network backbone. The selection of BAM is based on its capability to simultaneously capture channel dependencies and spatial relationships, which is essential for accurately recognizing subtle printing defect patterns in small-sized text. As a result, the enhanced model (YOLOv8-BAM) demonstrates a significant improvement in key performance metrics compared to the baseline YOLOv8 model, namely: an increase in Recall of 4.17%, Precision of 3.23%, and F1-score of 3.16%. These findings validate that YOLOv8-BAM is a more robust and reliable solution for automated industrial label quality inspection.

Experimental Study on Steering Wheel Pulse Measurement-Based Active Safety Systems for Collision Prevention

C DineshKumar — 2026-04-27

Active safety system plays a vital role in automobile to enhance the safety to prevent the accidents and fatalities. Therefore, accidents are increased day by day and rider safety is striving hard to improve in an automobile. Researchers reported that more than 20 percentage of accidents occurs due to fault of the driver during driving like inattention and health related issues are the predominate one in accidents. The above issues are reduced by using the preferred system and analyze the driver condition while driving. The system shows the health condition in dashboard and displays the heart rate and warns the driver whenever the pulse range is varied more than the normal range. The measuring device mounted in driving wheel and pulse is measured using throb measuring device. The driver positioning his hand on driving wheel and considered as clock position “10 and 12”. The system warns not only to the driver and it warns to the co passengers for indicates the driver is in health issues. The research is carried out especially to prevent the fatality, severe injury and damage to the vehicle in the event of collision.

Design and Optimization of Micro-Blade Passive Flow Control for Total Pressure Loss Reduction in an Axial Compressor Cascade

Oussama Hachelfi — 2026-04-27

The present paper investigates the aerodynamic influence of a micro-blade (MB) flow control in a low-speed axial compressor cascade, through a combined two-dimensional (2D) optimization and three-dimensional (3D) validation framework. The influence of the MB’s key geometric parameters, including chord length, camber angle, stagger angle, and maximum thickness, as well as spatial positioning, was systematically analyzed to identify the optimal configuration. Results reveal a size-wake trade-off; small elements fail to sustain meaningful momentum exchange, whereas large elements introduce excessive blockage and wake losses. An intermediate chord of 20% of the main blade, combined with a medium element maximum thickness (5% of the MB chord) and 50° camber, provides the most favorable compromise, achieving a 55.56% reduction in total pressure loss coefficient (TPL) under stall conditions. The optimized MB was subsequently evaluated in a 3D cascade configuration to examine its influence on corner separation and 3D flow structures across design and off-design inlet flow angles. The 3D results showed a clear reduction in the corner separation spanwise extent, especially at high incidences, though a TPL penalty is observed at the nominal design angle, where the baseline flow is already attached. Overall, the findings demonstrate that carefully optimized MB can effectively mitigate losses and enhance compressor operability when geometry and placement are properly designed.

Enhancing Biodiesel Yield from Canola Oil Through Response Surface Methodology: In-Depth Study of Reaction Parameters and Engine Efficiency

Sangeetha Krishnamoorthi — 2026-04-27

The response surface methodology (RSM) was employed to ascertain the most favorable operating conditions for the transesterification process of canola oil. These conditions encompassed the stoichiometric ethanol-to-oil ratio, reaction temperature, and reaction duration, all of which contributed to the maximization of biodiesel output. With an R2 value of 91.43%, the RSM model was statistically significant. Fourier transform infrared spectroscopy (FTIR) and gas chromatography (GC)-mass spectrometry confirmed that the majority of triglycerides were formed as methyl ester. When compared to pure petro-diesel and biodiesel blends, the performance and emissions of the biodiesel produced were found to be promising in a diesel engine

A Neutrosophic Generalized Gamma Approach to Modeling PRL, LH, and Testosterone in Beagles

Bevara Kondala Rao — 2026-04-28

The Classical Generalized Gamma Distribution (GGD) has been widely applied in survival analysis, hydrology, and
reliability engineering due to its flexibility in handling diverse data patterns. To incorporate uncertainty and
imprecision commonly encountered in practical scenarios, this study introduces an extended form of the distribution,
termed the Neutrosophic Generalized Gamma Distribution (NGGD). The NGGD is employed to investigate the
secretion behavior of prolactin (PRL), luteinizing hormone (LH), and testosterone in male dogs. Analysis reveals
that PRL exhibits a non-pulsatile, constitutive release pattern, which contrasts with the mixed pulsatile secretion
observed in humans. The NGGD-based probability density functions further enable more accurate differentiation
between control and drug-treated groups, aligning with established physiological evidence. These findings highlight
the effectiveness of the NGGD in capturing biological variability under uncertainty, with potential applications
extending to risk assessment, quality monitoring, and supply chain analysis.

QW-CNN: Secure and Robust Image Compression Using Quincunx Wavelet Decomposition, Lightweight CNN Modeling, and Selective Encryption over Noisy Channels

Haouari Benlabbes — 2026-04-28

With the rapid growth of digital imaging and multimedia applications, efficient image compression while ensuring data security has become a critical challenge. This paper proposes a novel framework, QW-CNN, which integrates Quincunx Wavelet Transform (QWT) for multiresolution image representation, a lightweight convolutional neural network (CNN) for adaptive coefficient prediction, and selective encryption to enhance security during transmission over noisy channels. The proposed method compresses images by predicting and thresholding wavelet coefficients using the CNN, followed by encrypting only high-magnitude coefficients to reduce computational overhead while maintaining confidentiality. Huffman-based entropy coding is applied to further reduce data size. The robustness of the framework is evaluated under additive white Gaussian noise (AWGN) channels. Experimental results on standard benchmark images (Lena, Pepper, Mandrill, Pirate, and Cameraman) demonstrate that QW-CNN achieves high compression efficiency, excellent visual quality, and strong security performance. The proposed method consistently delivers peak signal-to-noise ratio (PSNR) above 40 dB and structural similarity index (SSIM) above 0.97 for SNR levels above 40 dB, highlighting its effectiveness in secure and reliable image transmission..

A HieArarchical LSTM Framework for Capturing Long- and Short-Term Preferences in POI Recommendation

Sarala Patchala — 2026-04-28

Point-of-Interest (POI) recommendation is crucial for improving user experience in location-based social networks (LBSNs). With the growing number of users checking in at various places personalized recommendations are necessary to provide relevant suggestions. Existing methods use long short-term memory (LSTM) networks to model user preferences. However, these methods either consider long- and short-term preferences separately or merge them into a single model without effectively capturing the interactions. This research revisits the problem of long- and short-term preference learning by proposing a hierarchical LSTM (HiLSTM) framework. The framework aims to enhance next POI recommendations by learning representations at two levels: POI-level and semantic-level. Instead of treating these factors independently, HiLSTM integrates them through a structured hierarchical learning approach. One of the key challenges in POI recommendation is handling the sparsity of check-in data. Many users frequently visit new locations. It makes difficult to rely solely on past visits. The proposed model addresses this by introducing a semantic filter. It provides recommendations based on a user’s categorical preferences. By filtering out irrelevant POIs at an early stage, the recommendation process becomes more effective and computationally efficient. To capture long-term user preferences, HiLSTM employs an attention mechanism. Meanwhile, short-term preferences are derived from recent check-ins. It confirms that immediate user intent is not overlooked. The combination of these two components results in a more balanced and accurate recommendation system. These datasets contain check-in records from location-based social networks, enabling rigorous evaluation. The hierarchical structure and attention mechanism contribute to a significant improvement in recommendation precision. This work introduces a novel hierarchical LSTM framework for next POI recommendation.

Deep Multimodal CNN Fusion Scheme for Accurate Stress Identification

Sarala Patchala — 2026-04-28

Stress is a major issue in today’s life. It harms health and lower work output. People sometimes do not notice when under stress. That is why early stress detection is important. This paper uses two types of body signals: ECG (Electrocardiogram) and EDA (Electrodermal Activity). Both are physiological signals and help measure stress levels. A deep learning model named CNN (Convolutional Neural Network) is used. CNN has many layers. Each layer captures different kinds of features—low-level, mid-level and high-level. These features are useful in identifying stress. Instead of using features from only one level, this paper combines all three levels. This process is named as hierarchical feature fusion. It helps in creating a strong and rich representation of the signals. The features from ECG and EDA are first extracted at different CNN layers. Then, a module termed MMTM (Multimodal Transfer Module) is used. This module helps combine features from both signals. It improves the way the model learns from the data. The model is tested using both raw data and features from selected frequency bands. Results show that using features from all three CNN levels gives better performance. The proposed model performs better than existing models when using frequency band features. This shows that combining low, mid and high-level CNN features with multimodal fusion is helpful. It improves the accuracy and generalization of stress detection. This method works better across different datasets and different people. The proposed system is a useful tool in real-world stress detection systems.

A3D: Joint Search for Robust DNNs and Adversarial Attacks via AutoML

Sarala Patchala — 2026-04-28

Deep neural networks (DNNs) are widely used today. These methods work well in many tasks but show weakness against adversarial attacks. Attacks use small changes in input images to fool DNNs. Because of this, many platforms exist to test the strength or weakness of a DNN. However, most current platforms face two problems. First, improving the structure of neural networks is not possible. Second, creating stronger attacks is not supported. As a result, these platforms do not help make DNNs more secure or smarter. To address this, a new system was proposed named A3D. It stands for Auto-Adversarial Attack and Defense. This platform does two main jobs. It finds strong DNN structures. It finds smart attack methods. This is done by using automatic search techniques. A3D uses different search tools. These tools help to build better DNNs that are harder to fool. A3D finds better ways to fool the DNNs. The system works in both directions. Stronger the attacks, better the defense models it creates. The stronger the models, the smarter the attacks it designs. A3D is tested on popular datasets. These include CIFAR10, CIFAR100 and ImageNet. The results show that A3D works well. It creates DNNs that are more secure. It creates attacks that are more effective. A3D is a powerful tool. It helps researchers test and improve DNNs in a smart and automatic way. It solves the limits of older systems. And it brings better security and performance to deep learning.

Incorporation of Native Fruits from Brazilian Biomes into Chocolate: A Literature Review

Elisangela Santos Reis de Oliveira — 2026-04-28

Chocolate, a food product widely appreciated for its flavor and versatility, has been undergoing innovation. The incorporation of native fruits from Brazilian biomes into its formulation has become a promising alternative to combine sensory pleasure with nutritional benefits. Thus, this study aimed to review recent scientific literature on the use of these fruits in chocolate production, focusing on their nutritional, functional, and technological impacts. The research was conducted through a systematic review following the PRISMA 2020 protocol, including articles published between 2020 and 2025 in the ScienceDirect database that presented information on fruit-based formulations and nutritional composition data. Fruits such as cupuaçu, palm, elderberry, and sacha inchi demonstrated potential to enrich chocolates due to their bioactive and nutritional compositions, in addition to contributing to sustainable practices through the use of agricultural by-products. The analysis revealed significant gains without interfering with sensory acceptability, as well as opportunities for regional innovation and valorization of Brazilian biodiversity. Therefore, it was evident that the integration of native fruits into chocolate strengthens the development of healthier and more natural foods and stimulates the circular economy and biodiversity.

Efficient Cyber Threat Detection in Smart Home IoT Networks Using Machine Learning and Explainable AI

Pritimayee Satapathy — 2026-04-28

Smart homes are increasingly common, making it more important than ever to safeguard them against cyber threats. In this work, we develop an improved anomaly-based Intrusion Detection System (IDS) for smart home IoT networks by combining machine learning (ML) techniques with advanced Feature Selection (FS) and Explainable Artificial Intelligence (XAI). We propose an FS approach that finds the intersection of important features identified by Recursive Feature Elimination with Cross-Validation (RFECV) using two different base learners: a Decision Tree (DT) and a Light Gradient Boosting Machine (LGBM). This yields a compact subset of 12 network traffic features. Using this significantly reduced feature set, our lightweight LGBM model achieves 99.86% detection accuracy, an F1-score of 99.93%, and a recall of 99.96% on the IoTID20 dataset. The false positive rate is greatly reduced, and computational cost is also minimized. To enhance model transparency, we integrate XAI tools, Local Interpretable Model-Agnostic Explanations (LIME) provide clear instance-level explanations, and Shapley Additive Explanations (SHAP), which provide global explanations of the classifier’s decisions. Experimental results demonstrate that combining FS with XAI can improve both the efficacy and the usability of IoT IDS. Our hybrid model outperforms several recent deep learning approaches in precision and recall, while being far more interpretable. Overall, the proposed method yields high detection rates with improved transparency, making it well-suited for resource-constrained smart home environments.

Input Current Ripple Reduction of Modified Interleaved Quadratic Boost Converter for Photovoltaic Applications

Abdulkarim Laith Abdulkarim — 2026-04-28

In photovoltaic (PV) systems, achieving high voltage gain and keeping low input current ripple is essential for reliable operation and ensuring that the PV source consistently provides its maximum power. Low input current ripple helps the PV module operate close to its maximum power point (MPP). Conventional DC converters including quadratic boost converter frequently exhibit significant input current ripple for constrained voltage gain. To improve performance under varying load and irradiance conditions, a modified interleaved quadratic boost converter (MIQBC) is suggested. The proposed converter is able to reduce the input current ripple by employing a two-phase interleaved structure that operates with a 180° phase shift, incorporating design modification. Using an ESP32 control unit and a perturb and observe (P&O) algorithm to control the maximum power point during operation. The converter was simulated with LTspice, designed, and tested practically to assess the performance under real solar irradiation. The theoretical analysis indicated that the voltage gain reached approximately four times the input voltage, and input current ripple reduced to less than 2% while the efficiency was about 96% at full load. The experimental tests indicated that the converter demonstrated reliable performance and effective power tracking by eliminating the input current ripple. The topology provides high reduction in input current ripple and a real enhancement in voltage gain compared to conventional topologies. The results ensure efficient use of the PV panel's maximum power capability and proves that the converter is suitable for photovoltaic applications.

Performance Valuation of Electrical Characteristics for MOSFET Device Using TCAD software

Zahra R. Mahmood — 2026-04-28

MOSFET transistor is one of the most important and efficient components in modern electronic circuits due to its high efficiency and low power consumption. The performance of N-MOS and P-MOS transistor has been studied and analyzed by evaluating key electrical characteristics such as threshold voltage, drain current, (gate and drain) voltages, LDD (Lightly Doped Drain functions), PLDD (P-type Lightly Doped Drain) and DIBL (Drain-Induced Barrier Lowering). Simulation using Silvaco TCAD Program tools was employed to extract the device structure and current-voltage characteristics for both transistors. For N-MOS, it was observed that drain current (Id) grew linearly with expanding gate voltage (Vg) for different values of drain voltages (Vd), while for P-MOS the values were opposite. The Threshold Voltage for N-MOS transistor (Vt= 0.5V) while for P-MOS (Vt= -0.5V), The maximum current obtained is Id= 0.515 mA at Vd= 3.35 V for N-MOS and the highest current obtained is Id= -0.335 mA at Vd= -3.35 V for P-MOS, so N-MOS was faster than P-MOS, while P-MOS is considered complementary to N-MOS. The two transistors were combined for more low power dissipation and high performance.

Investigating the potential of Quantum Inspired Machine Learning approaches for mental health detection of young adults

Mrutyunjaya Panda — 2026-04-28

Medical Health Informatics has envisaged quantum computing as flurry of promising solutions to deal with mental health related issues of young adults in recent times. Even though traditional machine algorithms try to find an amicable solution to healthcare service providers, still the efficient detection and classification of the highly complex patterns that emerges from the healthcare datasets needs further investigations. Quantum machine learning, inspired from the principle from Quantum computing is found to be a transformative alternative in redefining the healthcare applications. In this article, Quantum inspired Support Vector Machines (QISVMs) and Quantum inspired Neural Networks (QINN) as two efficient quantum machine learning approaches are proposed by combining the traditional yet very popular and powerful machine learning algorithms such as Support Vector Machine (SVM) and Artificial Neural Network (ANN) with the power of quantum computing principles. The proposed Quantum inspired machine learning approaches are evaluated with several figure of merit and then, compared with the performance of the traditional algorithms counterpart for analyzing the most complex nature of the healthcare data with proper identification of minute wise data patterns and biomarkers in order to detect the early-stage healthcare issues pertained to a patient. Experimental study shows the efficacy of these quantum machine learning methods in comparison to the existing literature with threats to validity, some hidden challenges, and ethical issues associated with these technological advancements.

Organizational Competence in Conditions of Sustainability – Interaction Between Structural Approaches, Entrepreneurial Vision, and Personal Potential

Inna Ippolitova — 2026-04-28

In the current context of digital transformation, global challenges and the need to achieve sustainable development goals, organizations are increasingly in need of new management models based on a competency-based approach. The relevance of the study is due to the need to integrate individual, entrepreneurial and organizational competencies to ensure the sustainability and adaptability of organizations. The purpose of the study is to find out the relationship between these levels of competencies in the context of forming a competent organization as a tool for ensuring sustainable development. The methodological basis of the study is a content analysis of publications of 2021–2025 covering the transformation of competency-based approaches in the field of management. As a result of systematizing sources and building an analytical model, five key thematic areas of integrated competencies development were identified: strategic management, entrepreneurial education, digital transformation, corporate training, and sustainable development. The study revealed the increasing role of the digital environment in the formation of professional competence, the growing importance of interpersonal and social skills, and the need to unify management models in the global context. The practical significance of the results lies in the possibility of their application to the formation of internal policies of organizations, the development of staff training programs and the improvement of management strategies. The results obtained can serve as a basis for the development of indicators for assessing competencies in the practice of personnel management. The author suggests directions for further research, including empirical testing of the effectiveness of models of the integrated competency approach in specific industries and regions.

Exploring User Experience and Preferences in Digital and Traditional Mental Health Screening at Cagayan State University Carig Campus

Arlen Bergante Calimag — 2026-04-28

Mental health screening in higher education is crucial for early detection of psychological issues among students. With digital technologies increasingly used in student support, this study evaluated the effectiveness and user experience of mobile-based versus traditional paper-and-pencil mental health screenings at Cagayan State University–Carig Campus. Thirty-three students completed assessments for depression, anxiety, and stress using both methods in a quasi-experimental design. Results revealed a significant difference in depression scores between methods, indicating that the screening format influences reporting of depressive symptoms. No significant differences were found for anxiety and stress, suggesting both methods are comparable for these conditions. Students preferred the mobile-based screening, highlighting greater comfort, privacy, ease of use, and accessibility. Overall satisfaction favored the digital method. The study concludes that mobile-based screening offers a user-friendly and efficient alternative that is psychometrically similar to traditional approaches. It recommends integrating mobile screening tools into university mental health services to enhance student support and accessibility.

Sustainable Self-Compacting Concrete: Investigating the Size Effects of Coal Bottom Ash as a Lightweight Aggregate

Boulahya Ibtissam — 2026-04-28

The mining and processing of coal generate significant waste, often disposed of in landfills, with coal slag heaps in southwestern Algeria forming artificial mountainous structures. This study investigates the potential of reusing coal bottom ash (CBA) as a lightweight aggregate in self-compacting concrete (SCC), fully replacing natural fine and coarse aggregates. The research focuses on the effect of CBA aggregate size on the fresh and hardened properties of SCC. Experimental results show that replacing natural coarse aggregate (NCA) with coarse CBA (CCBA) enhanced self-compactability, while fine CBA (FCBA) reduced workability. The use of CBA resulted in lower dry densities (1920.6–2134 kg/m³), qualifying the mixtures as lightweight SCC according to EN 206-1 standards. Although compressive and flexural strength decreased and porosity increased due to the lightweight and porous nature of CBA, strength values met ACI standards for structural applications. Microstructural examination indicated a rise in porosity within the concrete matrix; however, a limited formation of supplementary calcium-silicate-hydrate (C-S-H) gel was observed, which contributed to partially offsetting the reduction in mechanical strength. This study demonstrates the viability of CBA, particularly as a coarse aggregate, for producing lightweight SCC, offering a sustainable solution for recycling industrial waste and developing eco-friendly construction materials with acceptable structural properties.

Agentic AI in Cloud-Based Credit Card Fraud Detection: Towards Autonomous Risk Mitigation

Sushil Prabhu Prabhakaran — 2026-04-28

Credit card fraud (CCF) still remains an ongoing concern for financial institutions because to the huge disparity between genuine and fraudulent transactions, as well as the ever-changing behavior of fraudsters. This paper proposes a cloud-based Agentic Artificial Intelligence system for real-time credit card fraud detection that utilizes autonomous multi-agent collaboration and deep temporal modeling. The system makes use of the publicly accessible CCF Detection dataset, commencing with secure cloud-level data ingestion, preprocessing, and normalization. An agentic transaction analysis layer made up of qualified agents accomplishes data validation, behavioral pattern analysis, and transaction history verification. To successfully capture both local spatial aspects and long-term temporal dependencies in transactional behavior, deep temporal fraud modeling employs a hybrid full-dimensional dynamic convolutional network mixed with shuffle attention and LSTM. Finally, an automated risk assessment and decision module generates fraud scores and initiates relevant mitigation steps such as alarms or transaction blocks. Experimental results show that the proposed framework outperforms baseline GRU-based models in detection performance, demonstrating its effectiveness, scalability, and appropriateness for real-time intelligent fraud prevention in cloud environments.

A Multimodal Graph Contrastive Learning for Human Activity Recognition Using Deep Learning Technique

Velantina V — 2026-04-28

The deep learning technique for Human Activity Recognition (HAR) systems has achieved remarkable improvements in recent years in recognition of complex action classes and real-world contexts.This research advances our unified deep learning framework, called the Hybrid Dense Temporal Transformer Network (HDTTN) to capture spatial, temporal, and semantic information to capture better human activity detection. We introduce Dense Net to improve spatial feature extraction from visual inputs, Temporal Convolutional Networks (TCNs) for learning short-term motion patterns, and Transformer encoders for learning long-range temporal dependencies that are crucial for processing complex and subtle activities. Thestudy employs early multimodal feature fusion strategy for further enhance representational coherence which makes it easier to incorporate heterogeneous cues at the feature level and to learn dynamic multimodal representations. Moreover, a hybrid optimization approach is integrated for parameter fine-tuning for efficiency, reduce overfitting, and boost model robustness. The proposed HDTTN framework is shown to be effective on the large scale and difficult Kinetics dataset containing a wide range of unconstrained human activities. Experimental results show that our proposed model is 93% accurate, compared to several existing state-of-the-art baseline approaches. Moreover, qualitative and quantitative analyses validate HDTTN's ability to identify intricate and nuanced activities across a multitude of environments.

Industry 4.0 Mobile Workflow Quality Assessment Using SmartPLS: Data Readiness, AI Accessibility, and Satisfaction Drivers

Mesith Chaimanee — 2026-04-28

Low-code platforms such as Google AppSheet enable industrial workflow digitization, where responsiveness and AI assistance shape operational reliability. Sustained adoption depends on measurable task performance and accessible AI features across heterogeneous mobile and web clients. Prior low-code acceptance research relied on self-reports and did not unify objective performance, data readiness, and AI accessibility into a single model. This work aims to quantify the data-to-performance-to-continuance mechanism for AppSheet deployments using PLS-SEM. Task scripts captured open, sync, and submit times, plus error events, and reduced them into a five-level objective performance index. Survey responses from 260 users were modeled in SmartPLS with 5,000 bootstrap resamples. Reliability and validity are supported (Cronbach’s alpha ranged 0.885–0.916; composite reliability ranged 0.925–0.947; AVE ranged 0.755–0.856). The model explains satisfaction and continued use intention (R² = 0.299; R² = 0.264). Objective performance drives perceived performance (β = 0.488), and AI accessibility drives ease of use (β = 0.349), while satisfaction, trust, and usefulness predict intention (β = 0.246, 0.232, 0.199). These results support deployment controls that couple data readiness validation, performance monitoring, and accessible AI interaction design for Industry 4.0 workflows. Future technology extends this framework with longitudinal telemetry, multi-indicator performance constructs, and role-stratified multi-group estimation.

Switching of Medium Voltage Capacitor Banks and the Impact of Overvoltages on the Components of an Electrical System

Aldi Mucka — 2026-04-28

The study aimed to determine the regularities of switching overvoltages during the operation of medium-voltage capacitor banks and to evaluate the effectiveness of methods for their reduction. The methodology was based on a combination of mathematical modelling in specialised environments, laboratory testing of condenser plant models and statistical analysis of operational data. The results showed that the amplitude of overvoltages in capacitor banks increased from 1.6-2.0 to 2.5-3.0 relative units depending on the power, and the duration of transients exceeded 2.8 milliseconds at large capacities, which created a risk of insulation damage. The study determined that more than 90% of capacitor failures were of an insulating nature, with daily switching reducing service life by a third, and overvoltage over 2.5 relative units halving service life. The study showed that in the secondary circuits of transformers during resonant interaction, the amplitude of overvoltages reached 4.0-4.5 relative units, which led to the failure of electronic devices and premature tripping of protections. The effectiveness of the technical equipment confirmed that synchronous circuit breakers reduced the level of overvoltage to 1.2 relative units, while pre-trigger resistors and overvoltage limiters reduced it by 40-50%, but required frequent maintenance. The conclusions of the study were to substantiate the need for the integrated use of protection methods that ensure both technical reliability and economic feasibility of operation, and the results obtained can be used by power engineers and design organisations to optimise operating modes and increase equipment durability.

A Novel and Robust Fractional-Order Proportional-Integral-Derivative Acceleration Controller for Electric Furnace Temperature Regulation

A. Idir — 2026-04-28

Electric furnaces are widely used in industrial thermal treatments because they combine high energy efficiency with precise temperature regulation. However, the nonlinear system dynamics, inherent time delays, and large thermal inertia often limit the achievable level of control performance in practice. A novel fractional-order Proportional–Integral–Derivative–Acceleration (PIαDA) controller is presented in this work, with its parameters tuned via a Modified Flower Pollination Algorithm (MFPA) to address the aforementioned limitations. The tuning procedure is formulated to explicitly improve transient response characteristics while enhancing robustness, thereby supporting reliable operation under varying operating conditions. Controller performance is assessed through comprehensive simulation studies that include reference tracking, step changes in the temperature setpoint, external disturbance rejection, and tracking under noisy reference signals. Compared with MFPA-tuned benchmark controllers (i.e., the MFPA-optimized PIDA and the conventional PID), the proposed MFPA-optimized PIαDA controller achieves higher tracking accuracy, shorter rise and settling times, reduced overshoot, and improved robustness against disturbances and measurement noise. Overall, these findings indicate that combining fractional-order control structures with advanced metaheuristic optimization can substantially enhance temperature regulation performance in industrial electric furnace applications.

Intelligent Technique for Electromechanical System Data by Using Model of NARX Neural Network

Mohammed Al-Abbasi — 2026-04-28

In this study, the electromechanical systems have been modeled using the neural network technique. One of the most important parts of engineering is system modeling, which is essential for both system analysis and control process implementation, especially when a sophisticated and accurate modeling approach like NARX neural networks is employed. In order to gather the vibration produced by the electromechanical system devices after a sinusoidal voltage between zero and forty volts was applied, a practical test was carried out in this study. One side of the actuator was fixed, while the other side was free. After double integration for the output signal, the displacement data was gathered using an accelerometer sensor. Following variable modification, the transfer function, which shows the system's dynamic behavior based on input and output data is produced. NARX neural network approach was utilized for modeling using the MATLAB application. The outcomes demonstrated how well the NARX neural network approach represented the system and produced an appropriate transfer function in the time and frequency domains. Prior to and throughout the modeling process, the system's behavior remained unchanged, and the minimal mean square error was 0.00001391.

Issues and Prospects of Integrating Relational Databases with Cloud Technologies

Vladyslav Kozub — 2026-04-28

Relevance. The integration of relational databases into cloud computing environments is characterized by the rapid growth of data volumes, the need to ensure flexibility, scalability and high performance of information management systems. Given that cloud technologies allow organizations to quickly adapt infrastructure to changing loads, integrate transactional and analytical processes, increase the efficiency of business processes and minimize operating costs; integration of RDBMS with cloud services is becoming a key factor in ensuring sustainability, security and competitiveness of organizations.

Objective. The purpose of this article is to systematically analyze the current challenges of integrating relational databases into cloud computing environments and to identify promising areas for their development, considering technological and organizational aspects.

Methods. The study applies to a set of methods that provides a systematic and reasonable assessment of the integration of relational databases into cloud environments. The forecasting method was used to outline the market prospects and growth dynamics of corporate data in the cloud; methods of synthesis, systematization and generalization were used to classify architectural models, analyze technical, economic and security aspects and identify key development trends; and a comparative analysis of cases of leading cloud RDBMS providers allowed to assess the effectiveness of real implementation; which provides a holistic view of the problem and forms a methodological basis for a reasonable choice of integration models.

Results. According to the report, the amount of enterprise data being stored in the cloud is increasing in tandem with the steady growth of the global cloud computing market. This demonstrates the importance of cloud solutions for strategic planning and the speed at which businesses are digitizing. Cloud-native, hybrid, multi-cloud, and microservice models that offer cost-effectiveness, technical agility, and security are all options for relational database integration into the cloud. AI-driven management, serverless architectures, and containerization all boost productivity by enabling real-time data flow optimization.

Conclusions. It is suggested that a successful integration approach for RDBs into the cloud can be established based on adaptability (flexibility), autonomy, and interoperability, ensuring strategic flexibility of organizations. Leveraging data fabric, data mesh, and autonomous AI-driven solutions, classic management logic is transformed from one of reactive administration to one of proactive load prediction. This multi-dimensional strategy drives digital transformation efficiency and forms the competitive capabilities of today’s enterprises.

The Role of Artificial Intelligence in Quality Control Automation: Impact of AI Technologies on Traditional Defect Detection Methods and their Integration into Production Processes

Kanan Mikayilov — 2026-04-28

The study aimed to assess the effectiveness of artificial intelligence (AI) technologies in automating quality control in production. The study analysed the application of AI technologies to automate quality control in production processes. The study analysed modern approaches to the implementation of computer vision, deep learning algorithms and predictive analytics to improve the accuracy of defect detection, reduce the influence of the human factor and ensure continuous product monitoring. The study determined that the use of computer vision, deep learning and predictive analytics systems helps to improve accuracy, optimise costs and reduce production defects. The study determined that the use of computer vision, deep learning and predictive analytics has the potential to improve the accuracy of defect detection, reduce the impact of the human factor and optimise the cost of product quality control. The analysed examples of the integration of smart technologies in various industries demonstrate the effectiveness of such solutions, provided they are properly adapted to the production environment. The study established that traditional quality control methods have limitations and need to be modernised by supplementing them with digital solutions. Modelling has demonstrated that the introduction of AI technologies can help to increase the efficiency of production processes, reduce reject rates and improve equipment maintenance. Comparative analysis demonstrated a potential reduction in product inspection time and the number of undetected defects using computer vision systems and deep learning algorithms. The possibilities of predictive analytics, which, according to the results of the literature analysis, can ensure timely maintenance of equipment and prevent its failures, were analysed. The study concluded that the combination of traditional control methods with intelligent technologies reduces production losses and can increase the overall productivity of enterprises if properly adapted.

Digital Twin for Integration of Control and Diagnostics of Electromechanical Systems Under Uncertainty

Dmytro Koshkin — 2026-04-28

The study aimed to create a digital twin for the integration of control and diagnostics of electromechanical systems under conditions of uncertainty, with minimal reliance on physical sensors. The research was conducted at Mykolaiv National Agrarian University. Physically based models were developed for thermal processes in windings, assessment of losses in magnetic conductors, and wear indicators for components, virtual sensors, signal filtering algorithms and degradation prediction were implemented, and verification was conducted on test benches and in computer modelling. Quantitative results were obtained, which constitute the main contribution of the work: the accuracy of reproducing hidden parameters was 93.6-97%, the relative error of reproducing losses in the transformer was 3%, the relative error of thermal estimates was 3.5-6.8%, the correlation with reference measurements reached 0.99; the reduction in the dispersion of noisy signals was 33-41%, the signal-to-noise ratio increased by 4.2-6.7 decibels, and the root mean square error decreased by 35-44% with an additional delay of no more than 0.04 seconds. The forecast of the time to failure of the hydraulic unit provided 92% correct estimates within a tolerance of ±10% for a 48-hour horizon; in the traction electric drive, a drop in efficiency (efficiency) by 6.7 percentage points under conditions of magnetic saturation was confirmed; in ship drives, peak torsional loads were reduced by 11%; in biogas plants, the energy balance error was 5.4%; in irrigation systems, energy consumption was reduced by 9%; in robotics, the accuracy of deviation detection was increased by 14%. The models can be used in ship drives, biogas plants, robotic lines, irrigation pumping stations, transformer substations, hydraulic drives and traction electric drives, reducing downtime and energy consumption without changing the existing control infrastructure.

Optimization of EV Charging Systems Using Hybrid GWO-PSO MPPT for PV Source

Venkatesh Reddy Birudala — 2026-04-28

Electricity is produced from variety of fossil and renewable sources and it is crucial that EVs have to be powered from renewables. With Electric Vehicle penetration, the combination of PV with EV drastically reduce our dependence on fossil fuel based electricity plants. The day to day power demand needed and anyone with a solar power system is likely to install a solar power charging station in their home in upcoming years. Solar power is variable as irradiance is changing during the day so to accommodate the changing irradiance MPPT algorithm has to switch or reach the maximum power point quickly. A GWO-PSO based MPPT is proposed in this paper and compared with P&O, PSO and GWO MPPTs and the hybrid GWO-PSO MPPT is used to charge the Electric Vehicle.

Next-Level Dynamic Queries Optimization: Smarter Joins, Faster Views

Salma Hanane — 2026-04-28

In modern database systems, query workload optimization through materialized views is crucial for achieving high performance. This paper introduces a novel intelligent framework that identifies frequent subexpressions from SQL workloads, selects candidate materialized views, and predicts their potential benefits using a Deep Neural Network (DNN) model.Unlike existing static heuristic methods, our approach adopts a dynamically learning mechanism with proper determination of properties required for effective views in line with distributed workloads and predicate normalization. By excluding an extra step needed in choosing views or query rewriting, query execution is reduced with the proposed framework, as illustrated in existing experimental results of proposed methods in comparison with existing ones.

Intelligent system for automated design of energy-efficient facilities in the agro-industrial complex with built-in diagnostics of electrical machines

Volodymyr Martynenko — 2026-04-28

The aim of the study was to substantiate the effectiveness of an intelligent system for automated design of energy-efficient facilities in the agro-industrial complex (AIC) with integrated diagnostics of electric drives. The methodology included experimental testing of asynchronous electric motors (1.5-7.5 kW), numerical modelling using finite element and finite difference methods, and optimisation based on a genetic algorithm and particle swarm. Data analysis was performed using Student's t-test, analysis of variance, and principal component analysis. The results showed that in faulty electric motors, the current increased by 20%, the power factor decreased from 0.87 to 0.74, the vibration level exceeded that of serviceable samples by three times, and the temperature of the windings increased from 65 to 92 degrees Celsius. Numerical modelling showed a 9% reduction in energy consumption, a 17-degree Celsius decrease in temperature, a 16% increase in power factor, a 47% reduction in vibrations, and an increase in efficiency from 85 to 92%. Statistical analysis confirmed the reliability of the results, and the “serviceable/damaged” classification model achieved high accuracy with an area under the curve of 0.94. The scientific novelty lies in the first-ever combination of structural optimisation of electric drive parameters with built-in technical condition analysis in a single CAD/FEM environment. The developed system provides early detection of deviations in motor operation, reduction of energy consumption and improvement of reliability without loss of productivity. The practical outcome is the creation of an integrated tool that increases the energy efficiency of AIC production lines and can be implemented in energy audit and maintenance processes.

Integration of control and measurement devices into CAD models of industrial power systems: Metrological aspects of accuracy and impact on energy efficiency

Larisa Vakhonina — 2026-04-28

The research relevance is determined by the growing need of industrial enterprises to improve modelling accuracy and energy efficiency, which directly depends on the integration of control and measuring instruments into Computer-Aided Design (CAD) models of industrial power systems. The study aimed to justify the possibility of using CAD systems for the integration of control and measuring instruments, incorporating the metrological aspects of accuracy and their impact on the reliability of digital models and the efficiency of energy consumption management. The research methodology included a classification analysis of control and measuring instruments according to their functional and metrological characteristics, the application of a formula for the relative error of quantitative assessment of instrument accuracy, as well as the modelling of various scenarios for data integration in CAD environments (AutoCAD, SolidWorks, Engineering PLAN). The classification of control and measuring instruments into high-precision (δ≤1%), medium-precision (δ=1-5%) and low-precision (δ>5%) was investigated. The study determined that the use of low-accuracy instruments led to a 12-15% reduction in the reliability of CAD models and caused additional energy losses of up to 8% in production processes. In contrast, high-precision instruments ensured deviations of no more than 0.8%, increased the accuracy of energy consumption forecasting by 10-12% and contributed to a reduction in operating costs by 6-7%. The integration of control and measuring instruments into digital twins of industrial facilities was analysed, which optimised equipment utilisation, reduced energy losses by 9% and increased the equipment utilisation rate from 82% to 89%. The study concluded that the correct selection and classification of control and measuring instruments is critical for the stability of digital models and their practical value. The study confirmed that the combination of metrological reliability of instruments and digital design technologies creates conditions for effective energy consumption management. The practical significance of the results is determined by the possibility of their application for optimising the design, modernisation and management of industrial power systems, incorporating quantitative indicators of accuracy and energy efficiency.

Intelligent Dynamic Switching Between IRS and Decode-and-Forward Relay: A Channel-Aware Optimization Approach for 6G Networks

Murtadha Ali Nsaif Shukur — 2026-04-28

This paper proposes a dynamic switching framework between Intelligent Reflecting Surface (IRS) and Decode-and-Forward (DF) relay transmission modes for next-generation wireless communication systems. Unlike existing studies that rely on static technology selection, the proposed approach enables channel-aware mode adaptation based on instantaneous channel conditions and energy efficiency considerations. The framework supports four operational modes: IRS-only, DF-only, hybrid IRS–DF, and direct SISO transmission. An efficient mode selection algorithm is developed to maximize achievable rate while accounting for power consumption and switching overhead. Simulation results demonstrate up to 20.2% improvement in spectral efficiency and 10.1% enhancement in energy efficiency compared to conventional static schemes, under realistic channel conditions. The proposed framework exhibits low computational complexity, making it suitable for practical 6G network deployment.

Integration of Flow Sensors into Cyber-Physical Systems for Intelligent Water Resource Management in Residential Environments

Roberto Higino Pereira da Silva — 2026-04-28

The growing scarcity of natural resources necessitates the development of more efficient technologies for water management in residential settings. This study introduces a solution based on Cyber-Physical Systems (CPS) for intelligent water monitoring, rooted in the RAMI 4.0 architectural model and the Asset Administration Shell (AAS) concept. The primary objective was to ensure interoperability and standardized communication between the physical and digital environments. Utilizing an experimental and quantitative methodology, a functional prototype was developed using the YF-S201B flow sensor integrated with an ESP8266 microcontroller, employing MQTT and OPC UA communication protocols. The AAS was implemented to serve as the digital representation of the asset, facilitating centralized visualization and data management. The results obtained validate the effectiveness of the proposed architecture, demonstrating accuracy in real-time monitoring of flow rates and accumulated volumes. It is concluded that applying Industry 4.0 technologies in residential contexts enables early leak detection and promotes sustainability, providing a robust and scalable solution for modern home automation.