Salp Swarm Optimized ARBS-Net-Enabled Framework for Fault Diagnosis in Electric Vehicles

Resumen

Electric Vehicles (EVs) represents an excellent possibility towards developing a sustainable transport sector with low carbon emissions, low noise levels, high energy-efficient and integrate with electrical grids flexibly. Thus, the process of EV begins with data collection which involves different types of data such as voltage, current, vehicle speed, temperature, vibration and humidity to create a dataset of critical parameters. Pre-processing involves normalizing the dataset to create a consistent value for every element of data and to increase the value of the informations set. Exploratory data analysis (EDA) then lead to explore the distribution of different features in the dataset and detect any underlying or hidden patterns among the data. In feature engineering, the features are determined as critical for the performance of EV extracted from the data and used to identify potential failure modes present in EV. The Salp Swarm Optimized Attentive Radial Basis Spiking Network (ARBS-Net) is developed using a combination of benefits associated with multiple feature types, including sparse convolution, attention mechanisms and evolutionary optimization, to establish high-performance prediction and classification of fault conditions present in EV systems. This model is trained in Python and attains the maximum accuracy, precision, recall and F1-Score metrics of 0.99, thus demonstrating the proposed classification's capability to effectively predict and classify the occurrence of various fault conditions within EV systems. This intelligent diagnostic framework serves as a complete solution to implement proactive maintenance and increase EV operational reliability.