Optimization of Power System Reliability in Wind Farms Using Hybrid Heuristic Artificial Neural Networks and Universal Moment Generating Functions

Abstract

This paper presents an innovative approach to optimising the reliability of electrical systems associated with wind farms by integrating advanced methods, including the Universal Moment Generating Function (UMGF), artificial neural networks (ANN), and the Harmony Search (HS) algorithm. The study focuses on modelling multi-state systems to address the complexity of wind energy production, which involves stochastic variability and degradation phenomena. The proposed methodology combines reliability assessment techniques with cost constraints to determine the most economical system configuration that meets specified reliability levels. Through simulations involving various subsystems (wind turbines, transformers, transmission lines), the authors demonstrate that their hybrid approach effectively balances investment costs and system reliability. The results indicate that maximizing reliability incurs higher costs, with the optimal compromise found at a reliability level around 97%, corresponding to a minimal investment of approximately 30,400 MDA (Million Dinars Algerian). Overall, this integrated optimization framework offers a practical, flexible solution for designing cost-effective, reliable wind energy systems, contributing valuable insights to renewable energy planning and reliability engineering.