Application of Parallel So-KELM Optimization Algorithm in Wind Power Generation Forecasting

Wind power generation is a critical component of smart grid systems. However, environmental factors introduce instability, leading to errors in wind power forecasting. This study addresses the challenge of improving prediction accuracy by employing an optimized Kernel Extreme Learning Machine (KELM) model, enhanced with the snake optimization (SO) algorithm. The proposed approach integrates reverse learning, somersault foraging, and a hybrid strategy to prevent local optima entrapment and enhance convergence precision. Through benchmark testing, the optimized Parallel SO-KELM algorithm demonstrates superior performance in wind power parameter prediction across various simulated weather conditions. Comparative analysis highlights the accuracy improvements achieved, with error rates significantly reduced, confirming the efficacy of the proposed methodology.