Abstract

This paper introduces Dual-Elite Social Group Optimization (DE-SGO), A novel version of the standard SGO algorithm for enhanced Maximum Power Point Tracking in photovoltaic systems. Under partial shading conditions, conventional MPPT algorithms often converge to local maxima rather than the Global Maximum Power Point (GMPP) due to multi-peak power-voltage characteristics, resulting in significant energy harvest losses. The proposed DE-SGO addresses this limitation by implementing structured knowledge transfer from both the best and second-best solutions simultaneously, creating a hierarchical learning structure that enhances exploration and exploitation capabilities. The novel proposed algorithm has been implemented in MATLAB Simulink using SPM0409 photovoltaic specifications and evaluated against standard SGO, Cuckoo Search Optimization CSO, Particle Swarm Optimization PSO, and Teaching-Learning-Based Optimization TLBO across three scenarios: rapid irradiance fluctuations, moderate partial shading with a centrally-located GMPP, and severe partial shading with a rightward-shifted GMPP. The simulation results demonstrate that DE-SGO consistently outperforms competing algorithms in both tracking efficiency (achieving up to 99.71% under severe partial shading) and convergence speed (0.18-0.23 seconds compared to 0.32-0.72 seconds for other methods). The algorithm exhibits superior performances in terms of dynamic response characteristics, with low oscillations around the MPPT, and more stable duty cycle control. The obtained results without additional hardware or complex implementation. Offering a robust solution for maximizing energy harvest in applications where non-uniform irradiance is common, particularly in systems experiencing frequent environmental variations.

Key words: Solar energy, Photovoltaic systems, DE-SGO, SGO, MPPT, optimization algorithms, Partial shading.