A novel enhanced Grey Wolf Optimizer for global optimization problems: Application to photovoltaic parameter extraction

Abstract

Complex optimization problems in engineering remain challenging due to their nonlinear objective functions, multimodal search spaces, and high-dimensional parameter spaces. In this paper, we present a new enhanced Grey Wolf Optimizer (NE-GWO) to effectively overcome these challenges. The NE-GWO introduces personal best memory mechanism for leading individual agents, probabilistic selection strategy of leaders for diversity preservation, and sinusoidal perturbation operator for local optima avoidance, and thus achieves an overall better exploration-exploitation trade-off. We rigorously test the algorithm on conventional benchmark functions and on one practical photovoltaic (PV) parameter identification application, which comprises the single-diode model (SDM), double-diode model (DDM), triple-diode model (TDM), and the Photowatt-PWP201 commercial module. Comprehensive comparison with nine of the state-of-the-art metaheuristics conclusively verifies that NE-GWO outperforms competitors decisively on precision of solutions, speed of convergence, and robustness. Statistical Wilcoxon signed-rank tests at the level of 5% confirm that performance improvements of NE-GWO are statistically significant. The results establish conclusively the consistent superiority of NE-GWO, which achieves the lowest root mean squared error (RMSE) for all PV models: 7.7487E−04 for the SDM, 7.6257E−04 for the DDM, and 7.5869E−04 for the TDM on the RTC France solar cell, and 2.06904E−03 for the Photowatt-PWP201 module. In addition, NE-GWO achieved this performance with competitive computation times, emphasized with a low standard deviation, which verifies its stability and accuracy. The results conclusively establish NE-GWO as an effective and reliable solver for difficult global optimization problems.

Topics & Concepts