A fuzzy-driven modified draco lizard optimizer for precise PV model parameter estimation

This paper introduces a Modified Draco Lizard Optimizer (MDLO) for parameter identification of Photovoltaic (PV) models utilizing diode-based equivalent circuits. In the proposed MDLO, three key enhancements are involved in improving adaptability, efficiency and stability. A fuzzy Gaussian-based controller enables a smooth probabilistic transition between exploration and exploitation, while an exponential steering vector decay refines the balance between large-scale search and local intensification. In addition, a boundary re-initialization strategy maintains population diversity by redistributing infeasible solutions within the feasible search space. To assess the feasibility of the suggested method, the proposed MDLO is carried out to extract the parameters of the single, double, and triple diodes models of two commercial PV modules, including RTC France and KC200. The outcomes indicate that the Root Mean Square Error (RMSE) is significantly reduced with the use of MDLO, which provides more accurate parameter estimations and enhances energy conversion efficiency in PV applications. Compared to traditional DLO and other methods, MDLO exhibits superior performance, making it a promising candidate for PV model parameter extraction issues. Moreover, an ablation study and an exploration–exploitation behavior analysis are conducted to elucidate the individual and combined effects of the proposed enhancement mechanisms. Population diversity and step-size evolution confirm that the proposed MDLO achieves a well-regulated transition from global exploration to local exploitation. Comprehensive non-parametric statistical tests further demonstrate that the performance improvements are statistically significant, robust, and consistent across independent runs.