Identification of solar panel parameters using artificial intelligence techniques and the Lambert-W function

Abstract

This work addresses the estimation of key parameters in photovoltaic (PV) models using advanced nature-inspired optimization algorithms, including the Mountain Gazelle Optimization (MGO) algorithm, Red-Tailed Hawk algorithm (RTH), Dung Beetle Optimization (DBO) algorithm, and Arctic Puffin Optimization (APO) algorithm. These algorithms are applied to both the single-diode and double-diode PV models, using experimental data from RTC France silicon solar cells. To enhance estimation accuracy and efficiency, the Lambert-W function is employed as a non-iterative alternative to the traditional Newton-Raphson method. The performance of each approach is evaluated on the basis of the root mean square error (RMSE) and the convergence speed. The results demonstrate that the RTH algorithm, when combined with the Lambert-W function, achieves the most accurate and efficient parameter estimation, consistently outperforming the Newton-Raphson method with lower RMSE values and faster convergence. The Lambert-W function proves particularly advantageous by allowing for an explicit, iteration-free formulation of the objective function, thereby reducing computational complexity, minimizing execution time, and enhancing solution stability.