Enhancing accuracy and convergence in triple-diode photovoltaic parameter extraction

Parameter extraction for the triple-diode photovoltaic (PV) model presents a complex, highly nonlinear optimization problem. It necessitates an optimization algorithm that has strong exploration and exploitation abilities to avoid getting stuck in local optima and to accurately determine the model’s parameters. Although various optimization methods have been used in the literature to address this problem, most of them produce poor results, show instability across different PV modules, have slow convergence, and/or require high computational costs. These limitations motivate us to introduce a new robust parameter identification technique called IGO, which can achieve more accurate results with fewer function evaluations. It is based on integrating the recently published growth optimizer (GO) with two new optimization strategies—the convergence improvement strategy and the ranking-based update strategy. The latter strategy steadily enhances the exploration operator throughout the optimization to prevent premature convergence to local optima. Simultaneously, it gradually boosts the exploitation operator in the late phases to accelerate convergence to the global optimum. The former strategy focuses on enhancing the exploitation operator during the optimization process to maximize convergence speed while strengthening the exploratory operator in late stages to mitigate the risk of settling in local optima. Integrating both strategies in the proposed IGO aims to balance exploration and exploitation throughout different phases of iteration, thereby preventing stagnation in local optima and encouraging rapid convergence toward the global optimum. The proposed IGO is tested on six popular PV modules and compared with several recently published optimizers using various statistical measures in addition to convergence speed to demonstrate its effectiveness and significance. The experimental results demonstrate that IGO outperforms all other methods in both parameter quality and convergence speed, confirming it as a reliable alternative for extracting the unknown triple-diode model parameters.