A Modified Lyrebird Optimization for Single, Double, and Triple Diode Parameters PV Cell Extraction

Abstract

Accurate modeling of photovoltaic (PV) cells is essential for performance assessment, control, and optimization of solar energy systems. The diode circuit models, particularly the single-diode (SD), double-diode (DD), and triple-diode (TD) structures, are widely adopted for characterizing PV behavior; however, extracting their unknown parameters is a challenging nonlinear, multimodal optimization problem. To address these challenges, this study proposes a modified lyrebird optimization (MLO) algorithm, an enhanced variant of the recently developed LO. It integrates a memory-based learn search strategy (MBLSS) to reinforce exploration and a diversity maintenance learn search strategy (DMLSS) to refine exploitation. The algorithm was employed to extract parameters of both the RTC France solar cell and the Solarex MSX-60 PV module under SD, DD, and TD models. Extensive simulations and statistical analyses demonstrated that the proposed MLO significantly outperforms the conventional LO and a wide range of metaheuristic and analytical methods in terms of root mean square error (RMSE), convergence speed, stability, and robustness across multiple runs. In the MSX-60 module tests, the proposed MLO reduced the RMSE by more than 55% compared to the conventional LO and achieved a stable mean RMSE of 1.75 × 10-3 over 50 independent runs. Similarly, for the RTC France solar cell, MLO achieved a minimum RMSE of 9.82 × 10-4, outperforming several recently reported metaheuristics. Moreover, the proposed MLO was extended and validated on the Shell S75 monocrystalline module under different irradiance and temperature conditions. The results demonstrated consistently lower RMSE values and near-zero variance across operating ranges, confirming the robustness and stability of MLO in practical PV environments. The strong agreement between simulated and experimental I–V and P–V characteristics confirms the reliability of the extracted parameters. These findings highlight the potential of MLO as a robust and accurate tool for PV modeling, with promising applications in solar system design, performance evaluation, and predictive energy management.