PV parameter estimation of dye-sensitized solar cells using metaheuristic algorithms

• PV parameters of DSSCs were estimated for three N719 dye concentrations (0.25, 1, and 2 mM) using experimental I–V data, with four optimization methods evaluated via error, consistency, and convergence analysis. • Physical bounds of parameters are justified, and TDM accuracy is validated through comparison with SDM for improved DSSC modeling. • The best-performing optimization algorithm is identified based on RMSE minimization and convergence stability, and benchmarked against existing methods to confirm reliability. • Wilcoxon and Friedman tests with mean ± standard deviation confirm statistically significant performance differences among algorithms • Sensitivity analysis with ± 20% perturbations is performed to evaluate parameter reliability and quantify their influence on RMSE across different dye concentrations. Dye-sensitized solar cells (DSSCs), a viable alternative due to its low manufacturing cost, simple fabrication process, and strong performance under low-light conditions, have gained significant attention. However, improving the adaptability of DSSCs necessitates large-scale system modelling. At the same time, accurate modelling, in turn, requires precise model parameter estimation. Given the nonlinear characteristics of DSSCs, the lack of standard I-V curve datasets, and use of various dye materials impose challenges. Although different equivalent circuit models coexist, a study on robust optimization algorithms for DSSC modelling is of high importance and inevitable, as the choice of model or method and its performance directly influence overall reproducibility. This study, therefore, examines the suitability of four different optimization algorithms, namely, the Walrus Optimization Algorithm (WaOA), Dynamic Control Cuckoo Search (DCCS), Enhanced Barnacle Mating Optimization (EBMO), and Black Widow Optimization with Gaussian mutation algorithm (BWOG), for DSSC parameter identification. A synergistic framework combining optimization techniques with a nonlinear curve fitting approach is applied to DSSCs with three different concentrations of N719 dye in absolute ethanol solution (0.25 mM, 1 mM, and 2 mM). In addition, sensitivity analysis of the extracted parameters was conducted applying ± 20% perturbations. Statistical validation using Wilcoxon signed-rank and Friedman tests was also conducted to identify the best–suited DSSC parameter estimation method. The results revealed that the WaOA method effectively balances exploration, migration, and exploitation to achieve the optimal solution across multiple independent runs. From the statistical data, WaOA achieved the lowest objective function values of 4.5482E-05, 3.1349E-05, and 4.8289E-05 at the respective concentrations. Furthermore, significant improvements in the values of RMSE up to 9.63% (0.25 mM), 10.37% (1 mM), and 4.04% (2 mM) were also noted. Benchmarking against literature algorithms validates WaOA reliability and stability.