RIME optimization with dynamic multi-dimensional random mechanism and Nelder–Mead simplex for photovoltaic parameter estimation

Abstract

Abstract Solar photovoltaic technology is efficient and clean, but extracting photovoltaic cell parameters is challenging due to various influencing factors. The rime optimization algorithm (RIME) is a recently proposed metaheuristic algorithm (MAs). This paper introduces the dynamic multi-dimensional random mechanism (DMRM) combined with the Nelder–Mead simplex (NMs) to propose an enhanced version of RIME, called DNMRIME. DMRM improves the convergence accuracy of RIME by random non-periodic convergence, and NMs accelerate convergence, enabling DNMRIME to escape local optima and perform better on hybrid and composite functions. To evaluate the performance of DNMRIME, a qualitative analysis and an ablation study were conducted on CEC 2017. To verify its effectiveness, DNMRIME was compared with 14 well-known MAs, including some champion algorithms, and the results of the Wilcoxon signed rank test showed that DNMRIME ranked first. To extract parameters on SDM, DDM, TDM, and PV, DNMRIME was applied, resulting in mean RMSE values of 9.8602188324E − 04, 9.8296993325E − 04, 9.8393451046E − 04, and 2.4250748704E − 03 respectively. Moreover, under varying temperature and irradiation conditions on three manufacturers (KC200GT, ST40, SM55), DNMRIME extracted parameters with simulation data matching the actual data. Therefore, unlike previous studies, this study proposes DMRM and DNMRIME, demonstrating the efficiency and practicality of DNMRIME and further highlighting potential value of DNMRIME in photovoltaic parameter extraction. The source code of DNMRIME is available at https://github.com/zyetpink/DNMRIME-Solar-Model-dataset.