On Solar Photovoltaic Parameter Estimation: Global Optimality Analysis and a Simple Efficient Differential Evolution Method

Shuhua Gao,Cheng Xiang,Ming Yu,Kuan Tak Tan,Tong Heng Lee

DOI: https://doi.org/10.1109/cdc49753.2023.10383303

2023-01-01

Abstract:The literature presents a vast array of advanced metaheuristic methods for photovoltaic parameter estimation. However, the focus of this study is not to introduce another new method into this already crowded field. Instead, we examine two important but often overlooked questions: (i) are existing results globally optimal, and (ii) can a simpler method achieve comparable performance. We conduct case studies using two widely used I-V curve datasets. To address the first issue, we develop a branch and bound algorithm that, despite its sluggishness, either certifies the global minimum or provides a tight upper bound. These values serve as useful benchmarks for fair metaheuristic evaluation and further development. To answer the second question, our extensive examination and comparison surprisingly reveal that a basic differential evolution (DE) algorithm can achieve the certified global minimum or obtain the best known result. Additionally, the DE algorithm's runtime is much shorter than that of current state-of-the-art metaheuristic methods, making it a great choice for time-sensitive applications. This remarkable finding suggests that using increasingly complex metaheuristics in ordinary PV parameter estimation problems might be unnecessary. Finally, we discuss the implications of these outcomes and propose the simple DE method as the premier choice for industrial applications.