Parameter extraction of photovoltaic model based on butterfly optimization algorithm with chaos learning strategy

Abstract

Accurate extraction of photovoltaic (PV) model parameter is significant for the evaluation of PV system. Most of existing methods for extracting the PV parameters suffer from low extraction accuracy and are prone to fall into local optima. To solve these shortcomings, a chaos learning butterfly optimization algorithm (CLBOA) is proposed. First, the introduced Cauchy mutation increases the perturbation to butterflies and helps the algorithm to jump out of local optima. Second, a chaos learning strategy is proposed to lead the butterflies within the population to learn from the optimal individual using the improved Tent chaos mapping, which strengthens the learning exchange of the population, enhances the ability of global exploration and local exploitation, and optimizes the convergence speed and accuracy. Third, the three worst individual positions were randomly initialized using a terminal elimination mechanism to increase population diversity. After evaluating CLBOA through the CEC2022 benchmark suite, it was used to extract parameters for a total of 12 PV cell and module models for 5 materials. Compared with 9 well-known algorithms, CLBOA performs excellent in terms of convergence performance, parameters extraction accuracy, running time and improvement index. Finally, CLBOA was applied on YL PV power station model of Guizhou Power Grid in China under different irradiations and temperatures. The I-V and P-V curves reconstructed by CLBOA are highly fit to the synthesized curves, and the individual absolute error (IAE) and relative error (RE) also confirm the satisfactory accuracy of parameters extraction. Comprehensive analysis shows that CLBOA can extract parameters better than the comparison algorithms, demonstrating its strong potential for PV parameters extraction.