Random learning gradient based optimization for efficient design of photovoltaic models

Abstract

How to effectively realize the parameter identification of different photovoltaic models has gradually become a research hotspot. This paper proposes an improved gradient-based optimizer (GBO) that combines a random learning mechanism, named RLGBO, to tackle the parameter identification problem in photovoltaic models. The GBO method is a recent swarm-based approach proposed in 2020, and it is exciting for us that it has no metaphor in its model as a step forward in optimization. This optimizer has two core procedures: gradient search rule (GSR) and local escaping operator (LEO). The new random learning mechanism is introduced into the original GBO, which effectively alleviates the shortcomings of falling into local optimum, and improves the convergence speed and accuracy. The random learning mechanism encourages the optimal individual to learn random communication results between different individuals continuously. In order to assess the performance of the suggested RLGBO, it is applied to the parameter evaluation of the single diode model, double diode model, three diode model, and photovoltaic module model. The experimental results demonstrate that RLGBO features well-heeled superiority and is highly competitive with recently reported technologies. Besides, RLGBO is applied in three different commercial photovoltaic models, including SM55, ST40, and KC200GT, to resolve the single diode model and double diode model's parameter identification problem under different temperature and light conditions, as well. The results verify that RLGBO can accurately estimate model parameters regardless of various environmental conditions. In general, the proposed RLGBO is expected to be a new reliable solver to evaluate the relevant parameters in photovoltaic models. A webpage at https://aliasgharheidari.com will provide an online service for any support regarding the algorithm in this paper.