Modeling Solar PV Efficiency: Machine Learning-Enhanced Algorithms for Diode Model Parameter Extraction

Abstract

Solar photovoltaic (PV) systems can be significantly enhanced through the use of accurate solar cell models. Unfortunately, the absence of precise parameters from manufacturers limits the accuracy of these models. Given the impossibility of reliable modeling without such parameters, this paper introduces a multi-objective optimization algorithm to estimate the necessary parameters effectively. The problem of suboptimal optimization results often arises due to local minima and premature convergence of the optimization algorithm, even though there are a number of optimization algorithms that address this issue. This paper is intended to examine the reliability of the proposed algorithm to determine if it is reliable. For the purpose of showing the proficiency of the proposed optimization algorithms, their performance is compared with that of some other well-known algorithms to show their superiority. The performance of the algorithm is validated by comparing experimental results, including analyses based on statistical data, with estimated parameters based on statistical analysis. Furthermore, the results obtained with the proposed algorithms indicate that they are better suited for estimating solar PV models than the other algorithms i.e., rmse of the proposed algorithm for three diode model is 4.21E−13 as well as 3.20E−13 for four diode model. A simple structure and high accuracy are the main characteristics of the proposed algorithm, which indicates its potential for a variety of applications in the solar energy field in the future. Moreover, the proposed algorithm is computationally efficient as well as easy to use and can be applied to a number of applications.