Optimization of Photovoltaic System Modelling: A Comparative Study and Experimental Validation Using Bond Graph Methodology and a Genetic Algorithm

Abstract

In this study, we present a methodology designed to model a solar photovoltaic plant. Our approach emphasizes simplicity in modeling procedures. Specifically, we employ the Bond Graph Approach to construct a solar cell model, which is represented by a circuit consisting of one diode and two resistors. To deduce the five unknown parameters of the solar cell, we employ the genetic algorithm. The primary objective of this modeling endeavor is twofold: first, to scrutinize and gain insights into the operational behavior of the PV plant, and second, to develop a more precise and dependable PV model capable of replicating PV panel performance under outdoor environmental conditions. To ascertain the accuracy of our proposed model, we conduct a comparative analysis with both the MATLAB PV model and empirical data. This comparative assessment encompasses monocrystalline PV technology. We conduct simulations using real-world temperature and solar irradiation data for four different days, representing various seasons throughout the year. These simulations are executed using the 20-Sim software platform. Our findings reveal that the proposed model exhibits greater accuracy than does the MATLAB PV model. Furthermore, the results demonstrated a high level of fidelity in reproducing experimental data under the specific operational conditions and for the considered PV technologies.