Performance Evaluation of Artificial Neural Network Maximum Power Point Tracking Algorithm for Photovoltaic Panel with Single Ended Primary Inductor Converter

Renewable energy sources are becoming increasingly important to replace fossil fuel-based production. Photovoltaic (PV) systems are one such resource, with their usage recently increasing due to net-zero targets. It is crucial to optimize solar panels to achieve maximum efficiency according to varying conditions. Maximum power point tracking (MPPT) algorithms have been developed to benefit from the panels with the highest performance. This study introduces a new MPPT algorithm based on artificial neural networks (ANN), utilizing the Single-ended primary-inductor converter (SEPIC), which offers numerous advantages. The proposed method enhances photovoltaic system efficiency by training a feed-forward artificial neural network for MPPT. This training utilizes 131 months of detailed temperature and irradiance data, collected hourly from Istanbul. The trained ANN uses temperature and irradiance data to estimate the instantaneous maximum power point (MPP) of the PV system in real time. With the help of these predictions, the power converter parameters are optimized to operate the PV system as efficiently as possible. The efficiency of the proposed system has been assessed through MATLAB/Simulink simulation of a 20-watt PV panel system, and the results are presented accordingly. This study is the first to demonstrate the effectiveness of the MPPT-ANN algorithm by using real temperature and irradiance data for the SEPIC converter.