Single-Phase Grid-Connected Photovoltaic Systems using aDeep Reinforcement based MPPT Algorithm with Grey-Wolf Optimization under Partial Shading Condition

Abstract

A detailed MATLAB model of a residential PV system that is linked to the grid is shown in this article. A state-of-the-art ILPAO approach was used to fine-tune the model. It has a unique conversion topology with two steps. An optimized current controller, a scanning-enabled Deep Reinforcement Learning based Maximum Power Point Tracking algorithm (SEDRL-MPPT), and a high-gain voltage regulator are all part of the suggested control scheme that attempts to solve the issue of multiple local maxima in the PV power-voltage characteristics caused by partial shading.The simulations showed that the stabilized current controller improvements significantly improved the responsiveness and stability of the DC connection's transient voltage, and oscillations and overshoot are reduced by the boost voltage regulator gains while operating under dynamic irradiance circumstances. If the irradiation is not uniform, SEDRL-MPPT will maximize energy extraction. Comparative analysis against a baseline control configuration confirms that the proposed method’s effectiveness in improving system stability, dynamic performance, and power quality under realistic environmental disturbances. These findings validate the proposed approach as a robust solution for efficient and reliable residential PV-grid integration.