Machine learning based optimal control of modular converter for PV assisted power supply systems

Abstract

This paper presents the topology and machine learning-based intelligent control of a single-stage grid-connected high-power photovoltaic (PV) system for quality power export to the grid and optimal net energy utilization. A nineteen-level bi-modular inverter is proposed for efficient single-stage PV power conversion. The proposed integrated intelligent machine learning-based control serves for power conversion control as well as supervisory control for hourly PV energy estimation and load demand control for optimal energy consumption. The objectives of power control are extracting maximum power from PV sources and exporting power to the grid at unity power factor. While the objectives for supervisory control are local load demand control for exporting power at higher export prices. The proposed system is implemented using MATLAB/Simulink to validate the efficiency of power conversion, effectiveness of machine learning for energy estimation, and load relay control for optimal energy pricing. The results proved efficient tracking of maximum power, unity power factor at grid terminals, and load relay control for PV energy availability and export cost function.