Reinforcement Learning based Energy Management System for Grid-Connected PV plants and Energy-Stored Quasi-Z-source Cascaded H-Bridge Multilevel Inverter

A multiagent reinforcement learning (RL) algorithm is used in this paper as the energy management system (EMS) of an energy stored quasi-Z-source cascaded H-bridge multilevel inverter with (ES-qZS-CHBMLIs) for grid-connected photovoltaic (PV) cells. The duty of the new EMS is meeting the requested power while sorting or releasing power between the battery energy storages (BES). A photovoltaic plant with 4.8 kW capacity in 3 cascaded modules configuration connected to a single-phase grid is simulated in MATLAB to assess the performance of the suggested RL algorithm. For each module, different battery parameters, variable grid references and varying irradiance values are considered to evaluate the suggested EMS. In order to control DC-link voltage, dq dimensional current components, quality of grid power, and optimal operation of the power sources conventional control methods are applied..