Power Flow Management with Q-Learning for a Grid Integrated Photovoltaic and Energy Storage System

This paper develops a fuzzy Q-learning (FQL) approach-based power flow management algorithm for a single-phase grid-connected photovoltaic (PV) system with an energy storage unit (ESU). The FQL coordinates the PV power generation which is based on the mission profile, state of charge (SOC) of ESU, and the load profile such that a power balance is achieved in the network. Further, the transition between standalone (SA) and grid-connected (GC) modes during the power flow management is achieved using a proportional capacitor current feedback. While operating in the SA mode, the control action for load voltage is achieved by an outer current loop. This combination of coordinated and transition control achieves power balance and smooth transition between the SA and GC modes. To verify the effectiveness of the proposed control strategy, a 4 kWp PV system is operated along with an ESU in both SA and GC modes for a varying mission and load profile. The simulation and experimental results validated the multifunctional features of the proposed method.