Proximal Policy Optimization-Based Coordinated Voltage Control for a Grid-Connected Microgrid Cluster Considering Photovoltaic Intermittency

Abstract

With the increasing penetration of distributed photovoltaic generation, intermittent photovoltaic output may cause voltage fluctuations in microgrid clusters. This paper proposes a coordinated voltage control strategy based on Proximal Policy Optimization for a grid-connected microgrid cluster consisting of three sub-microgrids with photovoltaic generation, energy storage systems, and ZIP loads. The voltage control problem is formulated as a Markov decision process with a 12-dimensional state space and a 6-dimensional continuous action space. The state variables include bus voltages, photovoltaic outputs, storage states of charge, and net loads, while the control actions consist of storage charging and discharging powers and photovoltaic inverter reactive powers. Simulation results show that the voltage qualification rates under clear, cloudy, and rainy conditions are 89.0 percent, 89.8 percent, and 86.2 percent, respectively. Compared with droop control, the proposed strategy reduces the mean voltage deviation by approximately 12 to 14 percent, providing an effective approach for maintaining voltage stability in microgrid clusters.