Voltage Control in Radial Distribution Systems using Deep Reinforcement Learning Technique

This paper proposes a novel approach for voltage control in radial power distribution systems using the state-of-the-art Double Deep Q-Network (DDQN) algorithm to optimize the operation of two capacitor banks within a 13-bus radial network under different load configurations. A Python-based framework facilitates interfacing between a power distribution simulation tool, OpenDSS, and a deep RL agent. The deep RL agent dynamically adjusts the capacitor bank operation through iterative learning to maintain bus voltages within predefined limits. Furthermore, the performance of the DDQN agent is compared with that of the DQN agent and the conventional capacitor controller embedded within OpenDSS.