Voltage Control on Distributed Generation Systems based on Multi-Agent Reinforcement learning approach

The voltage control problem due to bidirectional power flows is more apparent when heterogeneous distributed generation systems (DGS) are integrated into the grid. In this paper, a novel method of voltage control in distributed generation systems based on a reinforcement learning technique is proposed. DGS incorporating renewable energy resources are highly complicated nonlinear dynamic systems. There are several challenges in employing the existing control methods. The novel method presented in this paper entrenches the Q learning algorithm into the voltage control problem of DGS. The Q-learning algorithm teaches agents responsible for decision taking in controlling the voltage and award the reward if the aim is achieved. The IEEE 9 bus test system with DG’s integrated is used with various controlling agents connected. The results show significant improvement in the reliability of agent communication and the efficiency of the proposed method.