Grid Voltage Control Method Based on Generator Reactive Power Regulation Using Reinforcement Learning

摘要

Too high or too low grid voltage will greatly affect the operation safety of the power system. This paper proposes a voltage regulation method based on generator reactive power regulation using reinforcement learning. First, the members of the agent are selected based on the reactive power of the generator, and then the Q tables are trained corresponding to the agent. According to the voltage amplitude state of the point to be adjusted, the regulation action is given. In order to solve the problem of complex combination of action sets and difficult convergence of Q table in a single agent, multi-agents are selected to decompose complex regulation actions into a series of continuous simple actions. At the same time, in order to improve the effectiveness of agent actions, the “progress” reward mechanism is proposed. The method based on reinforcement learning proposed in this paper can effectively regulate the reactive voltage of the system.