基于强化学习算法的静止同步补偿电压控制器

APPLICATION OF REINFORCEMENT LEARNING TO STATCOM CONTROLLER

将强化学习算法应用于静止同步补偿(STATCOM)电压控制器,克服了常规 STATCOM 电压控制器对系统数学模型的依赖性,同时根据来自系统的强化学习信号,采用自适应启发评价算法更新其参数。在控制器中采用局部可测信息量,以确保其可实现性。仿真实验结果表明,基于强化学习算法的 STATCOM 电压控制器可以在紧急情况下对系统电压进行调节,从而在一定程度上确保了系统的安全稳定运行。

The application of reinforcement learning algorithm to static synchronous compensator (STATCOM) controller could overcome the dependence of traditional STATCOM controller on mathematical model of power system, at the same time, the controller's parameters could be updated by use of adaptive heuristic assessment algorithm according to the reinforcement signal from power system. In the controller the local measurable information was used to ensure the realizability. The simulation results show that the STATCOM controller based on reinforcement learning algorithm can adjust the system voltage under emergent situation, so the secure and stable operation of power system can be improved to a certain extent.