A reinforcement learning approach to STATCOM controller

The advanced static synchronous compensator (STATCOM) is a device that can provide reactive support to bus voltage. A novel control scheme based on reinforcement learning algorithm for the STATCOM is presented in this paper. Design of STATCOM controller presented does not depend on structure and parameters of power system. The adaptive heuristic critic algorithm is adopted in this paper and the parameter of the controller are updating by given reinforcement signal of system. Only local information is required. A simulation example of single-machine infinite power system is given, and the result shows the effectiveness on maintaining voltage and preventing voltage collapse.