Control of advanced static VAr generator by using recurrent neural networks

According to the dynamic characteristic of the advanced static VAr generator (ASVG), a recurrent neural network (RNN) based inverse dynamic controller is constructed in this paper, and its training algorithm is given. Taking the single machine infinite bus power system, a three-phase short circuit is used for the test of the proposed RNN controller. Simulation results show the RNN controller can learn the inverse dynamic of a controlled power system and has better performance than the PID controller.