Superconducting magnetic energy storage based power system control using ADP

Active power oscillation becomes a critic hurdle for bulk power transmission between large-scale interconnected power grids. Damping controllers based on energy storage device (ESD) could provide effective solution to address this issue. In this paper, superconducting magnetic energy storage (SMES) based power system oscillation damping controllers are developed to increase the system transient stability, where one is the traditional linear matrix inequality (LMI) technique based design and the other is on-line reinforcement learning (RL) based design. The proposed RL based design employs adaptive dynamic programming (ADP) algorithm, which uses multiple-layer neural networks as the action network and the critic network. The proposed two controllers are tested on an IEEE 39-bus benchmark system under various system disturbance conditions. Simulation results demonstrate the satisfied control performance of the proposed controller.