Deep Reinforcement Learning-based Approach for Online Tuning SMES Damping Controller Parameters

With the penetration of the power electronics dominated large power system, the design of the damping controller needs to ensure the stability of the power system at any disturbance condition. In this context, the closed-loop control agent trained via deep deterministic policy gradient (DDPG) by interacting with massive simulations, for tuning the superconducting magnet energy storage (SMES)-based damping controller parameters, is proposed. It can make autonomous damping mitigation strategy support the grid operator via the current system state. Numerical simulation is tested on the SG-SMES system with a 25MW wind farm, demonstrates the promising performance of the proposed approach.