A Hybrid Method of SVG Parameter Calibration Based on Particle Swarm Optimization and Soft Actor-Critic

Static var generators (SVGS) are essential for compensating reactive power and maintaining bus voltage levels before and after large disturbances. Accurate modeling of SVG controller parameters is critical to ensure the reliability of dynamic simulation in power system operation. This paper introduces a novel SVG parameter identification method, which integrates the benefits of particle swarm optimization (PSO) and reinforcement learning (RL) to improve the efficiency and accuracy of parameter calibration. We use simulation data generated by PSD BPA software combined with actual power grid data as the environment to optimize the parameter settings of SVG. This study combined PSO with the Soft Actor-Critic (SAC) algorithm to optimize and calibrate these parameters. Compared with the simulation results, our method shows significant advantages in calibration accuracy and processing speed and effectively improves the simulation quality and practicability of the SVG model.