Deep reinforcement learning based parameter self-tuning control strategy for VSG. (August 2022)

• Record Type: Journal Article
• Authors: Xiong, Kang; Hu, Weihao; Zhang, Guozhou; Zhang, Zhenyuan; Chen, Zhe
• Is Part Of: Energy reports. Volume 8(2022)Supplement 5
• Page Start: 219
• Page End: 226
• Publication Date: 2022-08
• Subjects: Distributed generation -- Virtual synchronous generator -- Adaptive control -- Deep deterministic policy gradient; Power resources -- Periodicals; Energy industries -- Periodicals; Power resources; Periodicals; Electronic journals; 621.04205
• DOI: 10.1016/j.egyr.2022.02.147
• Languages: English
• ISSNs: 2352-4847

Abstract

Abstract: With the development of new energy technology, the distributed generation has attracted more and more attention. In order to enhance the inertia of distributed generator system to improve its stability, the control technology of virtual synchronous generator (VSG) is proposed. However, the traditional VSG control technology often has poor flexibility and long dynamic adjustment time. In this context, a deep deterministic policy gradient (DDPG) algorithm based adaptive controller is designed to realize the adaptive control of inertia and damping coefficient in the system, so that the parameters can be adjusted adaptively under different operating conditions. Next, a simulation model of VSG isolated island single machine operation model is built in MATLAB Simulink, and the implementation of DDPG algorithm is given and verified by simulation. The results show that the VSG parameter adaptive system controlled by DDPG has stronger ability to resist disturbance and achieves better performance than the traditional VSG adaptive system. The DRL model only takes 0.448 s to return to stable, but the other two models need 0.632 s and 0.818 s respectively.