Deep Reinforcement Learning-Based Additional Damping Control for Grid-Forming HVDC Power System

This paper tackles the stability issues in grid-forming HVDC power systems by applying deep reinforcement learning (DRL) techniques. This paper introduces an additional damping control method that utilizes Deep Deterministic Policy Gradient (DDPG). This approach leverages the adaptive capabilities of DRL, particularly the strengths of the DDPG algorithm in handling continuous action spaces, deterministic policies, and high-dimensional state spaces, to address complex, non-linear control problems, thereby improving grid-forming HVDC power system stability and performance. Simulations demonstrate that the DDPG-based control method significantly improves the stability of grid-forming HVDC power systems, effectively suppressing oscillations and maintaining stability during disturbances in the receiving power grid. Meanwhile, it confirms the feasibility and effectiveness of DRL in grid-forming HVDC power systems, paving the way for future advancements in intelligent control strategies for modern power grids.