Resilient Optimal Virtual Synchronous Generator Control Under DoS Attacks

This brief leverages a novel resilient reinforcement learning (RL) technique to tackle the model-free optimal transient problem for virtual synchronous generators (VSGs) under denial-of-service (DoS) attacks. An optimal control strategy is established from real-time data of the VSGs under DoS attacks using the iterative algebraic Riccati equation (ARE). Furthermore, with a theoretical upper limit for DoS attack durations, system resilience is analyzed using the Lyapunov method. The resilient hybrid iteration (RHI) algorithm introduced in this brief eliminates the assumption of initial stable control in previous approaches, removes the need for precise system dynamics, and reduces computational requirements. Finally, the effectiveness of this approach is demonstrated through simulation experiments.