Data-Driven Event-Triggered H-infinity Load Frequency Control with Security Against DoS Attacks

Load frequency control (LFC) is critical for stabilizing maintaining grid frequency, especially in power systems with high wind power penetration and increasing vulnerability to denial-of-service attacks. This paper proposes a data-driven dynamic event-triggered reinforcement learning (RL) algorithm to realize the constrained H$_\\infty$ LFC in multi-area power systems with wind power integration under denial-of-service attack conditions. The proposed approach reformulates the control problem as a min-max optimization task and introduces a dynamic event-triggered mechanism to alleviate computational  and communication burdens. A neural network-based RL framework is developed to approximate the near-optimal event-triggered control strategy without requiring explicit system dynamics. Additionally, to mitigate the adverse effects of frequency-based denial-of-service attacks, an attacks compensation mechanism is designed. Theoretical analysis establishes the input-to-state stability and ensures the convergence of neural network weights. Simulation results validate the proposed control strategy in maintaining grid stability and alleviating transmission burdens under adverse conditions.