Event-Triggered Optimal Security Control for a Class of Nonlinear Systems Under Hybrid Network Attacks

This paper presents an event-triggered optimal security control strategy for nonlinear systems under hybrid network attacks on sensor channel. And first, the attacks are modeled based on a Bernoulli framework, simplifying the model while ensuring applicability to various attack scenarios. By using an observer, states can be reconstructed. Second, an actor-critic reinforcement learning framework is adopted to design the optimal control strategy, ensuring that both communication resource utilization and energy consumption are minimized. Finally, a simulation verifies the feasibility of the proposed method.