Event-Triggered single-network Control for Nonlinear Continuous-Time Zero-Sum Games with Partially Unknown Dynamics

This paper proposes an event-triggered integral reinforcement learning algorithm to solve the Hamilton-JacobiIsaacs (HJI) equation related to the zero-sum game problem. Then, we design a triggering rule to constrain the control updates. Secondly, the HJI partially differential equation is approximately solved using a single-critic neural network structure. The control law and disturbance law are modified when an event occurs. The system state and the network weights error are uniformly ultimately bounded (UUB) by Lyapunov theory. Finally, an engineering example is utilized to confirm the performance of the algorithm.