空间非合作目标安全追踪的动态事件触发 学习控制

A dynamic event-triggered tracking method based on integral reinforcement learning is proposed to address the safe tracking problem of non-cooperative space targets. By incorporating a control barrier function and hyperbolic tangent terms into the value function，active obstacle avoidance and continuous handling of asymmetric control input constraints are achieved. On this basis，a dynamic event-triggered control framework is constructed to reduce communication and computational burdens. Furthermore，a critic-only neural network is employed to estimate the value function and the optimal control policy，and the Hamilton-Jacobi-Isaacs equation is formulated within the integral reinforcement learning framework to alleviate the reliance of weight updates on an accurate system model. Subsequently，an experience replay mechanism is introduced to relax the requirement of the persistent excitation condition. Finally，based on Lyapunov theory，the uniform ultimate boundedness of the system states and the weight estimation errors is established. Numerical simulation results demonstrate that the proposed method can achieve stable tracking of non-cooperative targets while satisfying safety and input constraints，and significantly reduce sampling and communication loads.