Self-Triggered Adaptive Dynamic Programming Suboptimal Control of Unknown Affine Nonlinear Systems With Full-State Constraints and Input Constraints

A self-triggered approximate suboptimal controller design method is proposed within the adaptive dynamic programming framework. By constructing a novel auxiliary value function and candidate Lyapunov function, this method utilizes the barrier Lyapunov function for the first time in designing an optimal controller for a broader class of affine nonlinear systems with full-state constraints, overcoming the traditional limitation that the barrier Lyapunov function is restricted to strict feedback nonlinear systems. Furthermore, a nonquadratic cost function is employed to simultaneously satisfy the input constraints. By constructing identifier neural networks, optimal performance control is achieved, even with unknown system dynamics. The proposed self-triggered mechanism enables the controller and network weights to be updated only at the predicted moment, which not only reduces communication resource consumption but also eliminates the need for system state monitoring, as required in the event-triggered mechanism. Rigorous convergence analysis establishes a strong theoretical foundation for the stability and safety of the proposed method in practical applications. Simulation experiments confirm the effectiveness of the algorithm.