Online Optimal Self-Triggered Sampling Control With Efficient Adaptive Critic Learning

A new online self-triggered sampling control method is presented to address the optimal adaptive critic regulation issue for discrete-time nonlinear systems. First, a self-sampling function is designed based on the stability of the controlled system to gather only the necessary data during the sampling process, and the influence of the control parameters on the self-triggering interval is proved in detail. Subsequently, a self-triggered-based offline iterative algorithm is developed to create a stability criterion condition, aiming to obtain an initial admissible control policy. In the online optimal control phase with the sampled data, the Hamilton-Jacobi-Bellman equation of the controlled system is solved by building the model, critic, and action neural networks. In addition, the stability proof of the self-triggered-based controlled system is provided, and the impact of control parameters on the self-triggered architecture is analyzed. Finally, an experimental plant with nonlinear characteristic is presented to demonstrate the comprehensive performance of the proposed online control method.