A Novel Framework for Game-Based Optimal Event-Triggered Control of Multi-Input Nonlinear Systems

• 成果类型: Article
• 署名作者: Xu, Wenqi; Wang, Tong; Qiu, Jianbin; Liu, Xiaoping
• 署名单位: Harbin Institute of Technology; Lakehead University
• 刊物名称: IEEE TRANSACTIONS ON AUTOMATIC CONTROL
• ISSN/ISSBN: 0018-9286
• DOI: 10.1109/TAC.2024.3358986
• 发表日期: 2024
• 页码: 4867-4874
• 关键词: games; game theory; Nonlinear systems; Stability criteria; mathematical models; dynamic programming; differential games; Adaptive dynamic programming (ADP); Event-triggered control (ETC); graphical game; multiplayer game; Zero-sum game

摘要

This article concentrates on the game-based optimal event-triggered control problem for a class of multi-input nonlinear continuous-time systems. The main objective is to improve the system performance. Furthermore, the optimal event-triggered conditions are derived to maximize the intersampling intervals. To simultaneously consider the optimality of control signals and event-triggered conditions, a novel two-layer game framework with redefined cost functions is developed. In the first layer, the continuous control and the threshold of the event-triggered condition are regarded as two players of the zero-sum differential game due to contrary benefits. In the second layer, event-based players game with others to reach the Nash-saddle equilibrium point in the framework of graphical game is studied. Furthermore, to tackle the issue that the analytical solution of Hamilton-Jacobian-Isaac equation is intractable to be obtained, an adaptive dynamic programming-based scheme using critic-only neural networks is considered following by a zeno-free behavior proof. Finally, a simulation example is displayed to validate the theoretical results.