Adaptive Reinforcement Learning Optimal Time-Varying Formation Control for Nonlinear Multiagent Systems via a Dynamic Event-Triggered Mechanism

作者：Zhenghua Li, Shuo Ding, Ben Niu, Ning Zhao, Xudong Zhao · 发表于：IEEE Systems Journal · 年份：2026 · DOI：10.1109/jsyst.2026.3678159 · 被引用次数：2 · 研究领域：Adaptive Dynamic Programming Control、Reinforcement Learning in Robotics、Distributed Control Multi-Agent Systems

This article investigates an event-triggered optimal time-varying formation (TVF) control problem for nonlinear multiagent systems with unmeasurable states and unknown nonlinear dynamics. A neural network-based state observer is first designed to reconstruct unmeasurable system states. Meanwhile, neural networks are employed to approximate unknown nonlinear functions. Subsequently, an adaptive reinforcement learning algorithm is proposed to address the optimal control issue within an identifier–critic–actor framework. To save system communication resources, this article designs a dynamic event-triggered mechanism in sensor-controller channels and develops an adaptive state event-triggered control scheme. Unfortunately, state-triggered control may cause nondifferentiability in virtual control laws, posing significant challenges for the design of an optimal TVF controller. To overcome this obstacle, we use observer output signals to construct virtual control laws. Then, an event-triggered controller is constructed by replacing estimated states with intermittent estimates. The results show that formation tracking errors can converge to a small neighborhood near the origin, and all closed-loop signals are semi-globally uniformly bounded. Finally, the effectiveness of the control scheme is verified by an example simulation.