Scholarlink

An integral reinforcement learning (IRL) distributed consensus optimal control issue is developed for nonlinear multi-agent systems (MASs) with partially unknown system dynamics. In control design, the IRL algorithm is used to learn the online solution for the Hamilton-Jacobi-Bellman (HJB) equation. Then, combined IRL and actor-critic neural networks (A-C NNs), a new adaptive distributed consensus optimal control approach is designed. To overcome the shortcomings of periodic sampling, a dynamic event-triggered (ET) mechanism is designed. The designed control algorithm ensures that the closed-loop system stabilizes and Zeno behavior can be avoided.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic Event-Triggered Distributed Optimal Control of Nonlinear Multi-agent Systems Based on Integral Reinforcement Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ying Xu,
  • Kewen Li,
  • Yongming Li

摘要

An integral reinforcement learning (IRL) distributed consensus optimal control issue is developed for nonlinear multi-agent systems (MASs) with partially unknown system dynamics. In control design, the IRL algorithm is used to learn the online solution for the Hamilton-Jacobi-Bellman (HJB) equation. Then, combined IRL and actor-critic neural networks (A-C NNs), a new adaptive distributed consensus optimal control approach is designed. To overcome the shortcomings of periodic sampling, a dynamic event-triggered (ET) mechanism is designed. The designed control algorithm ensures that the closed-loop system stabilizes and Zeno behavior can be avoided.