欢迎访问《兵工学报》官方网站,今天是 分享到:

兵工学报 ›› 2023, Vol. 44 ›› Issue (S2): 223-234.doi: 10.12382/bgxb.2023.0664

所属专题: 群体协同与自主技术

• • 上一篇    

动态事件触发下一般线性多智能体系统完全分布式一致性控制

张继雄1,2, 李宗刚1,2,*(), 宁小刚1,2, 陈引娟1,2   

  1. 1 兰州交通大学 机电工程学院, 甘肃 兰州 730070
    2 兰州交通大学 机器人研究所, 甘肃 兰州 730070
  • 收稿日期:2023-07-15 上线日期:2024-01-10
  • 通讯作者:
    * 邮箱:
  • 基金资助:
    国家自然科学基金项目(61663020); 甘肃省高等学校产业支撑计划项目(2022CYZC-33)

Fully Distributed Consensus Control of General Linear Multi-agent System Based on Dynamic Event-trigger

ZHANG Jixiong1,2, LI Zonggang1,2,*(), NING Xiaogang1,2, CHEN Yinjuan1,2   

  1. 1 School of Mechanical and Electrical Engineering, Lanzhou Jiaotong University, Lanzhou 730070, Gansu, China
    2 Robotics Institute, Lanzhou Jiaotong University, Lanzhou 730070, Gansu, China
  • Received:2023-07-15 Online:2024-01-10

摘要:

针对事件触发下一般线性多智能体系统一致性问题研究中,现有算法多采用静态触发方式、一致性协议设计中存在全局信息、无法实现完全分布式控制的问题,提出一种基于动态事件触发的一致性协议。通过引入时变耦合权重自适应地调整图的标量增益,实现一致性问题的完全分布式求解。针对事件触发控制中需要持续检测触发条件的问题,引入基于个体的状态估计器,用于在触发间隔期内对自身状态进行估计,减少了智能体间的通信次数。基于所选择的Lyapunov函数,设计一种耦合系统状态估计值、测量误差和自适应参数的动态触发函数,显著地减少了系统的触发次数、降低了系统的能耗。数值计算表明所提一致性协议能够实现渐近一致性,且不发生Zeno行为。仿真结果表明所提协议能有效降低系统能耗、节省通信资源。

关键词: 多智能体系统, 一致性问题, 一般线性系统, 动态事件触发, 完全分布式算法

Abstract:

For the consensus control problem of general linear multi-agent system under event-trigger, the existing algorithms mostly adopt static triggering method, and there is global information in the design of consensus protocol, which can not realize fully distributed control. A consensus protocol based on dynamic event-triggering is proposed to solve this problem. The scalar gain of the graph is adaptively adjusted to realize the fully distributed solution of consensus problem by introducing the time-varying coupling weight. For the problem of continuously detecting the trigger conditions in event-triggered control, an individual-based state estimator is introduced to estimate its own state during the trigger interval, which reduces the number of communications between agents. Based on the selected Lyapunov function, a dynamic trigger function is designed to couple the state estimation value, measurement error and adaptive parameters of the system, which significantly reduces the number of triggers and the energy consumption of the system. Numerically calculated results show that the proposed consensus protocol can achieve asymptotic consensus without Zeno behavior. Simulated results show that the proposed protocol can effectively reduce system energy consumption and save communication resources.

Key words: multi-agent system, consensus problem, general linear system, dynamic event-trigger, fully distributed algorithm

中图分类号: