基于动态事件驱动的多智能体系统预测控制  

Dynamic event-triggered-based predictive control of multi-agent systems

作　　者：路平立[1] 骆文城 杜长坤 LU Ping-li;LUO Wen-cheng;DU Chang-kun(School of Automation,Beijing Institute of Technology,Beijing 100081,China;Intelligent Science&Technology Academy Limited of CASIC,Beijing 100143,China)

机构地区：[1]北京理工大学自动化学院,北京100081 [2]航天科工集团智能科技研究院有限公司,北京100143

出　　处：《控制与决策》2024年第12期3981-3988,共8页Control and Decision

基　　金：国家自然科学基金项目(61433003,60904003,11602019);国防科技重点实验室基金项目(6142208200308)。

摘　　要：针对一类具有时变时延和非周期性拒绝服务(denial-of-service, DoS)攻击的离散时间线性多智能体系统,提出一种基于动态事件驱动的预测控制方案.通过引入网络化预测控制方法,该方案可有效降低时变时延和非周期性DoS攻击对系统性能造成的负面影响.此外,各智能体的传感器侧均设计动态事件驱动控制机制,以减少各智能体反向通道的数据传输频率,从而节约系统的网络资源消耗.随后给出闭环多智能体系统渐近稳定且输出实现一致的充要条件.最后通过数值仿真结果验证所提出方案的有效性和正确性.For a class of discrete-time linear multi-agent systems subject to time-varying delays and aperiodic denial-of-service(DoS)attacks,a dynamic event-triggered-based predictive control scheme is proposed.The proposed scheme can effectively reduce the adverse impacts of time-varying delays and aperiodic DoS attacks on system performance by introducing the networked predictive control approach.Moreover,the dynamic event-triggered control mechanism is designed on the sensor side of each agent to reduce the data transmission frequency in the feedback channel of each agent,thereby saving the network resource consumption of the system.Subsequently,the sufficient and necessary conditions to guarantee asymptotic stability and to achieve output consensus of closed-loop multi-agent system are given.Finally,the effectiveness and correctness of the proposed scheme are verified by the numerical simulation results.

关 键 词：多智能体系统 事件驱动控制 预测控制 时延 拒绝服务攻击 渐近稳定 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]