Study On Output Regulation And Event-triggered Consensus Problem Of Multi-agent Systems

With the rapid development of computer technology and communication technology,multi-agent systems(MASs)have become an important advanced subject of artificial intelligence(AI).Through the exchange of information and cooperative control among the agents,multi-agent systems can complete specific tasks that cannot be accomplished by a single agent.Multi-agent systems have the characteristics of information sharing,easy extension and high reliability.The output regulation and consensus problem are the basis for the cooperative control of multi-agent systems.The output regulation,containment control and consensus problem are considered.The main contributions are summarized as follows:Firstly,the cooperative output regulation of discrete-time linear multi-agent systems is addressed,and a new control law is proposed.Since the state information of the agent is not detectable,a full-order state observer is introduced to provide its own estimated states,the estimated states of its neighbor agents and that of the neighbor net of its neighbor agents for each agent.The system can track the reference input or reject the disturbance faster under the dynamic output feedback control law which is synthesized by the controller and the observer.By using Kronecker product,regulation equation is converted into the form of a non-homogeneous equation AX=B,the sufficient and necessary condition that the tracking error converges to zero is provided by proving that the matrix A is full row rank.Secondlly,for the containment control problem of general linear multi-agent systems,distributed adaptive scheme is used in the thesis.Under directed fixed topologies,a containment control protocol is proposed based on the relative states of neighboring agents.By using matrix theory and Lyapunov techniques,some sufficient conditions guaranteeing containment control are established.Each agent is assigned a time-varying coupling weight such that no global information is required for designing parameters of the controller.The proposed protocol can effectively be implemented in a fully distributed fashion to drive the states of all followers to asymptotically converge to the convex hull formed by those of the leaders.Finally,an event-triggered mechanism is employed to study the tracking control problem of leader-following multi-agent systems.The control input of each agent is triggered by its own states and the states of its neighboring agents.The procedure of the distributed adaptiveevent-triggered strategy is given by Lyapunov method and Linear Matrix Inequality(LMI)toolbox.The follower agents can track the leader agent asymptotically without continuous communication among agents.