Distributed Observers Based Consensus Control Of Multiagent Systems Over Clustered Networks

Research on consensus control of multi-agent systems(MASs)is currently at the forefront of cluster intelligent technology,and it has important application value in national economy and national defense construction.The problem of consensus control in MASs refers to a group of autonomous agents achieving a common task through local information sharing and distributed control strategies.However,due to the difficulty of directly obtaining the state information of agents in practical systems,it is necessary to utilize distributed observer technology to estimate and observe the local states,and then design corresponding distributed control strategies to achieve the consensus of the state or output of MASs.This dissertation aims to investigate several key issues in distributed observer-based consensus control,including communication delays,cluster network topology,dynamics of agents,hybrid control strategies,and event-triggered mechanisms:1.Consensus control based on distributed observers with communication delays is investigated.Assuming that the state of the agents cannot be fully measured,the reducedorder observers are used to estimate the unmeasured states of each agent.Three types of communication delays are considered in the network:homogeneous delays,heterogeneous delays,and event-triggered communication.For each scenario,a distributed control algorithm is proposed to ensure that all agents achieve state consensus or output consensus.By using methods such as Jordan decomposition,topological decomposition,Zeno behavior analysis,and joint inequalities,the consensus achievement process of closedloop MASs under different communication delay scenarios are analyzed and the upper bound of delays to guarantee consensus achievement are obtained.2.The output consensus control of homogeneous MASs based on distributed reset observers in cluster networks is studied.A graph-theoretic description of cluster networks is developed and the algebraic properties of cluster networks are analysed.Considering the cluster network has a hybrid communication mechanism and assuming that the states of the agents cannot be fully measured,the reset full-order observers and reduced-order observers are proposed to estimate the unmeasured states of each agent,and design corresponding distributed output feedback control strategies.The common Lyapunov function method is used to analyze the global output consensus achievement of the closed-loop MASs.3.The output consensus control of heterogeneous MASs based on distributed reset observers in cluster networks is studied.To address the problem of output consensus in heterogeneous MASs,an internal reference model is established for each agent.Considering that the states of the agents cannot be fully measured,a reset reduced-order observer is designed for hybrid communication to estimate the unmeasured states of the agents.Based on this,an output feedback control method based on the agent’s own output,observer states,and internal model states is proposed.The exponential stability of the hybrid closed-loop system is analyzed by using a common Lyapunov function method,and obtain sufficient conditions for output consensus of heterogeneous MASs on cluster networks.4.The output consensus control based on event-triggered impulsive observers in cluster networks is studied.In order to avoid excessive resource consumption caused by real-time updates of the agent’s controller,the agents in the cluster adopt an eventtriggered mechanism for communication,and inter-cluster communication is performed only by the leader at the time of event triggering.When only output information is available,heterogeneous reduced-order observers are designed for the leader and followers to estimate their unmeasured states.Furthermore,an event-triggered hybrid control strategy based on the reduced-order observer is proposed for each agent.The stability of the hybrid closed-loop MAS is analyzed using a common Lyapunov function method,and sufficient conditions for achieving output consensus are obtained.Additionally,the Zeno behavior of the event-triggered mechanism is analyzed to ensure the feasibility of the event-triggered hybrid control strategy.