Distributed Consensus Research Of Heterogeneous Multi-agent Systems

In the past decade,with the development of communication technology and computer technology,cooperative control of multi-agent systems has been broadly applied in the fields of military,satellite,transportation,medical and so on.There are abundant researches on homogeneous multi-agent systems,but few researches on heterogeneous multi-agent systems.For this reason,this thesis focuses on heterogeneous multi-agent systems.The factors of input saturation,competitivecooperative interaction and convergence time are considered in this thesis and corresponding controllers are designed to achieve consensus.The main contents are listed as follow.For heterogeneous multi-agent systems composed of second-order linear and nonlinear dynamics,the consensus problem is analyzed with and without input satiation.Through applications of graph theory,Lyapunov technique,Lasalle's invariance principle as well as other mathematical methods,a saturated control protocol and an unsaturated control protocol are designed,and the conditions of achieving second order consensus are given for both cases.For heterogeneous multi-agent systems with different control parameters and state dimensions,bipartite output consensus under competitive-cooperative interaction is studied.Based on graph theory,Lyapunov stability theory,output regulation method and other theoretical methods,an adaptive state feedback control protocol that does not rely on global leader information and global communication topology information is designed to achieve bipartite output consensus.After that,an output feedback control protocol is designed to achieve bipartite output consensus.For the bipartite output consensus issue of heterogeneous multi-agent systems subject to external disturbance,two algorithms are proposed.First,through applications of finite-time control approach,output regulation theory as well as other theoretical methods,a finite-time algorithm is presented to achieve bipartite output consensus.But this settling time is related to the initial value.Later,a fixed-time controller is designed for consensus seeking and the obtained settling time is only related to the parameters not to the initial value of the system.