Research On The Consensus Of Multi-agent Systems In Cooperation Competition Network

In recent years,with the rapid development of artificial intelligence technology and distributed control technology,more and more scholars pay attention to the problem of multi agent-systems.As the basic problem of multi-agent systems coordination control,with the deepening of its research,some related problems,such as asymptotic consensus,grouping consensus and bipartite consensus,are gradually derived.Therefore,the study of bipartite consensus and grouping consensus has more practical significance.However,most of the existing researches on multi agent-systems are based on cooperative networks,the connection weight between agents is positive,which has great limitations and conservatism.Therefore,this paper mainly studies the problems of bisection consensus and group consensus of first-order and second-order multi-agent systems under the cooperation competition network.The main contents include the following aspects:Firstly,the fast bipartite consensus problem of first-order multi-agent systems in cooperation competition network is studied.Based on the first-order system,the control protocol is given.In order to make the system converge quickly,the predictive method is used to design the state predictor.By predicting the future state of the neighboring agents,the individuals in the system can adjust their own state in real time,so that the whole system can achieve the effect of rapid convergence.Compared with the first-order system,the second-order system contains both position information and velocity information,which can accurately describe the moving object.Therefore,the research on the second-order system has more practical significance.The state predictor is added to the bipartite consensus control of the second-order multi-agent systems to make the second-order system converge quickly.Secondly,the group consensus traction control of multi-agent systems in cooperation competition network is studied.Different from the state that agents can only converge to the same size and the opposite direction in bipartite consensus,the group consensus in cooperative competitive networks can converge to the case that the sizes of groups are different.Based on this,for the first-order multi-agent systems,the leader in the system is selected by the traction control method,and the intelligent body in the system converges to the predetermined value group in the lead traction by the way of leader follow.Because the second-order multi-agent systems contains two variables of position and velocity and converges to two states.Therefore,the traction control protocol is designed again.And the proposed traction control protocol can converge the whole system to the expected value and achieve the group consensus under cooperative competition only by controlling some nodes in the system.