Study On The Consensus Of Nonlinear Multi-Agent Systems Considering Time-Varying Delay

As the computer networks and artificial intelligence develop rapidly,the consensus control of nonlinear multi-agent system has become a key research direction in distributed cooperative control,and has extensive applications in many fields,such as UAVs,mobile robots,smart grid and so on.In practical application,the phenomenon of time delay is almost inevitable,which may have a negative impact on the system performance and even result in losing the consensus of system.Therefore,it is very necessary to consider the time-varying delay in the system model when studying the multi-agent system.In this dissertation,the distributed consensus problem is investigated for several groups of nonlinear heterogeneous multi-agent systems,considering the effects of time-varying state delay and nonuniform time-varying input delay respectively.The main research work is summarized as follows:(1)The distributed consensus problem is studied for a group of nonlinear heterogeneous multi-agent systems where multiple time-varying state delays and unknown nonuniform control coefficients are considered.The nonlinear functions with delayed states are heterogeneous and completely unknown,and they are compensated by constructing some appropriate Lyapunov-Krasovskii functionals.The enhanced Nussbaum function is used to address the problem caused by unknown control coefficients,and remove the constraints that only specific Nussbaum functions can be used in the existing related research results.By incorporating finite-time command filters into backstepping technology,a distributed adaptive control protocol independent of state delay is proposed.The proposed control method is proved to achieve the synchronization of the follower’s output and the leader.(2)The distributed consensus problem is investigated for a group of nonlinear heterogeneous multi-agent systems with nonuniform time-varying input delays,where the agent’s states are restricted by nonuniform asymmetric time-varying functions.Firstly,by utilizing the information of the local and neighbors,the finite time observer is designed to estimate the leader’s information.Next,by introducing a nonlinear coordinate transformation,the multi-agent system with state constraints is transformed into an unconstrained system with new variables,and the requirement that the state constraint boundary function is strictly greater than or less than zero is removed.The time-varying input delay is compensated by developing an auxiliary system,and the conditions on the input delay are relaxed.Finally,by utilizing dynamic surface control,neural network technology and the proposed observer,the appropriate control protocol is developed,which can make the system reach a consensus without the violation of full-state constraints.(3)The self-triggered finite-time consensus problem is studied for a class of nonlinear heterogeneous multi-agent systems with nonuniform time-varying input delays.For the sake of conserving communication resources among agents,a distributed finite-time self-triggered observer is constructed without exhibiting the Zeno behavior.The auxiliary system is modified so that it can better compensate for the nonuniform time-varying input delays.By combining dynamic surface control,neural network techniques and the proposed self-triggered finite-time observer,a finite-time control protocol is developed,which ensure that the system achieves a finite-time consensus only through intermittent communication among the agents.