The Consensus Of Multi-agent Network System

The consensus of multi-agent network system is one of the most famous problems in thefield of distributed coordination, and received extensive attention of research scholars in theworld. However, whether multi-agent network can achieve the consensus is affected by manyobjective factors, such as the difference between the model and real structure, the changes ofthe topologies and the environmental noise interference during the communication betweenthe agents. The early work is mainly to control some of the first-order linear system.Therefore, the dynamics of each individual behavior in general is not a constant, whichimplies that the dynamics is time-varying even with nonlinear, meanwhile, the analysis tendto be more complicated. The dynamic information cannot be ignored in the process ofresearch. Based on the predecessors’ work, this paper is concerned with the control strategiesof consensus for nonlinear multi-agent systems, the main contents are as follows:Chapter1mainly introduces the background and the progress in the study of multi-agentnetwork consistency, and our work is briefly introduced.The second, third, fourth chapters are the focus of this paper. In chapter2, theinvestigation on the intrinsic nonlinear multi-agent systems with impulsive control ispresented. Firstly, the consensus of first order nonlinear multi-agent network with impulsiveapproach is considered. Then the consensus of second order nonlinear dynamics with a leaderand switching topologies is investigated. Based on algebraic graph theory, Lyapunov stabilitytheory, sufficient conditions for the convergence to consensus are established. Numericalsimulations are provided to illustrate the theoretical results.In chapter3, the consensus of nonlinear multi-agent network under a novel distributeddelay control is investigated. Considering the impact of the former behaviors of nonlinearagents, we put forward a novel type of integral distributed delay protocol which is formed as aweighted sum of historical information exchange over all time interval [t τ, t].The weightedfunction can be decaying exponential weighted function, can also be a piecewise function.Firstly the consensus of nonlinear multi-agent in directed network with distributed delaycontrol is considered. Then the mean square bounded consensus problem of multi-agentsystems with intrinsic nonlinear dynamics under new protocol is studied, where each agent isaffected by stochastic noises. The notable thing is that the protocol of this chapter not onlydescribe many typical phenomena such as signal transmission among agents, but also explainthat the distant past has less influence compared to the recent behavior of the state, and simplypart traditional protocols with the changes of the weighted function. Many traditionalprotocols can be considered as its special cases. Therefore, the results in this chapter makemuch progress comparing with previous ones and have some research value.In chapter4, the mean square bounded consensus problem for the nonlinear multi-agentsystems is presented to investigate the impact of stochastic disturbances. In practicalapplication, there are different types of delay such as discrete delay and distributed delay,which is studied in previous chapter. In this chapter, we consider the most general case, thetime-varying delay in agreement. By means of the method of stochastic differential equation theory and matrix theory, the consensus of nonlinear dynamics with fixed topology andswitching topology is discussed. Based on a Lyapunov-based approach, some sufficientconditions are given to achieve the mean square bounded consensus. Finally, numericalsimulations are provided to illustrate the theoretical results.In chapter5, the innovation of this paper is summarized, meanwhile, the improvementwork is given.In the end of this dissertation, references are listed.For all the above studies, we use Matlab to program and obtain simulation results. Thesimulation results are in agreement with the theoretical analysis.