Consensus And Multi-consensus Evolution Of Complex Multi-agent Systems

As one of the emerging scientific research fields within past decades,the complex network has received more and more attentions from all around the world.In general,the research related to complex networks can be roughly divided into two research directions: the one is developed from the micro perspective and the other is expended from the macro perspective.The first one mainly studies the dynamic relationship between each node and network characteristics.On the contrary,this second one discusses the general and dynamic/static characteristics of the complex network based on the graph theory,probability and statistics theory.In the past half century,researchers have studied the micro and macro characteristics of the complex network deeply and have obtained many meaningful and effective results by means of the algebraic graph theory,the probability theory,the control theory,the matrix theory and related knowledge.In this thesis,the relationship between nodes and network characteristics in view of network vulnerability of complex network is studied.As well as the consensus and multi-consensus evolution control and anti-control problems of complex multi-agent systems are further discussed from the micro perspective.The research contents of this thesis are summarized as follows:1.Vulnerability analysis algorithm design of the complex network.Considering that nodes of the network are affected by its direct neighbors and indirect neighbors,an analytical algorithm for evaluating the vulnerability of complex network is proposed based on analytic hierarchy process and fuzzy discriminant method.In this algorithm,multiple factors which influence complex network topology,such as nodes,edges,direct neighbors and indirect neighbors,etc.,are considered comprehensively,and the comprehensive weight vector of nodes is established.Then,the calculation method which can computes the influence of network capability existed multi-node parallel failures is constructed to assess the vulnerability of the complex network.Finally,the effectiveness of the proposed algorithm is verified by an example about the high-speed railway network in China where the network capability is influenced by each important node and node groups.2.Control and anti-control of multi-consensus and consensus evolution.An impulsive bounded consensus and multi-consensus interaction evolution protocol with adjustable intelligence degree is designed by using the location and velocity sampling information of each agent.This protocol can achieve the consensus and multi-consensus interaction evolution only adjusting intelligent degree of each agent instead of changing the communication topology of the multi-agent network.Moreover,the sufficient and necessary conditions of the leader-following bounded consensus tracking are given by using the Hurwitz stability theory and the properties of Laplace matrix.Finally,a multi-agent tracking problem is given as an example to show the effectiveness of the proposed impulsive control protocol and also obtain better obstacle avoidance performance.3.Coupling behavior analysis of multi-agent system under opinion evolution.In order to study the influence of the dynamic behavior of the multi-agent systems with limited intelligence degree when opinion of agent changes dynamically as opinion propagating among agents,the coupling relationship between multi-agent dynamic behavior and opinion evolution governed by conformity rules are discussed by applying with the consensus theory of multi-agent system and the improved SIS model.Then the static and the dynamic consensus protocols based on opinion evolution are designed respectively.Moreover,the sufficient and necessary consensus conditions for the multi-agent system are obtained by using the Hurwitz stability theory,and the threshold value of the improved SIS(Susceptible-Infected-Susceptible)propagation model is also obtained.Finally,a conclusion that the evolution of opinions only affects the dynamic behavior of the multi-agent system,but does not affect the stability conditions of the multi-agent systems is deviated.At the end of this thesis,some research directions which can be paid more attentions in the near future are summarized after summarizing the whole thesis and prospecting cooperative control of the multi-agent systems.