Resilient Group Consensus Of Multi-Agent Systems Under Network Attacks Of XIDIAN UNIVERSITY

With the popularization and application of intelligent technology,distributed multi-agent systems have received extensive attention from scholars in various fields due to their characteristics of autonomy,scalability and coordination.The consensus of distributed multi-agent systems is the most basic and important problem has been studied by many scholars,and there are countless results.However,with today’s increasingly complex network environment,the problem of the network security cannot be ignored.Therefore,scholars begin to study the resilient consensus of the multi-agent system.It is worth noting that in many application scenarios,many interconnected components in the network are growing increasingly more complex,and the number of consensus state values may vary depending on the task,workspace or time.Thus,agents in a multi-agent system eventually converge to multiple consensus states with respect to multiple quantities of interest for some reason.Therefore,considering the existence of malicious agents in the network,the resilient group consensus of second-order continuous-time multi-agent systems and the resilient group consensus of heterogeneous multi-agent systems under impulse control are studied by using mathematical tools such as graph theory.This thesis is mainly as follows:Firstly,considering the communication delay between agents in the network,the resilient group consensus of second-order continuous-time multi-agent systems under synchronous impulse control is studied.Assuming that the upper limit of the number of malicious agents in the neighbors of normal agents in the network is known,based on the idea of deleting extreme values in neighbor information,a synchronous impulse resilient group consensus algorithm based only on position information is designed.When the communication topology satisfies certain robustness conditions,it is demonstrated that the normal agents of the multi-agent system can finally achieve resilient group consensus,and the effectiveness of the theoretical results is verified by simulation examples.Secondly,considering that in the synchronous impulse resilient group consensus algorithm,all normal agents need to have the same update time,which is difficult to apply in real life.Therefore,this thesis further designs a resilient group consensus algorithm based only on position information under asynchronous impulse control,in which the state update time of each normal agent is independent from other agents.When the communication topology satisfies certain robustness conditions,it is demonstrated that the normal agents of the multi-agent system can finally achieve resilient group consensus,and the effectiveness of the theoretical results is verified by simulation examples.Finally,considering the diversity of agents in the groups,the resilient group consensus of heterogeneous multi-agent systems composed of first-and second-order agents is studied.A resilient group consensus algorithm under impulse control is designed for normal agents.When the communication topology satisfies certain robustness conditions,it is demonstrated that the normal agents of the multi-agent system can finally achieve resilient group consensus,and the effectiveness of the theoretical results is verified by simulation examples.