Research On Node-to-Node Consensus Of Multi-agent Networks

At present,the consensus of multi-agent networks is still a hot research topic in the fields of mathematics,physics and engineering.For example,unmanned aerial vehicles,autonomous diving systems,communications,transportation,and power grids are all applications of the consensus issues of multi-agent networks in real life.With the application and development of technology,some scholars have done relevant research on node-to-node consensus.The node-to-node consensus of multi-agent networks is based on the network of the leaders' layer and the followers' layer.Node-to-node consensus issue studies how the corresponding agents in the two-layer network can achieve the common goal and accomplish a certain task together.In this dissertation,event-triggered control and switching control methods are used to study the node-to-node consensus of multi-agent networks,and how to achieve node-to-node consensus in the network environment with packet loss is studied.The main innovations are described as follows:(1)Each layer in the network model is a nonlinear structure and the agents have different nonlinear dynamics.In order to reduce calculation,communication and save resource,the event-triggered control method is used to study the node-to-node consensus of multi-agent networks,and the discrete communication between adjacent nodes is considered.Finally,based on the graph theory and Lyapunov stability theory,the sufficient conditions for node-to-node consensus under event-triggered control are obtained.(2)By using a control method that switches between periodic sampled-data control and event-triggered control,the node-to-node consensus of multi-agent is studied.Consider that the network model is a linear model and the state matrix of each agent is non-identical.According to the Lyapunov stability theory,the sufficient conditions for the node-to-node consensus problem of multi-agent networks are given.Simulation results show that this switching method can avoid Zeno phenomenon and effectively reduce the update frequency of control protocol.At the same time,the triggering time does not need to be at fixed observational moments,then the triggering time is more flexible.(3)There may exist phenomena such as actuator suspension,network traffic congestion,and packet transmission failure in the network environment,these problems may result in packet loss between agents.Aiming at this problem,based on the event-triggered control scheme,using graph theory,matrix theory and constructing the appropriate Lyapunov function,the sufficient conditions for the node-to-node consensus of multi-agent networks via pinning a follower and pinning multiple followers are provided respectively.Finally,it is proved that Zeno phenomenon will not occur,and some numerical simulations are given to illustrate the effectiveness of the theoretical results.