Consensus For Multi-agent Systems Based On Predictors And The Event-triggered Strategy

As everyone knows,the information is broadcasted between agents through the communication network in multi-agent systems,thus some shortages of the network can become the important factors to affect the performance of multi-agent systems,such as the finite bandwidth,limited energy and communication delays.In view of this reason,this thesis investigates the event-triggered consensus problem and the compensation for unknown communication delays of multi-agent systems.The state predictors of agents are designed for different types of multi-agent systems.Then some different consensus protocols are designed based on the state predictors.In this thesis,the predictive scheme is used to solve the event-triggered consensus problem and the compensation for unknown communication delays of different types of multi-agent systems.Because it is difficult to estimate the control input of agents in continuous-time multi-agent systems,the existing event-triggered consensus protocol which is designed based on the predictive method only use the state of agents at the triggering instants to estimate the current state of agents roughly,but the effection of control input is not taken into account.This kind of predictive method may increase the state estimation error and the triggering times of agents,it can lead to the increase waste of communication resources.For agents in the multi-agent system,the estimation problem of control inputs is transformed into the estimation problem of state differences between agents is this thesis.Both the control input predictor and the state predictor of agents are designed and the accuracy of state estimation is improved.Then an event-triggered consensus protocol is designed based on the state predictors of agents.Furthermore,a necessary and sufficient condition for consensus is obtained and the inexistence of Zeno behavior in the multi-agent system is proved.Compared with two existing event-triggered consensus protocol,it can be known that the event-triggered consensus protocol based on the predictor designed in this thesis can make multi-agent systems achieve consensus by virtue of lower frequency of communication.To the best of my knowledge,there are few works which consider the event-triggered consensus problem of multi-agent systems with directed communication topologies and parameter uncertainties.So the above predictive scheme of state is extended to the multi-agent system with the parameter uncertainty in this thesis.An event-triggered consensus protocol which is appropriate for multi-agent systems with parameter uncertainties is de-signed based on predictors.A necessary and sufficient condition for consensus and a method to design the feedback gain matrix are presented.And the inexistence of Zeno behavior in the multi-agent system is proved for different types of parameter uncertainties.Furthermore,the event-triggered consensus problem of discrete-time multi-agent systems with parameter uncertainties is also investigated.The state predictor for discrete-time multi-agent systems is designed.And an event-triggered consensus protocol is also designed based on the state predictor.Then a necessary and sufficient condition for consensus and a sufficient condition for the existence of the feedback gain matrix are derived.In addition,the definition of Zeno-like behavior for discrete-time multi-agent systems and a sufficient condition for the inexistence of Zeno-like behavior are given.The distributed tracking problem of multi-agent systems is widely-used in the field of control,but the researches on the event-triggered consensus problem of leader-following multi-agent systems,especially on the event-triggered consensus problem of leader-following multi-agent systems with parameter uncertainties are also quite few at present.So this thesis investigates the event-triggered consensus problem of different leader-following multi-agent systems.First,the state predictor of agents is designed for the linear continuous-time leader-following multi-agent system.Then an event-triggered consensus protocol is also designed based on the state predictor.A necessary and sufficient condition for consensus is obtained and the inexistence of Zeno behavior in the leader-following multi-agent system is proved.After that,the above predictive scheme of state for the linear continuous-time leader-following multi-agent systems is extended to the continuous-time leader-following multi-agent system with the parameter uncertainty.And an event-triggered consensus protocol which is appropriate for leader-following multi-agent systems with parameter uncertainties is designed based on predictors.A necessary and sufficient condition for consensus without continuous communication between agents is given.Then a method to design the feedback gain matrix is also presented.And the inexistence of Zeno behavior is proved for different types of parameter uncertainties.Furthermore,the event-triggered consensus problem of discrete-time leader-following multi-agent systems with parameter uncertainties is also investigated.A necessary and sufficient condition for consensus and the method to design the feedback gain matrix are presented.And a sufficient condition for the inexistence of Zeno-like behavior is also given.The consensus problem of discrete-time multi-agent systems with unknown commu-nication delays is considered.The state predictor of agents is designed for the discrete-time multi-agent system based on the predictive scheme,then a consensus protocol based on the predictive scheme is designed.Under the consensus protocol based on the predictor,the unknown communication delays between agents are compensated effectively.Then the consensus problem of discrete-time multi-agent systems with unknown communication delays is transformed into the stability problem of the discrete-time system with multiple delays.A sufficient condition for the consensus is obtained based on the Lyapunov method.Compared with the traditional consensus protocol without the compensation for delays,the consensus protocol based on the predictive scheme can make discrete-time multi-agent systems achieve consensus faster.