Event-based Consensus Of Multi-agent System With Discrete-time

Cooperative control of multi-agent systems(MASs)has been widely used in many fields,including unmanned aircraft control,future independent combat system control and transportation control etc..Consensus is the fundamental and important problem for the cooperative control of MASs.In order to reduce the frequency of communication and consumption of information between agents,the occupancy of computing resource and controller update times,the event-based control method is applied to deal with the consensus of MASs in this thesis.In details,the consensus of discrete-time multi-agent systems with event-based control method is studied for first-order,second-order and general linear multi-agent systems.In the premise of ensuring the stability and consensus of the system,event-based control method tries to minimize the communications and controller update times.The main contributions are as follows:Firstly,the consensus problem of first-order discrete-time multi-agent systems with time delay by an event-based approach is studied.Based on the algebraic graph,matrix theory and stability theory of discrete time dynamic systems,a sufficient condition for achieving consensus is given under the assumption that the interconnection topology among agents has a directed spanning tree.Furthermore,the Zeno-like behavior for discrete-time multi-agent systems is defined and is excluded for the closed-loop system.Then,based on the algebraic graph,matrix theory and stability theory of second order systems with discrete time,event-based consensus of second-order discrete-time MASs is investigated.The event-based controller is designed for each agent,which is based on the state measurement error among neighborhood agents at its own triggering time instants.Some sufficient conditions are obtained for achieving consensus.To avoid observing the current states of its neighbor agents as well as its own state at every discrete-time,a self-triggered approach is presented to determine the triggering time sequence.Furthermore,certain conditions are given to avoid controller update at every discrete time.Finally,the consensus problem of discrete time multi-agent systems with linear dynamic models is investigated.In order to reduce the controller update times and communication frequency among agents,an event-based dynamic feedback scheme isproposed.For each agent,the controller update instants are determined by a state-dependent threshold.It is shown that the controller update for each agent is only dependent on its own event times.Some sufficient conditions are obtained for consensus.Observing the current states of its neighboring agents as well as its own state at every discrete-time is also avoided.Furthermore,certain conditions are given to exclude the Zeno-like behavior of the triggering time sequence.