Consensus Of Multi-Agent Systems Based On Sampled-Data Control And Event-Triggered Mechanism

In recent decades,cooperative control of multi-agent systems has wide applications in different fields such as biology,economy,industry and so on.As one typical problem of cooperative control,consensus is devoted to a large amount of attention.Consensus means that all agents converge to a common state by sharing information among neighbors.Generally,continuous-time communication is commonly used to deal with consensus problems of multi-agent systems,i.e.,information among agents is transmitted all the time.Obviously,it is impractical to apply continuous-time communication to real-world systems.Because traditional analog controllers are replaced by the digital ones and information is broadcast discretely,sampling schemes are introduced into the consensus protocol design of multi-agent systems.However,traditional periodic sampling communication over a digital network may cause wastes of energy,bandwidth and communication resources,especially for the wireless networks whose devices are often battery-operated.Thus,energy efficiency need to be considered crucially.Event-triggered communication is an effective way to improve control efficiency.The general research of event-triggered communication focuses on precisely determining when the communication events should occur and when the controllers should be updated.In recent years,event-triggered communication as an alternative of continuous or traditional periodic communication has received major attention in the field of multiagent systems.Combined with sampling communication scheme,this dissertation studies the event-triggered consensus problems for multi-agent systems.The main contributions of this dissertation are summarized as follows.Firstly,for leaderless and leader-following consensus of second-order multi-agent systems,asynchronous periodic sampling consensus protocols are proposed with event-triggered algorithm.For the leaderless case,stochastic matrix theory is employed to analyze the static consensus of the closed-loop multi-agent systems.By choosing parameters of the proposed consensus control protocol appropriately,the states of all agents achieve consensus asymptotically.For the leader-following case,if the topology contains a directed spanning tree,we reach two conclusions:(1)The positions of the followers converge to a nonzero constant and the velocities of the followers converge to zero,when the velocity of the leader is zero;(2)The errors of the positions and velocities between the followers and the virtual leader converge to a small region containing 0,respectively,when the velocity of the leader is a nonzero constant or time-varying.Secondly,the observer-based leader-following output consensus problem is investigated for high-order nonlinear multi-agent systems with a distributed event-triggered mechanism.The system dynamics are described as a strict-feedback form and satisfy Lipschitz condition.Three control protocols are considered to solve the output consensus problem.First,a distributed event-triggered strategy is developed by using dynamic output-feedback control method and we achieve that the inter-event times are lower-bounded by a positive constant,that is,the Zeno behaviour is excluded.Second,an event-triggered output consensus protocol is proposed with a sampled-based control scheme.Third,when the states of the multi-agent systems are not measured,we design a consensus protocol for the multi-agent systems with sampled-data control and event-triggered mechanism.By using the Lyapunov stability theorem,it is proved that all the output tracking errors converge to zero exponentially for the three cases by choosing the design parameters properly.Finally,the cooperative global robust output consensus problem based on event-triggered control is considered for a class of heterogeneous nonlinear multi-agent systems with external disturbances.Two cases are considered under directed topology:(1)For the case that each agent is required to know the system matrix of the leader system,we design a novel distributed dynamic compensator and a distributed event-triggered algorithm.(2)For the case that the matrix of the leader system is only known by its neighbors,a distributed observer is designed to estimate both the state and the system matrix of the leader system.Also,a distributed event-triggered algorithm is proposed.For these two results,we get that the output tracking errors converge to zero with exponential rate and the states of all agents need not to be triggered synchronously,that is,each agent whether broadcasts its states to its neighboring agents or not is only dependent on the state information about itself rather than its neighbors.It is further shown that with the unified frame work of the distributed eventtriggered strategy,all the inter-event times are guaranteed to have a positive lower bound and the Zeno behaviour is thus excluded.