Event-triggered Consensus For Nonlinear Multi-agent Systems

The consensus problem of multi-agent systems has attracted much attention due to a wide range of applications,such as satellite attitude alignment,distributed optimization,and distributed parameter estimation.As a basic problem of multi-agent collaborative control,consensus means that a group of agents achieve the same state by interacting with neighbors.In the traditional consensus control setting,it is assumed that the multi-agent system can obtain continuous communication and control signals,but this process consumes a large amount of communication resources,requiring agents to have sufficient computing resources and support for an ideal communication environment.Obviously,this is unrealistic in some practical applications,especially in the case of communication bandwidth and channel limitations.The introduction of event-triggered control provides a positive solution to the above problems.For multi-agent systems with nonlinear and uncertain parameters,how to design a suitable event trigger mechanism to avoid continuous communication between multiple agents is studied.The main research contents and results are as follows:A new event triggered mechanism is introduced into consensus control algorithm for uncertain multi-agent systems.Different from the traditional event trigger control,this algorithm introduces an event-triggered condition to the reference model of the system.The event-triggered controller does not rely on continuous communication among agents.For uncertain parameters,it is solved by an adaptive algorithm.A state-dependent trigger condition under an undirected graph is designed,where each agent only needs to monitor its own state continuously to determine when to trigger an event and broadcast its states to its neighbors.Moreover,a state-independent trigger condition under a directed graph is designed.Compared to the state-dependent trigger condition,the state-independent trigger condition has a simpler form and has lower requirements on the system topology.The time each agent broadcasts its state to a neighbor depends only on changes in its own state,not on the state of its neighbors.Since the threshold is completely independent of the system state,the sampling behavior does not clearly reflect the nature of the system dynamics.Another challenge to introduce event triggering mechanisms for consensus algorithms is that designed algorithms not only ensure that the agent can achieve consensus,but also ensure that the event trigger interval is strictly greater than zero(i.e.there is no Zeno behavior).The original intention of the design is to reduce the communication frequency between the agents.If this is not guaranteed,the proposed design method will have no meaning.The correctness of the algorithm was verified by Matlab simulation,and the advantages of the event-triggered controller proposed in this article are illustrated.