Research On Leader-following Consensus Of Multi-agent Systems Based On Event-triggered Control

Multi-agent systems are topology abstraction of many real systems,consisting of some semi-autonomous or autonomous subsystems with life cycles that act on themselves and environment through network interconnection.The research of multi-agent systems involves a lot of comprehensive theoretical knowledge,which is usually used to analyze and deal with various complicated engineering problems.As a current research focus,consensus problem is the premise for realizing cooperative control of multi-agent systems,which is the concentrated embodiment of autonomy and swarm intelligence through the cooperation between individuals to show the overall behavior beyond the scope of ability.In view of realistic factors such as limited network bandwidth and scarce computing resources,this paper will discuss leader-following consensus problem of multi-agent systems based on event-triggered mechanism to alleviate the communication pressure of the system without affecting the consensus performance.The main contents are outlined as follows:Aiming at the problem that high frequency data transmission can lead to communication congestion and thus affect the stability of system,an event-triggered mechanism which relies on the agent’s own position and velocity information is introduced for multi-agent systems.Through the derivation of the formula,the concrete implementation process of the event-triggered control strategy for studying the consensus problem under a fixed directed topology is illustrated.When the state error satisfies the specific triggering conditions,the following agents transmit and update the data information while executing the control task,without considering the triggering moment of the neighbor nodes.Simulation experiments show the validity of the proposed theory.In order to enhance the convergence speed of consensus,the fixed-time consensus problem of a class of fuzzy multi-agent systems with nonlinear dynamical behavior is discussed.To further decrease the frequency of controller update,a dynamic variable which can be adjusted adaptively is introduced,and combined with new dynamic variable obtained online,the event-triggered condition of the following agents is designed.Nonlinear controller and measurement error are designed by using hyperbolic tangent function to exclude Zeno behavior of the controlled system.Based on mathematical tools such as algebraic graph theory and stability theory,sufficient conditions for multi-agent systems to achieve fixed-time consensus are derived.The simulation results show that the convergence time of fixed-time consensus is independent of the initial state value,which verifies the feasibility of the proposed control method.Aiming at the situation that multi-agent systems are vulnerable to the threat of network attack in open communication environment,the security consensus problem with leader is studied.Considering a recoverable attack that occurs frequently,the communication topology breaks down and control signals cannot be input when the attack occurs.An auxiliary dynamic variable is introduced for each agent,and the event-triggered rules are dynamically adjusted according to state updates of itself and neighbor.In addition,due to the difficulty of obtaining the global information of the system,a distributed dynamic eventtriggered consensus protocol is designed.All agents trigger independently,avoiding continuous information transmission.The correctness of the theoretical analysis is confirmed by numerical simulation.