The Design And Implementation Of Consensus Control Of Multi-agent System Based On Event-trigger Mechanism

Multi-agent system, has been widely used in many fields, is a hot direction in the field of control theory. Consensus problem is a fundamental problem of multi-agent system. All the improvements in consensus problem have an important effect on multiagent system in formation, flocking, tracking and so on. ”Consensus” refers to the states which are relevant to the system tend to be the same, while ”consensus problem” is the rule of interaction and information communication between different agents. Theoretical basis of consistency has been developed to be relatively complete, but all of these studies are in continuous time or in the discrete time. In the practical application, the excessive usage of communication will cost 70% of the energy for sensors. In order to reduce consumption of energy and occupancy of computing power, this thesis will introduce an event-triggered mechanism to improve the existing consensus protocol – in the premise of ensuring the stability and consensus of the system, it tries to minimize the computing and communications. The main contents are given as follows:1. We introduce an event-trigger mechanism and design a first-order system eventtrigger control protocol on the basis of the first-order multi-agent system consensus protocol. A triggering function of event-trigger mechanism ensuring the stability and consensus of first-order system is designed by using Lyapunov stability. By calculating the theoretical shortest trigger interval, it proves Zeno behavior does not exist in the system, and the stability and consistency is ensured to be achieved. Through MATLAB simulation, the system is verified to demonstrate consensus under the proposed protocol.Compared to the original continuous control protocol, event-triggered protocol reduces communications of the system during reaching consensus.2. We design the event-trigger control protocol of second-order multi-agent system without leader. Through matrix transforming, the system is divided into two subsystems.We design the triggering function ensuring the stability and consensus of second-order system by using Lyapunov stability in subsystems. Similarly, it proves Zeno behavior does not exist in the system under the control protocol. Through MATLAB simulation,the system is verified to reach consensus under this control protocol. Compared with the original continuous protocol, event-triggered protocol reduces communications of the system during reaching consensus.3. We design the event-trigger control protocol of the second-order leader-follower system. By using variable trasformation and Lyapunov stability, the triggering function is designed. By scaling, the minimum interval of communication is acquired and it proves Zeno behavior does not exist in the system. The system consensus is verified to be finally reached under the protocol. Compared to the continuous-time control, event-triggered protocol reduces the communications of the system.4. We design and develop a multi-agent system simulation program. The program implements inputing the initial conditions of the multi-agent system simulation by widget, simulation, dynamic displaying simulation results and showing important data. In addition, this program can extend the multi-agent algorithms through some simple steps to adapt to different situations. Programs can run on 64-bit Windows systems.