Research On Consensus Control Of Multi-agent Systems Based On Iterative Learning

In a multi-agent system,agents can better solve the problems that individuals can't solve through mutual coordination and cooperation.Therefore,the research on cooperative control of multi-agent system has become a hot spot in many fields and is widely used in the fields of UAV,satellite formation,microgrid and so on.For the multi-agent system with periodic motion,the iterative learning control has been gradually applied to the multi-agent system because of its simple structure,low requirement on the accuracy of modeling and good robustness.At present,in the research of multi-agent iterative learning control,the convergence speed of the system is slow,while in the engineering application,the convergence speed of the system is higher.In the actual control process,the communication condition and environmental factors of the agent themselves lead to the change of the system topology,which will affect the stability of the systemFor the multi-agent system under different conditions,in view of the above problems,the consistency control problem of multi-agent system based on iterative learning is studied in this thesis.In the repeated time interval,the designed iterative learning control law can make the system completely track to the ideal state trajectory.The main contents are as follows.(1)Under the fixed topology diagram,to solve the consistency problem of the linear multi-agent system with periodic motion in the interval of repetition time,the iterative learning control method was used to design the multi-agent consistent iterative learning control protocol for the first-order integrator multi-agent system and the general linear multi-agent system.By using the compression mapping method,the convergence of the tracking error is analyzed,and the sufficient conditions for the system to completely track the trajectory of the desired state are obtained,so that the multi-agent system can fully track the leader in the repeated time interval.(2)An iterative learning control algorithm is proposed considering the requirements of system convergence in real engineering applications to solve problem of consensus iterative learning control in multi-agent systems in finite time.The finite time algorithm is used to deal with the tracking errors between the agents and the virtual leader in the previous iteration,and the error convergence rate is improved.Based on this,a new finite time iterative learning law is constructed,and the improved learning law significantly reduces the number of iterations required for the system error convergence.Graph theory and Lyapulov stability theory are used to prove the stability of the learning law in a finite time,and the convergence conditions of the learning law are obtained based on the norm theory.Finally,the validity of the method is verified by numerical simulation results.(3)Under the condition of joint connected topology switching,for the iterative learning consistency control problem of multi-agent systems,a topology switching iterative learning control law is designed.This control law enables the system to solve the consistency problem under the condition that the topology changes with time and some agents cannot obtain the external information.Based on the compression mapping theory,the sufficient conditions for tracking the trajectory to the desired state are given,and the stability of the system is proved by using the graph theory of Lyapunov stability theory.Finally,the effectiveness of the control law is verified by numerical simulation?...