Point-to-Point Iterative Learning Control And Performance Optimization Of Multi-Agent Systems

In Multi-agent Systems(MAS),agents can accomplish complex tasks that cannot be completed by a single agent through coordination and cooperation among agents.Therefore,the coordination and control of MAS have been applied to many fields,such as mobile robots,sensor networks,and satellite formations.Consensus problem is a fundamental issue in MAS control,which refers to the use of local information by agents to achieve global control and ultimately make the state variables of the agents tend to consensus.The assumptions for this problem usually include the stability of agents,the fixed topology of the network,and the reliability of network communication.For multi-agent systems with periodic motion characteristics,Iterative Learning Control(ILC)has gradually been applied to multi-agent systems performing repetitive tracking tasks due to its inherent repetition mechanism.By using the input-output information of the previous trials of the system,it continuously corrects and updates the current trail of control input signals,and can finally achieve complete tracking of the reference trajectory in a finite time.In practical industrial processes,it is often not necessary to track the complete reference trajectory,but only to track the given target value at specific times.This tracking control problem is called point-to-point tracking problem.Since point-to-point tracking tasks relax the requirements for tracking,it greatly improves the freedom of design.Therefore,this paper studies the point-to-point iterative learning control of multi-agent systems and its performance optimization while ensuring tracking accuracy and improving the overall performance of the system.The main research contents and innovations include:1.A distributed point-to-point iterative learning control method is designed for linear discrete multi-agent systems with periodic motion characteristics under a fixed topology structure,and its sufficient and necessary condition for convergence is provided.Due to the existence of factors such as equipment aging and measurement deviation in practical applications,there are deviations between the actual system model and the nominal system model.Therefore,the robustness analysis under model uncertainty is further discussed.Finally,the effectiveness and rationality of the proposed method are verified through simulation results.2.A centralized norm-optimal point-to-point iterative learning control method is proposed by combining optimization theory with iterative learning control for the point-to-point tracking problem of multi-agent systems.Its sufficient and necessary condition for convergence is provided,and the robustness analysis of the system under model uncertainty is further discussed.As the proposed algorithm is a centralized control form,there are many restrictions.To address this,the Alternating Direction Method of Multipliers(ADMM)is used to achieve a distributed solution for centralized control.On this basis,the robustness of the approach is analyzed.Finally,the effectiveness and rationality of the proposed method are verified through simulation results.3.For the point-to-point iterative learning control problem of multi-agent systems,when the tracking time instants in the point-to-point tracking task are different,the system output trajectory and the corresponding system input energy are also different.Therefore,by treating tracking time instants as variables and using coordinate descent method,a point-to-point iterative learning control method with optimized tracking time instants is designed,which searches for the optimal tracking time instants iteratively in order to further reduce the energy loss of the system.Its sufficient condition for convergence is provided,and the robustness analysis of the system under model uncertainty is further discussed.Finally,the effectiveness and rationality of the proposed method are verified through simulation results.