Ressearch Of Cooperative Control For Multi-Agent System

Cooperative control is the basis of the research of multi-agent system. Distributed control is also an important branch of cooperative control whose aim is to decompose a multi-input-multi-output (MIMO) system into several single-input-single-output (SISO) systems. By communicating with each other, all SISO systems reach consensus. In recent research, the stable region of low-order for cooperative control has not been solved very well. Thus, in this paper, the research of cooperative control for multi-agent system is presented.Multi-agent system consists of several subsystems with multiple delays. The stabilization of each subsystem is the basis of the research of cooperative control. Thus, firstly for a given transfer function of single plant, the design method of P controller for a single plant with multi-delays is proposed and then for a further research, it is extended to PI and PID controller.Secondly, on the basis of the method to stabilize the plant, distributed control strategy is presented in this paper. Transform the consensus problem into stability problem of the whole system. Then according to control theory, decompose the whole system into several subsystems based on the eigenvalues of Laplacian matrix. Thus, the stability problem of the whole system has been changed into the stability problem which may reduce the complexity of the system. By computing the intersection of stable region of each subsystem, the stable region of the whole system can be achieved. While the distributed control method may not work when subsystems are not stable. So by extending the distributed control method the two-freedom-of-degree controller is designed which consists of individual controller and coupling controller. By the cooperative control of two kinds of controllers, the multi-agent system can reach consensus under all of its agents are stable with time moving.Finally, we verity the algorithms mentioned above through smart car plane. Smart car plane confirm the algorithms, which improve the performance of multi-agent system.