The Problem Research About Distributed Predictive Control Based On Agent Coordination

With the rapid development of computer network, control is no longer limited to centralized control and replaced by distributed control. Predictive control as an advanced process control method has been widely used in various industrial processes. There may exists a large number of complex high-dimensional large-scale systems in actual system, but also limited by the device scale, geographically dispersed, communication bandwidth. So we can dispersed the online problems of a complex large-scale system to each subsystem, each subsystem complete their specific function of local control and work. At the same time, each subsystem accomplishes the task of the entire system through the communication to each other to exchange information and share resources to accomplish the task of the entire system.In this paper, a distributed model predictive control (DMPC) strategy is proposed based on agent coordination. At first, the initial feasible solution of each agent can be achieved by solving local optimization problems in which the state constraints of neighbor subsystems are considered at each sampling time. In the negotiating process, the innovative global optimization objective is determined for the sake of reducing iteration time and improving the convergence speed efficiently.In order to ensure the stability of the predictive control system, this paper design an iterative approximation method to obtain the convex sequence of maximal terminal set of states, when the number of iterations tends to infinity step, this set converge to the maximal terminal set of states, the maximal domain of attraction. In the meantime construct the terminal weighting matrix in order to guarantee the stability of the local subsystem. In order to guarantee the stability of the closed-loop system then use the maximal terminal set of states and the terminal weighting matrix to distributed predictive control strategy based on multi-agent coordination. Finally, the accuracy and efficiency of the proposed scheme is put to test through simulation.