Research On Distributed Predictive Control Of Multi-Agent System

With the development of the information age,more and more large-scale network control systems have emerged in actual industrial production processes.Centralized control pattern is clearly no longer able to meet the performance requirements of large-scale network control systems,and the large communication burden has become one of the drawbacks of centralized control.In order to cope with the deficiency of centralized control,distributed control comes into being.The distributed control pattern can achieve better performance by coordinating communication among various sub controllers,but frequent communication problems between subsystems can cause excessive communication burden and affect the real-time performance of control,even leading to data packet loss.Therefore,this thesis studies the predictive control problem of distributed multi-agent system.The main research content is as follows:(1)For a class of multi-agent system,a predictive control strategy based on priority allocation of multi-agent system is proposed.Firstly,graph theory is used to describe the priority allocation and the optimal priority allocation strategy is given based on minimizing system performance index.Secondly,a feasibility and stability analysis are conducted on the priority-based distributed predictive control algorithm.Finally,the simulation results show that the proposed priority-based distributed predictive control method can effectively reduce the number of information interactions between subsystems and reduce network resource utilization while achieving good performance.(2)For a class of multi-agent system,an event triggered distributed predictive control strategy based on priority allocation of multi-agent system is proposed.This strategy enables each agent subsystem in the distributed network topology to only exchange information with the agent subsystem with higher priority than itself at the event triggered time,which can significantly reduce computational and communication consumption.Secondly,theoretical proof is provided to ensure the feasibility and closed-loop stability of this control strategy.Finally,the proposed method can be validated by a numerical simulation,which indicates the priority-based distributed event-triggered model predictive control system can effectively reduce communication pressure between subsystems while ensuring system performance index.(3)For a class of uncertain multi-agent system,a distributed robust predictive controller design method based on priority assignment of multi-agent system is proposed.Firstly,according to the complex and changeable real systems in industry,the uncertain system models for a class of multi-agent systems are considered,and the definition of coupling constraints between agents is given.Secondly,based on the above system,a priority-based distributed robust controller for multi-agent system is designed by transforming the optimization problem of robust predictive control with infinite-time domain into the calculation problem of linear matrix inequality.Finally,the effectiveness of the proposed algorithm was verified through numerical simulation.