Research On Communication Strategy Of Cooperative Distributed Model Predictive Control Algorithm

Distributed control is an ideal control method to balance the performance of centralized control and the safety of decentralized control in the control of large-scale systems.Because of the good ability to deal with constraints and to predict the future behavior of the system,model predictive control(MPC)is widely used in distributed control,called distributed model predictive control(DMPC).Cooperative distributed model predictive control algorithm can achieve the Pareto optimality with the global optimization range But the realization of cooperative DMPC needs a high communication burden,especially when the amount of subsystem is large,the communication burden will be extremely high,which hinder the development of cooperative DMPC in the practical application.For cooperative DMPC,the paper tries to reduce the communication burden,the main research includes:(1)A DMPC algorithm based on hierarchy decomposition is proposed.Firstly,according to the structure of communication system,the system is grouped into several connected sets using the interpretation model method and loop decomposition algorithm.Then the hierarchy the connected set is defined.Finally,subsystems solve the optimization problem to optimal inputs in a sequence.The subsystem only needs to communicate with the subsystems inside the connected set,and does not need to communicate with the subsystems outside the connected set,which reduces the communication burden.(2)A DMPC algorithm based on event trigger is proposed.The algorithm introduces 0-1 variables which indicate whether the coupling between two subsystems is needed or not,and constructs a dynamic coupling model.A multi objectives optimization problem which balances the control performance and communication burden is constructed,the optimization of coupling structure will change the original coupling structure,and the communication burden can be reduced by ignoring the coupling between subsystems,for which communication is not needed anymore.In order to reduce computational consumption,considering the stability of the system,the algorithm constructs an event trigger,when the trigger is triggered,then solve an optimization problem contains coupling structure decision and control performance to update the coupling structure,otherwise,the coupling structure remains unchanged,then solve the optimization control to calculate the optimal input.Compared with the DMPC algorithm based on the dynamic coupling model,the proposed algorithm reduces the computational complexity.