Research On Networked Distributed Model Predictive Control Strategy For Chemical Process

The enlargement of the process industry is an effective way for modern industry to realize energy saving,consumption reduction,emission reduction,and economical production costs.Because of its revolutionary significance,it is getting more and more attention from the engineering session.The industrial process consisting of Reactor-Storage Tank-Separator(RSS)is a Large Scale Industrial Process(LSIP)with various of complex factors.The core of LSIP is to realize the optimization of the entire process.The key point to LSIP's optimization problem is to make the process industry get the most benefit.For the RSS process,the centralized Model Predictive Control(MPC)strategy cannot cope with the increase in the dimension of the variable,and thus resulting in poor real-time,flexibility and fault tolerance of the system.The Networked Distributed Model Predictive Control(DMPC)strategy transforms large-scale online optimization control problems into small-scale distributed optimization control strategys of multiple subsystems.At the same time,multiple subsystems communicate and share information through networks.Optimize the operation of the entire system in a coordinated manner,improve the overall control performance of the system,and maximize the benefits of process industrial production.This article focuses on the Networked Distributed Model Predictive Control strategy that combines optimization and control as below.Illustrate the significance and research status of optimization control strategies for process industries,the basic principles of MPC and DMPC strategies,and elaborate the control effects and economic benefits of optimization control technology strategies for process industries briefly.Model description of industrial process RSS process.Based on the process flow,and dynamic differential equation of RSS process,establish its mathematical model and dynamic response in MATLAB/Simulink;Devide the RSS procecc into subsystems according to process flow and geographic location distribution of RSS process,and perform linearization and dynamic performance testing.Research on DMPC Strategy of RSS Process.Aiming at the problem of real-time,rapidity and flexibility in the effect of RSS process control,the centralized MPC strategy proposes the DMPC control strategy of RSS process,that is,the centralized optimization control problem of RSS process is transformed into the distribution of multiple subsystems.The simulation results show that compared with the centralized MPC control strategy,the proposed DMPC control strategy has the advantages of faster response,smaller overshoot,and stronger stability when adjusting the RSS process.Research on NC-DMPC strategy for RSS process.Aiming at the problem of operational flexibility and poor real-time performance in the traditional information transmission mode of RSS process,a distributed predictive control algorithm based on network information transfer is proposed.Considering the mutual coupling between subsystems,several sub-model predictive controllers are proposed to perform coordinated global performance optimization.Simulation results show that the adopted NC-DMPC strategy improves the operational flexibility and real-time performance when adjusting the RSS process.Research on NDMPC Strategy of RSS Process.Combining the characteristics of communication networks and the stability mechanism of distributed networks,an NDMPC strategy based on RSS process is proposed.Considering the network predictive control system,the Control Prediction Generator(CPG)and Network Delay Compensator(NDC)are used to optimize the time-delay and data packet loss problems of the RSS process under the NDMPC strategy.The simulation results show that the NDMPC strategy adopted in this paper can effectively compensate for the drop of control performance caused by network time delay and data packet loss.