Research And Application Of Networked Distributed Model Predictive Control

Upsizing and streamlining of industrial processes can effectively save costs in the production process.Traditional control method is difficult to meet the control requirements of the LSIP because it cannot solve the problems of nonlinearity,coupling and constraints,Choosing appropriate advanced control strategies is the key to realize optimal operation and safe operation of LSIP.Networked distributed control provides an opportunity for the optimal operation of LSIP.The method based on networked distributed control brings new opportunity to solve the optimization operation of LSIP,but when controlling LSIP,the high dimensionality brought by increased number of subsystems,the coupling of dynamic behavior among subsystems and the time lag and packet loss caused by the network bring challenges for controlling LSIP.Network control technology combine with distributed predictive control of subsystem collaborative optimization can effectively improve the control flexibility of LSIP and achieve the goal of optimizing system performance and reducing costs.By learning the distributed model predictive control strategy and its development process in recent year,the article studies the networked distributed model predict ive control(NDMPC)of LSIP and its application in Reactor-Storage Tank-Separator(RSS)process.The main work is as follows:1)Introduce the research of NDMPC and illustrate the prospects in the industrial process.Describe the RSS process and analyze the i mpact of different input values on the output variables.2)Research on NDMPC strategy of RSS process.Considering that different subsystems will be affected by neighborhood systems,the networked distributed model predictive control(NDMPC)exchanges state i nformation among sensors,controllers and actuators through communication network to improve the flexibility and effectiveness,and reduce error between set point.The simulation results show that when considering the influence of neighborhood system,the output of the system is closer to the reference value,and the rapidity of the system is improved.3)Networked DMPC strategy under multi-sampling rate of RSS process is researched.Reasonable multi-sampling rate model can improve the network utilization.For different subsystems,using correlation function to determine sampling time to capture system behavior more accurately and improve control performance.The simulation shows that the network DMPC with multi-sampling rate improves the steady-state and dynamic performance of the system.4)Considering the economy in the production process,the traditional quadratic index function is replaced by the economic index,and the networked distributed economic model predictive control(EMPC)of RSS is researched.Based on the subsystem model,the controllers are designed,at the same time,the loss of data in the process of information transmission is considered.Under the control of economic model prediction,it is necessary to ensure the optimal dynamic performance of the process and maximize the economic benefits under the change of price factors of key production variables.