Research And Implementation Of Model Predictive Control Based On Hardware-in-the-loop Simulation Technology

Along with the improving of the complexity in process industry process control system,many problems, such as production quality, economic benefit and the environmental safetyare introduced. The appearance and development of model predictive control technique ishelpful to solve the problems. Model predictive control with its lower request to themathematical model, combining with the feedback correction method could recedue horizoncontrol in the limited time domain. The reseach of model predictive control relies oncomputer simulation methods to analyze the effect, but it is obstacled for the controllerapplication when it is from a real industrial environment. On the other hand, model predictivecontrol strategy is usually accompanied by the use of distributed control system. Consideringthe economic results, security and other problems of enterprises, the design and implement ofthe model predictive control needs a lot of preliminary simulation and testing. Besides,dynamic matrix control has been widely applied in process industry. Dynamic matrix controluses output step disturbance to correct prediction model, when facing the problem ofunmeasurement disturbance, which solve the problem of disturbance rejection. However, thismethod lack of the ability to predict disturbance. So it can not rapid, effective rejectdisturbance during the situation of other types of the disturbance or existed noise. In order tosolve above problems, the research of model predictive control based on hardware-in-the-loopsimulation and combining with kalman filter and model predictive control to improve theability of disturbance rejection is introduced. The main research is as follows:Using real hardware controller and industrial control network to build distributedcontrol system. Developing computer simulation model as controlled object based onmethanol production process. Therefore, this paper builds the process industry integrate-d automation simulation system by taking use of this pattern that real control systemcombines with virtual computer simulation model. This paper analysis the dynamic matrix control algorithm in detaile in view of the SISO and MIMO system and understand contol performance with the help of comuputer simulation for typical unit of tankand distillation column. Take use of the result of simulation, hardware controller a-nd industrial control network, collect the process data of computer simulation model, t-hus to design controller. Dynamic matrix contoller has been applied to the computersimulation model of the flow control unit and distillation column unit in the simulation environment, which could simulate model predictive control of actual control performance and fully verify its feasibility at the hardware-in-the-loop environment.In order to solve the limitation problem of the model predictive control in disturb-ance rejection, the state space model is used to predict controlled system of dynamicperformance in the future limited time domain. Kalman filter, which is effective to estimate the state variables, is used to estimate the augmented state variables that are o btained by the disturbance variable. Thereby, the ability to predict disturbance perfor-mance is achieved, and the impact of disturbance action on the controlled system coul-dbe fully considered. The same time, integrating receding horizon control strategy to improve the ability of model predictive control deal with unmeasurement disturbance.The simulation results show that model predictive control based on kalman filter has significant advantage for disturbance rejection during the situation of input step distur-bance.