A Nonlinear Model Predictive Control Algorithm For The Crude Oil Distillation Unit

Crude oil distillation unit(CDU)is the first and the most important process of the petroleum refining industry.It not only provides many kinds of raw materials for down stream,but also consumes a lot of energy.Hence,it is a priority for the petroleum refining industry to improve the control performance and increase the economic benefit of the CDU.Aimed at the performance improvement of the CDU,this project has achieved following goals:At first,in this paper,the process of the CDU,the characteristics of crude oil and the characteristics of products are analysed.The steady simulation and the dynamic simulation are simulated in Aspen simulation software.the accuracy of prediction model determines the effectiveness of the NMPC scheme.Therefore,to obtain a simple but precise prediction model,a support vector regression(SVR)based prediction model is employed to describe the dynamic characteristics of the CDU.The SVR-based prediction model has simple form and good predictive effect.In view of the drawbacks of the existing model predictive control(MPC)scheme when facing the strong coupling of the CDU,an improved nonlinear model predictive control(NMPC)with coupling weighting factors is proposed to optimize the CDU.To solve the problem that the PID scheme can not consider economic benefit directly,an economic-oriented tracking control(EoTC)is proposed to improve economic benefit and maintain control performance.In this paper,several simulation results demonstrate the effectiveness of the improved NMPC controllers and the EoTC controllers.