Research On Model Predictive Control And PID Control Of Industrial Processes

As a promising advanced control algorithm, model predictive control (MPC) hasacquired significant applications since it was proposed to show better performancethan conventional proportional-integral-derivative (PID) control in dealing withprocesses with time delay, dynamics coupling and so on. However, the application ofMPC is less convenient than traditional PID control in terms of simpler structures,cost and hardware in the practice.Dynamic Matrix Control (DMC) is the most widely used MPC algorithm that isonly suitable for the asymptotically stable process. In terms of the integrating systemsin industries, the application of DMC algorithm is not effective because there will bea prediction error caused by the cutoff of model.In this thesis, two main points are addressed. A modified DMC is proposed forintegrating processes where the shifting matrix of DMC algorithm is corrected toavoid the prediction error caused by the model cutoff and a novel error correctionmethod is derived to facilitate the application of DMC algorithm to integratingprocesses. Simulation results verify the effectiveness of the improved DMC algorithmfinally. Another issue is the combination of advantages of predictive control andtraditional PID controllers, where predictive control and traditional PID control arecombined to form a novel PID controller that has the same performance as predictivecontrol and the simple structure as traditional PID controller simultaneously.Simulation results on industrial coke processes show the effectiveness of the new PIDcontroller.