Study Of DMC-PID Cascade Control And GPC Algorithms

Model Predictive Control is a kind of novel computer-controlled algorithms which is just developed in recent years. Due to the control strategies multiple test, moving optimization and feedback correction, the effect of its manipulation is obvious in the application of that the establishment of accurate digital model is unavailable, and that the difficulty in the complex industrial productive process. These features make it be paid attention by worldly engineers and applied in management in the industrial fields such as fossil oil, chemical industry, electric power, metallurgy, machinery, etc.The algorithms of predictive control we use today are include:model algorithmic control(MAC), dynamic matrix control(DMC), generalized predictive control(GPC), generalized predictive pole(GPP), internal model control(IMC), inference control(IC), etc.In this paper, the algorithms and application of predictive control system are studied. This paper involves the background, history and current situation of predictive control system and the briefs of some typical research in predictive control. An set of basic knowledge in predictive control is introduced comprehensively firstly, include the basic principles and features(three segments:model prediction, moving optimization and feedback correction). Furthermore, we research in the principle and algorithmic description of DMC, and describe the work both online and offline in detail. We also show our studies of the parameter designing and tuning, and the principle and basic algorithm of GPC in this paper.In the paper, we propose an improved method of DMC-POC cascade control. It has good quality in dynamic and static tuning and shows great robustness. In simulate experiment, it is proved that by using DMC-PID cascade control the control system has a great ability to anti-interference and obviously improved dynamic response and response speed. To test this method, we applied it in evaporator superheat control system in laboratory to get the report. Moreover, we tested the anti-interference ability via Matlab simulation.At the fifth part, an improved multivariable generalized predictive control algorithm is proposed. This algorithm has implicit doubling control function, that means the algorithm can overcome the intricacy and flaw of traditional distributed control and doubling control. It mainly based on the simulation of mono-variable implicit GPC algorithm, turn multivariable to mono-variable via pre-decoupling the influences of functional distribution and input/output exchange. Then establish a multivariable implicit GPC by Matlab, and simulate the system.At last, the data is analysed in simulate graph. The great capability of this algorithm is shown in robustness, traceability, anti-interference, and controlling unstable non-minimum phase systems.