Research And Application Of Multivariable Generalized Predictive Control

This paper aims at the advanced control and it's application. The plant object is the petrochemical process, such as atmospheric and vacuum towers etc. Mainly it discusses the following problems: the development of advanced control, strategy of advanced control; identification of multivariable dynamic process model, the current methods for system identification, the closed loop system identification problem, the design of identifier; the application of genetic algorithms to closed loop system identification; multivariable generalized predictive control, the design of software controller and it's application.The paper mainly researches the multivariable system identification, learns all kinds of methods of system identification, deeply studies the model decomposition algorithm. And develops the dynamic process model identifier on model decomposition algorithm. The results of experiments with classic process models are very accurate. The paper discusses the closed loop system identification, and the identifier can identify closed system flexibly. The application of genetic algorithms to closed loop system identification is also discussed.Multivariable generalized predictive control is the important focus in this paper too. The paper studies the theory of multivariable generalized predictive control, and realizes optimum predictor and controller in the computer. The results of experiments is satisfying. Finally, the paper deeply researches the combine system of multivariable generalized predictive control and PID control, discusses the influence of parameters to control effect. And these are the bases for the applying on the industrial process.