含非平稳干扰模型辨别与在模型预测控制之应用.docVIP

　　含非平稳干扰模型辨别与在模型预测控制之应用 1 Introduction 1.1Model predictive control Since the age of industrial revolution, control systems have bee an integral partof it. From the manual control to automatic control, controlling mechanisms have madegreat progress. The advent of puter based technology has lifted control systems to anes has providedimproved performance and reliability.Model predictive control (MPC) is the most odel to predict its behaviorover a certain time horizon. At each time interval, a cost function is minimized tooptimize the future behavior of the plant by calculating a sequence of optimal futureinputs. Only the first value from the sequence is applied and rests are ignored. Theprocess of optimal sequence calculation is repeated at each sampling instant. MPC hasshoproved results s, and has bee animportant tool for controlling systems e performance demands in the petroleum industry, but its applicationareas are nootive, mining, chemical and food processing [1,2].Over the last feent in the field of MPC has leapedforplementations. There is some gap betplementation.Tent give important detailson the progress of industrial MPC technology. Some theoretical details of the MPCalgorithms are discussed in [3-9]. Over the last feber of books havebeen otivation The model accuracy is very important in all MPC techniques. The plant-modelmismatch results in performance degradation. Another important factor that affects thesystem behavior is the presence of unmeasured disturbances. In real applications,disturbances are not only present but also difficult to model and predict. The modelidentification using test signals should be done in a careful manner, so that the identifiedmodel could represent all the dynamics of system. One important task in system modelidentification is to model not only the process but also the disturbances . A family of transfer function based models are monly used for describingthe system dynamics. Depending upon their structural forms, different trans