Font Size: a A A

Research Of Predictive Control Based On Complex Principle Model

Posted on:2003-07-27Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y J LinFull Text:PDF
GTID:1118360062975629Subject:Thermal Engineering
Abstract/Summary:PDF Full Text Request
Model predictive control has been developed and applied to practical control problems in many process industries. However, if the control objects or systems are very complex, such as a power unit, which often represents to be non-linear, multivariable or time-varying, the routine design method only consider some part of dynamic characteristics existed but ignore more dynamic and static characteristics of the system. Although many applications have been studied, especially to the non-linear system control, there is not a very ideal solution yet. In this paper, the universal method of model predictive control to complex system has been studied and a corresponding support system to such control put forward. It is pointed out in the paper that the model used for model predictive control of complex system would be built based on mechanism and analytical relations of mass, energy and momentum and other physical property. This kind of model will be able to represent all characteristics of the complex system both in dynamic and static. The model setup by this may be able to smoothly describe the characteristics of those parameters, which would be control under any operation conditions, naturally include the non-linearity and relativity of the system. Therefore, the shortcoming of the routine design method mentioned above could be avoided. Based on the principle of optimum control, making use of the predicted controlled parameters, the paper points out that the solution of predictive control problem is turned into problem of calculations to the parameters of routine P1 controller, model predictive time and filter persistence; the formula of selecting the P1 controller parameters is also derived. In this paper the controller design is based on continuous-time system design strategy, which avoids the problem sampling time is the controller parameter in routine predictive controls. It is attested that the control strategy in the paper does correspond to sub-optimum control. It also avoids the process of linearization and non-linear programming to the model needed in routine model predictive control system design. A man-machine interactive tool is also designed in the paper, by which, the controller parameters can be selected and adjusted conveniently by means of amplitude and phase frequency. In the paper a support system for the complex processes modeling and predictive control is designed. The modular modeling is adapted, this method decompose the production process in according to equipment or the subsystem and set up the predictive model based on the modules. By means of GUI the support system provide many functions such as complex mechanism modeling, debugging, predictive control implementing and so on. With this support system, the predictive control I can operate non-real-time. In the paper, an algorithms library for modeling of power umt equipment or sub-system is established. With these algorithms and the support system mentioned above, a simulation test has been taken to the main steam temperature control of boiler by means of model predictive control using complex mechanism model. The test shows that the strategy of using complex mechanism model for model predictive control of this paper is very correct. It has outstanding advantages in controlling the non-linear, multivariable complex system. At the same time, the support system and the simulation test system designed...
Keywords/Search Tags:model predictive control, nonlinear model predictive control, complex mechanism model, modeling support system, optimum control, boiler main steam temperature
PDF Full Text Request
Related items