| Summary: | 碩士 === 國立中正大學 === 化學工程研究所 === 87 === Model predictive control is a control method which uses an explicit process model to predict future system behavior and selects an user-defined objective function to determine the optimal input to the system. Therefore, it is clear that predictive controller design mainly consists of two parts : prediction and minimization. To present exactly the future behaviors of the system, a proper model is selected beforehand. Early the development of model predictive controller is based on linear models. However, industrial processes usually contain complex nonlinearties ,and linear model predictive control may be only applied effectively when the nonlinear system operation around some particular equilibrium points . It takes granted that the nonlinear model describes the system better than linear model.
In this thesis the models combine the nonlinear models from theory viewpoint while disturbance models are used. Beside the proper models to describe the nonlinear systems , and the proper optimal method is important. The aim of this thesis is to investigate the use of Differential evolution algorithms(DEA) for optimization in nonlinear model predictive control. DEA are optimization methods inspired by natural biological evolution. They have been successfully applied to a variety of complex optimization problems where other techniques have often failed. Simulations are presented to illustrate its usefulness
for set-point tracking and disturbance rejection when using the corresponding models obtained by DEA compared with the adaptive generalized predictive control and the another optimal method based on gradient in IMSL .
|
|---|
