Research On Nonlinear Predictive Control Based On Support Vector Machine

As a successful algorithm of advanced process control, model predictive control has been gain increasing attention from academicians and engineers. More and more attention has been focused on nonlinear model predictive control. This thesis is devoted to predictive control strategies for nonlinear systems. The main research works are as follows:1. A survey of model predictive control on both theory and application is introduced. The main research results are presented.2. The basic theory of support vector machine is introduced and the method for choosing parameters of support vector machine is presented.3. A new predictive functional control based on support vector machine is presented for the strongly nonlinear industrial process in the chemical process. The simulation result for a CSTR process shows that the modeling method is useful to model a nonlinear process and has the good generalized ability. Predictive functional control based on the predictive model shows satisfactory control performance.4. A new nonlinear predictive control based on least square support vector machine is proposed. A least square support vector machine is used for modeling the nonlinear process. The simulation result for a pH neutralization process shows that nonlinear predictive control based on the predictive model shows satisfactory control performance.The thesis concludes with a summary and perspectives of future research of nonlinear predictive control.