Predictive Control Based On Nonlinear Separation Model And Its Application

This dissertation puts forward a method of designing the suboptimal strong tracking filter, which is based on the approximation error upper bound of nonlinear separation model and applied to design feedback corrected controller in predictive control.Main contents are as followed:(1) A nonlinear separation mode with simple structure is used to describe the nonlinear dynamic process. Asymptotic identifying method is applied to identify nonlinear separation mode and achieve perfect effect, because the method focuses on the designation of identifying experimentation and identify parameters of high-order model at first, then make efforts to reduce the orders of model to receive fewer parameters and more precise model.(2) According to the principle of feedback in predictive control, it is very difficult for state estimate method of traditional predictive control to revise the error when predictive model has bigger error, even the system stability can not be guaranteed. This paper puts forward a method of designing the suboptimal strong tracking filter, which is based on the approximation error upper bound of nonlinear separation model. It not only preserves the advantage of strong tracking filter, but also independently introduces modeling approximation error, the dissertation gives the difference between modeling approximation error and state estimation error. The filter for state estimation of nonlinear system is improved. In feedback emendation, the capacity and quality of revising predictive error is enhanced.(3) The dissertation introduces that traditional online optimization was replaced by active set to solve inequality limitation optimum method. It can ensure the optimization of input and output in the extent of limitations. This method is more suitable for the implementation of industrial predictive control scheme.Simulation results and case study prove the validity and applicability of above method.