Analysis of nonlinear predictive control with extended dynamic matrix control

In this paper a procedure for nonlinear predictive control is analyzed. To demonstrate the performance of the procedure a process described by a nonlinear differential equation (NDE) is used. The predictive algorithm operates with an empirical model of the process. Due to the nonlinear behavior of the system, a Multilayer Percepton (MLP) is used for identification. In this paper a brief characterization of the identification with the MLP is given. After this the predictive algorithm is analyzed and explored with simulations. The predictive controller is compared with a PID- controller. Further, the behavior of the controller in presence of disturbances and various parameter adjustments is examined.