Model predictive control of a nonlinear unstable process

In this work, conventional (fixed model) dynamic matrix control (DMC), and adaptive (variable model) DMC which combines on-line parameter estimation via a recursive least squares algorithm and control are carried out on a nonlinear, unstable process. The controlled process has multiple steady states one of which is unstable. The control around the unstable point using DMC is done in a cascade arrangement. In the inner loop the unstable process is stabilized with an analog proportional only controller. This stabilized process is then controlled by the outer loop DMC. With this configuration stable control is achieved with the conventional DMC; whereas the stability and the performance of the system with the adaptive DMC is found to be dependent on the controller parameters. (C) 1998 Published by Elsevier Science Ltd. All rights reserved.