Adaptive Model Predictive Control for Unstable Nonlinear Processes

Abstract

An Adaptive Model Predictive Control (AMPC) scheme for unstable nonlinear processes is presented. The main idea is to design the AMPC under a closed loop system which is stabilized by state or output feedback gains. The final control inputs are the summation of the feedback outputs and the control action of the AMPC. The characteristic features of this methodology are easy implementation to existing processes, robustness, and easy handling of unstable nonlinear MIMO processes. The parameter identification technique is used to trace the stabilized process model parameters assuming the process is slowly time varying. The control performance of the proposed scheme is verified by simulation studies on a jacketed continuous stirred tank reactor (CSTR) system and two jacketed CSTRs in series with a separator.