Multi-Model Adaptive Integration of Real Time Optimization and Model Predictive Control

This paper presents a multiple model adaptive approach to integrate Real Time Optimization and Linear Model Predictive Control in the three-layer structure. The proposed approach is an extension of a previous integrating structure with guarantee of nominal stability. The Real Time Optimization targets are sent to the subsequent layers that calculate the control action. Assuming that a nonlinear system can be locally represented by a set of linear models, we propose to compute a control action for each model simultaneously and to implement a linear combination of those control actions. The weight of each control action is obtained at each time step and its computation is based on the distance from the actual process output to the output where each linear model was obtained. The approach was developed for nonlinear systems. The performance of the optimization/control system is tested through the simulation of a polymerization reactor, a chemical process with nonlinear behavior. The results are compared to a previous multi-model robust approach. The results show computational, economic and dynamic advantages when disturbances affect the process. (C) 2019, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.