A robust gradient-based MPC for integrating Real Time Optimizer (RTO) with control

AGUSTINA D'JORGE; ANTONIO FERRAMOSCA; ALEJANDRO HERNAN GONZALEZ

Cordoba

Conferencia; XVI Reunión de Trabajo en Procesamiento de la Información y Control; 2015

Recently, a gradient-based model predictive control (MPC) strategy was proposed to reduce the computational burden of integrating real time optimization (RTO) and control: the main idea is to obtain the on-line controller solution by means of the convex combination of a feasible solution and a solution of an approximated (linearized) problem. This formulation, however, is developed only for the nominal case, which significantly reduce its applicability. In this work, an extension of the gradient-based MPC to include bounded additive disturbance is made. Based on the concept of robust set-interval, the uncertainty is explicitly accounted for, while economic performance and stability is maintained. Several scenarios are simulated to show the benefits of the proposal in contrast to the nominal controller.