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

D' JORGE, AGUSTINA; FERRAMOSCA, ANTONIO; GONZÁLEZ, ALEJANDRO H.

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

Recently, a gradient-based model predictivecontrol (MPC) strategy was proposed to reducethe computational burden of integrating real timeoptimization (RTO) and control: the main idea is toobtain the on-line controller solution by means of theconvex combination of a feasible solution and a solutionof an approximated (linearized) problem. Thisformulation, however, is developed only for the nominalcase, which significantly reduce its applicability. Inthis work, an extension of the gradient-based MPC toinclude bounded additive disturbance is made. Basedon the concept of robust set-interval, the uncertaintyis explicitly accounted for, while economic performanceand stability is maintained. Several scenarios aresimulated to show the benefits of the proposal in contrastto the nominal controller.