CONICET | Buscador de Institutos y Recursos Humanos

Recently, a linear Model Predictive Control (MPC) suitable for closed-loopre-identication was proposed, which solves the potential conflict between thepersistent excitation of the system (necessary to perform a suitable identi-cation) and the control, and guarantees recursive feasibility and attractivityof an invariant region of the closed-loop. This approach, however, needs tobe extended to account for a proper robustness to moderate-to-severe modelmismatches, given that re-identications are necessary when the system is notclose to the operating point where the current linear model was identied. Inthis work, new results on robustness are presented, and an exhaustive appli-cation of the new MPC suitable for closed-loop re-identication to a nonlinearpolymerization reactor simulator is made to explore the diculties arising froma real life identication. Furthermore, several closed-loop re-identications areperformed in order to clearly show that the proposed controller provides un-correlated input-output data sets, which together with the guaranteed stability,constitute the main controller benefit.