Infinite horizon MPC applied to batch processes. Part II

GONZÁLEZ, ALEJANDRO HERNÁN; ADAM, EDUARDO; ODLOAK, DARCI; MARCHETTI, JACINTO LUIS

Rosario Argentina

Congreso; XIIIº Reunión de Trabajo en Procesamiento de la Información y Control (RPIC 09); 2009

In this part II, it is presented a new infinite horizon model predictive controller (IHMPC), under closed-loop paradigm, with learning properties applied to batch processes. When a batch process is attempted to be controlled two convergence analyses are necessary: the convergence into a given iteration or batch run (intra-run stability, presented in González et al., 2009) and the convergence from run to run (inter-run stability, considering infinite batch run here considered). To account for the first one, the proposed strategy uses a virtual horizon that matches the traditional idea of infinite receding horizon of MPC with the finite duration of the run batch. To account for the second convergence analysis, a learning scheme based on the closed-loop paradigm of the IHMPC, is developed. To evaluate the proposed controller, a numerical example corresponding to batch reactor is shown, where the learning properties of the algorithm can be clearly seen.