Model predictive control for nonlinear batch processes with asymptotically perfect tracking

Abstract

We propose a novel model predictive control(MPC) algorithm designed specifically for nonlinear batch or other repetitive processes. Unlike existing MPCs which duplicate the same control error in the repeated batches, the proposed algorithm can achieve perfect tracking (for square systems) despite model uncertainty as the number of batch runs increases. As a special case, we also propose a maximum horizon algorithm which ensures convergence just as infinite horizon MPC does for continuous processes. Numerical examples are provided to demonstrate the performance of the proposed batch MPC algorithm.