Robust cooperative distributed MPC: A multi-model approach

Abstract

In this work, we propose a robust cooperative distributed MPC based on a multi-model approach. A new method for obtaining a state-space model in the velocity-form of inputs is presented. This model is suitable for predicting steady-state outputs within an offset-free setpoint tracking framework in which a terminal constraint is relaxed to guarantee problem feasibility under any setpoint change. The proposed strategy also deals with the zone control problem, which is widely applied in the process industry. Moreover, an alternative algorithm is proposed in order to improve convergence over iterations. Recursive feasibility, convergence and stability properties are proved. Two numerical examples are provided with the application of the proposed algorithms.