Off-Line Method for Improving Robustness of Model Predictive Control Laws

Abstract

This paper presents a unified off-line state-space methodology for enhancing the
robustness of MPC controlled systems using the Youla parametrization. An initial stabilizing MPC
controller is firstly designed in the state-space representation and further robustified off-line via
LMI techniques to find the best trade-off between stability robustness and nominal performance.
This robustification method, that represents the original contribution of the paper, is finally
applied to the velocity control of an induction machine to reduce the influence of additive and
multiplicative unstructured uncertainties, while respecting a time-domain template for disturbance
rejection.