Improved robustness of multivariable model predictive control under model uncertainties

This paper presents a state-space methodology for enhancing the robustness of multivariable MPC controlled systems through the convex optimization of a multivariable Youla parameter. The procedure starts with the design of an initial stabilizing Model Predictive Controller in the state-space representation, which is then robustified under modeling errors considered as unstructured uncertainties. The resulting robustified MIMO control law is finally applied to the model of a stirred tank reactor to reduce the impact of measurement noise and modelling errors on the system.