PolyRMPC: A new Matlab toolbox to robustify predictive controllers under unstructured and structured uncertainties

摘要

This paper proposes a new Matlab toolbox to simplify the task of designing robustified Model Predictive Controllers under both unstructured and structured uncertainties. With a user-friendly design, the toolbox offers the possibility to develop an initial stabilizing predictive controller to handle both Single-Input Single-Output or multivariable systems. Using the Youla-Kučera parametrization, this initial control law can be further robustified under unstructured and polytopic uncertainties, while respecting time-domain specifications for disturbances rejection. Due to its visualization facilities, the performances of the developed controllers can be easily analyzed and compared in both time and frequency domains. Combining Object Oriented Programming with Matlab Graphical User Interface, this toolbox is built on top of a clear solid framework offering the possibility to extend and update the software design.