TP Model Transformation in Control Design : Plenary Talk

Abstract

The motivation behind of applying alternative representations of models of a given dynamic system is that the structure of the models considerable influences the control design, rather than the approximation accuracy. A good selection of an alternative model representation leads to an an effective control design and good control performance, while a “bad” selection of an alternative model representation leads to a hard or even unfeasible control design. The TP modell transformation is capable of numerically reconstruct the HOSVD based canonical form of qLPV dynamic models. Further extensions of the TP model transformation such as pseudo-, multi-TP model transformation were proposed to easily derive alternative representations of a model at hand. It is proven that the convex hull defined by the TP model transformation is most crucial point in linear matrix inequality based control design. Finally, it turned out it is inevitable to derive model alternatives to find the optimal control solutions. The plenary lecture discusses the advances of the TP model transformation based control design.