Parametrische Modellordnungsreduktion für hierarchische selbstoptimierende Systeme

Parametric model-order reduction for the reduction of Pareto optimal systems is presented within this thesis. The developed method can be used to simplify complex models which describe the dynamical behavior of self-optimizing systems. The close interrelation of parametric model-order reduction with both hierarchical optimization as well as the structuring concept and hierarchical modeling of mechatronic systems is an outstanding feature of the proposed method.Two types of parametric model-order reduction are considered. A particular Arnoldi algorithm for manual rational interpolation has been implemented on the one hand. On the other hand, a combination of matrix interpolation and H2-optimal tangential interpolation has been investigated. Both types considerably benefit from a method for interpolation of Pareto optimal systems which has been developed within this thesis. Optimal system configurations and corresponding dynamical systems are encapsulated by means of suitable parameterizations and automatic discretization of the Pareto set. Additionally, the complexity of the reduction problem itself is reduced as the number of parameters only depends on the number of objective functions.A test rig of an active suspension system representing the active suspension of the RailCab serves as an application example. The reduced models computed by means of the proposed method possess a very small reduction error