COUPLING OF STRUCTURES USING MEASURED FRF'S BY MEANS OF SVD-BASED DATA REDUCTION TECHNIQUES

摘要

The theory of calculating the Frequency Response Functions (FRF) of a structure composed of several substructures using the FRF's available on the substructures is well established. The basic formulation of the method is very simple, involving only elementary matrix algebra - matrix inversion. The method is attractive, for no parametric modeling is required: the responses can be used directly. However, due to inaccuracies on the measured data, the matrix inversion may be ill-conditioned and the computed FRF's may be noisy and incorrect. This paper explores some data reduction techniques, based on Singular Value Decomposition, in order to reduce the dimension of the FRF data of the substructures, in the spatial domain (number of degrees of freedom used for coupling) as well as in the frequency domain. The final result should be an approximating estimate of the global dynamic behaviour of the coupled structure.