An application of discrete wavelet analysis and connection coefficients to parametric system identification

Most system identification methods are based on a time or frequency domain analysis of data that were measured in dynamic tests. In this article, an algorithm for the identification of a finite element model's parameters is presented that utilizes wavelet coefficients of the measured data and their integrals or derivatives, respectively. How the wavelet coefficients of derivatives and integrals can be related to each other by connection coefficients is shown. These relations are then applied within the proposed parameter identification algorithm. The performance of the method is demonstrated by means of numerical and experimental investigations.