Extraction of impulse response data via wavelet transforms for multiple input structural system identification
Amy N. Robertson, K. C. Park, K. F. Alvin · Dynamics Specialists Conference · 1996
This paper presents an improved wavelet transformbased method for determining the Markov parameters, or impulse response, of a structural dynamic model from measured multi-input and multi-output data. Auto and cross-correlations of the ensembles are determined and averaged in the wavelet domain. This alleviates the problem of solving an underdetermined convolution equation for the Markov parameters and filters out some of the noise present in the signals. Comparisons of the wavelet transform approach with standard Markov parameter extraction methods show the proposed wavelet ensembling procedure is in general more effective in filtering signal noise.