Nonstationary signal classification using pseudo power signatures: The matrix SVD approach

Non-Stationary Dynamics Data Analysis With Wavelet-SVD Filtering

“…The SVD filtering scheme to be described encodes the features of time-bandwidth product, frequency-time dependence, and number and location of signal components. It is invariant to shifts of the signal in time or frequency [5,7].…”

“…Localisation of the energy density describes energy density concentrations at specific locations in the time-frequency plane. Dominant and important concentrations need to be accurately discriminated to provide descriptors relating to the location in time, time duration, frequency location and local bandwidth of principal energy density [5]. Minimising the number of these descriptors while preserving salient information from the energy distribution for modal estimation is desired for efficient dynamics analysis.…”

“…However, it requires user intervention unless regions are well-defined a priori. For this reason, an automated filtering procedure is proposed using the singularvalue decomposition (SVD) of the time-frequency data [5,7].…”

Determination of wavelet ridges of nonstationary signals by singular value decomposition

“…In this section, the wavelet ridge-detection algorithm based on singular value decomposition of the scalogram of the mono-component and multicomponent nonstationary signals in the presence of noise and measurement errors has been explained. The proposed algorithm is different than the ones given in [13], [22] which apply SVD directly to the data matrix, but it is similar to the algorithms which apply the SVD method to the WT coefficient matrix [23], [24]. In [23], the first left singular vector called as pseudo power signature is used in the classification of the seismic signals.…”

“…The proposed algorithm is different than the ones given in [13], [22] which apply SVD directly to the data matrix, but it is similar to the algorithms which apply the SVD method to the WT coefficient matrix [23], [24]. In [23], the first left singular vector called as pseudo power signature is used in the classification of the seismic signals. The left and right singular vectors are used to characterize the speech signals in [24].…”