Frequency slice wavelet transform for transient vibration response analysis

This paper introduces a new kind of time-frequency signal analysis method, called frequency slice wavelet transform (FSWT), by means of extension of short-time Fourier transform (STFT) defined directly in frequency domain. The original signal can be decomposed by frequency slice function (FSF), which is similar with the wavelet base but can be designed very freely. At the same time, the original signal can be reconstructed by a FSWT representation in an easy way without the strict limitation of wavelet theory. Some new characteristics of its time-frequency window will be shown. Due to these features, FSWT is more flexible to fit ever-changing signals, and convenient to analyze and control in application. Next, the frequency resolution ratio of signal and Dirac function, etc., are employed to study FSWT, and to select a new scale parameter. The new scale is a good balance factor between time and frequency resolution. Moreover a fast discrete algorithm of FSWT is completed. Its application is focused on transient vibration signal analysis in this paper. FSWT can not only individually represent each modal signal in frequency domain, but also correctly show its details in time domain. FSWT helps to discover some new features of the experimental signal obtained from a small laboratory bridge monitoring system. By using FSWT, the filtering under high noise, and the segmenting of signal with high damping and close modes of frequency, will be discussed. Finally, the summary shows that this paper will be able to provide a more available tool for signal analyzing simultaneously in time-frequency domain, and further to refine the wavelet theory. Crown Copyright (C) 2009 Published by Elsevier Ltd. All rights reserved.