The analysis of axle box vibration signal based on frequency slice wavelet transform

In order to extract useful features from axle box vibration signals, an exact feature extraction method using the frequency slice wavelet transform (FSWT) is presented. Firstly, the vibration signal is decomposed with the FSWT and the full band of its time-frequency distribution is obtained. After that, time and frequency intervals are chosen to get zoom analysis of time-frequency feature on the basis of the characteristic of the signal energy distribution, time-frequency regions containing the fault characteristics are separated. Finally, after filtering, the effective fault features are separated from the signal through the FSWT reconstructions. The presented method is shown to be efficient to isolate the desired components from axle boxvibration signals, it can provide a basis for time-frequency feature extraction.