OOR and Track Defects: Analyzing Transient data with the Wavelet Transform

Frequency analysis of train passby noise and vibration is usually limited to one-third octave analysis for analyzing assessment criteria for human comfort and instrument sensitivity. However narrowband frequency analysis is essential for certain purposes such as identifying pure tones, periodic signals, highly resonant oscillation, and identifying sources. Acquiring meaningful narrowband frequency information from train passby measurements is challenging due to the transient nature of the excitation and non-stationary behavior. Spectrogram plots can be used to visualize how the frequency spectrum varies with time. The spectrogram is usually generated by Fourier analysis and/or one-third octave analysis. These analyses, although fast and computationally cost-effective, do not provide a full representation of the signal in the frequency and time domains. The Wavelet Transform provides more detailed frequency and time domain representations of transient non-stationary data than Fourier or third octave analyses. In this paper, The Wavelet Transform is used to analyze field data of short events within a distributed transient vibration signature. The main goal is obtaining more detailed information contained in the measured data and characterizing the spectrum of the vibration to identify maintenance and state of good repair issues such as out-of-round (OOR) wheels and various track defects.