Taner Onsay
Alan G. Haddow
Dept. of Mech. Eng., Michigan State Univ., East Lansing, MI 48824
Short-time Fourier (STFT), Gabor, and wavelet transforms are applied to the analysis of vibration signals obtained from mechanical systems. Nonstationary vibration signals resulting from impacts, friction, and transient dynamic interactions are studied by employing the transforms to demonstrate the time-evolution of their spectra. In the application of the STFT, the time-window length and the overlap in window translation are varied and their effect on the resolution of the time-frequency information is discussed. Due to its optimal time-frequency localization characteristics, the Gaussian window used in Gabor transform is found to be particularly useful in the analysis of time-decaying sinusoidal signals, whereas for vibration signals having sharp variations the logarithmic zooming properties of wavelets result in better localization in the time-frequency domain. The use of different analyzing wavelets are considered for detection of transient vibrations and characterization of the associated dynamic behaviors. Based on applications and examples of synthetic signals, the advantages and the limitations of each transform technique are discussed and their conceptual similarities are emphasized.