Sound and Vibration Signal Analysis using Improved Short-Time Fourier Representation

Abstract

Time-frequency imaging provides a straightforward means to understanding machinery conditions. The methods of short-time Fourier transform (STFT), Wigner-Ville distribution (WVD) and smooth-windowed Wigner-Ville distribution (SWWVD) are applied to the condition monitoring of rotating machines. The sound and vibration signals of a rotating fan are tested, and time-frequency images are illustrated in terms of STFT, WVD and SWWVD. The results show that the resolution of STFT is low, and the resolution of WVD is high but with interference. To overcome the interference in the WVD image, a variable smooth-windowed weighting function is applied. The smoothing window function resulted in interference attenuation but also in reducing the concentration. The SWWVD is a compromise between STFT and WVD. The SWWVD exhibits better resolution than STFT and has less interference than WVD.