Comparison of different techniques for time-frequency analysis of internal combustion engine vibration signals

Abstract

In this study, we report an analysis of cylinder head vibration signals at a steady engine speed using short-time Fourier transform (STFT). Three popular time-frequency analysis techniques, i.e., STFT, analytic wavelet transform (AWT) and S transform (ST), have been examined. AWT and ST are often applied in engine signal analyses. In particular, an AWT expression in terms of the quality factor Q and an analytical relationship between ST and AWT have been derived. The time-frequency resolution of a Gaussian function windowed STFT was studied via numerical simulation. Based on the simulation, the empirical limits for the lowest distinguishable frequency as well as the time and frequency resolutions were determined. These can provide insights for window width selection, spectrogram interpretation and artifact identification. Gaussian function windowed STFTs were applied to some cylinder head vibration signals. The spectrograms of the same signals from ST and AWT were also determined for comparison. The results indicate that the uniform resolution feature of STFT is not necessarily a disadvantage for time-frequency analysis of vibration signals when the engine is in stationary state because it can more accurately localize the frequency components excited by transient excitations without much loss of time resolution.