Time‐Frequency Domain Analysis

The time-frequency domain has been used for nonstationary waveform signals, which are very common when machinery faults occur. This chapter introduces signal processing in the time-frequency domain and provides an explanation of several techniques that can be used to examine time-frequency characteristics of the time-indexed series signal, which can be obtained more effectively than the Fourier transform and its corresponding frequency spectrum features. These techniques include: short-time Fourier transform (STFT), wavelet transform, Hilbert-Huang transform, empirical mode decomposition, local mean decomposition, Wigner-Ville distribution, and spectral kurtosis. The fast kurtogram algorithm uses a series of digital filters rather than the STFT. Unlike the window used with the STFT, the wavelet function is scalable, which makes it adaptable to a wide range of frequencies and time-based resolution; the three main transforms in wavelets analysis are the continuous wavelet transform, discrete wavelet transform, and wavelet packet transform.