Haar wavelet for machine fault diagnosis

Continuous wavelet transform (CWT) is a kind of time–frequency analysis method commonly used in machine fault diagnosis. Unlike Fourier transform, the wavelet in CWT can be selected flexibly. In engineering application, there is a problem of how to select a suitable wavelet. At present, the selecting method mainly depends on the waveform similarity between the signal required to extract and the wavelet. This method is imperfect. For example, Haar wavelet possesses the rectangular waveform in its supporting field and dissimilarity to any component in the machine signal. It is rarely used in machine diagnosis. However, the time–frequency periodicity of Haar wavelet continuous wavelet transform (HCWT) should be useful in revealing the features in signals. In addition, Haar wavelets under different scales have good low-pass filter characteristic in frequency domain, particularly under larger scales, and that can allow HCWT to detect the lower frequency signal. These merits are presented in this paper and applied to diagnose three types of machine faults. Furthermore, in order to verify the effect of Haar wavelet, the diagnosis information obtained by HCWT is compared with that by Morlet wavelet continuous wavelet transform (MCWT), which is popular in machine diagnosis. The results demonstrate that Haar wavelet is also a feasible wavelet in machine fault diagnosis and HCWT can provide abundant graphic features for diagnosis than MCWT.