Application of Harmonic Wavelet and Its Time-Frequency Profile Plot to Diagnosis of Rotating Machinery
Gao Qiang, He Zhengjia
(Xi'an Jiaotong University, Xi'an 710049, China)

Abstract: Harmonic wavelet definition and characteristics are introduced. In harmonic wavelet analysis, wavelet time-frequency map and contour plot are ususlly used to show the harmonic wavelet decompositon. They can illustrate the distribution of signal energy over time-frequency directly, furthermore, identify weak singularity in a pure signal clearly. When the signal possesses noise, these two methods become ineffective on identifying singularity, so they are almost useless in practice. A new method, wavelet time-frequency profile plot(TFPP) is proposed, which can suppress the effect of noise and recognise weak singularity. With TFPP, vibration data of a key rotating machine in an oil refinery are analysed. After synthesizing fast Fourier transform,singular feature of the signal is extracted and weak fault is diagnosed successfully.
Keywords: harmonic wavelet;time-frequency profile plot;rotating machinery;fault diagnosis