Bearing Fault Diagnosis Based on Hilbert Envelope and Continuous Wavelet Transform

Rolling element bearings (REBs) are the commonly used components of rotating machines such as rotors, turbines, motors, and pumping systems. Faults in REBs induce periodic impulses. The faults may lead to improper operation of machines and breakdown of rotating machinery. Preventive maintenance requires the early identification of bearing defects. The present work provides a combined method for detecting bearing faults based on the Hilbert envelope (HE) and the continuous wavelet transform (CWT). The CWT with Coiflet wavelet of order one is used for the analysis. Hilbert envelop removes the high-frequency oscillations from the vibration signal. The periodicity of these impulses is detected using CWT to obtain the bearing characteristics, fault frequency and harmonics. The wavelet analysis is a convolution of the signal and the mother wavelet function that has been dilated and translated. Wavelet transform possesses unique characteristics for extracting irregularities from a noisy signal. The wavelet transform (WT) examines a signal into various frequencies at different resolutions. In the present work, the CWT coefficients are obtained by selecting a suitable range of scaling. The scalogram of the CWT coefficients at a suitable range of scales is plotted. A comparison is made between the CWT scalogram of the input signal with and without the HE. CWT of the HE output gives precise peaks corresponding to the fault frequencies. The number of harmonics corresponding to the defect frequency is more in the case of a CWT with the HE. These harmonics give a clear picture of the fault in the scalogram. The present work is demonstrated using a simulated bearing fault signal and experimental bearing dataset.