A wavelet decomposition analysis of vibration signal for bearing fault detection

This paper presents a study of vibrational signal analysis for bearing fault detection using Discrete Wavelet Transform (DWT). In this study, the vibration data was acquired from three different types of bearing defect i.e. corroded, outer race defect and point defect. The experiments were carried out at three different speeds which are 10%, 50% and 90% of the maximum motor speed. The time domain vibration data measured from accelerometer was then transformed into frequency domain using a frequency analyzer in order to study the frequency characteristics of the signal. The DWT was utilized to decomposed signal at different frequency scale. Then, root mean square (RMS) for every decomposition level was calculated to detect the defect features in vibration signals by referring to the trend of vibrational energy retention at every decomposition. Based on the result, the defective bearings show significant deviation in retaining RMS value after a few levels of decomposition. The findings indicate that Wavelet decomposition analysis can be used to develop an effective bearing condition monitoring tool. This signal processing analysis is recommended in on-line monitoring while the machine is on operation.