Thrust bearing groove race defect measurement by wavelet decomposition of pre-processed vibration signal

Abstract

Nature of vibration signal from a ball bearing system reveals location of fault and its severity. A proper signal processing technique is required to extract relevant information from a vibration signal. In this work measurement of groove race defect width in thrust bearing is presented by processing vibration signal using wavelet based technique. In the first step, pre-processing of vibration signal is done by multiplying the signal amplitude with its own absolute values. It helps to overcome the problem of detection of weak burst in normal vibration signal retaining sign of the original signal. In the second step, Symlet5 wavelet based decomposition is applied to the pre-processed signal to measure the size of seeded defect in outer groove race of thrust bearing. The decomposition based on Symlet wavelet is preferred because it reduces the ambiguity in detecting the entry point and exit point of the defect as it is having linear phase in nature. Measurement has been done for the groove defect width over the range of 0.4399–1.4854mm. Measurement of defect has been also verified using image examination. Maximum deviation in result has been found in the cases which have minimum axial load (48N) for every size of defect. With increase in axial load to 73N and further to 98N the bearing system gets properly stressed, which increases the accuracy of measurement. With further increase in axial load to 126N and above, it has been observed that the burst disappears due to prevailing over-stressed condition of the loading disk and additional stiffness and damping of the flexible shaft at increased load.