Elsevier

Tribology International

Volume 35, Issue 12, December 2002, Pages 793-800
Tribology International

Application of discrete wavelet transform for detection of ball bearing race faults

https://doi.org/10.1016/S0301-679X(02)00063-4Get rights and content

Abstract

Bearing race faults have been detected by using discrete wavelet transform (DWT). Vibration signals from ball bearings having single and multiple point defects on inner race, outer race and the combination faults have been considered for analysis. The impulses in vibration signals due to bearing faults are prominent in wavelet decompositions. It is found that the impulses appear periodically with a time period corresponding to characteristic defect frequencies. It has been shown that DWT can be used as an effective tool for detecting single and multiple faults in the ball bearings.

Introduction

Detection of rolling element-bearing faults has been gaining importance in recent years because of its detrimental effect on the reliability of the machines. Bearing defects can be classified as distributed or local. The distributed defects are surface roughness, waviness, misaligned races, and off-size rolling elements. The local defects include cracks, pits, and spalls on the rolling surfaces. McFadden and Smith [1], [2] have developed the models for high-frequency vibration produced by a single and multiple point defects on the inner race of the rolling element bearing under radial load.

Bearings frequently develop localized defects in the raceways, rollers, and cage. Periodic impacts are generated when rollers pass upon these defects with the exception of cage defects. The periodic impacts occur at ball-passing frequency (characteristic defect frequencies), which can be estimated from bearing geometry and rotating speed. There are several methods to detect bearing local faults using vibration signals as given in an excellent review [3]. Theoretically, in Fast Fourier Transform (FFT) spectra, the characteristic defect frequencies should present corresponding to the bearing defect. But sometimes these frequency components are not present in the spectra because the impulses generated by the defects are masked by noise. To overcome this problem, signal processing techniques such as a processing technique based on averaging technique [4], adaptive noise canceling [5] and high-frequency resonance technique (HFRT) [6] have been developed to improve signal-to-noise ratio for more effective detection of bearing local defects. Among these [4], [5], [6] signal-processing techniques, the high-frequency resonance technique is more popular for bearing fault detection. However, it requires many computations and also several runs of impact tests to be performed to find the bearing resonance frequency. Hence, extra instruments such as impact hammers or vibration exciters and their controller are needed for HFRT.

Recently, the application of wavelets has emerged in the context of damage detection, and an excellent review of this is given in [7]. Mori and et al. [8] have predicted the spalling on the ball bearing by applying discrete wavelet transform to vibration signals. Jing and Qu [9] have proposed a denoising method based on Morlet wavelets for feature extraction and they have successfully applied it to inner race fault detection of the roller bearing. However, the previous works [8], [9], [10] dealt with the detection of one fault in a bearing using wavelet transform. In the present study, the diagnosis of single and multiple ball bearing race faults has been investigated using discrete wavelet transform.

Section snippets

Discrete wavelet transform

Wavelets provide a time-scale information of a signal, enabling the extraction of features that vary in time. This property makes wavelets an ideal tool for analyzing signals of a transient or non-stationary nature. The continuous wavelet transform (CWT) of f(t) is a time-scale method of signal processing that can be defined as the sum over all time of the signal multiplied by scaled, shifted versions of the wavelet function Ψ(t). Mathematically,CWT (a,b) = 1|a| −∞f(t) Ψt−ba dtwhere Ψ(t)

Experimental measurements

In the present study, the faults were introduced in the inner race and outer race of the ball bearings by making a scratch mark with an electric pulse, the bearing components were then assembled as different bearings at a bearing manufacturer’s plant. All scratch marks are of the same size with 2 mm length, 500 μm width and 300 μm depth. Table 1 shows the type of faults introduced. In the present investigation of the bearings, the balls and cage were not scratched.

The ball bearings were then

Results and discussion

The time domain vibration signals of good and defective bearings as given in Table 1 have been considered for analysis with the following data: bearing classification = 6203 series deep groove ball bearings; ball diameter (Bd) = 6.747 mm; pitch diameter (Pd) = 28.7 mm; number of balls (Nb) = 8 and the contact angle (α) = 0. The characteristic bearing defect frequencies of inner and outer race are then calculated by using the following formulae:Ball Pass Inner Race (BPIR) =Nb21+BdPdcosα fBall

Conclusions

The impulses due to bearing faults are clear in wavelet decompositions of the defect bearings. In case of single defects and two defects on outer race at 180° apart, the impulses appear periodically with a time period corresponding to characteristic defect frequencies. But, in the case of two defects, one on the inner race and the other on the outer race, one set of impulses is running at a time period corresponding to the inner race defect frequency while the other set of impulses is running

Acknowledgements

The authors are grateful to the R&D department of TATA STEEL, Bearings Division, Kharagpur for providing the test facility and bearings for the present investigations.

References (13)

There are more references available in the full text version of this article.

Cited by (0)

View full text