RECENT APPROACHES AND MEASUREMENT TECHNIQUES FOR …€¦ · Recent Approaches and Measurement Techniques for the Bearing Fault Analysis – A Review 136 [email protected]

http://www.iaeme.com/IJMET/index.asp 134 [email protected]

International Journal of Mechanical Engineering and Technology (IJMET) Volume 9, Issue 13, December 2018, pp. 134–146, Article ID: IJMET_09_13_015

Available online at http://www.iaeme.com/ijmet/issues.asp?JType=IJMET&VType=9&IType=13

ISSN Print: 0976-6340 and ISSN Online: 0976-6359

© IAEME Publication Scopus Indexed

RECENT APPROACHES AND MEASUREMENT

TECHNIQUES FOR THE BEARING FAULT

ANALYSIS – A REVIEW

J.J. Jayakanth

Department of Mechanical Engineering, Vels Institute of Science, Technology & Advanced

Studies (VISTAS), Chennai, Tamil Nadu, India.

M. Chandrasekaran*

Department of Mechanical Engineering, Vels Institute of Science, Technology & Advanced

Studies (VISTAS), Chennai, Tamil Nadu, India.

*corresponding author

ABSTRACT

In recent years manufacturing sector uses different types of motors for its production

of parts and components. Majority of machining applications uses induction motors,

single phase motors for light duty application and three phase motor for heavy duty

application. The need for the industry is to run the motors failure free so that the

production capacity is meet out. To achieve this, motor components has to perform

trouble free operations, but in a due course of time when the motors are made to run for

a longer period of time, it develops wear among the parts. When considering the faulty

parts bearing wear is one major issue. Due to this the motor also gets damaged, if this

happens the entire motor has to be repaired or replaced. Many fault detections methods

and techniques have been implemented; different methods of bearing fault identification

are being conducted. Here a detailed review about bearing fault analysis and its recent

developments are described in this paper.

Keywords: induction motors, bearing fault identification

Cite this Article: J.J. Jayakanth and M. Chandrasekaran, Recent Approaches and

Measurement Techniques for the Bearing Fault Analysis – A Review, International Journal

of Mechanical Engineering and Technology, 9(13), 2018, pp. 134–146

http://www.iaeme.com/IJMET/issues.asp?JType=IJMET&VType=9&IType=13

1. INTRODUCTION

In any rotating machinery ball bearing is an important component. Most of the mechanical

hardware and motors uses ball bearings, which gives a supporting mechanism for the

machineries. The ball bearing makes the motor or any rotating machinery to run smoother with

minimal friction and mechanical losses in its movement. If the ball bearing runs smoother then

there won’t be any vibration or defects in it. Hence, fault diagnosis using vibration has become

J.J. Jayakanth and M. Chandrasekaran


major contributing area and reported more number of [1–4]. Due to presence of noise in the data

measured, difficulties raised in bearing fault diagnosis from vibration signal analysis [5, 6].

Hence, de-noising the acquired vibration signal become more important in the area of signal

conditioning and fault diagnostics using vibration signal [6,7]. Whenever bearing is used for

longer period of time its starts developing internal wear, cracks, and hardening of the grease and

develops vibration in the bearing which results in wear of the rotating machinery. Due to this

reason the ball bearing in rotating machinery has to be replaced over a period of time. Some

methods have to be implemented to identify the defects caused in the ball bearing, so that the

rotating machinery does not get damaged by ball bearing. Many researches are exploring in depth

identification of bearing defects. Due to dynamic load, speed and external interference detection

of bearing faults operated in variable frequency drives becomes challenging [2]. Some of the

methods like time domain analysis, frequency analysis, wavelet transform, spectral analysis,

vibration and current monitoring are implemented in ball bearing analysis. This paper gives a

review of the various recent ball bearing defect analysis methods and techniques.

The common ball bearing defects are inner race defect, outer race defect, ball bearing defect,

hardening of grease and the vibration caused by it. From the reviewed papers few methods have

been classified which can give a detailed study on it. The following are the methods which can

detect the above mentioned defects.

• Time domain analysis

• Frequency domain analysis

• Vibration analysis

• Wavelet transform

• Spectrum analysis

• Current monitoring and

• Mathematical Modelling

(a) (b)

Figure.1. (a) Rolling element bearing and (b) Internal view shows the parts of ball bearing [10].

1.1. Time & Frequency domain analysis:

Reza Golafsha,.et.al, [7] has analysed the fault signature modelling and detection of inner race

bearing faults in their paper on “SVD and ‘Hankel’ matrix based de-noising approach for ball

bearing fault detection and its assessment using artificial faults”[1]. In this method, Singular

Value Decomposition (SVD), analyze the peak spectra using the phase coupled sidebands

occurring at spacing, has been predicted in this model. First the fault signature magnitude and

Recent Approaches and Measurement Techniques for the Bearing Fault Analysis – A Review


load is tested. The results of the test consist of the machine vibration due to the radial movement

of the bearing, air gap variation due to relative motion between the raceways, variation in rolling

resistance due to load torque. The ball bearing radial movement confirms the machine vibration

with a faulty frequency. The measurements are taken through vibration transducers. From the

results a model is developed with time domain convolution of white noise with that of the

mechanical frequency response of the mechanical function [1].

Figure 2.Shown in the above figure is: downward radial load applied to the inner race of four bearing

surfaces as well as the resulting load zone [1].

To a large extent, the magnitude of this radial movement depends on the magnitude of radial

load applied to the inner raceway. Bearing illustrated in fig. 2. Shows, defect impact causes a

minimal effect in a rolling element, in the absence of radial load. This is because of racial

separation between the inner and outer raceways maintained in the neighboring rolling elements

and is able to pass over the defects. However, rolling element exhibits elastic deformation under

application of a radial load and experiences a compressive force. This forces rolling element into

the defects and causes impact effect. It also increases the inner race defects due to race way and

rolling element couplings. When the characteristic fault frequency of inner-race is an integer

multiple of rolling element characteristic frequency the rolling elements are easy to identify.

Hence the defects identification in rolling element or cage, this model is can be easily extendable.

Xiaoxi Ding.et.al.[3], proposed time-frequency manifold reconstruction of sparse method as a

transient method for identifying the bearing fault features. From this method reconstruction of

image sparse is computed with analysis framework. The transient characteristics of signal from

low signal-to-noise ratio (SNR) are studied jn this model. Time-domain averaging method [4],

band-pass filtering [5], frequency-domain thresholding [6], empirical mode decomposition

(EMD) [7], wavelet transform (WT) [8], time–frequency analysis (TFA)[9] are various de-

noising methods reported for fault signature extraction.

However, in acquired signals, the presence of background noise affects the fault diagnosis

and causes difficulty in fault diagnosis. This makes incapability in fault analysis; it makes the de-

noising process one of the most essential steps in signal condition monitoring and fault

identification. This study reports de-housing process with Singular Value Decomposition (SVD)

and ‘Hankel’ matrix successfully applied to vibration analysis of ball bearing in time domain

spectrums for the elimination of the background noise and the improvement there liability of the

fault detection process. The effectiveness of this method is compared experimentally as well as



the simulated vibration signal tests conducted confirmed this de-housing approach suitability for

the fault identification in ball bearings.

S.A.McInerny et.al. [10], describes a laboratory module on fault detection in rolling element.

This module considering the basic operational characteristics of bearing element also provided

theoretical formulas for fault characteristic frequencies. Detection of bearing faults is carried out

using conventional spectral analysis of generated vibration signals using program developed in

MATLAB. This generated vibration signals includes important signatures obtained in bearing

housings. Also, described the inter relationship between fault signatures of bearing and amplitude

modulation and demodulation with the support of envelop analysis. Graphical user friendly

interface (GUI) software utilities developed in MATLAB allows exploring envelope analysis

using measured data / the generated signal. With GUI utilities obtained spectra with Hanning

window implemented FFT and user selectable parameters. In this method the envelop data has

bandwidth half of the acquired data, hence the enveloped data down sampled by a factor of 2

prior to spectrum analysis. The fault frequency of bearing and it’s harmonics are clearly visible

in the final processed spectrum.

1.2. Vibration Analysis:

Vibration analysis is one of the most powerful tools for the bearing fault detections. It would be

a simple process if noiseless vibration signatures are measured on bearings to identify fault in it.

But practically the acquired vibration signals from bearing element and it’s housing, buried in

noise due to random vibration caused by friction, misalignments and imbalances. These random

vibrations dominate the fault signature in the measured spectrum and actual expected vibration

spectrum are lost in the higher harmonics and its spectral noise floor.

Bearing failure occurs due to several mechanisms, including damages caused by mechanical,

crack, wear, lubricant deficiency, and corrosion [3]. Nicks and dents are due to abusive handling

of berating elements. When higher stress on the contact impair the smooth surface, i.e., marred

and reduces the life time of bearing element drastically. ‘Brinelling’ is another harmful condition

caused in rolling element due to its overload. Cyclic loading and over loading create crack in

bearing, manufacturing defects also leads to crack. In a bearing element gradual deterioration due

to wear leads to dimensioning related faults. Normally metal-to-metal friction occurs and

increases due to inadequate or poor lubrication also increases the temperature of the bearing

element, reduces bearing life time. Plastic deformation occurs due to operating forces. Scuffing

or scoring is lighter adhesive damage; sizing or galling is highly intense damage. Abrasive wear

occurs if any hard particle intervene the smooth contact surface. Higher humidity environment

due to surface oxidation produce rust and pits and raises the stress and ultimately leads to abrasion

and rapid wear. In an example of an outer race of a ball bearing is flawed because of one of the

failure mechanisms in which one of the balls rolls over the flaw, an impulsive force is incurred

that causes the bearing to vibrate. The bearing responds by “ringing” at its Fig. 3.



Figure 3. Vibration signal plot, amplitude as a function of time of a bearing shows ringing excitation

due to fault in outer bearing race [10].

The response excitation occurs each time one of the balls rolls over the flaw, hence the

fundamental frequency of the response waveforms is the rate at which the elements roll over the

flaw, is of interest in the detection of bearing faults, not the resonance frequency. Depending on

nature of bearing fault like outer race, inner race, balls and cage, different characteristic frequency

are generated in S.A.McInerny.et.al. [10].

Four characteristic frequencies are given here under of a rotating bearing element in which

the outer race is stationary and its rotating inner race.

Train Frequency or Cage Frequency (FTF)

FTF (Hz) = S (1/2) (1 – B/P Cos ϕ). (1)

Ball pass frequency, Outer race (BPFO)

BPFO (Hz) = S (N/2) (1 – B/P Cos ϕ). (2)

Ball pass frequency, Inner race (BPFI)

BPFI (Hz) = S (N/2) (1 + B/P Cos ϕ). (3)

And Ball Spin frequency (BSF)

BSF (Hz) = S (P/2B) (1 + B2/P2 Cos2 ϕ). (4)

Where B is ball diameter, P is pitch diameter, N is number of balls, and S is rotation rate of

the shaft (in hertz). But in thrust load condition the characteristic frequencies have some slippage

and needed appropriate correction factors. For these purpose software modules incorporated

required correction parameters are available commercially [11].

With the above theoretical calculations attempts were made to identify the characteristic

frequencies of bearing under test [10]. In this work a synthetic signal was constructed with a

square wave approximation and random noise, ringing pulse sequence and their sum has been

carried out and obtained simultaneous spectra and composite waveform. But in this analysis, the

higher harmonics in pulse sequence spectrum higher harmonics in the spectrum disappearing

slowly compared to square wave approximation. The peaks in the composite signal spectrum are

lost. Finally they used band pass filtering in “envelope analysis” rejects the high-amplitude low-

frequency signals associated with misalignment and imbalance to eliminate random noise [8, 9].

Analyzers and data loggers used in envelope analysis usually have user selectable band pass

settings of required frequencies. Final step in the envelope analysis process is calculation of the

spectrum of the rectified band pass filtered signal with Hilbert transform. In this work the

software utility developed in graphical user interface (GUI) with MATLAB utility for the

envelope analysis.



Figure.4. GUI based data processing flow diagram of Envelop Analysis with MATLAB utilities [10].

1.3. Spectrum Analysis

In S.A.McInerny et.al. Report [10], bearing fault frequency and its harmonics are distinctly visible

in the spectrum obtained with MATLAB based graphical user interface (GUI) program. This

confirms effectiveness of this spectrum analysis method in bearing fault deduction analysis.

JJ.Jayakanth.et.al.,[13], brings out the very simple way of detecting bearing faults, that too

in a static condition, as a first time, using impulse excitation technique and implemented with

power spectrum in LabVIEW graphical user interface (GUI) program with the graphical icon

auto computed FFT Power Spectrum module. In which acquired vibration signal from MEMS

accelerometer as a function of time, is auto computed online Power Spectrum (PS) through

LabVIEW Palette (i.e., Power Spectrum.vi), for 10 impulses to the bearing under test, computes

the averaged auto power spectrum of acquired time signal from MEMS accelerometer.

The power spectrum provides clear signature of faulty and good bearing in any size or make

of tested bearings. Since it reflects the energy spent by fault bearing is larger compared to the

smooth and lesser energy consumed good bearing for the given excitation, i.e, for an impulse.

Defective bearings increase the amplitude of the peaks in the power spectrum after FFT is carried

out which indicates the presence of the defect (as shown in fig.5).

Figure.5. (a) Acquired Vibration Signal from MEMS Accelerometer for excited impulses, (b) Auto

computed FFT Power Spectrum acquired Vibration Signals in LabVIEW for two bearings of same size

and make [13].



1.4. Wavelet transforms:

S.Khanam.et.al. [14] diagnose fault analysis of bearing and access its severity using wavelet

technique. Generally speaking limited duration waveforms are called wavelets, in wavelet

analysis breaking up of a signal into shifted and scaled versions of the original wavelet that has

an average value of zero [15]. Wavelets are ideal tool for analyzing transient signals and provide

time-scale information and facilitate extraction of varying parameters with respect to time.

Wavelets transform are of two types, viz., continuous and discrete. Convolution of signal and

wavelet function calculation is involved in continuous wavelet transform A wavelet function is a

small oscillatory wave contains both the analysis and the window function. In discrete wavelet to

extract the genuine frequency content of the signal in various sub bands filter banks are used, for

signal analysis and synthesis. To reduce the computational time, discrete wavelet transform is

derived from continues wavelet transform discretization by adopting dyadic scale and translation

[16].

�� , � = ��

�� ∗ �� (7)

Where j, k are integers, 2j and 2jk are the scale and translation parameter respectively, ψ is

the ‘mother’ wavelet and ψ * is the complex conjugate of ψ. The signal s(t) to be processed

passes through low pass and high pass filters emerging as low frequency and high frequency

signals (with approximations ai and details di respectively) at each decomposition(of level ‘n’).

Therefore, signal s (t) can be written as [17]

�� = �� + ∑ � ! "� (8)

Daubechies [15] proposed ‘Symlets’ supported wavelets with least asymmetry and highest

number of vanishing moments for a given support width, associated scaling filters are near linear-

phase filters, which make them easier to deal with small discontinuity present in the signal

without affecting the original information, The continuous and discrete wavelet transform uses

this approach due to its perfect reconstruction and cancellation capability. The sym5 wavelet

based decomposition gives raise defect size indication by precisely locates leading edge and

trailing edge impacts. The outer race defect size of the bearing element can be evaluated by

recording the raising (entry event) and trailing edge (exit event) of the signal after decomposition

for the sym5 wavelet. The following equation provides information about size of the defect using

duration between set of events at the leading and trailing of the defect edge:

�#$#%� �&'# = () × +) × ∆� = -./01 21 − 567

-.7%8� 9: × ∆� (9)

Where ‘rb’ is radius of the ball, ‘ωb’ is the ball spinning speed, ‘ωs’ is the rotational speed of

shaft, Dp is the bearing pitch diameter, db is the diameter of the ball, α is the deep groove ball

bearing’s contact angle equal to zero and Δt is the duration between the raising and trailing impact

pulse.

S.Khanam.et.al.,[14] have been performed experiments with an accelerometer using an

undamped natural frequency for acquiring the vibration signal using accelerometer mounting on

the top of test bearing housing. The captured signal was stored in Fast Fourier Transform (FFT)

Analyzer and transferred to an intelligent processor system where the post processing was carried

out in the MATLAB environment. Experiments have been performed at user selectable shaft

rotational speed of and radial load. With varying size of the defect on the outer race of the test

bearing, random for the healthy case, periodic impulses generated as the ball passes over the

defect at a rate equal to characteristic defect frequency were observed experimentally. So the

impulses in the high frequency range of wavelet decomposition are covered in initial

decomposition. An average of the data points between the two events for all the bursts present in



the signal enable to obtain bearing outer race defect size precisely by time interval between two

subsequent events.

The decomposition of the vibration signal using Symlet wavelet measures precisely the fault

size on the ball bearing outer race. The bearing fault less than the ball diameter with significant

depth has peak at leading edge and at trailing edge. The ball contact with the base generates

multiple events due to shallow faults. These measurements of defect sizes are verified with optical

microscope confirm the technique: a way easier and precise estimation bearing fault size.

Wavelets are competent tool for bearing fault diagnosis [18-24], due to its efficient

computational implementation and flexibility. Main implementation of wavelets in bearing fault

diagnosis by Peng and Chu [18] involves time frequency signal analysis, demising and extraction

of the weak signal, fault feature extraction, singularity detection, compression of vibration signal

and identification of system. Discrete wavelet transform are used in Prabhakar et al. [19] to

detect and identify single, multiple and combination ball bearing race faults. Shi et al. [20] uses

wavelet transforms and envelop spectrum for extracting bearing defect signatures. ‘Morlet’

wavelets are used for demodulating defect components in the bearings in Nikolaou et al., [21]

work, Qiu.et al. [22] reported a comparison of de-noising with wavelet filter and of de-noising

with wavelet decomposition and concluded that wavelet filter based de-noising and wavelet

decomposition based de-noising with the conclusion that wavelet filter based de-noising is

finding its suitability in weak signature detection. Junsheng et al. [23] reported the extraction of

bearing fault features from vibration signals by constructing impulse response wavelet in

continuous wavelet transform. The recent work by Kumar et al. [24] reported signal

decomposition using ‘Symlet wavelt’ for the extraction of outer race of taper roller bearing fault

size. The confident applications of ‘Symlet’ wavelets in many other fields are reported in

literature Literatures like noise reduction from ECG signals [25] and speech signal de-noising

[26].

1.5. Kurtosis coefficient a Time-Domain Statistics:

From the time domain data, vibration signal, evaluation of kurtosis coefficient [27-30] is

effectively used in the bearing fault detection. A zero mean signal, variance, σ2, is the mean

square value [i.e., the square of the root-mean-square, RMS value] of an AC coupled

accelerometer signals are zero mean signals. Its rectified waveform has a nonzero average or

mean value. In this case, the variance is given by

; = �< ∑ �=�� − > < "� (5)

Where N is the number of data points in the data sequence, and µ is the average value. The

kurtosis coefficient is a fourth-order statistic normalized by the square of the variance σ2

? = �< ∑ �@��A�BC

DC< "� (6)

This statistical moment is sensitive to the pulses induced by bearing defects and effective in

the rolling element bearing diagnostics.

The localized defects in the bearing element outer race, inner race and rolling element cause

faults in rolling element or the cage, which generates impacts and impulse and excite the total

system including bearing, sensor mounted on it and beating housing. The characteristic

frequencies of bearing elements and its order can be identified by using signal processing

techniques.

A de-convolution and de-noising technique is reported in Lianhuan Hong et.al. [31] As a

maximum correlated adaptive ‘kurtosis’ de-convolution and multi wavelet transform. In this

paper This paper propose a compound fault analysis combined with Hilbert envelope spectrum

analysis based on maximum correlated ‘kurtoris’ de-convolution theory. a parameter range



estimation method is given for its parameter optimization. Standard flexible multi wavelets are

used in signal post processing to enhance the de-noising effect. An M-shift correlated kurtosis

composed fault characteristics index with square envelop spectrum entropy is used to construct,

particle swarm optimization based, customized multi wavelets. The usefulness of this method is

demonstrated by both simulated signal and practical vibration signals of a rotor in an experimental

rig with different compound faults. The reliability and superior effectiveness of this method are

confirmed by comparison with other fault detection methods.

1.6. Current monitoring:

S.Mitra et.al. [32] Carried out comparative measurements on fault bearing analysis with three

different techniques say: current monitoring technique, sound signal technique and vibration

analysis technique. In its work using current clamp, three phase induction motor current signal

was recorded. Audible sound or noise of the machine also measured using standard digital sound

level meter kept near motor under test. An accelerometer with sensitivity of 1 mV/g, and

bandwidth of 2 kHz was kept on the top the motor with to acquire the vibration signal. Fig.6.

shows the acquired spectrum of the current, vibration and sound sensors.

(a) (b)

(c)

Figure.6. Frequency as function of power spectral density plot of the (a) motor current signal (b) sound

signal acquired (c) vibration amplitudes of rotating bearings when the mixer motor (under load) in the

presence of external noise [32] .

In the comparative study, it has been reported first, the current signals are too weak, hence

the fault identification is too difficult in real time back ground, hence, it is really a challenging to

identify healthy and faulty bearings. Fault identification with acoustic and vibration signals

techniques are reasonably good compared to current monitoring technique.. Mounting acoustic

microphone also very simple than the vibration sensors, but artefact effect due to outside noise

in acoustic method is very much sensitive and affects the actual measurements. Hence it confirms,

vibration signals technique dominates the other methods for fault identification of the industrial

actuators.



1.7. Mathematical models:

Nabhan.A.et.al. [33] developed a mathematical model for the ball bearing defect analysis and

vibrations due to defect on the bearing race. Using a commercial package ABAQUS/CE,

investigations are carried out at the outer race and its housing is modeled in three-dimension.

Comparison has been made with an experimental vibration results with the simulated acceleration

signal obtained from the dynamic model. The influence of housing structure with different height

is investigated and corresponding vibration response has been evaluated using the root mean

square parameter RMS. It was found that, the housing with less height has lowest fluctuation in

the output response. Also, stability of bearing increased with the decrease of the bearing height,

where the bearings with less height showed the best results.

A finite element contact bearing model [34] establishes a high-precision elastic based on a

contact algorithm, in which all the important bearing geometry like internal clearance, roller and

race crowning, race width and thickness, and dimensions of the raceway shoulders were included.

This method is presented for calculating and analyzing the radially loaded double row bearing

with a raceway defect of varying depth, length, and surface roughness [35], also validated by

comparing experimental results to existing analytical models. The resonance frequency in the

first vibration mode of mechanical system was studied [36]. According to this study, the envelope

detection method for the first vibration mode resonance frequency could be effectively applied

in the signal processing for the bearing defect diagnosis. It is also found that, the vibration signal

amplitude for outer race defect is more than that of the inner race defect and the ball defect. The

theoretical model was aimed to study the effect of defect size, load and speed on the bearing

vibration and predict the spectral components.

To differentiate between vibrations signatures, Finite element model can be effectively used

for defects of different sizes in the bearing [37]. Experimental measurement result has been taken

for the analysis of the signal that has been obtained through the use of FFT analyzer. The vibration

signal pattern obtained from the simulation was found to have similar characteristics with

experimental data. A dynamic loading model simulates the distribution of load in the outer race

due to transfer load from the ball. Time domain analysis is performed to evaluate the output result

of vibration analysis from the finite element software. RMS and peak to peak value is used as the

time signal descriptors and can be used as a parameter for condition monitoring purposes. The

vibration response of healthy and defected bearing is compared with the simulated vibration

pattern and observed similar characteristics obtained from experimental results, [38].

2. CONCLUSION

The SVD and ‘Hankel’ matrix based de-noising approach based fault-signature model and

detection scheme, time - frequency analysis increases the possibility of inner-race defects

identification easily, also it can identify rolling-element defects when the inner-race characteristic

fault frequency, is an integer multiple of characteristic fault frequency due to rolling element.

This fault-signature model and detection scheme can be further extended for rolling-element and

cage defects. In the time–frequency manifold sparse reconstruction model studies the transient

characteristics of signal from low signal-to-noise ratio (SNR). Among the various other methods,

de-noising performance of TFM has a good result, since it combines the de-noising and atomic

decomposition merits in image sparse reconstruction. This method is effectively verified by

experimental analysis for bearing fault feature extraction.

A laboratory instructional module presented with conventional spectral analysis illustrated with

a synthetic signal generated in MATLAB, also a graphically driven MATLAB software utility

effectively explores the envelope analysis of measured data and the generated signal. Theoretical

calculations were made to identify the characteristic frequencies of bearing under test. Also

construction of synthetic signal using a square wave approximation, random noise, ringing pulse



sequence and their sum has been carried out and obtained spectra each sequence and is composite

waveform. Using band pass filtering in “envelope analysis” and Hilbert transform for rectification

rejects the low-frequency, high-amplitude harmonics, associated with imbalance and

misalignment and to eliminate random noise outside the pass band to prevent loss of peaks in the

composite signal. Spectrum analysis reported with GUI utilities in both MATLAB and LabVIEW

implemented development programs confirms effective user friendly approach for fault

diagnosis.

In very simple way the designer directly using auto computed FFT Power Spectrum virtual

instruments program icon in LabVIEW of acquired vibration Signals and obtained the power

spectrum, which clearly identifies fault in bearings, that too in static environment , not like all

other reports uses only dynamic measurements technique..

The flexibility and efficient computational implementation in wavelets approach has proved

to be a competent tool for bearing fault detection. Kurtosis coefficient in a time-domain statistics

supports enhancing the de-noising effect further in the compound faults. The reliability and usage

of this method are compared with other fault detection methods.

Excellent comparative study of faulty bearings were carried out successfully with current

monitoring technique in bearing fault identification, and further confirmed that, the vibration

signals technique is predominant than other methods for fault identification of the bearing

elements.

Mathematical model approach provides high-precision elastic with a finite element contact

bearing model and includes all the important bearing geometry like: internal clearance, roller and

race crowning, race width and thickness, and dimensions of the raceway shoulders. Mathematical

model also supports calculating and analyzing the radially loaded double row bearing with a

raceway defect of varying depth, length, and surface roughness and validated successfully by

comparing experimental results.

REFERENCES:

[1] Jason R. Stack, Senior Member, IEEE, Thomas G. Habetler, Fellow, IEEE, and Ronald G.

Harley, Fellow, IEEE, Fault-Signature Modelling and Detection of Inner-Race Bearing

Faults, IEEE Transactions on Industry Applications, Vol. 42, No. 1, January/February 2006

[2] S. Mitra, C. Koley, “Different Measurement Techniques for Detection of Bearing Faults in

Industrial Actuators – Comparative Study”, IEEE Calcutta Conference (CALCON) 2017

[3] X.Ding,Q.He,Time–frequency manifold sparse reconstruction: A novel method for bearing

fault feature extraction, Mech.Syst.Signal Process.(2016).

[4] S.Braun,The synchronous (time domain) average revisited, Mech. Syst. Signal Process. 25

(4) (2011) 1087–1102.

[5] Y.Wang, Z.He, Y.Zi, Enhancement of signal denoising and multiple fault signatures detecting

in rotating machinery using dual-tree complex wavelet transform, Mech. Syst. Signal Process.

24 (1) (2010) 119–137.

[6] E.G. Strangas, S.Aviyente, S.S.H.Zaidi, Time–frequency analysis for efficient fault diagnosis

and failure prognosis for interior permanent-magnet AC motors, IEEE Trans. Ind. Electron.

55 (12) (2008) 4191–4199.

[7] R.Golafshan, K.YuceSanliturk, SVD and Hankel matrix based de-noising approach for ball

bearing fault detection and its assessment using artificial faults, Mech.Syst.SignalProcess.

(2015)

[8] J.Antoni, Fast computation of the kurtogram for the detection of transient faults, Mech. Syst.

Signal Process.21 (1) (2007) 108–124.

[9] M. Amar, I.Gondal, C.Wilson, Vibration spectrum imaging: a novel bearing fault

classification approach, IEEETrans. Ind. Electron.62 (2015) 494–502.



[10] S.A. McInerny and Y. Dai, Basic Vibration Signal Processing for Bearing Fault Detection,

IEEE Transactions on Education, Vol. 46, No. 1, February 2003. p149-156.

[11] MATLAB Signal Processing Toolbox User’s Guide, vol. 4, The MathWorks, Inc., Natick,

MA, 1996.

[12] Y.Yang, D.J.Yu, J.S.Cheng, A roller bearing fault diagnosis method based on EMD energy

[13] Entropy and ANN, J.Sound Vib. 294 (2006) 269–277. 13. J.J.Jayakanth and

M.Chandrasekaran, “Virtual Instrumentation Programming and PSoC

[14] Embedded Design for Beating Fault Deduction: Uses Impulse excitation Technique”.

International Journal for Engineering &Technology. 7(3.27) 2018. P.170-173. 14. S.

Khanam, N. Tandon, J. K. Dutt. Fault size estimation in the outer race of ball

[15] Bearing using discrete wavelet transform of the vibration signal. Procedia Technology 14

(2014) 12 – 19, http://visl.technion.ac.il/documents/wavelet_ug.pdf.

[16] Prabhakar, S., Mohanty, A.R., Sekhar, A.S., 2002. Application of Discrete Wavelet

[17] Transform for Detection of Ball Bearing Race Faults, Tribology International 35, and p.793.

17. Kumar, R., Singh, M., 2013. Outer Race Defect Width Measurement in Taper Roller

[18] Bearing using Discrete Wavelet Transform of Vibration Signal, Measurement 46, p.537. 18.

Peng, Z.K., Chu, F.L., 2004. Application of the Wavelet Transform in Machine

[19] Condition Monitoring and Fault Diagnostics: A Review with Bibliography, Mechanical

Systems and Signal Processing 18, p. 199. 19. Prabhakar, S., Mohanty, A.R., Sekhar, A.S.,

2002. Application of Discrete Wavelet

[20] Transform for Detection of Ball Bearing Race Faults, Tribology International 35, and p.793.

20. Shi, D. F., Wang, W. J., and Qu, L. S., 2004. Defect detection for bearings using

[21] Envelope spectra of wavelet transform, ASME Journal of Vibration and Acoustics 126 (4), p.

567. 21. Nikolaou, N. G., Antoniadis, I. A., 2002. Demodulation of Vibration Signals

[22] Generated by Defects in Rolling Element Bearings Using Complex Shifted Morlet Wavelets,

Mechanical Systems and Signal Processing 16(4), p. 677. 22. Qiu, H., Lee, J., Lin, J., Yu, G.,

2006. Wavelet Filter-Based Weak Signature

[23] Detection Method and its Application on Rolling Element Bearing Prognostics, Journal of

Sound and Vibration 289, p. 1066. Junsheng, C., Dejie, Y., Yu, Y., 2007. Application of an

Impulse Response Wavelet

[24] To Fault Diagnosis of Rolling Bearings, Mechanical Systems and Signal Processing 21, p.

920. 24. Kumar, R., Singh, M., 2013. Outer Race Defect Width Measurement in Taper Roller

[25] Bearing using Discrete Wavelet Transform of Vibration Signal, Measurement 46, p 537.

[26] 25. Awal, M. A., Mostafa, S. S., Ahmad, M., 2012. Quality Assessement of ECG Signal Using

Symlet Wavelet Transform, Proceedings of International Conference on Advances in

Electrical Engineering, p.129-134. Chavan, M. S., Mastorakis, N., Chavan, M. N., Gaikwad,

M. S., Implementation of

[27] SYMLET Wavelets to Removal of Gaussian Additive Noise from Speech Signal, Recent

Researches in Communications, Automation, Signal Processing, Nanotechnology,

Astronomy and Nuclear Physics, p. 37-41. L. Ott, An Introduction to Statistical Methods and

Data Analysis.North Scituate, MA:

[28] Duxbury Press, 1977. 28. H. Prashad, “Functional performance of roller bearings for

acceptance in routine

[29] Applications,” J. Inst. Eng. (India), pp. 105–113, Aug. 1995. 29. F. Honaravar and H. R.

Martin, “New statistical moments for diagnostics of rolling

[30] Element bearings,” J. Manuf. Sci. Eng., vol. 119, pp. 425–432, Aug. 1997. R. B. W. Heng

and M. J. M. Nor, “Statistical analysis of sound and vibration signals for monitoring rolling

element bearing condition,” Appl. Acoust., vol. 53, Jan-Mar 1998. pp. 211–226.

[31] Lianhuan Hong, Xiaobo Liu1 and Hongyan Zuo. Compound fault diagnosis of rotating

machinery based on adaptive maximum correlated kurtosis deconvolution and customized



multiwavelet transform. IOP Publishing Ltd. Measurement Science and Technology, Volume

29, Number 11. 4th October 2018.

[32] S. Mitra et.al. Different Measurement Techniques for Detection of Bearing Faults in

[33] Industrial Actuators – Comparative Study. IEEE Calcutta Conference (CALCON), 2017. 78-

1-5386-3745-6/17. Nabhan. A. et.al.,EGTRIB Journal of The Egyptian Society of Tribology.

Volume.12, No. 2,

[34] April 2015, p.15 – 29 (ISSN 2090 – 5882). 34. Yi G., Robert G. P., “Stiffness matrix

calculation of rolling element bearings using a finite

[35] Element/contact mechanics model”, Mechanism and Machine Theory, 51, pp. 32–45, (2013).

35. Dick P., Carl H., Nader S., Alireza M. A., Sarabjeet S., “Analysis of bearing stiffness

[36] Variations, contact forces and vibrations in radially loaded double row rolling element

bearings with raceway defects”, Mechanical Systems and Signal Processing, 50-51, p.139–

160, (2015). Yuh-Tay S., “An envelope analysis based on the resonance modes of the

mechanical system

[37] For the bearing defect diagnosis”, Measurement, 43, pp. 912–934, 2010. 37. Patil M. S., Jose

M., Rajendrakumar P. K., Sandeep Desai, “A theoretical mode to predict the

[38] Effect of localized defect on vibrations associated with ball bearing”, International Journal of

Mechanical Sciences, 52, pp. 1193–1201, (2011). 38. Patel U. A., Shukla R., “Vibrational

Analysis of Self-Align Ball Bearing Having a Local

[39] Defect through FEA and its Validation through Experiment”, International Journal of Modern

Engineering Research, 2, pp. 1073-1080, (2012). 39. Purwo K., Zahari T., “Vibration

Analysis of Defected Ball Bearing using Finite Element

[40] Model Simulation”, the 9th Asia Pasific Industrial Engineering & Management Systems

Conference, (2008).

Documents

RECENT APPROACHES AND MEASUREMENT TECHNIQUES FOR …€¦ · Recent Approaches and Measurement Techniques for the Bearing Fault Analysis – A Review 136 [email protected]