INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 10, ISSUE 06, JUNE 2021 ISSN 2277-8616

A Review of Fault Diagnosis Techniques of Rolling

Element Bearings for Rotating Machinery
Mohsin Hassan Albdery, István Szabó

Abstract: Rolling element bearings are critical components of rotating machines, and fault in the bearing can cause the machine to fail. Bearing failure is
one of the leading causes of failure in various rotating machines used in industry at high and low speeds. As a result, early detection of such defects
could prevent failure of the industrial sector or machinery by replacing rolling element bearing and the severity of damage under operating conditions of
the bearing, which may help avoid machine malfunctioning. Defective bearings cause vibration, and these vibration signals can be used to evaluate the
faulty bearings. This article provides a brief overview of recent trends in bearing fault detection techniques. Finally, it is concluded that vibration analysis
technique and other fault diagnostics and condition monitoring of rolling element bearings, fault detection techniques produce better results.

Index Terms: Fault detection, Rotating Machinery, Rolling Element Bearing, Frequency, Vibration Measurement, Bearing defect, Artificial Intelligent
Method
——————————  ——————————

1 INTRODUCTION
Rotating machines are widely used in industrial sectors due to
their economy, performance, and robustness. However, they
are frequently operated in harsh and hazardous conditions,
resulting in component failure during operation, jeopardizing
worker safety and resulting in economic loss. Over 42% of
mechanical failures are due to bearing failure [2]. Condition-
based maintenance is applied to make proactive maintenance
decisions based on data collected from CM [3]. Additionally,
condition monitoring (CM) and fault detection and diagnosis
(FDD) of rotating machinery (RM) have received significant
attention in recent years [8], [9]. As a result, CM and FDD
have emerged as the most critical and vital aspects of
industrial life [10]. Faulty rolling element bearings are
frequently the source of problems in rotating machinery. Based
on a literature review, rather than relying solely on vibration Fig 1. The amplitude of the anomaly measure versus the time
analysis, it has been determined that a combination of other point for a whole-life bearing [5].
condition monitoring techniques is more accurate for accurate
fault analysis and rolling condition monitoring. [11]. This has The purpose of this article is to provide a brief overview of
rapidly become a critical subject area due to the ease with recent trends in research on rolling element bearing fault
which the health of rolling element bearings can be diagnosis techniques.
determined using specific techniques such as vibration
monitoring. The vibration signature reveals, there is a 2 Modes of Bearing Failure
symptom indicating early- [prolonged note] that information, it A bearing is made up of several components: an inner race, an
serves as a crucial indication of a problem within them. Given outer race, balls, and rolling elements. Each bearing has a
the severe implications of this, it is important to gain a deeper unique set of characteristic frequencies that are dependent on
understanding of the early stages of bearing defect through the geometry of the bearing. Bearing failure can be caused by
the use of these techniques[5].However, the early fault signal various factors, including cracks, mechanical damage, wear
is often too low due to the fact that small, localized damage and tear, corrosion, insufficient lubrication, and so on [4]. The
may generate small periodical impulses. In the meantime, bearings components are gradually deteriorated due to wear.
strong noise interferes with the measured vibration signal in Friction between the contact surfaces increases as a result of
the early stages [13]. For example, Fig. 1 shows a real whole poor lubrication which lead to increase of the temperature of
life bearing run-to-failure process. As seen in the figure, there the bearing components [23]. In below, most of the failure
are three stages in the whole life including the normal modes happen in rolling element bearing as shown in fig.2:
operation stage (I), early fault stage (II), and failure stage (III). 1.Fatigue:
———————————————— It starts with a small crack on the bearing surface (rollers or
 Mohsin Hassan Albdery is currently pursuing PhD degree program in races) due to a change in the material structure caused by
Mechanical Engineering in Doctoral School, Hungarian University of repeated stress in the contact areas.
Agriculture and Life Sciences, Hungary E-mail: 2. Wear:
mohsinhassaniq@gmail.com
 István Szabó is currently Professor and Head of Institute of
Wear is produced by dirt or foreign particles inside the rolling
Technology in Hungarian University of Agriculture and Life in or against the seal rollers because of a deficiency of
Sciences, Hungary lubrication or compression.
3.Electric erosion:
It is damage (in the form of craters) to one of the bearing
components (rollers or races) caused by an electric current
170
IJSTR©2021
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 10, ISSUE 06, JUNE 2021
passing through the bearing.
4.Corrosion:
Corrosion occurs when water or other corrosive agents enter
the bearing through damaged seals, acidic lubricants, or a
rapid change in operating temperature.
5. Plastic deformation:
Plastic deformation occurs primarily when the bearing is
subjected to an excessive load, causing the raceways to
indent.
6.Fracture and cracking:
Fracture and cracking are the results of stress caused by
abrasive
treatment
(impacts) or
cyclic
stress. Additionally, high heating can result in fracture and
cracking.
a. Fatigue b .Wear c.Electrical
erosion
d.Corrosion e. Plastic deformation f.Fracture
3. FAULTS DETECTIONS TECHNIQUES:
Various techniques are used to detect defects in rolling
element bearings; in the following section, we will discuss the
fault detection technique in this field:
3.1 Vibration Measuring Methods
Since machine vibration is an essential component of
structural health monitoring, there is a great deal of interest in
acquiring, analyzing, and quantifying
this parameter to improve reliability, life, quality control,
Fig. 2 Modes of
failure of rolling element bearing [4].
productivity, and safety against catastrophic failure. Although
there are numerous causes of vibrations, unbalancing or
misalignment of rotating parts accounts for approximately 90%
of all problems [1]. Vibration can be defined by three
parameters: amplitude, velocity, and acceleration. The
sensitivity of the sensors used to measure these parameters
varies with the vibration frequency. The general rule uses
IJSTR©2021
www.ijstr.org
ISSN 2277-8616
amplitude sensors to detect low-frequency signals, velocity
sensors for mid-frequency signals, and accelerometers for
high-frequency signals [2]. In Fig. 3 we can see the basic
principle of vibration measurement procedure.
Fig. 3 Basic principle of vibration measurement [23]
Machine vibration measurement is a reliable, non-intrusive
technique for monitoring the condition of a machine during
startup, shutdown, and normal operation [17].
3.1.1Time Domain Data Analysis
The time domain is used to describe the visualization or
analysis of vibration data as a function of time. This format's
primary advantage is that little or no data is lost prior to
inspection. This enables extensive detailed analysis. However,
the disadvantage is that there is frequently too much data to
make fault diagnosis simple and straightforward.
3.1.2 Frequency Domain Data Analysis
The term "frequency domain" refers to the visualization or
analysis of vibration data in terms of frequency. To convert the
time domain vibration signal to the frequency domain, a
Fourier transform is typically used, typically in the form of a
fast Fourier transform (FFT) algorithm. For rolling element
bearing we have four types of frequency measurement as
follow:
A. Ball pass frequency outer race (BPFO):
The term "ball pass frequency outer race" refers to the
fundamental vibration frequency produced when the ball
passes over a defect in the bearing's outer race, such as a
crack [23]. The frequency of ball passes in the outer race can
be calculated as:
(1)
B. The ball pass frequency inner race (BPFI):
vibration frequency produced by the ball passing over a defect
in the bearing's inner race, such as a crack. The frequency of
ball passes in the inner race can be calculated as:
(2)
C. Ball spin frequency (BSF):
The rate of pulse repetition that occurs each time a defective
roller or ball passes over other rolling element bearing
elements is known as the ball spin frequency (BSF).The ball
spin frequency (BSF) can be calculated in the following
171
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 10, ISSUE 06, JUNE 2021
manner:
(3)
D. Fundamental train frequency (FTF):
The fundamental train frequency (FTF) is defined as a defect
that occurs in the cage of a rolling element bearing. The
fundamental train frequency (FTF) can be obtained as:
(4)
Where D represents the pitch diameter, d denotes the ball
diameter, N represents the shaft rotation speed in revolutions
per minute, n represents the number of balls, and θ is the
contact angle.
3.1.3 Time–frequency-domain technique
A few time–frequency domain systems have been developed
that demonstrate the possibility of detecting and diagnosing
bearing problems in some of the more complicated rotating
machines. The sound level to vibration signal ratio is low and a
large number of frequency components are available.
[11],[13],[14]. The time–frequency technique can display the
frequency segments of a vibration signal and distinguish their
time variation characteristics. Procedures in the time-
frequency domain can deal with both non-stationary and
stationary vibration signals [6],[19]. This is the primary benefit
of time-domain techniques over frequency-domain techniques.
3.2
Temperature
Monitoring
Extreme
heat is
generated by distributed defects in the rolling element-bearing
components. As a result, the most straightforward technique
for detecting a fault in rotating machines is to monitor the
temperature of the bearing housing or lubricant. The
advantage of this method is Ability to detect anomalous
working areas. Still, it is inefficient since a substantial rise in
temperature occurs in the last stages of life difficult to detect
impending failure [12].
3.3 Motor current signature:
The frequency spectrum that results from the motor current
can be used to keep an eye on the running condition of the
machine. The change in the various mechanical components
is associated with an increase in electric background noise;
therefore, electric motor current signal-processing techniques
are effective for fault signature detection.
3.4 Wear Debris Analysis:
In this method that incorporates sensitive sensors, the metallic
particles can be located in the lubricants and the source of the
problem, or location of the fault, is established by the
spectrographic analysis[10]. The main benefit of this approach
is simplicity and supply-wise because it is straightforward to
describe and is achievable. The other significant
disadvantages are that it carries the risk of oil contamination
and is hard to find and in the supply system [12].
3.5 Beering Faults Diagnosis Based On Artificial
Intelligent Method (AI)
3..5.1 Artificial Neural Network (ANN)
Artificial neural networks (ANN) have recently gotten a lot of
IJSTR©2021
www.ijstr.org
ISSN 2277-8616
attention in the industrial world. [187, 188]. Additionally, ANN is
used to process and classify data. Similarly, [18] presents an
AI self-adaptive FDD system inspired by genetic algorithms
(GA) and nearest neighbor (NN). Thermography infrared (IRT)
is used to diagnose a variety of roller element bearing
conditions. To find approximation coefficients during the
feature extraction stage, a two-dimensional discrete wavelet
transform (2D-DWT) is used in conjunction with Shannon
entropy. Additionally, GA and nearest neighbour techniques
are used to determine the histograms of selected coefficients
for use as inputs to the feature space selection method. The
cost-effectiveness, non-contact, and non-intrusive nature of
this method is the primary advantages. The multilayer
perceptron (MLP) [10] is a supervised learning neural network
with multiple layers. The authors in [15] describe an approach
for identifying FDD bearing faults using ANN for IM.
Additionally, the proposed pattern identification approach
makes use of two current sensors. Thus, a multilayer
perceptron (MLP) is used, which has one and two hidden
layers.
3.5.2 Deep learning diagnosis Technique
Typically, deep learning techniques are based on complex
architectures of computational elements. A well-known and
widely used example of this type of element is the ANN [15],
which can construct a deep neural network (DNN) with a deep
architecture. Numerous deep learning algorithms have been
successfully applied in fault diagnosis, including autoencoders,
stacked autoencoders [17], DBM, and DBN [16]. Lei et al. [13]
recently demonstrated the use of stacked autoencoders to
directly learn features from mechanical vibration signals. To
classify the health conditions, soft max regression is used.
Combining stacked autoencoders and soft max regression
enables high accuracy in diagnosing bearing faults. Li and
Wang [16] initialize the MLP's initial weights and offsets with
stack autoencoders and provide expert knowledge about
spacecraft conditions. [15] To diagnose rotating machinery
faults, use frequency spectra to train a stacked autoencoder.
The number of conditions determines the size of the output
layer. The effectiveness of the stacked autoencoder is tested
using four roller bearing datasets. Due to the fact that deep
learning-based methods for fault diagnosis are still in their
infancy, they are not as widely used as they are in other fields
[19]. Nonetheless, it holds great promise as a result of its
outstanding performance thus far. As a word of caution, in
practice, the number of parameters increases due to the deep
architecture, increasing the risk of over-fitting. Numerous
techniques have been developed to circumvent this issue,
including early stopping, regularization, and drop out.
3..5.3 Convolution Neural Network (CNN)
CNN is a highly effective deep learning technique that is also
used for fault diagnosis. Due to the fact that the most
frequently encountered data type is a time-domain signal, the
one-dimensional CNN was applied to real-time motor fault
diagnosis. In some cases, data can be presented in a two-
dimensional format (for example, as a time–frequency
spectrum), and then classified using image processing
techniques [18]. However, these methods rely on the expertise
of experts as well. According to recent research on bearing
diagnosis methods, all methods incorporate feature extraction
and fault detection, as illustrated in Fig. 4. When we choose
the BPNN as the classifier, we first reduce the dimensions of
172
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 10, ISSUE 06, JUNE 2021
the data to minimize computational time and avoid the fitting
phenomenon. The majority of fault diagnosis models employ
the stacked auto-encoder when data-driven feature extraction
is used. While the dimension reduction step is eliminated,
signal pre-processing is still required[23].
Fig. 4 Convolutional neural networks [5].
3.5.4 Fault Diagnosis Method Based on SVM
It frequently combines the wavelet method, empirical mode
decomposition [6], [25], and spectral analysis [8] to determine
the time domain and frequency domain statistical properties of
vibration signals, such as root mean square value, kurtosis,
power spectral density, and information entropy. and sideband
index, as the main identification targets. Each Statistical
characteristics of vibration signals can be used to infer their
overall characteristics [15]. However, the local information
contained in the signal data will be masked by the statistical
features, which cannot accurately reflect the signal's local
details. The misclassification ratio is higher for problems
involving multiple classifications and insensitive fault features.
When dealing with large amounts of data, it's all too easy to
fall victim to the disaster of dimensionality.
3.5.5 Combined ANN and SVM
To achieve high diagnostic performance, it has been proposed
to combine ANN and SVM CM and FDD techniques [17]. More
precisely, [20] proposes an FDD approach for rolling element
bearings that is based on the extraction of statistical features
from vibration signals. In this section, statistical characteristics
are derived using advanced signal processing techniques and
central limit theory. Notably, the output feature vector
(statistical feature vector) is used as an input vector for a
classifier that uses ANN and SVM to categorize various types
of faults. As a result, the authors argued in this study that ANN
and SVM could not provide an analytical guarantee for the
FDD classifier's accuracy. Additionally, [23] introduces an FDD
method for ball bearings that incorporates both ANN and SVM.
Additionally, statistical techniques are used to reconstruct the
time domain characteristics of vibration signals. Following this,
the classification stage employs ANN and SVM [24].
4 Conclusions
The purpose of this review article is to provide a general
descriptive overview of the fault detection techniques currently
used in rolling element bearing diagnostics for rotating
machinery. Some methods is discussed in along with its
advantages and disadvantages. Numerous researchers have
IJSTR©2021
www.ijstr.org
ISSN 2277-8616
also demonstrated that using ANN can detect bearing faults at a
high level. It was demonstrated that it is more accurate to
combine other detection techniques for fault analysis and rolling
element bearing condition monitoring rather than relying solely
on vibration analysis. Researchers have developed several
techniques for measuring vibration and are still attempting to
improve signal- processing techniques.
5 Acknowledgments
This work was supported by the Stipendium Hungaricum
Scholarship Programme and the Mechanical Engineering
Doctoral School, Hungarian University of Agriculture and Life
Sciences, Gödöllő, Hungary.
REFERENCES
[1] Bengio, Y., Courville, A., & Vincent, P. (2013).
Representation learning: A review and new
perspectives. IEEE transactions on pattern analysis and
machine intelligence, 35(8), 1798-1828.
[2] Li, K., & Wang, Q. (2015, October). Study on signal
recognition and diagnosis for spacecraft based on deep
learning method. In 2015 Prognostics and System Health
Management Conference (PHM) (pp. 1-5). IEEE.
[3] Chen, X., Wang, S., Qiao, B., & Chen, Q. (2018). Basic
research on machinery fault diagnostics: Past, present,
and future trends. Frontiers of Mechanical
Engineering, 13(2), 264-291..
[4] Yu, W., Yuqing, L., Minqiang, X., & Wenhu, H. (2019). A
Review of Early Fault Diagnosis Approaches and Their
Applications in Rotating Machinery [J]. Entropy, 21(4).
[5] Xu, Z., & Saleh, J. H. (2021). Machine learning for
reliability engineering and safety applications: Review of
current status and future opportunities. Reliability
Engineering & System Safety, 107530.
[6] Liu, Z., Jia, Z., Vong, C. M., Han, J., Yan, C., & Pecht, M.
(2018). A patent analysis of prognostics and health
management (PHM) innovations for electrical
systems. IEEE Access, 6, 18088-18107.
[7] Li, Y., Li, G., Yang, Y., Liang, X., & Xu, M. (2018). A fault
diagnosis scheme for planetary gearboxes using adaptive
multi-scale morphology filter and modified hierarchical
permutation entropy. Mechanical Systems and Signal
Processing, 105, 319-337.
[8] Li, Y., Wang, X., Si, S., & Huang, S. (2019). Entropy based
fault classification using the Case Western Reserve
University data: A benchmark study. IEEE Transactions on
Reliability, 69(2), 754-767.
[9] Zhou, S., & Wang, J. (2018). Erratum:―Dual serial vortex-
induced energy harvesting system for enhanced energy
harvesting‖[AIP Advances 8, 075221 (2018)]. AIP
Advances, 8(8), 089901.
[10] Liang, X., Zuo, M. J., & Feng, Z. (2018). Dynamic
modeling of gearbox faults: A review. Mechanical Systems
and Signal Processing, 98, 852-876.
[11] Jammu, N. S., & Kankar, P. K. (2011). A review on
prognosis of rolling element bearings. International
Journal of Engineering Science and Technology, 3(10),
7497-7503.
[12] Jia, F., Lei, Y., Lin, J., Zhou, X., & Lu, N. (2016). Deep
neural networks: A promising tool for fault characteristic
mining and intelligent diagnosis of rotating machinery with
massive data. Mechanical Systems and Signal
Processing, 72, 303-315.
173
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 10, ISSUE 06, JUNE 2021 ISSN 2277-8616

[13] Lei, Y., He, Z., & Zi, Y. (2011). EEMD method and WNN for
fault diagnosis of locomotive roller bearings. Expert
Systems with Applications, 38(6), 7334-7341.
[14] Li, S., Xin, Y., Li, X., Wang, J., & Xu, K. (2019, May). A
Review on the Signal Processing Methods of Rotating
Machinery Fault Diagnosis. In 2019 IEEE 8th Joint
International Information Technology and Artificial
Intelligence Conference (ITAIC) (pp. 1559-1565). IEEE.
[15] Wenyi, L., Zhenfeng, W., Jiguang, H., & Guangfeng, W.
(2013). Wind turbine fault diagnosis method based on
diagonal spectrum and clustering binary tree
SVM. Renewable Energy, 50, 1-6.
[16] Scheffer, C., & Girdhar, P. (2004). Practical machinery
vibration analysis and predictive maintenance. Elsevier.
[17] Alfredson, R. J., & Mathew, J. (1985). Frequency domain
methods for monitoring the condition of rolling element
bearings. Transactions of the Institution of Engineers,
Australia. Mechanical engineering, 10(2), 108-112.
[18] Taylor, J. I. (1980). Identification of bearing defects by
spectral analysis.
[19] McFadden, P. D., & Smith, J. D. (1984). Vibration
monitoring of rolling element bearings by the high-
frequency resonance technique—a review. Tribology
international, 17(1), 3-10.
[20] Yu, W., Yuqing, L., Minqiang, X., & Wenhu, H. (2019). A
Review of Early Fault Diagnosis Approaches and Their
Applications in Rotating Machinery [J]. Entropy, 21(4).
[21] Simon, L. M. (2011). Fault detection: Theory, methods and
systems. Nova Science Publishers.
[22] Gupta, P., & Pradhan, M. K. (2017). Fault detection
analysis in rolling element bearing: A review. Materials
Today: Proceedings, 4(2), 2085-2094.
[23] Bertolini, M., Mezzogori, D., Neroni, M., & Zammori, F.
(2021). Machine Learning for industrial applications: a
comprehensive literature review. Expert Systems with
Applications, 114820.
[24] Zhang, S., Zhang, S., Wang, B., & Habetler, T. G. (2020).
Deep learning algorithms for bearing fault Diagnosticsx—
A comprehensive review. IEEE Access, 8, 29857-29881.
[25] Zhang, J., Yi, S., Liang, G. U. O., Hongli, G. A. O., Xin, H.
O. N. G., & Hongliang, S. O. N. G. (2020). A new bearing
fault diagnosis method based on modified convolutional
neural networks. Chinese Journal of Aeronautics, 33(2),
439-447.

174
IJSTR©2021
www.ijstr.org