A Review of Condition Monitoring of Motors-May 2017

PUBLISHED BY
World's largest Science,

Technology & Medicine
Open Access book publisher
101,000+
3,000+ INTERNATIONAL 98+ MILLION
OPEN ACCESS BOOKS AUTHORS AND EDITORS DOWNLOADS
AUTHORS AMONG
BOOKS TOP 1% 12.2%
DELIVERED TO AUTHORS AND EDITORS
MOST CITED SCIENTIST FROM TOP 500 UNIVERSITIES
151 COUNTRIES
Selection of our books indexed in the

Book Citation Index in Web of Science™
Core Collection (BKCI)
Chapter from the book Fault Diagnosis and Detection

Downloaded from: http://www.intechopen.com/books/fault-diagnosis-and-detection
Interested in publishing with InTechOpen?

Contact us at [email protected]
Chapter 4
An Assessment on the Non-Invasive Methods for

Condition Monitoring of Induction Motors
Muhammad Irfan, Nordin Saad, Rosdiazli Ibrahim,

Vijanth S. Asirvadam, Abdullah S. Alwadie and
Muhammad Aman Sheikh
Additional information is available at the end of the chapter
http://dx.doi.org/10.5772/67917
Abstract
The ability to forecast motor mechanical faults at incipient stages is vital to reducing
maintenance costs, operation downtime and safety hazards. This paper synthesized the
progress in the research and development in condition monitoring and fault diagnosis of
induction motors. The motor condition monitoring techniques are mainly classified into
two categories that are invasive and non-invasive techniques. The invasive techniques
are very basic, but they have some implementation difficulties and high cost. The non-
invasive methods, namely MCSA, PVA and IPA, overcome the disadvantages associ-
ated to invasive methods. This book chapter reviews the various non-invasive condition
monitoring methods for diagnosis of mechanical faults in induction motor and concludes
that the instantaneous power analysis (IPA) and Park vector analysis (PVA) methods are
best suitable for the diagnosis of small fault signatures associated to mechanical faults.
Recommendations for the future research in these areas are also presented.
Keywords: condition monitoring, fault diagnosis, mechanical faults, bearing distributed

faults, gear faults
1. Introduction
Induction motors are the industry workhorse due to the fact that they are rugged, reliable
and economical. Induction motors are used in industry for conversion of electrical energy
into mechanical energy [1]. As shown in Figure 1, induction motors are being used in various
applications, such as in the nuclear power plants, aviation industry, transportation industry,
mining industry, chemical processing plants, paper mills and the petroleum industry [1].
© 2017 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons
Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.
88 Fault Diagnosis and Detection
Figure 1. Applications of induction motors.
The operation of induction motor in harsh industrial environment affects its reliability and
has been a critical issue. The unexpected breakdown of induction motor might result in the
disturbance of critical services such as medical applications, transportation military opera-
tions and aviation. An unexpected breakdown of a motor might result in costly maintenance
or loss of life in applications where continuous process is needed and where down-time is
not tolerable. The induction motors require only basic maintenance and have a very low fail-
ure rate. However, the burning of the motors causes a great deal of unacceptable production
loss. Consequently, diagnosing incipient faults will prevent the problems of unexpected
breakdowns of the machines and it helps in reducing the maintenance costs. As reported in
[2], 50% of operating cost of manufacturing and processing plants is related to maintenance.
Therefore, this is a major area of concern in industries. The researchers are constantly look-
ing for new techniques to minimize the unexpected machine failures and maintenance cost.
The scheduled replacement, scheduled maintenance and condition-based maintenance are the
basic methods used in the industry for reliable operations of the machines. The scheduled
replacement is a simple but expensive method as the replacement of machine parts is conducted
on a regular basis. In the scheduled maintenance method, the checking and/or overhauling of
the equipment has to be done on a regular basis and this method is widely used in industry
as it is less expensive as compared to the scheduled replacement. The condition-based main-
tenance method determines the machine condition by taking measurements using sensors.
By this method, the time periods between maintenance can be increased and machines can be
monitored continuously so that maintenance can be scheduled on a needed basis. Condition
monitoring (CM) is a method used in condition-based maintenance (CBM). It is an effective
type of predictive maintenance (PM). The main steps involved in PM are shown in Figure 2.
Normally, the condition monitoring maintenance process would be monitoring the specific
parameters like vibration, overheating, over current of equipment for early sign of coming
An Assessment on the Non-Invasive Methods for Condition Monitoring of Induction Motors 89
Figure 2. Activities in predictive maintenance.
failures and to predict the need of maintenance before rigorous failures. CM could be per-
formed through visual inspection or through sophisticated fault diagnosis system. CM is suit-
able for continuous process plants where machine breakdowns can be very costly.
The methods of condition monitoring are categorized into two primary classifications, namely
the offline tests and the online tests. The offline tests are performed by isolating the machine
from main AC power supply. Although this is a comprehensive approach, sometimes causes
unnecessary shutdowns on machines. Alternatively, the sensors which are installed on the
machine are used to detect faults for online condition monitoring and fault diagnosis in an
induction motors during the operation of the machine. The online tests cause fewer distur-
bances than offline tests but the results produced from online testing are more complicated
and their interpretation is difficult than the offline tests. Over the past two decades, there has
been an abundance of research work done in the online condition monitoring techniques for
diagnosing problems in induction motors. For the detection of various faults usually affect-
ing machines, several different techniques have been not only proposed but also used suc-
cessfully. However, a good understanding of the mechanical and electrical properties of the
machine in healthy and faulty conditions dramatically influences the accuracy and reliability
of the online condition monitoring methods [3, 4]. Online condition monitoring techniques
can be classified into two categories: firstly the classical method and secondly the digital
method [4]. In classical method, electromechanical devices are used to protect the motors.
The electromechanical devices are expensive, less efficient, having very slow response and
not reliable as some of the devices have even shorter life than the motor itself. The digital
method is the latest method for the condition monitoring and involves integrated circuits,
micro-controllers, micro-processors and programmable logic controllers.
2. Overview of induction motors
An induction motor is normally composed of the following parts: frame, stator, winding, rotor
and bearings. The structure of induction motor is shown in Figure 3. Induction motors are
Figure 3. The structure of induction motor [5].
used in almost all types of industries. Induction motor is an asynchronous machine made up
of a magnetic circuit that is joined to two electrical circuits; these circuits rotate with respect
to each other. Electromagnetic induction is used to pass power from one circuit to the other.
These electric motors are used to convert electrical energy into mechanical energy [5]. The
conversion of energy is dependent upon the natural presence of the phenomena connecting
magnetic and electrical fields in one side while motion and mechanical force are connected
into the other side. On the basis of types of rotor winding, induction motors can be placed
into two categories. They are the wound-rotor induction motors and squirrel-cage motors [6].
The squirrel cage induction motor is made up of conducting bars that are placed in slots of
the rotor body. These conducting bars are short circuited through end rings. Magnesium, cop-
per, or aluminium alloys are the materials usually used in manufacturing of the rotor bars.
Another kind of rotor is known as a form-wound rotor since it possesses a poly-phase wind-
ing much like that of the winding of the stator. There are three slip rings which are joined to
the winding of the rotor shaft. In a form-wound rotor, the slip rings are joined to a variable
resistance and can restrict the current as well as the heating of the rotor [6].
The squirrel-cage induction motor is economical and robust than the wound-rotor induc-
tion motor. At constant supply of voltages and frequency, squirrel-cage induction motor runs
at a constant speed. In this motor, if there is an increase in the load torque, the speed will
decrease slightly. Therefore, it is appropriate to be used in drive systems that run at a con-
stant speed [5, 6]. However, a variety of applications used in industrial areas need adjustable
speeds drives. Traditionally, it is a direct current (DC) motor that is utilized in adjustable
drive system. However, DC motors are expensive and possess carbon brushes that must be
frequently maintained. As squirrel cage induction motors have no brushes so they are cheap
and are preferred for high speed applications. Furthermore, due to the availability of solid
state controllers, mostly high speed drive systems use squirrel cage induction motors. This
type of induction motor is extensively utilized in drive applications of both low and high
performance due to its versatility and ruggedness. Induction motors are suitable for almost
all commercial and industrial applications due to their construction being so simple and they
have only a few parts, which reduce the cost of maintenance. Applications in both adjustable-
speed drive (ASD) and constant-speed drive are the main uses of induction motors.
Induction motors are often exposed to operating environments that may not be ideal and in
some cases are even harsh. These situations could be due to insufficient cooling, inadequate
lubrication, structure vibration, overload, frequent motor starts and stops, etc. In such situ-
ations, induction motors are put under detrimental stresses which can lead to failure [7, 8].
Because of the significant role that motors play in various applications, improvement in the
reliability of motors is required. The reasons why electric motors fail in industry have been
commonly reported as follows [9–11]
• Wrong-rated power, voltage and current
• Mistakes during repairs
• Unstable supply voltage or current source
• Post the standard lifetime
• Overload or unbalanced load
• Electrical stress from fast switching inverters or unstable ground
• Residual stress from manufacturing
• Harsh application environment
3. Faults in induction motor
As induction motor is most often symmetrical, so faults in the motor normally disturb the
symmetry of the motor. Burning of motors in the industry could be due to following reasons:
thermal overloading, overloading due to undesirable stresses, air-gap eccentricity, speed
oscillations, stator winding failure, broken rotor bars, bearing failure, coupled gear failures
and unbalanced voltages. A concise discussion is made regarding these defects based on how
important they are with regard to the condition monitoring of induction motors. According to
a survey conducted in 2005 by the Electric Power Research Institute, more than 40% burning
of AC motors is due to the failure of bearings (more than 50% burning of the motor is due to
mechanical defects, i.e. bearing defects, gear defects, belt and pulley defects). The summary
of the survey report is shown in Figure 4 [2, 3, 12].
As stated in Peter Vas [13] and P.J. Tavner and J. Penman [14], the defects of the motor are cat-
egorized into two groups.
• Mechanical defects
• Electrical defects
Air gap eccentricities, bearing defects, shaft misalignment and abnormalities at the mechani-
cal transmission system are included in the list of mechanical defects. Broken rotor bar and
Figure 4. EPRI survey report 2005 [2, 3, 12].
stator winding defects are included in the list of electrical defects. The detail of these defects is
discussed in following subsections.
3.1. Mechanical defects
The typical examples of mechanical defects that usually occur in an induction motor are
presented as below.
3.1.1. Bearing defects
Bearings are widely used in rotating machinery across various industries that include paper,
textile, aerospace, nuclear power plants, oil refineries, offshore pumping stations, steel, rail-
ways, construction, mining and renewable energy. The defects in the bearings cause break-
down of rotating machinery, which results in significant economic losses and in certain
situations loss of human lives; for example, when a train derails or an aircraft engine fails
due to a bearing defect. Bearings are typical components found in the motors that are used to
allow for the shaft rotation. Majority of the motor failure is due to the bearing malfunctions.
Bearings as shown in Figure 5 are made up of inner and outer races. Several rolling elements
(balls) are placed in between these two races. Cage is used to keep the balls moving at equal
distance from each other. Normally, stresses developed in the motor causes fatigue in the
bearing races. This fatigue causes localized defects (single-point defects like spalling or pitting
or dents or holes) and distributed defects (generalized surface roughness) in the bearings of
the motor [15–17].
Figure 5. The structure of ball bearing.
The localized defects are induced mainly due to the operational wear (contact fatigue) of
bearings. If the bearings are properly installed, kept free of contamination and well lubricated
during operation then bearing failure will occur only after the pre-determined life (millions
of cycles) of the bearing. This type of failure initially starts in a subsurface at micro-scale level
at a single-point and then due to continuous stress cycles, they eventually cause material to
break. These single-point defects are characterized as spalling or pitting or localized defect
[15–21]. Littman [22, 23] characterized as micro-scale subsurface defects as spalls and macro-
scale surface originated defects as pitting. These types of faults produce impulsive type of
vibration and serve as indication of incipient failure and due to this reason many fault diag-
nosis techniques have been developed to diagnose these types of faults.
The distributed defects are induced due to the manufacturing errors or due to contamination,
improper lubrication, corrosion, electrical fluting and misalignment during running opera-
tion of the bearing. In these types of faults, the magnitude of the contact force between race
and ball varies continuously during shaft rotation. Distributed faults create continuous type
of vibration and cause premature failure of the motor bearings and thus it is important to
develop a suitable condition monitoring scheme to diagnose these types of faults. Figure 6
shows the example of typical localized and distributed defects in bearings of induction motor.
Vibration and noise are generated in the motor due to these defects [24]. The failure of bear-
ings can also occur due to high bearing temperature. The temperature of the bearings should
not increase beyond specific degrees at rated conditions. An example of this would be in
the petroleum and chemical industries where the IEEE 841 standard specifies that the rise in
temperature of the bearings under a rated load should not go above 45°C. Rise in the winding
temperature, improper lubrication, the distribution of the temperature within the motor and
Figure 6. Example of bearing (a) localized defects and (b) distributed defects.
the operating speed of the motor are the main factors for the rise of the bearing temperature.
Consequently, the bearing temperature measurements can give helpful data to estimate the
health of the bearings as well as health of the motor [25, 26].
3.1.2. Air-gap eccentricity
Air-gap eccentricity is a typical defect found in the rotor of the motor. Noise and vibration
in the motor structure are usually produced due to eccentricity. For the healthy motors, the
centre of the rotor and stator bore is perfectly aligned. Moreover, the centre of rotation for the
rotor is the same as the stator bore centre. If the rotor is not aligned centrally, radial forces or
magnetic pull will be developed, which causes the rotor-stator rub resulting in the damage
of rotor and stator [27]. As demonstrated in Ref. [28] air-gap eccentricities are of three kinds
• Dynamic eccentricity
• Static eccentricity
• Mixed eccentricity
A dynamic eccentricity as shown in Figure 7 results in an unbalanced magnetic pull (UMP)

that acts on the rotor. The rotation of the UMP is similar to the rotation of the motor. The UMP
can be easily monitored via current or vibration analysis. On the other hand, eccentricity that
is static possesses a constant pull in only one direction, which also causes a UMP. However, it
is not easy to detect this type of UMP [29].
Sometimes, dynamic and static eccentricities often exist together. Moreover, one can never
assume that ideal centric conditions exist. Therefore, a certain amount of eccentricity is always
Figure 7. Example of (a) static eccentricity and (b) dynamic eccentricity [29].
expected in any real motor. This combination of eccentricities such as dynamic and static is
known as mixed eccentricity.
3.1.3. Abnormalities in mechanical transmission system
Detecting abnormalities at the mechanical transmission system has been a vital area of study
for quite a long time. Mechanical loads and gears are frequently connected to motors and a
variety of faults such as defected gearing system, belt-pulley system and coupling misalign-
ment are possible in these mechanical arrangements. An example of the gear fault is shown
in Figure 8. In some applications like in aircraft, the condition monitoring of gears coupled
with drive system is very much important to enhance the reliability of the gear [30]. Gear
failures tend to occur when a gear is working under high stress conditions. The common gear
faults are related to gear tooth irregularities namely chipped tooth, root crack, spalling, wear,
pitting, tooth surface damage and broken tooth. Gear faults usually have significant effects
on power transmission. They create disablement of the drives, which often causes damage to
Figure 8. Example of the gear defects [30].

other gearbox components such as the shaft, bearing, etc., by pieces of the broken tooth. The
tooth breakage can be due to overload or impact and damage.
3.2. Electrical defects
The following subsections give some of the typical examples of electrical defects in induction
motors.
3.2.1. Stator winding defects
The general belief is that damage in the insulation of the winding turns contributes majority
of the defects related to stator. This kind of defect is known as a ‘stator turn defect’ [31]. In a
symmetrical induction motor, a stator turn fault produces a huge amount of current to flow
through turns which creates too much heat in the shorted turns. The motor will burn if this
heat, which is in direct proportion to the square of the current, is more than the threshold
value [32]. An example of the stator related fault is shown in Figure 9.
In induction motors, the insulation used in stator winding is subjected to degradation due to
contamination, transient voltage stresses on the insulating material, mechanical stresses and
thermal overloading. Notably, thermal stresses are the primary cause for the deterioration of
the insulation in the winding of the stator. Insulation even of the best quality could experience
a rapid failure if the motor is run at a temperature greater than its threshold value. Generally,
the life time of the insulation is decreased by 50% for every 10°C increase over the threshold
value of the temperature of stator winding [33]. Therefore, monitoring the temperature of the
winding of the stator is vital so that the motor will not run at a level greater than its thermal
capacity. To accomplish this, several methods have been introduced. However, these methods
could not identify the exact heating-point at the earliest stage [34, 35].
Some factors that speed up the deterioration of the insulation include the defected bearings,
broken rotor bars, vibrations of the rotor, movement of a coil and misalignment of the rotor
and air-gap eccentricity [36]. Consequently, these mechanical failures should be identified
before they cause the failure of the insulation in the stator winding [37, 38]. Another problem
Figure 9. The stator fault in induction motor [31].

for the insulation is foreign materials such as dust and bearing grease which can cause con-
taminations over the insulation of the stator. This contamination could possibly cause the
reduction in dissipation of heat from the stator winding [39]. Due to this reason it is recom-
mended that all the motors should be kept clean and dry.
3.2.2. Rotor fault
Most often, the rotor bars in lower rated motors are produced by methods of die casting.
However, the rotor bars of high rating motors are manufactured using copper. Producing rotor
bars using methods of die casting has been found to cause a variety of technologically problems.
Asymmetries in the rotor of the induction motors have been found which were caused by either
technological problems, or because of the melting of end rings or bars. On the other hand, there
is abundance of other factors causing the failure of rotors. Some of these are listed below [12, 13]
• Metallurgical stresses that are non-uniform could possibly be created in the assembly of
the cage during manufacturing process and these stresses could result in a failure while the
motor is operating later on.
• When thermal stresses are put on the rotor bar at the start-up of the motor, the rotor might
not be capable of moving longitudinally in its slot.
• Some stresses could be developed on the rotor bars due to heavy end rings.
The reasons mentioned above could cause damage to the bars of the rotor and at the same
time cause the rotor to become unbalanced. Furthermore, asymmetrical rotor currents are
produced due to asymmetry on the cage of the rotor. Because of this, damage to just one rotor
bar could result in damage to the surrounding bars. This damage could then spread, result-
ing in fractures in several rotor bars. Cracks in bars cause overheating of the bar due to which
the bar may break. Consequently, the bars in the surrounding area will begin to carry higher
currents subjecting them to even greater mechanical and thermal stresses. These stresses can
start cracking in rotor bars and rotor lamination will be damaged as well [13]. The distribu-
tion of the temperature throughout the lamination of the rotor is also altered because of the
asymmetry of the rotor. Bar fractures can take place at different areas on the rotor. During fre-
quent starts of the motor, the chances of fracture in the rotor end rings increase [28]. Typical
example of rotor faults is shown in Figure 10.
Figure 10. Example of rotor fault in induction motor [13].

4. Condition monitoring techniques
Continuously evaluating the health status of an industrial plant and its machinery
throughout the entirety of its service is known as condition monitoring. Incipient fail-
ure detection is a vital process by which detection of defects in the early stages of their
development is possible [13]. Fault diagnosis of the induction motor with some com-
prehensive condition monitoring system is becoming even more vital. An early alert
about forthcoming failure is possible through the use of condition monitoring system.
Furthermore, scheduling of preventive maintenance of the machines is also possible.
Optimal preventive maintenance schedules are the result of this and also lead to the least
amount of down-time on the machines [14]. Moreover, condition monitoring system
gives indication to maintenance staff to arrange the required spare parts before s erious
breakdown occur on the machine, thus reduces overall down-time. Consequently, to
improve productivity, reliability and safety of electric machines, a suitable condition
monitoring system is essential. A tremendous significance has been put on by condition
monitoring system in the environment of business because of several reasons that are
listed below [13, 14]:
• To decrease the maintenance cost
• To determine the failure of machinery
• To enhance the reliability of both the machines and their parts
• To optimally use manpower and machine spare parts
• To maximize the performance of the machinery
• To enhance the failure prediction accuracy
The usage of condition monitoring for both mechanical and electrical machinery is not new.
While there have been a variety of techniques developed and improved over time, invasive
techniques such as acoustic emission analysis, noise analysis, thermal analysis, chemical
analysis and vibration analysis, and non-invasive techniques such as motor current signa-
ture analysis (MCSA), stator current Park vector analysis and instantaneous power analysis
(IPA) are considered as the most prominent methods in steady state operation conditions.
The structure representing various motor faults and fault diagnosis techniques is shown in
Figure 11.
It has been observed that even though invasive condition monitoring and fault diagnosis
techniques are optimal for the diagnosis of bearing localized and distributed defects, how-
ever, costly sensors and their associated wiring is the major disadvantage of these methods
[40–45]. For example, vibration sensors (accelerometers, velocity transducers) which are
integral part of this technique are too much expensive. As a result, the use of invasive fault
diagnosis methods is restricted in a variety of applications. This is particularly so with
Figure 11. The structure representing various motor faults and fault diagnosis techniques.
applications using small sized motors as cost has an important role to play when making
the decision as to which technique of condition monitoring is to be used. Another disad-
vantage of these techniques is that sensors need to be installed on the machine, and so it
needs access to machine which is not possible in every application. Sensor needs to be
mounted on the machine rightly for accurate results. As sensors also have some life period
after which they fail, in this aspect, bearing life period is more significant than sensor life
period [46–51].
Thus, the focus of this chapter is on the development of non-invasive condition monitoring
and fault diagnosis method for induction motors.
5. Overview on non-invasive fault estimation parameters
Induction motor defects can be diagnosed via its terminal quantities, such as voltage,
current, discharge and power, measurable outside the motor to give an indication of
its condition. A fault in motor (i.e. in bearings or gears) produces a distortion of the
electromagnetic field in the radial and circumferential plane due to which harmonic fre-
quencies appear in the stator current and in instantaneous power. The stator current or
instantaneous power of the healthy motor would have a single component of supply
frequency (fundamental component). Motor faults will modulate the air-gap flux distri-
bution which causes forward and backward rotating magnetic field and as a result side-
band harmonic frequencies appear around fundamental component. Faults in the motor
will generate one-side band below the supply frequency which is called lower-side band
(LSB) and one side band above the supply frequency which is called upper-side band
(USB).
Even though thermal and vibration monitoring have been utilized for decades, most of the
recent researches have been directed towards electrical monitoring of the motor with empha-
sis on inspecting the stator current of the motor.
5.1. MCSA for bearing localized defects
The MCSA method was utilized for diagnosis of the motor and inverter defects using infor-
mation from the motor stator current. In majority of applications, the stator current of an
induction motor is easily obtainable because it is utilized in the protection of the motors from
over-currents, ground currents. Thus, for the condition monitoring and fault diagnosis via
MCSA, no extra sensors are required. It is a non-invasive method as it does not require direct
access to the motor for its implementation [25, 52–58]. Thus, MCSA for the condition moni-
toring of the induction motors finds its application majorly in nuclear power plants, offshore
pumping stations and defence industry where access to the motor is not possible.
Due to its non-invasive feature, MCSA has received the attention of many researchers and
intensive research has been conducted on the MCSA. Initial efforts in MCSA can be credited to
Schoen et al. [25, 53, 54]. They present a method to diagnose the motor faults based on the spec-
tral analysis of stator current signal. Artificial neural networks were trained to learn the char-
acteristic defect frequencies in current spectrum for the online fault diagnosis. The detection
algorithm was implemented on a custom-designed test rig and its performance was verified on
various fault types. Benbouzid et al. [55–57] addressed the application of MCSA for the diag-
nosis and localization of electrical and mechanical faults of induction motors. The initial steps
taken to investigate the efficiency of MCSA for the purpose of motor fault diagnostic were
discussed. Experimental results clearly illustrate the stator current spectral analysis sensitivity
to induction motor faults. Later, Duque et al. [58] validated the bearing fault detection capabil-
ity of MCSA on inverter-fed induction motor. In the current spectrum analysis of the motor,
running values are compared with baseline values. In real time applications, baseline values
are dependent on the operating conditions. To tackle this issue, Stack et al. [59] proposed new
method which keeps track on baseline data at various operating conditions of the motor. They
used different load conditions and for each case they compare the baseline values with run-
ning values to estimate the health condition of the motor. Along with the FFT method for the
analysis of the spectrum, they used some advance signal processing and pattern recognition
techniques for defecting analysis of the motor. An assessment of monitoring methods used for
detection of bearing localized faults of induction motors was presented by Refs. [38, 60–62].
They proposed stator current monitoring as a very applicable technique in an industrial envi-
ronment. Their research presented the current spectrum analyses along with noise cancellation
method to detect mechanical faults of a variety of motors. They also propose a new method to
damage the bearings via shaft current. The interpretation of the spectrum of the motor current
indicates that this technique can successfully diagnose the bearing localized defects.
An improved stator current-based monitoring scheme was reported in Ref. [63], which
perfectly blends Fourier transform, self-adaptive filter and rotor slot harmonics-based slip
estimation techniques together. The experimental results on the bearing outer race localized
defects demonstrate that the presented scheme is effective. In order to diagnose the incipient
faults in rotor bars, a bilinear time-frequency analysis of stator current along with adaptive
filter technology was performed by Boqiang et al. [64]. The results indicate that the pro-
posed method has the capability to diagnose broken rotor bar faults at their incipient stages.
Application of MCSA for the detection of artificially damaged rolling bearings in asynchro-
nous machine was confirmed by Trajin et al. [65]. Blodt et al. [66] and Terra et al. [67] uti-
lized the stator current analysis and vibration analysis techniques to detect bearing localized
defects. They also proposed a new model for the investigation of the effect of load torque
variations on the stator current spectrum. The experimental results indicated that oscillations
of the torque produce the varying frequency contents which can be observed in the stator
current spectrum. Bayindir et al. [68] and Ioannides et al. [69] presented an automated fault
detection system for the induction motors based on programmable logic controllers. They
used the speed sensor, temperature sensor and current sensor to measure the motor speed,
motor temperature and running current. The system was developed to automatic turn-off the
motor if any one of the measured variable exceeds the preset values. However, the proposed
condition monitoring system proves to be very expensive due to involvement of the sensors.
A novel method for the use of the stator current and efficiency of the motors as pointer of
rolling-bearing defects was proposed by Frosini et al. [70, 71]. Their work describes the exper-
imental results of bearing localized faults in outer and inner race of bearings. The analysis
of the reduction in the efficiency of the motor due to defected bearings was also reported.
Ebrahimi et al. [72] investigated the efficiency of the MCSA for the detection of stator faults
in permanent-magnet synchronous motors. Experimental results indicate that the introduced
method can detect the short-circuit fault incisively. Mehala [73] used the MCSA scheme to
identify the bearing localized defects. FFT spectrum of the motor stator current was obtained
and analysed. Experiments were conducted on the two defect levels (hole sizes 2 and 4 mm)
in outer and inner race of the bearing under no-load and full-load conditions. The results
obtained through the experiments indicate that under no-load condition, the change in ampli-
tude values at characteristic defect frequencies is very small (<5 dB) however for full-load
conditions, the change in amplitude is detectable (>5 dB to <8 dB). Along with FFT analysis of
stator current, the researcher also performs wavelet and Park vector analysis of stator current
for diagnosis of various motor faults.
A novel method was proposed by Romero et al. [74] which merged information entropy
analysis with fuzzy logic inference to diagnose and classify faults like broken rotor bars,
bearing malfunctions, stator unbalance and combinations of faults by analysing stator
current signal. The proposed method shows satisfactory results that prove its suitability
for online detection of single and multiple faults in an automatic way through its hard-
ware implementation. Seera et al. [75] presented a novel approach to detect and clas-
sify comprehensive fault conditions of induction motors using a hybrid fuzzy min–max
(FMM) neural network and classification and regression tree (CART). MCSA method was
applied to form a database comprising stator current signatures under different motor
operating conditions. Comparison of stator current analysis via FFT and wavelet trans-
form was shown in Ref. [76]. Experimental results indicates that stator current a nalysis via
FFT can diagnose bearing localized defects only under steady state conditions, while, it
is unable to diagnose the faults when motor is operating in transient condition. However,
stator current analysis via wavelet transform has the capability to diagnose the faults
under steady state and transient operating conditions of the motor. Recently, a stand-
alone multi-sensor wireless system for continuous condition monitoring of induction
motors has been proposed by Ref. [77]. The proposed wireless system provides a low-
cost alternative to an expensive condition monitoring technology available through data
acquisition equipment.
In a recent study, the fault frequencies inside the stator current spectrum were analysed
through iterative method which provides an efficient fault diagnosis in the non-linear
motor operations [78]. The winding function approach was used by Ref. [79] to diagnose
rotor faults. The theoretical derivations were validated with experimental results. To
enhance the reliability of the fault diagnosis system, Choi et al. [80] proposed a condition
monitoring scheme which is based on speed feedback error management. In Ref. [81] a
cross-correlation-based condition monitoring system was used to diagnose the induction
motor eccentricity faults. The validity of the proposed method was confirmed through real
time experiments performed on digital signal processing (DSP)-based motor drive. In a
recent paper, Aydin et al. [82] proposed a hybrid approach based on multiple wireless sen-
sor systems for fault diagnosis of induction motors. The proposed method was shown to
be useful for analysing and monitoring signals from multiple induction motors. A novel
analytical model based on stator current monitoring was developed by Blodt et al. [83]
for the analysis of bearing defects. The results obtained through spectral analysis of the
measured quantities validate the proposed theoretical approach. Stack et al. [84] presented
various techniques to simulate artificial defects in the bearings of induction motor. Motor
asymmetrical fault signatures were identified by Benbouzid et al. [85] through spectral
analysis of motor current signal. An envelope analysis of the vibration signal was used to
diagnose bearings localized defects [86]. The results achieved with proposed technique are
shown to be strong function of selection signal bandwidth [86]. A wavelet-based condi-
tion monitoring system using analysis of the acoustic emission signal was proposed by
Kang et al. [87]. A generalized fault analysis system in multi-phase machines was pro-
posed by Choi et al. [88].The fault diagnosing capability of the proposed scheme has been
verified mathematically through simulations. Wang et al. [89] has used an improved com-
bination of the Hilbert and wavelet transforms to analyse the incipient bearing localized
defects. Experimental results indicate that the extraction capability of bearing localized
fault frequencies is greatly enhanced by the proposed method. An intelligent fault diagno-
sis scheme based on an independent component analysis was presented by Widodo et al.
[90]. An integrated system for motor bearing fault identification is presented in Ref. [91].
Hwang et al. [91] proposed an integrated fault diagnosis system based on cepstrum coef-
ficient method for feature extraction from motor vibration signals. The experimental results
indicate the effectiveness of cepstrum in diagnosing the bearing health. Neural networks
models with a decision structure are presented in Ref. [92] to analyse the bearing localized
defects. The results show good performance of the implemented model and its ability to
identify the bearing localized faults.
5.2. MCSA for gear defects
Gears are the main part of an electromechanical power transmission system in majority of
industrial applications [93]. Unexpected failures can be prevented through detection of incip-
ient gear faults. This will also help to reduce machine down-time and minimize financial
consequences of gear damages. This cannot be realized without using an efficient fault diag-
nosis and condition monitoring system [94]. Although vibration analysis has been shown to
be the reliable method for gear fault detection but the sensitivity to the installation position
and the background noise due to external mechanical excitations are main drawbacks of the
vibration measurement for gear and researchers are constantly looking to have non-invasive
method for gear fault diagnosis [95, 96]. The gear fault detection using non-invasive method
offers great advantages over invasive techniques principally due to its effective cost and the
need of minimum changes in the system installation. In this context, an extensive research
has been performed during recent years for the gear fault diagnosis using non-invasive
techniques, example, using current and voltage measurements in the vicinity of motor drive
systems [97–101].
As related to this, Blodt et al. [102] examined the detection of mechanical fault-related load
torque oscillations in induction motors using a stator current monitoring. They developed
a theoretical model to show the link between torque oscillation and modulation of a stator
current component. However, they neglect the impact of the gear stiffness on the stator
current. The effect of gearbox characteristic frequencies in the stator current of induction
machine was studied by Ref. [95]. It was shown that the input shaft, layer shaft and output
shaft frequencies in a multi-stage gearbox appear in the electromagnetic torque spectrum.
Also, these harmonics appear as the sideband frequencies around the electric supply fre-
quency of the stator current. The mesh frequencies were also well identified in the mea-
sured stator current. Their experimental investigation indicates that magnitudes of some
rotation and mesh related frequencies are sensitive to the gear tooth fault [95, 103, 104]. A
more rigorous attempt was carried out by Kia et al. [105, 106] to analyse the impact of motor
coupled gear on the stator current spectrum. Their theoretical and experimental investiga-
tions indicate that harmonics appeared in motor current spectrum at mesh frequencies and
mesh-related frequencies for healthy gears. For defected gear, additional harmonics related
to fault-induced mechanical impacts are produced at the rotational frequency. The experi-
mental results indicate that mesh and mesh-related frequencies have very weak signatures
and are suppressed in the noise. The diagnosis of damaged tooth of the gear through MCSA
at motor full-load conditions was reported in Refs. [107–109]. They also presented the com-
parison of diagnosis capabilities of vibration, acoustics and stator current analysis tech-
niques. Stator current has been shown to be the suitable non-invasive method to diagnose
gear faults. A simplified dynamic model considering a realistic behaviour of gear with mini-
mum number of gear mechanical parameters was used to study the effect of gear torsional
vibrations on motor current spectrum [110–112]. The impact of transmission error in gear
was shown to be related to pinion and wheel eccentricities and tooth profile abnormalities
which produces pinion and wheel rotation sidebands around the fundamental and mesh
frequencies.
An attempt has been performed by Ottewill et al. [113] to diagnose the tooth pitting fault in a
multi-stage gear. They have used a low-degree of freedom model for gear dynamic model. It was
verified by Girsang et al. [114] that the gear teeth faults may generate mechanical impacts which
can be observed in the torque and hence in machine electrical signatures. In Ref. [115], Huh et
al. focused on extracting operating point independent fault signatures by using a kinematic
error observer, spatial domain sampling methods and spatial domain signal filtering methods
for gear fault diagnostics of electromechanical actuators. Bogiatzidis et al. [116] reported the
identification of mechanical vibrations due to backlash phenomena appearing between the pin-
ion gear and the girth gear rim of the kiln is realized using the MCSA. The proposed diagnostic
method was tested on under-scale laboratory test rig. It was shown that due to fault in pinion
gear, the pinion rotation frequencies appear around fundamental supply frequency.
An online gear fault diagnosis system was reported in Refs. [108, 117] using a non-invasive
stator current space vector analysis. The proposed algorithm is based on the computation of
the fault index by using fault-related frequencies in the current space vector instantaneous
frequency spectrum. Tests performed with different load levels demonstrate a possible online
condition monitoring and fault diagnosis of gear tooth surface damage fault detection. It has
been demonstrated that incipient faults in gears produce small fault signatures that are some-
times buried in environment noise and thus could not be reliably detected through MCSA.
Recently, a statistical tool called spectral kurtosis with reference has been proposed to define
the electromechanical system healthy state reference. This technique proves its effectiveness
in case of load torque oscillation fault only [118].
Although MCSA has the capability to diagnose gear defects and bearing localized defects,
however, very small change in amplitude occurs at characteristic defect frequencies under
no-load condition. Also in an arbitrary noise condition, the small fault signatures are buried
in noise and it is difficult to discriminate between change in amplitude value due to fault and
random noise [38, 53–72]. Another disadvantage of the MCSA is that sometimes two side
band components (LSB and USB) lie near to the fundamental component and amplitude of
the side band components is suppressed by the highest peak at fundamental component. This
can create misdetection in an online fault diagnosis system [107–118].
6. IPA for motor electrical defects
Since bearings are used to support the rotor during rotation, hence any defect in bearing will
affect the radial movement of rotor. Due to this radial movement, air-gap between rotor and
stator will change to cause magnetic flux variations. These flux variations create modulations
(oscillations) in the instantaneous power which could be observed as two sideband components,
at (2fe – fv), (2fe + fv), and one fault component appears directly at the vibration frequency (fv). The
location of these harmonic frequencies allows the identification of abnormalities in the bearing.
The instantaneous power analysis (IPA) method has been used by several researchers to
detect rotor and eccentricity defects. Ahmad [119] uses the motor current, voltage, flux and
IPA methods to investigate the broken rotor bars and eccentricity defects of the motor. His
study shows that although the rotor bars and eccentricity defect of the motor can be detected
through current, voltage and flux analysis, however, these techniques are not effective under
no-load conditions of the motor. On the other hand, the instantaneous power analysis scheme
diagnosis the rotor and eccentricity defects effectively even under no-load condition of the
Figure 12. The stator current spectrum under no-load conditions of the motor (a) healthy motor and (b) defected bearing
with 2 mm outer race defect.
motor. This is due to the reason that the motor power is the product of the supply voltage
and current so it contains more information than the current and voltage only. In later work,
the superiority of the IPA method over MCSA was confirmed in [120–123] for the diagnosis of
faults in rotor, stator and eccentricity.
The comparison of the MCSA and IPA for 2 mm bearing outer race defect is shown in Figures
12 and 13, respectively. It has been observed from Figures 12 and 13 that the MCSA is not
Figure 13. Instantaneous power spectrum under no-load conditions of the motor (a) healthy motor and (b) defected
bearing with 2 mm outer race defect.
s uitable for detection of defects at lower load levels because very small change in the ampli-
tude value (2 dB) appears at characteristic defect frequencies of 28 and 128 Hz in stator cur-
rent spectrum. However, comparatively large change in amplitude value (7 dB) appears at
characteristic defect frequencies of 21, 79 and 179 Hz in instantaneous power spectrum. Also,
it has been observed that the IPA carries an additional characteristic frequency component
that provides an extra piece of information that can be utilized in a reliable intelligent condi-
tion monitoring system. Thus, IPA has more capability to detect bearing defects at low-load
conditions as compared to MCSA.
7. Impact of environmental noise on decision making of existence of fault

signatures
Incipient defects are defined as slowly developing defects or small unpredictable variations
in the system. They are characterized by small amplitude compared to the useful signal.
Considering the definition of incipient defects, as a slowly developing defect, it is important
to address the issue of detecting these defects at the earliest possible stage meaning smallest
amplitude (smallest severity). It has been observed that the low energy harmonics are pro-
duced due to incipient bearing localized faults and stator current analysis technique proves
to be less efficient to detect low energy fault signatures especially under noisy environment.
The IPA method has the ability to detect low energy fault signatures but in an online testing
system it is difficult to discriminate the amplitude changes due to environment noise and
due to existence of the fault. Therefore, it is unlikely to make decisions without considering
the noise variations. As related to this, in an online fault diagnosis system the environmen-
tal noise modelling has been a practical issue and many studies have been performed to
deal with it.
The averaging of multiple spectrums of stator current was used to eliminate the noise dis-
turbances [124]. Recently, in Ref. [125], Wiener filter is used to eliminate all signatures that
are irrelevant to the motor fault signatures. To achieve the high performance of the condition
monitoring system, a statistical process control (SPC) is used [63]. Golafshan et al. [126, 127]
presented the Singular Value Decomposition (SVD) and Hankel matrix-based de-noising pro-
cess for the elimination of the background noise and the improvement in the reliability of the
fault detection process. However, the proposed method is proven to be computationally inten-
sive. A classical multiple signal classification method has been proposed by Kia et al. [128] to
suppress the noise during the fault diagnosis of the motor. However, the proposed method
takes long computation time to find fault signatures and is affected by the low signal-to-noise
ratio. To solve this problem, an algorithm that is based on zooming in a specific frequency
range of the FFT spectrum was proposed in [129]. Kim et al. [130] proposed the idea for sepa-
rating rotor faults and load oscillations to reduce the false alarm rates in an online fault diag-
nosis system. It was reported in Ref. [131] that Wigner–Ville spectrum analysis based on cyclic
spectral density (CSWVS) was able to separate the bearing fault patterns from random noise
in the vibration signal. As compared with the envelope analysis for rolling element bearing
diagnosis, the strongest element of CSWVS may be its diagnostic ability for bearing with
distributed faults. In the experimental study, the results of envelope analysis, no matter what
filtering techniques are used, fail to extract bearing fault features and might be misinterpreted
as a gear fault. However, CSWVS utilizes the second order cyclo-stationary property of the
vibration signals produced by bearing distributed fault, and clearly extracts its fault features
reducing the masking effect of additive stationary noise. Wang et al. [132] addressed chal-
lenging issues on de-noising and identification of the incipient fault deterioration grade from
the noisy vibration signal for aircraft engine rotor prognostics. The enhanced robust methods
include an adaptive wavelet de-noising technique for weak signature enhancement and corre-
lation dimension for performance degradation assessment. The experimental results verified
that the weak vibration signal features of rotor are successfully revealed and enhanced using
adaptive wavelet de-noising method, and correlation well identify the rotor rub-impact fault
deterioration grade.
Eccentricity faults in induction motor were analysed by Refs. [133, 134] using a cross-cor-
relation method. The pre-determined threshold levels that have been used are as follows:
amplitude less than –60 dB healthy motor, amplitude –60 dB to –40 dB tolerable defects and
amplitude higher than –40 dB severe defects. A statistically derived adaptive threshold defini-
tion was proposed by Toliyat et al. [135]. The experimental results confirm the validity of the
proposed approach to diagnose eccentricity and rotor fault at 70% loading condition of the
motor. Although, the proposed approach considers the noise variations in real time applica-
tions; however, the performance of the proposed approach was not tested under variable
operating points of the motor. In a similar work, Rajagopalan et al. [136] use an adaptive
threshold scheme based on percentage of the fundamental current signal. The advanced sig-
nal processing algorithms are used to diagnose fault signatures [137]. In this study, the thresh-
old is pre-determined based on prior tests before the diagnostic procedure starts. Although,
the statistical performances of the instantaneous noise and bias with respect to motor operat-
ing points are not provided in that study but the suggested solutions present an intuitive way
to design a performance oriented adaptive threshold scheme for an online fault diagnosis
system. An analytical model to detect the incipient faults in the plant process in presence of
an arbitrary noise was recently presented in Refs. [138–141]. Their modelled threshold scheme
that is based on the probability distribution function has shown higher efficiency for incipient
fault detection in noisy environment.
Notably, the pre-determined threshold schemes for reliable diagnosis of electrical faults in
induction motor faults using MCSA presented in Refs. [133–137] could be used to estimate the
fault severity, but these threshold schemes tend to detect sufficiently strong fault signatures
to avoid noise interference. Thus the scheme could not be used to diagnose small fault signa-
tures associated to motor mechanical faults under no-load conditions.
Fournier et al. [142] defined the threshold references in healthy condition of the motor based
on the statistical spectral kurtosis measurements. The threshold scheme produces efficient
results at specific speed and load levels of the motor. However, this threshold scheme does
not consider instantaneous noise variations with respect to motor operating point (load,
speed), so errors in the detection occur when operating point of the motor changes. Picot et
al. [143] proposed a threshold scheme for the detection of small fault signatures related to
bearing cage defects in permanent magnet synchronous motor (PMSM). The noise variance of
the measured motor current spectrum was minimized by integration process and threshold
was designed by calculating mean and standard deviation of the frequency bins. Although,
the proposed method was shown to produce good performance at various speeds (variable
supply frequency) of the PMSM, however, it has some limitations. Firstly, the performance of
their designed threshold was not measured at various load conditions. Secondly, the designed
threshold scheme is dependent on the sampling size and window size. The large window
size may attenuate the amplitude of bearing fault signatures. The impact of noise variations
with respect to load variations on the reliable decision making of the existence of small fault
signatures needs to be investigated. Notably, this work provides intuitive way towards the
design of performance-based decision making system independent of environment impacts
and motor operating points.
Limitations of the existing non-invasive fault diagnosis techniques considering environment
impacts are described in Table 1.
Reference Fault type Threshold design Limitations

technique
[133, 134] 2008, Eccentricity Pre-determined • Does not count non-linear noise variations
2011 threshold with respect to motor speed and load
• Detection with two sideband fault signa-
tures have ambiguities
[135] Eccentricity, Rotor Statistically derived • Does not count non-linear noise variations
2012 threshold with respect to motor speed
[136] Eccentricity Pre-determined • Does not count non-linear noise variations

2006 threshold with respect to motor speed and load
[137] Bearing, Rotor Zero input test-based • Does not count non-linear noise variations
2006 statistical analysis with respect to motor speed and load
[142] Unbalance Reference-based • Does not count non-linear noise variations

2013 statistical analysis with respect to motor load
(spectral kurtosis)
[143] Bearing cage Defect Reference-based • Does not count non-linear noise variations
2014 statistical analysis (mean with respect to motor load
and standard deviation)
Table 1. Limitations of previously used MCSA-based fault diagnosis techniques considering environment impacts.
8. Park vector analysis for bearing localized defects
Park vector analysis (PVA) of motor current which is mainly used for field oriented control of
induction motor has received a growing attention in the recent years in condition monitoring
applications. Few researchers have used PVA with emphasis on the bearing localized faults,
rotor faults, eccentricity faults and stator faults [144–147]. In Park’s transformation process,
the three balanced AC quantities (Ia, Ib, Ic) are reduced to two DC quantities (Id, Iq). In analysis
of three-phase synchronous machines, Park’s transformation transfers three-phase stator and
rotor quantities into a single rotating reference frame to eliminate the effect of time varying
inductances [144, 145].
The Park vectors under ideal conditions of the motor represent a circular pattern centred at
the origin. However, the shape of the circle changes in presence of the faults inside the motor
and the various types of faults generate various types of shapes. Thus by detection of various
patterns, one can analyse the type of fault inside the motor. The Park’s transform is a graphi-
cal method to analyse the status of motor and does not require analytical expressions to cal-
culate any specific defect frequency.
Zarie et al. [144] performed experiments on induction motor to diagnose bearing localized
and extended defects (multiple holes in inner or outer race). Classification of the various
faults was achieved through utilization of neural networks. It was concluded in their study
that the proposed method provides a powerful and general approach to incipient fault
detection. Spyropoulos et al. [145], Laughman et al. [146] and Parra et al. [148] reported
that impact of the motor stator related faults could be examined through the Park vector
transformation approach. It was claimed that by using this approach, effects of stator faults
can be differentiated from time-varying loads. Salem et al. [147] measure the thickness of
Park vector curve using curve splitting factor. The experimental results on bearing outer
race defect indicate that thickness of the Park vector curve increases due to fault. Rezig et
al. [149] and Salem et al. [150, 151] conducted experimental study to diagnose the bear-
ing localized defects and eccentricity faults through Park vector transform of motor stator
current. Experimental results indicate that Park vector transform method can detect the
occurrence of faults but unable to identify the fault locations. In a recent study, Kuruppu
et al. 152] demonstrated that Park vector analysis method has capability to diagnose stator
inter-turn faults in inverter-fed field-oriented control motors. Proof of accurate fault detec-
tion capability for a wide speed range is presented through simulation and experimental
results.
The example of Park vector analysis of stator current for bearing localized defects is
shown in Figure 14. It can be observed from Figure 14 that bearing localized defects
cause an increase in thickness of the Id, Iq curves. The increase in curve thickness is due to
the harmonics induced by the localized defects inside the bearing of the motor. Therefore
the analysis of the Id, Iq curves can be a useful means for the fault diagnosis of induction
motors.
Figure 14. Example of Park vector analysis of stator current for (a) healthy bearing, (b) outer race localized defect and
(c) inner race localized defect.
9. Non-invasive method for diagnosis of bearing distributed defects
Most of the published work on the bearing condition monitoring via non-invasive techniques
as discussed in focused on diagnosis of bearing localized defects. It is based on some char-
acteristic fault frequencies that appear in spectrum of the stator current or instantaneous
power. However, the absence of clear characteristic fault frequencies should not be inter-
preted as a completely healthy condition of the bearing. On the other hand, the bearing dis-
tributed faults produce unpredictable broadband effects which are not necessarily related
with specific fault frequencies. These faults are common in industry, while they are often
neglected in the research literature.
Time-domain analysis is a useful feature extraction tool for condition monitoring and fault diag-
nosis of electrical motors. Time domain averaging (TDA) is a traditional and typical method
to detect fault signals in electrical motors. It extracts a periodic component of interest from a
noisy compound signal. Data-clustering techniques are used to extract an average pattern that
serves as the mechanical imbalance indicator. Zhang et al. [153] proposed the idea of creating
artificial distributed defects in bearings of induction motor via externally applied shaft current.
This idea was important as the defects would be sorted according to the kind of signatures
generated by defects instead of where the defects were located physically. Most of the condi-
tion monitoring approaches concerning fault detection found in the literature have been mainly
focused on identifying single-point defects. Indeed this class of defects is quite important, how-
ever, an approach that is comprehensive and robust must have the capability of detecting not
only single-point defect but also distributed defects in the bearing. The data obtained from 10
bearings failed by their proposed method, was analysed in time domain. It was shown that
drastic variations in machine vibration and stator current occur in case of defected bearings. A
method to segregate the bearing localized and distributed defects based on the time domain
analysis of vibration, stator current and acoustic emission signals was presented by Navarro et
al. [154]. The RMS values of the multi-sensory signal for healthy bearing, bearing localized and
distributed defects were collected and faults were classified based on fuzzy inference analysis.
In Refs. [155, 156], it was reported that generalized roughness in bearings of the motor pro-
duces a frequency spreading of the characteristic fault frequencies, thus making it difficult to
detect with MCSA method. In the papers, it was proposed to use a statistical analysis of typical
bearing faults in order to identify the spreading bandwidth related to bearing surface rough-
ness faults, relying on current and vibration measurements only. A diagnostic index based on
computation of energy in the previously defined bandwidth was used to diagnose bearing
surface roughness faults. The proposed method was validated experimentally with vibration
and current signals, with robust and reliable results. However, implementation of their pro-
posed method needs computational efforts and implementation of this method for inner and
outer race surface roughness was not given. In a recent study, Dalvand et al. [157] analyse bear-
ing distributed defects in inner and outer race of bearing simultaneously. They use statistical
analysis of the instantaneous frequency of motor voltage space vector (MVSV) and vibration
signal. The statistical indices like mean value (M), RMS value, standard deviation (SD), global
kurtosis (GK), skew factor (SF) and crest factor (CF) were utilized to analyse the measured sig-
nal. It was shown that mean value and RMS are not proper indices to analyse the distributed
defects in bearings because the values of these indices for healthy and defected bearings are not
significantly different. However, the GK, SF and CF are proper indices to diagnose bearing dis-
tributed faults. Experimental results indicate that although the statistical time domain analysis
of MVSV and vibration signals have capability to diagnose the bearing distributed defects;
however the reliability of MVSV is better than vibration monitoring. The analysis of individual
defect in bearing elements was not presented in their study.
10. Condition monitoring issues
The literature review has shown that the invasive methods like vibration analysis, acoustic
emission, noise analysis, chemical analysis and temperature measurements were applied to
protect motors from various faults. However, sensors used in these methods are very expen-
sive. Another disadvantage of these methods is that they require access to machine for sensor
installation which is not possible in every application. Also it requires special expertise for
the proper installation of sensors on the machine for accurate results. Therefore sensor-based
condition monitoring methods are not reliable.
Review papers presented in Refs. [158–163] on the condition monitoring techniques for the
induction motors demonstrate that extensive research has been carried out in the last decade
on the development of a non-invasive condition monitoring system for diagnosis of gear
defects and bearing localized faults. It has been found that the mechanical vibration produce
characteristic frequencies in stator current spectrum. The magnitude of the specific character-
istic frequencies increases with the increase of vibration. Every type of the motor defect has
its own unique characteristic defect frequencies. The MCSA has been extensively used to find
out these frequency modulations.
The implementation of MCSA for motor condition monitoring is not so complex. However,
low amplitude fault signatures are induced due to incipient faults under no-load condition and
thus could not be reliably diagnosed through MCSA. Recent developments in non-invasive
condition monitoring scheme demonstrate that PVA of motor stator current has the capability
to diagnose the motor faults at incipient stages. It was shown that the proposed method would
give good analysis of machine faults even if accurate fault frequency information is unavail-
able. In recent years, IPA method to diagnose rotor, stator and eccentricity faults at incipi-
ent stages conditions was proposed by few researchers. However, applicability of the IPA for
mechanical fault diagnosis in induction motor has not been reported in literature [164–170].
The incipient defect diagnosis in the machinery with unknown distribution of measured sig-
nals and unknown changed parameters is an important issue. However, most of the fault
diagnosis techniques need a priori knowledge on the signal distribution, changed parameters
and the changed amplitude. Fault detection plays a key role in enhancing today’s technologi-
cal system’s high demands for performance, productivity and security. The sensitivity of the
condition monitoring and fault diagnosis methods depends on the application of main goals.
When productivity is the main goal then the sensitivity requirement of condition monitoring
and fault diagnosis system is weak and only large defects should be detected. However, when
security is the main goal, undetected defects even with very small severity assessment may
result in catastrophic growing failures. For online fault diagnosis, when safety is the main
priority, it is crucial to be able to diagnose incipient faults in presence of environment noise.
Although IPA has some implementation advantages over MCSA for incipient fault detection
but the signatures of IPA are affected by industrial noise perturbations. There is much scope
for research to improve the capabilities of IPA in reliable online fault diagnosis in an arbitrary
environment noise.
Notably, the substantial number of research studies has been found focusing into detection of
bearing localized faults via non-invasive condition monitoring methods. Nonetheless, bear-
ing distributed defects does not produce localized defect frequencies and thus could not be
analysed via these methods. Recently, few studies have been found focusing on time domain
analysis of stator current signal and extracting the features like RMS and kurtosis values.
However, complexity involved in time domain analysis of stator current signal is the major
drawback. As PVA performs the graphical analysis of stator current and does not require
frequency information for fault detection thus its capability to diagnose bearing distributed
defects should be further researched.
The advantages and drawbacks of the existing non-invasive condition monitoring methods
are presented in Table 2.
No CM technique Advantages Drawbacks Reference

1. MCSA • Inexpensive • Unable to detect defects at incipient [25, 38, 52–73]
stages
• Non-invasive
• Fault detection affected by environment
noise
2. PVA • Inexpensive • Further investigation required to test its [144–152, 171]

capability for bearing distributed defects
• Non-invasive
3. IPA • Inexpensive • Fault detection affected by environment [119–123]

noise
• Non-invasive
• Capability to diagnose mechanical faults
• Can detect rotor needs to be tested
and eccentricity
defects at incipi-
ent stages
Table 2. Advantages and drawbacks of non-invasive condition monitoring methods.
11. Future directions
Based on the findings, there are further problems to be considered both in the development of
the technique and in the experimental design. These include
• Development of non-invasive condition monitoring scheme to diagnose faults in an arbi-
trary environment noise conditions
The non-invasive IPA has shown some implementation advantages over MCSA to
reliably diagnose mechanical faults. However, small fault signatures under low load
conditions are buried in noise and it is difficult to discriminate amplitude changes
due to fault and due to noise. Thus, there is a need to develop a robust non-invasive
condition monitoring scheme to make reliable decisions on the existence of fault
signatures.
• Development of non-invasive condition monitoring scheme to diagnose bearing distrib-

uted faults
Most of the published work on the bearing condition monitoring via non-invasive tech-
niques is focused on diagnosis of bearing localized defects. It is based on the some
characteristic fault frequencies that appear in spectrum of the stator current or instanta-
neous power. Nonetheless, bearing distributed defects does not produce characteristic
defect frequencies and thus could not been analysed via MCSA or IPA. These faults are
common in industry, while they are often neglected in the research literature. As PVA
performs the graphical analysis of stator current and does not require characteristic
defect frequencies information for fault detection thus its capability to diagnose bear-
ing distributed defects should be further researched.
• Development of hand-held instrument for on-site machine condition monitoring
The IPA and PVA algorithms could be embedded to develop a hand-held instrument for
on-site condition monitoring of induction motors. This will help to reduce the installation
cost of the data acquisition system especially for small and medium scale industrial use.
• Reliability tests for mainstream industries
The developed IPA and PVA algorithms should be expanded and utilized to analyse
faults in turbines and generators. For consideration to be acceptable by the main stream
industries, a reliability test of the developed algorithms should be conducted on large
size motors (more than 40 HP).
• Multiple bearing faults analysis
Even though the two techniques as proposed i.e. the instantaneous power analysis and
Park vector analysis can monitor the conditions of induction motors from various scenar-
ios, however these are not sufficient for most applications because outcomes of the analy-
sis are based on the assumptions that each fault occurs independently. The extension of
the approach as proposed in this work, to understand how each approach reacts to a com-
bination of several faults (e.g. stator, rotor, eccentricity, bearings, etc.), would be useful.
12. Conclusion
The ability to forecast motor faults at incipient stages is vital to reducing maintenance costs,
operation down-time and safety hazards. This paper synthesized the progress in the research
and development in condition monitoring and fault diagnosis of induction motors. The
review presented in this paper has contributed to an improved understanding of the devel-
opment of practical non-invasive condition monitoring and diagnostic tool for induction
motors. The non-invasive methods, namely MCSA, PVA and IPA overcome the disadvantages
associated to invasive methods. The MCSA, PVA and IPA can successfully diagnose the gear
and bearing localized defects. However, further research is required to analyse bearing dis-
tributed faults via non-invasive condition monitoring methods. It has been observed that IPA
has some implementation advantages over MCSA for incipient fault detection but IPA can be
affected by environment noise perturbations. There is much scope for research to improve the
capabilities of IPA in reliable online fault diagnosis system operating in an arbitrary environ-
ment noise. Furthermore, to realize the greatest economic and implementation benefits of IPA
and PVA, it is important to design a hand-held condition monitoring system based on IPA
and PVA techniques by considering the asset management perspective.
Acknowledgements
The authors acknowledge the support from Universiti Teknologi PETRONAS for providing
Universiti Research Innovation Fund (URIF) and the Ministry of Higher Education (MOHE)
Malaysia for the award of the Prototype Research Grant Scheme (PRGS).
Author details
Muhammad Irfan1*, Nordin Saad2, Rosdiazli Ibrahim2, Vijanth S. Asirvadam2, Abdullah S.

Alwadie1 and Muhammad Aman Sheikh2
*Address all correspondence to: [email protected]
1 Faculty of Engineering, Electrical Engineering Department, Najran University, Kingdom of

Saudi Arabia
2 Electrical and Electronics Engineering Department, Universiti Teknologi PETRONAS,

Bandar Seri Iskandar, Tronoh, Perak, Malaysia
References
[1] Irfan M, Saad N, Ibrahim R, Asirvadam VS, Magzoub M. An online fault diagnosis system
for induction motors via instantaneous power analysis. Tribology Transactions. 2016.
[2] Irfan M, Saad N, Ibrahim R, Asirvadam VS, Hung NT, Magzoub MA. Analysis of bear-
ing surface roughness defects in induction motors. Journal of Failure Analysis and
Prevention. October 2015;15(5):730–736
[3] Irfan M, Saad, Ibrahim R, Asirvadam VS, Magzoub M. An intelligent fault diagno-
sis of induction motors in an arbitrary noisy environment. Journal of Nondestructive
Evaluation. March 2016;35(12)
[4] Irfan M, Saad N, Ibrahim R, Asirvadam VS, Hung NT. Analysis of bearing outer race
defects in induction motors. 5th International Conference on Intelligent and Advanced
Systems (ICIAS); 2014
[5] Kia H, Henao SH, Capolino GA. Trends in gear fault detection using electrical signature
analysis in induction machine-based systems. IEEE Workshop on Electrical Machines
Design Control and Diagnosis (WEMDCD); 2015
[6] Sen PC. Principles of Electric machines and Power Electronics. John Wiley and Sons;
1989
[7] Irfan M, Saad N, Ibrahim R, Asirvadam VS. Development of an intelligent condition
monitoring system for AC induction motors using PLC. IEEE Business Engineering and
Industrial Applications Colloquium (BEIAC); 2013.
[8] Pineda M, Puche R, Riera M. Motor condition monitoring of induction motor with
programmable logic controller and industrial network. IEEE Proceedings on Power
Electronics; September 2011.
[9] IAS Motor Reliability Working Group. Report of large motor reliability survey of indus-
trial and commercial installations—Part I. IEEE Transactions on Industry Applications.
July/Aug 1985;IA-21:853–864
[10] IAS Motor Reliability Working Group, Report of large motor reliability survey of indus-
trial and commercial installations—Part II. IEEE Transactions on Industry Applications.
July/Aug 1985;IA-21:865–872
[11] IAS Motor Reliability Working Group. Report of large motor reliability survey of indus-
trial and commercial installations—Part III. IEEE Transactions on Industry Applications.
Jan. /Feb. 1987;IA-23:153–158
[12] Improved motors for Utility applications. Vol. 5. EPRI Publication EL-2678; October 2005
[13] Vas P. Parameter Estimation, Condition Monitoring, and Diagnosis of Electrical Mac
hines. Oxford: Clarendon Press; 1993
[14] Tavner PJ, Ran L, Pennman J, Sedding H. Condition Monitoring of Rotating Electrical
Machines. Letchworth, England: Research Studies Press Ltd.; 2008
[15] Fatigue and Fracture, ASM Handbook. Vol. 19. ASM International; 1996.
[16] Irfan M, Saad N, Ibrahim R, Asirvadam VS. A non-invasive method for condition moni
toring of induction motors operating under arbitrary loading conditions. Arabian
Journal for Science & Engineering. September 2016;41(9):3463–3471
[17] Suresh S. Fatigue of Materials. Cambridge, UK: Cambridge University Press; 1998
[18] Tallian TE. Failure Atlas for Hertz Contact Machine Elements. New York: ASME Press;
1992
[19] Olver AV. The mechanism of rolling contact fatigue: an update. Proceedings of the
Institution of Mechanical Engineers. Part J: Journal of Engineering Tribology. 2005;
219(5):313–330
[20] Halme J, Anderson P. Rolling contact fatigue and wear fundamentals for rolling bearing
diagnostics—state of the art. Proceedings of the Institution of Mechanical Engineers.
Part J: Journal of Engineering Tribology. 2009;224(4)377–393
[21] Sadeghi F, Arakere NK, Jalalahmadi B, Slack TS, Raje N. A review of rolling contact
fatigue. Journal of Tribology. 2009;131(4)
[22] Littmann WE. The Mechanism of Contact Fatigue. In: Ku PM, editor. Interdisciplinary
Approach to the Lubrication of Concentrated Contacts. NASA SP-237. Washington, DC:
National Aeronautics and Space Administration; 1970. pp. 309–378
[23] Littmann WE, Widner RL. Propagation of contact fatigue from surface and subsurface
origins. Journal of Basic Engineering. 1966; 88(3):624−636
[24] Nandi S, Toliyat HA. Condition monitoring and fault diagnosis of electrical machines—
a review. 34th Annual Meeting of the IEEE Industry Applications. 1999. pp. 197–204
[25] Schoen RR, Habetler TG, Kamran F, Bartheld RG. Motor bearing damage detection
using staor current monitoring. IEEE Transactions on Industry Applications. 1995;31(6):
1274–1279
[26] Eschmann P, Hasbargen L, Weigand K. Ball and Roller Bearings: Their Theory, Design,
and Application. London: KG Heyden; 1958
[27] Dorrell DG, Thomson WT, Roach S. Analysis of air-gap flux, current, and vibration sig-
nals as function of a combination of static and dynamic eccentricity in 3-phase induction
motors. IEEE Transactions on Industry Applications. 1997;33:24–34
[28] Wu S, Chow TWS. Induction machine fault detection using SOM-based RBF neural net-
works. IEEE Transactions on Industrial Electronics. February 2004;51(1),183–194
[29] Bradford M. Unbalanced magnetic pull in a 6-pole induction motor. IEEE Proceedings.
1968;115(11):1619–1627
[30] Dalpiaz G, Meneghetti U. Monitoring fatigue cracks in gears. NDT & E International.
1991;24(6):303–306
[31] Sottile J, Kohler JL. An on-line method to detect incipient failure of turn insulation
in random-wound motors. IEEE Transactions on Energy Conversion. December,
1993;8(4):762–768
[32] Sheikh MA, Nor NM, Ibrahim T. A new method for detection of unbalance voltage sup-
ply in three phase induction motor. Jurnal Teknologi. 2016;78(5)
[33] Sheikh MA, Nor NM, Ibrahim T. Detection of unbalance voltage supply through rotor har-
monics symbolic dynamics. The 6th International Conference on Intelligent & Advanced
Systems, August 2016.
[34] Farag SF, Bartheld RG, May WE. Electronically enhanced low voltage motor protection
and control. IEEE Transactions on Industry Applications. 1994;29(1):45–51
[35] Lee SB, and. Habetler TG. An online stator winding resistance estimation technique for
temperature monitoring of line-connected induction machines. IEEE Transactions on
Industry Applications. May 2003;39(3):685–694
[36] Bonnet AH, Soukup GC. Cause and analysis of stator and rotor failures in three-phase
squirrel case induction motors. IEEE Transactions on Industry Application. July/August
1992;28(4)921–937
[37] Roux WL, Harley RG, Habetler TG. Detecting rotor faults in permanent magnet syn-
chronous machines. SDEMPED’03; August 2003; Atlanta, GA; pp. 198–203
[38] Stack JR, Habetler TG, Harley RG. Bearing fault detection via autoregressive stator cur-
rent modeling. IEEE Transactions Industry Applications. May 2004;40(3):740–747
[39] Cash MA. Detection of turn faults arising from insulation failure in the stator windings
of AC machines [thesis]. USA: Georgia Institute of Technology; 1998
[40] Kankar PK, Sharma SC, Harsha SP. Nonlinear vibration signature analysis of a high
speed rotor bearing system due to race imperfection. Journal of Computational and
Nonlinear Dynamics. January 2012;7
[41] Babu CK, Tandon N, Pandey RK. Vibration modeling of a rigid rotor supported on the
lubricated angular contact ball bearings considering six degrees of freedom and wavi-
ness on balls and races. Journal of Vibration and Acoustics. February 2012;134
[42] Immovilli F, Cocconcelli M, Bellini A, Rubini R. Detection of generalized-roughness bear-

ing fault by spectral-kurtosis energy of vibration or current signals. IEEE Transactions
on Industrial Electronics. November 2009; 56(11)
[43] Dolenc B, Boškoski P, Juričić D. Distributed bearing fault diagnosis based on vibration
analysis. Mechanical Systems and Signal Processing. 2015;66-67:521–532
[44] Petersen D, Howard C, Sawalhi N, Ahmadi AM, Singh S. Analysis of bearing stiff-
ness variations, contact forces and vibrations in radially loaded double row rolling
element bearings with raceway defects. Mechanical Systems and Signal Processing.
2015;50:139–160
[45] Petersen D, Howard C, Prime Z. Varying stiffness and load distributions in defective
ball bearings: Analytical formulation and application to defect size estimation. Journal
of Sound and Vibration. 2015;337:284–300
[46] El-Thalji I, Antunen EJ. Fault analysis of the wear fault development in rolling bearings.
Engineering Failure Analysis. 2015;57:470–482
[47] Prieto MD, Cirrincione G, Espinosa AG, Ortega JA, Henao H. Bearing fault detection
by a novel condition-monitoring scheme based on statistical-time features and neural
networks. IEEE Transactions on Industrial Electronics. August 2013;30(8)
[48] Jin X, Zhao M, Tommy WSC, Pecht M. Motor bearing fault diagnosis using trace ratio lin-
ear discriminant analysis. IEEE Transactions on Industrial Electronics. May 2014;61(5)
[49] Shah DS, Patel VN. A review of dynamic modeling and fault identifications methods for
rolling element bearing. Procedia Technology. 2014;14:447–456
[50] Singh S, Howard CQ, Hansen CH. An extensive review of vibration modelling of rolling
element bearings with localised and extended defects. Journal of Sound and Vibration.
June 2015
[51] Tanver PJ. Review of condition monitoring of rotating electrical machines. IET Electric
Power Applications. 2008;02(4):215–247
[52] Zhou W, Habetler TG. Bearing condition monitoring methods for electric machines.
IEEE Transaction on Diagnosis for Electric Machines, Power Electronics and Drives,
2007.
[53] Schoen RR, Lin BK, Habetler TG, Schlag JH, Farag S. An unsupervised, online system for
induction motor fault detection using stator current monitoring. IEEE Transactions on
Industry Applications. 1995;31(6):1280–1286
[54] Schoen RR, Habetler TG. Evaluation and implementation of a system to eliminate arbi-
trary load effects in current-based monitoring of induction machines. IEEE Transactions
on Industry Applications. 1997;33(6):1571–1577
[55] Benbouzid MEH, Nejjari H, Beguenane R, Vieira M. Induction motor asymmetrical

faults detection using advanced signal processing techniques. IEEE Transactions on
Energy Conversion. June 1999;14(2)147–152
[56] Benbouzidi MEH, Viera M, Theys C. Induction motors faults detection and localiza-
tion using stator current advanced signal processing techniques. IEEE Transactions on
Power Electronics. January 1999;14(1):14–22
[57] Benbouzid MEH. A review of induction motors signature analysis as a medium for
faults detection. IEEE Transaction on Industry Electronics. 2000;47(5)984–993
[58] Duque O, Pérez M, Moríñigo D. Detection of bearing faults in cage induction motors
fed by frequency converter using spectral analysis of line current. IEEE Conference on
Electric Machine Drives. May 2005;17–22
[59] Stack JR, Habetler TG, Harley RG. Bearing fault detection via autoregressive stator cur-
rent modeling. IEEE Transactions on Industry Applications. May/Jun. 2004;40(3):740–747
[60] Bellini A, Immovilli F, Rubini R, Tassoni C. Diagnosis of bearing faults in induction

machines by vibration or current signals: A critical comparison. IEEE IAS Annual
Meeting. Edmonton, Alberta, Canada; October 2008; pp. 1–8.
[61] Zhou W, Habetler TG, Harley RG. Bearing fault detection via stator current noise can-
cellation and statistical contro. IEEE Transaction on Industrial Electronics. December
2008;55(12):4260–4269
[62] Stack JR, Habetler TG, Harley RG. Bearing fault detection via stator current modeling.
IEEE Transactions on Industry Applications. 2004;40(3):740–747
[63] Liling S, Boqiang X. An improvement of stator current based detection of bearing fault in
induction motors. IEEE Conference Record, Industry Applications Conference and 42nd
IAS Annual Meeting 2007.
[64] Boqiang X, Huihuan Z, Liling S, Junzhong S. Weak-signal detection and the applica-
tion in detection of electric motor faults. In: Proceeding of International Conference on
Electrical Machines and Systems; Seoul, Korea. Oct. 8-11, 2007
[65] Trajin B, Regnier J, Faucher J. Indicator for bearing fault detection in asynchronous motors
using stator current spectral analysis. IEEE International Symposium on Industrial
Electronics; 2008.
[66] Blodt M, Granjon P, Raison B, Rostaing G. Models for bearing damage detection in induc-
tion motors using stator current monitoring. IEEE Transaction on Industrial Electronics.
April 2008;55(4)
[67] Terra JI, Castelli M, Fossati JP. Fault detection and remote monitoring system for induc-
tion motors using MCSA technique. IEEE Transaction on Transmission and Distribution.
October 2008;11(2)
[68] Bayindir R, Sfa I, Colak I. Fault detection and protection of induction motors using sen-
sors. IEEE Transaction on Energy Conversion, September 2008;23(3)
[69] Ioannides MG. Design and implementation of PLC-based monitoring control system for
induction motor. IEEE Transaction on Energy Conversion. September 2004;19(3)
[70] Frosini L, Bassi E. Stator current and motor efficiency as indicators for different types
of bearing faults in induction motors. IEEE Transactions on Industrial Electronics.
2010;57(1):244–251
[71] Frosini L, Bassi E, Fazzi A, Gazzaniga C. Use of the stator current for condition monitor-
ing of bearings in induction motors. ICEM; September 2008; Vilamoura, Portugal; 1–6
[72] Ebrahimi BM, Faiz J. Feature extraction for short-circuit fault detection in permanent-
magnet synchronous motors using stator-current monitoring. IEEE Transactions on
Power Electronics. October 2010;25(10)
[73] Mehala N. Condition Monitoring and Fault Diagnosis of Induction Motor using Motor
Current Signature Analysis [PhD Thesis]. Kurukshetra, India: National Institute of
Technology; 2010.
[74] Romero-Troncoso RJ, Saucedo-Gallaga R, Cabal-Yepez E, Garcia-Perez A, Osornio-Rios

RA. FPGA-based online detection of multiple combined faults in induction motors
through information entropy and fuzzy inference. IEEE Transactions on Industrial
Electronics. November 2011;58(11)
[75] Seera M, Lim CP, Ishak D, Singh H. Fault detection and diagnosis of induction motors
using motor current signature analysis and a hybrid FMM–CART model. IEEE Trans
actions on Neural Networks and Learning Systems. January 2012;23(1)
[76] Singh S, Kumar A, Kumar N. Motor current signature analysis for bearing fault detec-
tion in mechanical systems. Procedia Materials Science. 2014;6:171–177
[77] Esfahani ET, Wang S, Sundararajan V. Multisensor wireless system for eccentricity and
bearing fault detection in induction motors. IEEE/ASME Transactions on Mechatronics.
June 2014;19(3)
[78] Choi S, Pazouki E, Baek J, Bahrami HR. Iterative condition monitoring and fault diag-
nosis scheme of electric motor for harsh industrial application. IEEE Transactions on
Industrial Electronics. 2014. DOI: 10.1109/TIE.2014.2361112
[79] Shi P, Chen Z, Vagapov Y. Modelling and analysis of induction machines under broken
rotor-bar failures. International Journal of Computer Applications. May 2013;69(14)
[80] Choi S, Akin B, Kwak S, Toliyat HA. A compact error management algorithm to mini-
mize false-alarm rate of motor/generator faults in (hybrid) electric vehicles. IEEE
Journal of Emerging and Selected Topics in Power Electronics. 2014. DOI: 10.1109/
JESTPE.2014.2302902
[81] Choi SD, Akin B, Rahimian M, Toliyat HA. Implementation of fault diagnosis algorithm
for induction machines based on advanced digital signal processing techniques. IEEE
Transaction on Industrial Electronics. March 2011;58(3)
[82] Aydin I, Karakose M, Akin E. Combined intelligent methods based on wireless sen-
sor networks for condition monitoring and fault diagnosis. Journal of Intelligent
Manufacturing. August 2013. DOI: 10.1007/s10845-013-0829-8
[83] Blodt M, Granjon P, Raison B, Rostaing G. Models for bearing damage detection in induc-
tion motors using stator current monitoring. IEEE Transaction on Industrial Electronics.
April 2008;55(4)
[84] Stack JR, Habetler TG, Harley RG. Fault classification and fault signature production for
rolling element bearing in electric machines. IEEE Transaction on Industry Applications.
2004;40(3)
[85] Benbouzid MEH, Nejjari H, Beguenane R, Vieira M. Induction motor asymmetrical

faults detection using advanced signal processing techniques. IEEE Transaction on
Energy Conversion. June 1999;14(2):147–152
[86] Patel VN, Tandon N, Pandey RK. Defect detection in deep groove ball bearing in pres-
ence of external vibration using envelope analysis and duffing oscillator. Measurement.
2012;45:960–970
[87] Kang M, Kim J, Kim JM. Reliable fault diagnosis for incipient low-speed bearings using
fault feature analysis based on a binary bat algorithm. Information Sciences. 2015;
294:423–438
[88] Choi SD, Akin B, Rahimian MM, Toliyat HA, Azadpour M. A generalized condition
monitoring method for multi-phase induction motors. IEEE International Conference
on Electric Machines and Drives; 2009
[89] Wang D, Miao Q, Fan X, Huang HZ. Rolling element bearing fault detection using
an improved combination of Hilbert and Wavelet transforms. Journal of Mechanical
Science and Technology. August 2009;23:3292–3301
[90] Widodo A, Yang BS, Han T. Combination of independent component analysis and
support vector machines for intelligent faults diagnosis of induction motors. Expert
Systems with Applications. February 2007;32(2):299-312
[91] Hwang Y, Jen K, Shen Y. Application of cepstrum and neural network to bearing fault
detection. Journal of Mechanical Science and Technology. May 2009;23:2730-2737
[92] Júnior AMG, Silva VR, Baccarini LMR, Reis MLF. Three-phase induction motors faults
recognition and classification using neural networks and response surface models.
Journal of Control, Automation and Electrical Systems. June 2014;25(3):330-338
[93] Kia SH, Henao H, Capolino GA. Development of a test bench dedicated to condition
monitoring of wind turbines. IEEE-IECON; Dallas (TX, USA); October 28–November
1; 2014
[94] Guzinski J, Diguet M, Krzeminski Z, Lewicki A, Abu-Rub H. Application of speed
and load torque observers in high speed train drive for diagnostic purpose. IEEE
Transactions on Industrial Electronics. January 2009;56(1):248-256
[95] Mohanty AR, Kar C. Monitoring gear vibrations through motor current signature
analysis and wavelet transform. Mechanical System and Signal Processing. January
2006;20(1):158-187
[96] Feki N, Clerc G, Velex P. An integrated electro-mechanical model of motor-gear units -
Applications to tooth fault detection by electric measurements. Mechanical System and
Signal Processing. May 2012;29:377-390
[97] Strangas EG. Response of electrical drives to gear and bearing faults—diagnosis
under transient and steady state conditions. In: Proceedings of Workshop on Electrical
Machines Design Control and Diagnosis (WEMDCD); invited paper, March 11–12,
2013; Paris (France), pp.289-297
[98] Henao H, Kia SH, Capolino GA. Torsional vibration assessment and gear fault diag-
nosis in railway traction system. IEEE Transactions on Industrial Electronics. May
2011;58(5):1707-1717
[99] Bogiatzidis I, Safacas A, Mitronikas E. Detection of backlash phenomena appearing in
a single cement kiln drive using the current and the electromagnetic torque signature.
IEEE Transactions on Industrial Electronics. August 2013;60(8):3441-3453
[100] Trajin B, Regnier J, Faucher J. Comparison between stator current and estimated
mechanical speed for the detection of bearing wear in asynchronous drives. IEEE
Transactions on Industrial Electronics. November 2009;56(11):4700-4709
[101] Ibrahim A, El-Badaoui M, Guillet F, Bonnardot F. A new bearing fault detection method
in induction machines based on instantaneous power factor. IEEE Transactions on
Industrial Electronics. December 2008;55(12):4252–4259
[102] Blödt M, Chabert M, Regnier J, Faucher J. Mechanical load fault detection in induc-
tion motors by stator current time-frequency analysis. IEEE Transactions on Industry
Applications. November 2006; 42(6):1454–1463
[103] Mohanty AR, Kar C. Fault detection in a multistage gearbox by demodulation of
motor current waveform. IEEE Transactions on Industrial Electronics. June 2006;53(4):
1285–1297
[104] Kar C, Mohanty AR. Vibration and current transient monitoring for gearbox fault
detection using multiresolution Fourier transform. Journal of Sound and Vibration.
March 2008; 311(1/2):109–132
[105] Kia SH, Henao H, Capolino GA. Analytical and experimental study of gearbox
mechanical effect on the induction machine stator current signature. IEEE Transactions
on Industrial Electronics. July 2009;45(4):1405–1415
[106] Kia SH, Henao H, Capolino GA. Gear tooth surface damage fault detection using
induction machine stator current space vector analysis. IEEE Transactions on Industrial
Electronics. March 2015;62(3):1866–2001
[107] Kia SH, Henao H, Capolino GA. A comparative study of acoustic, vibration and stator
current signatures for gear tooth fault diagnosis. International Conference on Electrical
Machines; 2-5 September 2012; Marseille.
[108] Kia SH, Henao H, Capolino GA. Gear tooth surface damage fault detection using
induction machine electrical signature analysis. 9th IEEE International Symposium on
Diagnostics for Electric Machines, Power Electronics and Drives (SDEMPED); August
27–30, 2013;Valencia.
[109] Kia SH, Henao H, Capolino GA. Trends in gear fault detection using electrical sig-
nature analysis in induction machine-based systems. IEEE Workshop on Electrical
Machines Design, Control and Diagnosis (WEMDCD); March 26-27, 2015; Torino.
[110] Kia SH, Henao H, Capolino GA. Torsional vibration effects on induction machine
current and torque signatures in gearbox-based electromechanical system. IEEE
Transactions on Industrial Electronics. November 2009;56(11):4689–4699
[111] Feki N, Clerc G, Velex P. Gear and motor fault modeling and detection based on motor
current analysis. Electric Power System Research. Feburary 2013;95:28–37
[112] Daneshi-Far Z, Henao H, Capolino GA. Planetary gearbox effects on induction machine
in wind turbine: Modeling and analysis. In: Proceedings of International Conference
on Electrical Machines (ICEM); 2–5 September 2012; Marseille (France); pp.1790–1796
[113] Ottewill JR, Orkisz M. Condition monitoring of gearboxes using synchronously

averaged electric motor signals, Mechanical System Signal Processing. July 2013;38(2):
482–498
[114] Girsang IP, Dhupia JS, Muljadi E, Singh M, Pao LY. Gearbox and drive train models
to study of dynamic effects of modern wind turbines. IEEE Transactions on Industry
Applications. 2014
[115] Huh KK, Lorenz RD, Nagel NJ. Gear fault diagnostics integrated in the motion servo
drive for electromechanical actuators. IEEE Transactions on Industry Applications.
July 2012;48(1):142–150
[116] Bogiatzidis IX, Safacas AN, Mitronikas ED. Detection of backlash phenomena appear-
ing in a single cement kiln drive using the current and the electromagnetic torque sig-
nature. IEEE Transactions on Industrial Electronics. August 2013;60(8)
[117] Kia SH, Henao H, Capolino GA. A real-time platform dedicated to online gear tooth
surface damage fault detection in induction machine. International Conference on
Electrical Machines (ICEM); 2-5 September, 2014; Berlin.
[118] Fournier E et al. Current-based detection of mechanical unbalance in an induction
machine using spectral kurtosis with reference. IEEE Transactions on Industrial
Electronics, March 2015;62(3):1879–1887
[119] Ahmad I. Investigation of single and multiple faults under varying load conditions
using multiple sensor types to improve condition monitoring of induction machines
[PhD Thesis], Australia: Univeristy of Adelaide; 2007.
[120] Drif M, Marques Cardoso AJ. Stator fault diagnostics in squirrel cage three-phase
induction motor drives using the instantaneous active and reactive power signature
analyses. IEEE Transaction on Industrial Electronics. May 2014;10(2)
[121] Cruz SMA. An active–reactive power method for the diagnosis of rotor faults in three-
phase induction motors operating under time-varying load conditions. IEEE Tran
sactions on Energy Conversion. March 2012;27(1)
[122] Rjakishmi Samaga LR, Vittal KP. Comprehensive study of mixed eccentricity fault
diagnosis in induction motors using signature analysis. Electrical Power and Energy
Systems. 2012;35:180–185
[123] Kim J, Shin S, Lee SB, Gyftakis KN, Drif M, Marques Cardoso AJ. Power spectrum-
based detection of induction motor rotor faults for immunity to false alarms. IEEE
Transactions on Energy Conversion. 2015
[124] Didier G, Ternisien E, Caspary O, Razik H. Fault detection of broken rotor bars in induc-
tion motor using a global fault index. IEEE Transactions on Industry Applications.
January 2006; 42(1):79–88
[125] Zhou W, Habetler TG, Harley RG. Incipient bearing fault detection via motor sta-
tor current noise cancellation using Wiener filter. IEEE Transactions on Industry
Applications. July 2009;45(4):1309–1317
[126] Golafshan R, Sanliturk KY. SVD and Hankel matrix based de-noising approach for ball
bearing fault detection and its assessment using artificial faults. Mechanical Systems
and Signal Processing. 2015
[127] Golafshan R, Sanliturk KY. The Effect of SVD based noise elimination methods applied
for ball bearing fault detection. TrC-IFToMM Symposium on Theory of Machines and
Mechanisms; June 14–17, 2015; Izmir, Turkey.
[128] Kia SH, Henao H, Capolino G. A high-resolution frequency estimation method for
three-phase induction machine fault detection. IEEE Transaction on Industrial
Electronics. August 2007;54(4):2305–2314
[129] Bellini A, Yazidi A, Filippetti F, Rossi C, Capolino GA. High frequency resolution tech-
niques for rotor fault detection of induction machines. IEEE Transactions on Industrial
Electronics. December 2008;55(12):4200–4209
[130] Kim H, Lee SB, Park SB, Kia SH, Capolino GA. Reliable detection of rotor faults under
the influence of low frequency load torque oscillations for applications with speed
reduction couplings. IEEE 10th International Symposium on Diagnostics for Electrical
Machines. Power Electronics and Drives; September, 2015; Portugal.
[131] Dong J, Chin J. Noise resistant time frequency analysis and application in fault
diagnosis of rolling element bearings. Mechanical Systems and Signal Processing.
2012;33:212–236
[132] Wang Z, Jiang H. Robust incipient fault identification of aircraft engine rotor based on
wavelet and fraction. Aerospace Science and Technology. 2010;14:221–224
[133] Akin B, Orguner U, Toliyat HA, Rayner M. Phase-sensitive detection of motor fault
signatures in the presence of noise. IEEE Transactions on Industrial Electronics. June
2008;55(6):2539–2550
[134] Akin B, Choi SD, Orguner U, Toliyat HA. A simple real-time fault signature moni-
toring tool for motor drive imbedded fault diagnosis systems. IEEE Transaction on
Industrial Electronics. May 2011;58(5)
[135] Toliyat HA, Nandi S, Choi S, Kelk HS. Electric machines, modelling, condition moni-
toring and fault diagnosis. CRC Press, Taylor & Francis Group; 2012.
[136] Rajagopalan S, Habetler TG, Harley RG, Restrepo JA, Alle JM. Non-stationary motor
fault detection using recent quadratic time-frequency representations. International
Conference Recording IEEE IAS Annual Meeting; October 2006; 5; pp. 2333–2339
[137] Jung JH, Lee JJ, Kwon BH. Online diagnosis of induction motor using MCSA. IEEE
Transactions on Industry Applications. December 2006;53(1):1842–1852
[138] Yousuf A, Delpha C, Diallo D. An optimal fault detection threshold for early detection
using Kullback–Leibler divergence for unknown distribution data. Signal Processing. 2015
[139] Harmouche J, Delpha C, Diallo D. Incipient fault detection and diagnosis based on
Kullback–Leibler divergence using principal component analysis: Part II. Signal Pro
cessing. 2015;109:334–344
[140] Delpha C, Diallo D. Incipient fault detection and diagnosis: A hidden information
detection Problem. IEEE 24th International Symposium on Industrial Electronics; June,
2015; Buzios.
[141] Chen J, Patton RJ. Robust Model Based Fault Diagnosis for Dynamic Systems. Springer
Science. Business Media; 2012.
[142] Fournier E, Picot A, Régnierl J, Yamdeu MT. On the use of spectral kurtosis for diag-
nosis of electrical machines. The 9th IEEE International Symposium on diagnostics for
electric machines, power electronics and drives (SDEMPED); 2013; Valencia
[143] Picot A, Obeid Z, Régnierl J, Poignant S, Darnis O, Maussion P. Statistic-based spec-

tral indicator for bearing fault detection in permanent-magnet synchronous machines
using the stator current. Mechanical Systems and Signal Processing. 2014;46:424–441
[144] Zarie J. Induction motors bearing fault detection using pattern recognition techniques.
Expert System with Applications. 2012;39:68–73
[145] Spyropoulos DV, Mitronikas ED. Induction motor stator fault diagnosis technique
using Park vector approach and complex Wavelets. The 20th IEEE International
Conference on Electrical Machines (ICEM); September, 2012; Marseille.
[146] Laughman C, Leeb SB, Norford LK, Shaw SR, Armstrong PR. A Park transform-based
method for condition monitoring of three-phase electromechanical systems. The 5th
IET International Conference on Power Electronics, Machines and Drives; April 2010;
Brighton, UK.
[147] Salem SB, Touti W, Bacha K, Chaari A. Induction motor mechanical fault identification
using park vector approach. International Conference on Electrical Engineering and
Software Applications (ICEESA); March 2013.
[148] Parra AP, Enciso AMC, Ochoa JO, Peñaranda JAP. Stator fault diagnosis on squirrel
cage induction motors by ESA and EPV. IEEE Workshop on Power Electronics and
Power Quality Applications (PEPQA); 2013; Bogota, July.
[149] Rezig A, N’Diaye A, Mekideche MR, Djerdir A. Modelling and detection of bear-
ing faults in permanent magnet synchronous motors. The 20th IEEE International
Conference on Electrical Machines (ICEM). September, 2012; Marseille.
[150] Bacha K, Salem SB, Chaari A. An improved combination of Hilbert and Park trans-
forms for fault detection and identification in three-phase induction motors. Electrical
Power and Energy Systems. 2012;43:1006–1016
[151] Salem SB, Bacha K, Gossa M. Induction motor fault diagnosis using an improved com-
bination of Hilbert and Park transforms. The 16th IEEE Mediterranean Electrotechnical
Conference (MELECON); March 2012; Yasmine Hammamet.
[152] Kuruppu SS, Kulatunga NA. D-Q current signature-based faulted phase localization
for SM-PMAC machine drives. IEEE Transactions on Industrial Electronics. January
2015;62(1)
[153] Zhang P, Du Y, Habetler TG, Lu B. A survey of condition monitoring and protec-

tion methods for medium-voltage induction motors. IEEE Transactions on Industry
Applications. January 2011;47(1):34–46
[154] Navarro L, Delgado M, Urresty J, Cusidó J, Romeral L. Condition monitoring sys-

tem for characterization of electric motor ball bearings with distributed fault using
fuzzy inference tools. IEEE Instrumentation and Measurement Technology Conference
(I2MTC); May, 2010; Austin.
[155] Immovilli F, Cocconcelli M, Bellini A, Rubini R. Detection of generalized-rough-

ness bearing fault by spectral-kurtosis energy of vibration or current signals. IEEE
Transactions on Industrial Electronics. November 2009;56(11):4710–4717
[156] Immovilli F, Bellini A, Rubini R, Tassoni C. Diagnosis of bearing faults in induction
machines by vibration or current signals: A critical comparison. IEEE Transactions on
Industry Applications. July 2010;46(4):1350–1359
[157] Dalvand F, Keshavarzi M, Kalantar A, Cheraghdar A. Detection of generalized-rough-

ness bearing fault using statistical-time indices of instantaneous frequency of motor volt-
age space vector. The 23rd Iranian Conference on Electrical Engineering (ICEE); 2015.
[158] Gao Z, Cecati C, Ding SX. A survey of fault diagnosis and fault-tolerant techniques part
I: fault diagnosis with model based and signal-based approaches. IEEE Transactions
on Industrial Electronics. 2015.
[159] Hurtado ZYM, Tello CP, Sarduy JG. A review on detection and fault diagnosis in
induction machines. Publicaciones en Ciencias y Tecnologa. July 2014;8(1)
[160] Heng A, Zhang S, Tan ACC, Mathew J. Rotating machinery prognostics: State of the art,
challenges and opportunities—A review. Mechanical Systems and Signal Processing.
2009;23:724–739
[161] Zhang P, Du Y, Habetler TG, Lu B. A survey of condition monitoring and protec-

tion methods for medium-voltage induction motors. IEEE Transactions on Industry
Applications. 2011;47(1)
[162] Gandhi A, Corrigan T, Parsa L. Recent advances in modeling and online detection of
stator interturn faults in electrical motors. IEEE Transactions on Industrial Electronics.
May 2011;58(5)
[163] Hameed Z, Hong YS, Cho YM, Ahn SH, Song CK. Condition monitoring and fault detec-
tion of wind turbines and related algorithms: A review. Renewable and Sustainable
Energy Reviews. 2009;13:1–39
[164] Irfan M, Saad N, Ibrahim R, Asirvadam VS. An online condition monitoring system for
induction motors via instantaneous power analysis. Journal of Mechanical Science and
Technology. April 2015; 29(04)
[165] Irfan M, Saad N, Ibrahim R, Asirvadam VS. Condition monitoring of induction motors
via instantaneous power analysis. Journal of Intelligent Manufacturing. February 2015
[166] Irfan M, Saad N, Ibrahim R, Asirvadam VS. An intelligent diagnostic condition moni-
toring system for ac motors via instantaneous power analysis. International Review of
Electrical Engineering. April 2013; 8(2):664–672
[167] Irfan M, Saad N, Ibrahim R, Asirvadam VS.. An approach to diagnose inner race sur-
face roughness faults in bearings of induction motors. IEEE International Conference
on Signal and Image Analysis (ICSIPA); October 2015; Kuala Lumpur, Malaysia.
[168] Irfan M, Saad N, Ibrahim R, Asirvadam VS.. Diagnosis of distributed faults in outer
race of bearings via Park’s Transformation method. The 10th Asian Control Conference
(ASCC); June 2015; Kota Kinabalu, Malaysia.
[169] Irfan M, Saad N, Ibrahim R, Asirvadam VS.. A non-invasive fault diagnosis system for
induction motors in noisy environment. IEEE International Conference on Power and
Energy (PECon); December 2014; Malaysia.
[170] Irfan M, Saad N, Ibrahim R, Asirvadam VS. An intelligent diagnostic system for condi-
tion monitoring of ac motors. The 8th IEEE Conference on Industrial Electronics and
Applications; June 2013; Melbourne, Australia.
[171] Zarei J, Poshtan J. An advanced Park’s vectors approach for bearing fault detection.
Tribology International. 2009;19:213–219

A Review of Condition Monitoring of Motors-May 2017

Uploaded by

Copyright:

Available Formats

A Review of Condition Monitoring of Motors-May 2017

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A Review of Condition Monitoring of Motors-May 2017

Uploaded by

Copyright:

Available Formats

PUBLISHED BY

World's largest Science,

Selection of our books indexed in the

Chapter from the book Fault Diagnosis and Detection

Interested in publishing with InTechOpen?

An Assessment on the Non-Invasive Methods for

Muhammad Irfan, Nordin Saad, Rosdiazli Ibrahim,

Additional information is available at the end of the chapter

Keywords: condition monitoring, fault diagnosis, mechanical faults, bearing distributed

Figure 1. Applications of induction motors.

Figure 2. Activities in predictive maintenance.

2. Overview of induction motors

Figure 3. The structure of induction motor [5].

• Wrong-rated power, voltage and current

• Mistakes during repairs

• Unstable supply voltage or current source

• Post the standard lifetime

• Overload or unbalanced load

• Electrical stress from fast switching inverters or unstable ground

• Residual stress from manufacturing

• Harsh application environment

3. Faults in induction motor

Figure 4. EPRI survey report 2005 [2, 3, 12].

3.1. Mechanical defects

3.1.1. Bearing defects

Figure 5. The structure of ball bearing.

Figure 6. Example of bearing (a) localized defects and (b) distributed defects.

3.1.2. Air-gap eccentricity

A dynamic eccentricity as shown in Figure 7 results in an unbalanced magnetic pull (UMP)

Figure 7. Example of (a) static eccentricity and (b) dynamic eccentricity [29].

3.1.3. Abnormalities in mechanical transmission system

Figure 8. Example of the gear defects [30].

3.2. Electrical defects

3.2.1. Stator winding defects

Figure 9. The stator fault in induction motor [31].

3.2.2. Rotor fault

Figure 10. Example of rotor fault in induction motor [13].

4. Condition monitoring techniques

• To determine the failure of machinery

• To enhance the reliability of both the machines and their parts

• To optimally use manpower and machine spare parts

• To maximize the performance of the machinery

• To enhance the failure prediction accuracy

5. Overview on non-invasive fault estimation parameters

5.1. MCSA for bearing localized defects

5.2. MCSA for gear defects

6. IPA for motor electrical defects

7. Impact of environmental noise on decision making of existence of fault

Reference Fault type Threshold design Limitations

[136] Eccentricity Pre-determined • Does not count non-linear noise variations

[142] Unbalance Reference-based • Does not count non-linear noise variations

8. Park vector analysis for bearing localized defects

9. Non-invasive method for diagnosis of bearing distributed defects

10. Condition monitoring issues

No CM technique Advantages Drawbacks Reference

2. PVA • Inexpensive • Further investigation required to test its [144–152, 171]

3. IPA • Inexpensive • Fault detection affected by environment [119–123]

Table 2. Advantages and drawbacks of non-invasive condition monitoring methods.

11. Future directions

Figure 1. Applications of induction motors.

Figure 2. Activities in predictive maintenance.

Figure 3. The structure of induction motor [5].

Figure 4. EPRI survey report 2005 [2, 3, 12].

Figure 5. The structure of ball bearing.

Figure 6. Example of bearing (a) localized defects and (b) distributed defects.

Figure 7. Example of (a) static eccentricity and (b) dynamic eccentricity [29].

Figure 8. Example of the gear defects [30].

Figure 9. The stator fault in induction motor [31].

Figure 10. Example of rotor fault in induction motor [13].