applied

sciences
Article
Broken Rotor Bar Detection in LS-PMSM Based on
Startup Current Analysis Using Wavelet
Entropy Features
Mohammad Rezazadeh Mehrjou 1,2 ID , Norman Mariun 1,2 , Norhisam Misron 2, *,
Mohd Amran Mohd Radzi 1,2 ID and Suleiman Musa 1,2
1 Department of Electrical and Electronic Engineering, Universiti Putra Malaysia, Serdang 43400, Selangor,
Malaysia; mehrjou.mo@gmail.com (M.R.M.); norman@upm.edu.my (N.M.);
amranmr@upm.edu.my (M.A.M.R.); sumusa115@gmail.com (S.M.)
2 Centre of Advanced Power and Energy Research (CAPER), Universiti Putra Malaysia, Serdang 43400,
Selangor, Malaysia
* Correspondence: norhisam@upm.edu.my; Tel.: +603-8946-6299

Received: 16 June 2017; Accepted: 9 August 2017; Published: 17 August 2017

Abstract: High-efficiency motors are being gradually introduced in many industrial applications because
of their positive impacts on the environment by reducing energy consumption and CO2 emission. In this
respect, line start permanent magnet synchronous motors (LS-PMSMs) have been introduced recently.
Due to their unique configuration, LS-PMSMs allow the obtaining of super premium efficiency levels,
accompanied with a high torque and power factor. However, since the use of LS-PMSMs in the industry
is in its infancy, no efficient scheme has yet been reported for broken rotor bar (BRB) fault detection in
this type of motor. Accordingly, the main aim of this research is to investigate the fault-related feature for
BRB faults on LS-PMSMs. In this regard, a simulation model and experimental setup for the investigation
of BRB in LS-PMSM are implemented. The detection strategy for BRB in LS-PMSM proposed here is
based on the monitoring of the start-up current signal and discrete wavelet transform. The entropy
features are used as fault-related features for BRB faults. Finally, the ability of these features is validated
for the detection of BRB in LS-PMSM through statistical analysis. In this research, the importance of the
starting load is also considered for BRB detection in LS-PMSMs.

Keywords: LS-PMSMs; broken rotor bars; transient; current signal; wavelet transform; entropy
features

1. Introduction
Induction machines that facilitate the production processes and related services lead to enormous
changes in human lifestyles. Induction machines are extensively used in all aspects of industrial,
commercial, domestic, utility and special-purpose commercial markets owing to their rugged
configuration, low cost, versatility, reasonably small size, and their operation with an easily available
power supply. However, induction machines suffer from a low efficiency and low power factor that
means the loss of energy is high. This issue is viewed as an important disadvantage because of the
energy cost and global energy concerns.
The improvement of induction machine efficiency was tried through an optimal design of these
motors. However, due to several inherent limitations, it is difficult to improve the efficiency of induction
machines significantly. An alternative solution is to replace the induction machines with high-efficiency
permanent magnet synchronous motors (PMSMs). An important obstacle for ordinary PMSMs is they
need an inverter to start that is not economical for many single speed applications. To overcome this
problem, the squirrel-cage bar-equipped PM motors—called line start permanent magnet synchronous

Appl. Sci. 2017, 7, 845; doi:10.3390/app7080845 www.mdpi.com/journal/applsci

Appl. Sci. 2017, 7, 845 2 of 20

Appl. Sci. 2017, 7, 845 2 of 20

motors (LS-PMSMs)—have been introduced. LS-PMSMs also allow for the reaching of super premium
permanent magnet synchronous motors (LS-PMSMs) —have been introduced. LS-PMSMs also allow
efficiency levels [1,2]. The structure and configuration of the three phase-four pole LS-PMSM are
for the reaching of super premium efficiency levels [1,2]. The structure and configuration of the three
depicted in Figure 1.
phase-four pole LS-PMSM are depicted in Figure 1.

Figure 1. The structure of three phase line start permanent magnet synchronous motors LS-PMSM
Figure 1. The structure of three phase line start permanent magnet synchronous motors LS-PMSM
(four pole).
(four pole).
An LS-PMSM is made up of a stator and a hybrid rotor with a squirrel-cage and pairs of
permanent
An LS-PMSM magnets.
is made In this
up ofmachine,
a statorwhen
and athe rotor rotor
hybrid is operated
with afrom a standstill,and
squirrel-cage squirrel-cage bars
pairs of permanent
generate a suitably high starting torque, enabling the direct-on-line movement
magnets. In this machine, when the rotor is operated from a standstill, squirrel-cage bars generate a of the rotor.
The high
suitably reverse-rotating fields of
starting torque, the air gap
enabling thereduces in a squirrel-cage
direct-on-line movement rotorofwhen there isThe
the rotor. an unbalancing
reverse-rotating
in the load condition or there is a fluctuation in the rotation speed. Otherwise, if there is no
fields of the air gap reduces in a squirrel-cage rotor when there is an unbalancing in the load condition
squirrel-cage in the rotor, these conditions lead to significant losses, and no reduction in the reverse-
or there is a fluctuation in the rotation speed. Otherwise, if there is no squirrel-cage in the rotor, these
rotating fields is observed [3].
conditions Inlead
the to significant
industry, losses, and
LS-PMSMs are no reduction
exposed in the reverse-rotating
to unavoidable stresses; for fields is observed
example, electrical,[3].
In the industry, LS-PMSMs are exposed to unavoidable stresses; for
environmental, mechanical and thermal stresses. These stresses produce some failures and imperfections example, electrical,
environmental, mechanical and thermal stresses. These stresses produce some
in different parts of the LS-PMSMs. Usually, the generated faults disrupt the operation of the failures and imperfections
in different parts degrade
LS-PMSMs, of the LS-PMSMs. Usually, quality,
the manufacturing the generated faults disrupt
and therefore result in thesignificant
operationcost of the LS-PMSMs,
penalties.
Maintenance
degrade costs can be
the manufacturing greatlyand
quality, reduced withresult
therefore the application of ancost
in significant efficient fault detection
penalties. Maintenance method,
costs can
as thisreduced
be greatly avoids spontaneous downtimes.
with the application Theefficient
of an application
faultofdetection
LS-PMSMmethod,
is growing gradually,
as this avoidsalthough
spontaneous
the industry
downtimes. The is still lackingofanLS-PMSM
application accurate fault detectiongradually,
is growing criterion for maintenance
although policies of
the industry is the
stillmotor.
lacking an
The aim of this paper is to extend a motor startup methodology to the detection of a broken rotor
accurate fault detection criterion for maintenance policies of the motor.
bar (BRB) in LS-PMSM. In this method, a novel approach of deriving the amplitude of the left sideband
The aim of this paper is to extend a motor startup methodology to the detection of a broken
harmonic (LSH) in a specific sub-band frequency is presented for BRB. Using discrete wavelet
rotortransform,
bar (BRB)LSH in LS-PMSM.
is extracted Infromthis method,
the a novel approach
current spectrum and is usedofasderiving
an input the amplitude
for entropy of the left
features.
sideband harmonic (LSH) in a specific sub-band frequency is presented for BRB. Using discrete wavelet
2. Background
transform, and Methodology
LSH is extracted from the current spectrum and is used as an input for entropy features.
The key for successful fault detection in electrical machines relies on the availability of accurate
2. Background and Methodology
information from them which allows for the understanding of the machine’s condition. Thus,
the basis
The key of
forany fault detection
successful system is precise
fault detection condition
in electrical monitoring.
machines reliesIt on
canthe
easily be realizedoffrom
availability accurate
the name that condition monitoring is an act designed to observe the performance
information from them which allows for the understanding of the machine’s condition. Thus, the basis of a device,
including
of any electric machines,
fault detection system with the purpose
is precise of a maintenance
condition monitoring. strategy.
It canThe reliability
easily of condition
be realized from the
monitoring techniques depends upon the best understanding of motor characteristics, including
name that condition monitoring is an act designed to observe the performance of a device, including
electrical and mechanical characteristics, in both healthy and faulty conditions. In this respect,
electric machines, with the purpose of a maintenance strategy. The reliability of condition monitoring
condition monitoring techniques have continuously been developed over the years, resulting in a
techniques
range ofdepends
availableupon the best
methods forunderstanding
failure diagnosis of in
motor characteristics,
electrical includingmonitoring
devices. Condition electrical and
mechanical characteristics,
techniques presented forinbroken
both healthy
rotor barand faultyin
detection conditions. In this respect,
induction machines can be condition monitoring
classified into the
techniques have
following continuously
categories [4,5]: been developed over the years, resulting in a range of available methods
for failure diagnosis in electrical devices. Condition monitoring techniques presented for broken rotor
bar detection in induction machines can be classified into the following categories [4,5]:

 Acoustic Emission [6];

 Air Gap Torque [7];
Appl. Sci. 2017, 7, 845 3 of 20
Appl. Sci. 2017, 7, 845 3 of 20


 Motor Current
Acoustic Emission Signature
[6]; Analysis (MCSA) [8];

 Air Gap Torque [7];
Electromagnetic Field Monitoring [9];

 Motor Current Signature
Instantaneous Analysis
Angular Speed [10];(MCSA) [8];


Electromagnetic
Power [11]; Field Monitoring [9];
 Instantaneous Angular Speed [10];
 Motor Circuit Analysis [12];
 Power [11];
 Thermal [13];
 Motor Circuit Analysis [12];

 Vibration
Thermal [14];
[13];

 Voltage [15].
Vibration [14];
 Voltage [15].
This section discusses the influence of BRB on the rotor magneto-motive force (MMF) and its
This
effects on section
the stator discusses
currentthe influence
signals. of BRB on
The forward the rotor of
component magneto-motive
this MMF, in aforce (MMF)
healthy case,and its at a
rotates
effects on the stator
synchronous speed,current signals.
ωsyn , with The forward
respects component
to the stator (or sωofsynthis MMF,
with in
regard a healthy
to the case,
rotor). rotates
When a bar
at a synchronous speed, , with respects to the stator (or with regard to
cracks, there will be no current flow in it, and therefore no magnetic flux is produced surrounding the the rotor). When a
bar cracks,
BRB. thereiswill
As there be no current
no magnetic fluxflow in it, and
around therefore
the BRB, no magnetic
a non-zero flux is rotating
backward producedfield
surrounding
is generated,
the BRB. As there is no magnetic flux around the BRB, a non-zero
and thus an asymmetry is created in the rotor MMF. It is worth mentioning that the backward backward rotating fieldrotating
is
generated, and thus an asymmetry is created in the rotor MMF. It is worth
field in a healthy rotor is zero. The non-zero backward rotating component produced due to the mentioning that the
backward rotating field in a healthy rotor is zero. The non-zero backward rotating component
existence of a BRB conducts an equal but reverse current in the healthy bars. Such a current causes
produced due to the existence of a BRB conducts an equal but reverse current in the healthy bars.
secondary failure. The rotation of non-zero backward MMF as a result of BRB is at a slip speed,
Such a current causes secondary failure. The rotation of non-zero backward MMF as a result of BRB
which is defined by sωsyn with regards to the rotor or by Equation (1) with respect to the stator [16].
is at a slip speed, which is defined by with regards to the rotor or by Equation (1) with respect
to the stator [16].
ωbMMF = −sωsyn + ωm = −sωsyn + (1 − s)ωsyn = (1 − 2s)ωsyn (1)
=− + =− + (1 − ) = (1 − 2 ) (1)
The non-zero backward MMF produces harmonics which are superimposed on the stator currents.
The non-zero backward MMF produces harmonics which are superimposed on the stator
These superimposed features can be used as signatures for the detection of BRB in the MCSA
currents. These superimposed features can be used as signatures for the detection of BRB in the
technique [17]. Figure 2 shows the effect of BRB faults on the MMF of rotor for a cage with 16 bars
MCSA technique [17]. Figure 2 shows the effect of BRB faults on the MMF of rotor for a cage with
per pair of poles. As explained before, broken bars cause harmonic components of (1 − 2ks) f in the
16 bars per pair of poles. As explained before, broken bars cause harmonic components of (1 − 2 )
stator
in the current. Among
stator current. these these
Among components,
components,the main one corresponds
the main one corresponds to the
to frequency thatthat
the frequency ( k = 2),
has has
called the left sideband harmonic (LSH) and defined
( = 2), called the left sideband harmonic (LSH) and defined as: as:

==
f LSH (1 (−1 2−)2s) f (2) (2)

Figure 2. Effect of a broken bar on the magneto-motive force (MMF) of a rotor [16].
Figure 2. Effect of a broken bar on the magneto-motive force (MMF) of a rotor [16].

The motor slip changes from one to zero during the start-up; at the beginning of the start-up,
The motor
the motor slip isslip
onechanges fromtoone
and it goes to when
zero zero during the reaches
the motor start-up;the
at steady
the beginning of the start-up,
state. Accordingly,
the motor slipofisLSH
the frequency one( and) it goesfrom
varies to zero when equal
the value the motor
to the reaches the steady
fundamental state.
frequency Accordingly,
to zero and
the
vicefrequency LSH ( f LSHof
versa. Theofappearance ) varies from theand
this harmonic value equal to the
its particular fundamental
evolution frequency
were also explainedto zero
[18]. and
vice versa. The
Squirrel-cage barsappearance
in LS-PMSMofimprove
this harmonic and its particular
start-up operation by allowingevolution
the rotorwere also
to have explained [18].
direct-on-line
Squirrel-cage bars in LS-PMSM improve start-up operation by allowing the rotor to have direct-on-line
 Sci. 2017,
Appl. Sci. 7, 845
2017, 7, 845 44 of
of 20
20

movement. In the stationary region, no current flows in the squirrel-cage bars and thus the value of
movement.
LSH becomes Inzero,
the stationary≈ 0.region, no current
The behavior flows
of the LSHinfrequency
the squirrel-cage bars and
for LS-PMSM thus the
during the value of
start-up
LSH becomes zero, A f LSH
transient is shown in Figure ≈ 0.
3. The behavior of the LSH frequency for LS-PMSM during the start-up
transient is shown in Figure 3.

Current [A]

t0 t1 t2 t3 Time [s] t4
s =1 s=0.5 s =0
Current [A]

fLSH fLSH fLSH fLSH A ≈0

(fLSH)

0 f 0 f 0 f 0 f 0 f
Frequency [Hz]

Figure 3.
Figure Behavior f LSH during
3. Behavior during the start-up for LS-PMSM.

The mainidea
The main ideaunderlying
underlying thisthis methodology
methodology is theistracking
the tracking of the characteristic
of the characteristic transient
transient evolution
evolution
of the fault-related feature from the LSH pattern [19]. Wavelet transform has been shownshown
of the fault-related feature from the LSH pattern [19]. Wavelet transform has been to be
to
one beofone
theofbest
the best
toolstools for non-stationary
for non-stationary signal
signal analysis
analysis [20].The
[20]. Theapplicability
applicability of of using
using wavelet
wavelet
transform
transform for for decomposing
decomposing aa signal signal and
and extracting
extracting itsits sensitive
sensitive bands
bands is is considered
considered to to bebe good
good for
for
the health monitoring of an electrical motor and fault diagnosis [21].
the health monitoring of an electrical motor and fault diagnosis [21]. Discrete wavelet transform Discrete wavelet transform
(DWT)
(DWT) is is proposed
proposed to to be
be used
used because
because it it provides
provides aa dyadic
dyadic band-pass
band-pass filtering
filtering ofof the
the signal
signal asas well
well
as
as the advantages of its simplicity, low computational requirements and the easy interpretation of
the advantages of its simplicity, low computational requirements and the easy interpretation of
the
the obtained
obtainedresults
results[18].
[18].The
The DWT
DWT is is
computed
computed through
through thethe
Mallat algorithm,
Mallat algorithm, in which
in which low-pass and
low-pass
high-pass
and high-pass filtering of aoftime-domain
filtering a time-domain signal is performed
signal is performed while
while the
thesampling
samplingrate ratechanges
changes [22].
[22].
DWT allows high frequency components to be analyzed with short time
DWT allows high frequency components to be analyzed with short time intervals, and low frequencyintervals, and low frequency
components
components to tobe
beanalyzed
analyzedwith withlong
longtimetimeintervals
intervals [23]. TheThe
[23]. decomposition
decomposition of aof
sampled
a sampled signal ( )
signal
using DWTDWT
i (t). using results
results + n1 +wavelet
1 waveletsignals,
signals,which
which areare
anan approximation
approximationsignal,signal, A(n (),t)and
, and n detail
detail
signals ( )with varying from 1
signals D j (t) with j varying from 1 to n [24].to [24].
( ) = ( ( ) + ( ) +∙∙∙ + ( )) (3)
i (t) = ( An (t) + Dn (t) + · · · + D1 (t)) (3)
The parameter is an integer known as the number of decomposition levels and it is set based
on theThe parameter
sampling raten( is),an integer
which known
is used for as the number
capturing ( ).ofThe
decomposition
n also can belevels and itbased
computed is seton
based
the
on the sampling
frequency intervalrate ( f s ), which
covered by the is used for capturing
fault-related component.i (t).Taking
The n into
alsoaccount
can be computed
whether the based on
second
the frequency
rule is considered,interval covered
n should by the
be less thanfault-related
predicted bycomponent.
the sampling Taking into approximation
rate. The account whether the
signal
( ) isrule
second is considered,
obtained using the n should
scalingbefunction,
less than deduced
predictedby by the sampling rate. The
father wavelet ( approximation
) and scaling
signal An (t) is: obtained using the scaling function, deduced by the father wavelet (ϕ j ) and scaling n
coefficients
coefficients anj :
An(()t)== ∑ anj ·∙ϕnj ((t)) (4)
(4)
i

The
The detail signalD (t() )is is
detail signal calculated
calculated using
using
j as a base;
di as a base; the wavelet
the wavelet coefficients
coefficients with
with level level is,
j, which
n
which is a scaled and time-expanded version of the wavelet functions, are deduced by the mother j
a scaled and time-expanded version of the wavelet functions, are deduced by the mother wavelet (ψi ):
wavelet ( ):

∑ di ·∙ψi (t() )
j j
D(n ()t)== (5)
i (5)
Appl. Sci. 2017, 7, 845 5 of 20
Appl. Sci. 2017, 7, 845 5 of 20

Figure
Figure 4 illustratesthe
4 illustrates thedyadic
dyadicwavelet
wavelet decomposition
decomposition algorithm
algorithmregarding
regardingthe thecoefficients of of
coefficients thethe
transform
transform at at
thethe differentlevels
different levelsaccording
accordingto
tothe
the description
description by
by [25].
[25].In
Inthis
thisfigure,
figure,the length
the of of
length those
those
coefficients and the frequency content at each level is shown, considering an original
coefficients and the frequency content at each level is shown, considering an original signal with signal with a a
sampling rate of samples/s.
sampling rate of f samples/s.
s

Figure4.4.Dyadic
Figure Dyadic wavelet
wavelet decomposition algorithm.
decomposition algorithm.

The
The analysisofofthese
analysis thesesignals
signals reveals
reveals the
the time
time evolution
evolutionofofthe
thecomponents
components of of
thethe signal( )i (t)
signal
which are contained within its corresponding frequency band. The approximation signal
which are contained within its corresponding frequency band. The approximation signal includes the includes
the low-frequency components of the signal, belonging to the interval:
low-frequency components of the signal, belonging to the interval:
( ) ∈ 0. 2 ( ) Hz (6)
f ( An ) ∈ [0.2−(n+1) f s ]Hz (6)
More concretely, the detail contains the information concerning the signal components with
frequencies included in the interval:
More concretely, the detail contains the information concerning the signal components with
frequencies included in the interval: ( ) ∈ 2 ( ) · 2 Hz (7)
There are many different mother wavelet −( functions that are categorized based on their real type,
f ( Dn ) ∈ [2 n+1) f s ·2−n f s ]Hz (7)
such as the Coiflet wavelet, and their complex type, such as the Morlet wavelet. In fault detection
algorithms, different types of mother wavelets have been used with satisfactory results [26,27]. In this
There are many different mother wavelet functions that are categorized based on their real type,
work, the DWT of the motor start-up current has been performed using a fourth-order biorthogonal
such as the Coiflet wavelet, and their complex type, such as the Morlet wavelet. In fault detection
Coiflet wavelet. The sampling frequency used to capture the data in this research is 5000 samples/s.
algorithms, different types of mother wavelets have been used with satisfactory results [26,27]. In this
To define the suitable level of decomposition, the main signal of the motor current was initially
work, the DWT of the motor start-up current has been performed using a fourth-order biorthogonal
decomposed into seven levels, where each level has its own range of frequencies. Table 1 presents
Coiflet wavelet. The sampling frequency used to capture the data in this research is 5000 samples/s.
the frequency bands equivalent to each level. Clearly, the fundamental frequency is located in the
To frequency
define therange
suitable level of decomposition,
corresponding the six.
to the detail of level main signal of the motor current was initially
decomposed into seven levels, where each level has
The qualitative pattern of LSH in the high decomposition its own rangelevelofof
frequencies.
the signals Table
can be1apresents
good
theindicator
frequency forbands equivalent
the fault detection,toaseach
shownlevel. Clearly,
in Figure 5. the fundamental
Hence, frequency
the reconstructed is located
signal in the
that comes
frequency range corresponding to the detail of level six.
from the detail and approximation of level seven, and also the approximation of level six, is used as
anThe
inputqualitative pattern of the
signal for calculating LSH in the high
fault-related decomposition
feature. The relevantlevel of the bands
frequency signalsofcan
D7 (be a good
Detail of
indicator for the
level seven), A7fault detection, asofshown
(approximation in Figure
level seven) and 5.
A6Hence, the reconstructed
(approximation signal
of level six) that comes
are shown in
from the1.detail
Table and approximation
Accordingly, the featuresof level seven,
evaluated and also theBRB
for monitoring approximation of levelare
faults in LS-PMSM six,based
is usedon as
an these
inputrelevant frequency
signal for bands.
calculating the fault-related feature. The relevant frequency bands of D7 (Detail
of level seven), A7 (approximation of level seven) and A6 (approximation of level six) are shown in
Appl. Sci. 2017, 7, 845 6 of 20

Table 1. Accordingly, the features evaluated for monitoring BRB faults in LS-PMSM are based on these
relevant frequency bands.
Appl. Sci. 2017, 7, 845 6 of 20
Table 1. Frequency ranges for the wavelet decomposition of the signal.
Table 1. Frequency ranges for the wavelet decomposition of the signal.
Frequency Ranges (Hz)
Decomposition Level Frequency Ranges (Hz)
Decomposition Level Detail (D) Approximation
Detail (D) Approximation (A) (A)
level 1level 1 2500–1250
2500–1250 1250–0
1250–0
level 2level 2 1250–625
1250–625 625–0625–0
level 3level 3 625–312.5
625–312.5 312.5–0
312.5–0
level 4level 4 312.5–156.2
312.5–156.2 156.2–0
156.2–0
level 5 156.25–78.12 78.12–0
level 5 156.25–78.12 78.12–0
level 6 78.12–39.06 39.06–0
level 6 78.12–39.06 39.06–0
level 7 39.06–19.53 19.53–0
level 7 39.06–19.53 19.53–0

Figure 5. Theoretical behavior of f LSH as a function of slip.

Figure 5. Theoretical behavior of as a function of slip.

Entropy
Entropy in in information
information theory
theory describes
describes the
the amount
amount of of information
information provided
provided by by aa signal
signal or
or
event.
event. It
It relates
relates the
the uncertainty
uncertainty of the signal
of the signal or
or event
event associated
associated with
with aa given
given probability
probability distribution.
distribution.
The
The concept
concept of of entropy
entropy has has found
found broad
broad applications
applications in in engineering,
engineering, including,
including, for
for instance,
instance, fault
fault
diagnoses
diagnoses [28–31]. A survey of recent methods for the fault diagnosis of rotating machinery using
[28–31]. A survey of recent methods for the fault diagnosis of rotating machinery using
entropy
entropy techniques
techniques waswas discussed
discussed in in [32].
[32]. According
According to to these
these studies,
studies, tntropy
tntropy has
has been
been used
used aa feature
feature
for
for the diagnosis of a failure in electrical machines. The features used in this research are log-energy
the diagnosis of a failure in electrical machines. The features used in this research are log-energy
entropy
entropy andand Shannon
Shannon entropy.
entropy. These
These two
two features
features are
are common
common concepts
concepts inin many
many fields,
fields, mainly
mainly inin
signal processing, and are listed as
signal processing, and are listed as bellow:bellow:

Entropy“Shannon”
Entropy ∑∑i Xn2 log
“Shannon”XEShannon ==−− log Xn2 (8)
(8)
with the convention 0log(0) = 0.
with the convention 0log(0) = 0.
Entropy “log-energy” = ∑ log( ) (9)
Entropy “ log −energy” XELog energy = ∑i log( Xn2 ) (9)
with the convention log(0) = 0.
where,
with is the reconstructed
the convention signal from
log(0) = 0. where, X is level n.
the reconstructed signal from level n.
n
Once
Once the
the features
features are
are determined,
determined, statistical
statistical techniques
techniques should
should be be used
used in
in exploratory
exploratory data
data
investigation.
investigation. The movement of features is surveyed to discover the overlapping of healthy and
The movement of features is surveyed to discover the overlapping of healthy and faulty
faulty
conditions
conditions at
at different
different levels
levels of
of load.
load. AA number
number of of methods
methods have
have been
been implemented
implemented for
for visualizing
visualizing
and
and checking normality in multi-level experiments to provide a good understanding of the overall
checking normality in multi-level experiments to provide a good understanding of the overall
characteristics
characteristicsofofthe
thedata.
data.Thus,
Thus,bybylisting
listingallalldata
datawith
witha multiplier, a boxplot
a multiplier, cancan
a boxplot be used to check
be used the
to check
the assumption of normality. Boxplots, in this research, are considered to analyse and compare
various features of parameter. The parameters investigated are two types of motor conditions,
healthy and faulty, and four different stages of starting load are considered for each condition.
The estimation of the statistical significance of differences was verified with a two-way Analysis of
variance (ANOVA) based on the motor condition, the starting load, and their interaction. Tukey’s
Appl. Sci. 2017, 7, 845 7 of 20

assumption of normality. Boxplots, in this research, are considered to analyse and compare various
features of parameter. The parameters investigated are two types of motor conditions, healthy and
faulty, and four different stages of starting load are considered for each condition. The estimation of the
statistical significance of differences was verified with a two-way Analysis of variance (ANOVA) based
on the motor condition, the starting load, and their interaction. Tukey’s honest significant differences
technique was also utilized for multiple comparisons after ANOVA. There are three sets of hypotheses
should beSci.
Appl. tested
2017, 7,with
845 the two-way ANOVA in this research. The null hypotheses for each of 7 ofthe
20 sets
are given below:
There are three sets of hypotheses should be tested with the two-way ANOVA in this research.
1. Thenull
The population means
hypotheses of the
for each first
of the factor
sets (motor
are given condition) are equal;
below:
2. The The
1. population means
population meansofof
the
thesecond factor
first factor (load)
(motor are equal;
condition) are equal;
3. There is no interaction between the two factors (motor
2. The population means of the second factor (load) are equal; condition and load).
3. There is no interaction between the two factors (motor condition and load).
Accordingly, the most important point of current research is to identify the features related to
BRB fault Accordingly,
detection in the most important
LS-PMSM. A briefpoint of current
overview of theresearch is to identify
methodology the features
employed for BRBrelated to in
detection
BRB fault detection in LS-PMSM. A brief overview of the methodology employed for
this research is depicted in Figure 6. In this paper, the performance of LS-PMSM under BRB during BRB detection
in this research is depicted in Figure 6. In this paper, the performance of LS-PMSM under BRB during
transient operation is scrutinized employing the finite element method (FEM).
transient operation is scrutinized employing the finite element method (FEM).

Figure 6. Methodology employed in the research design.

Figure 6. Methodology employed in the research design.
3. Simulation and Experiment Process

3.1. Simulation
This study looks into the dynamics of LS-PMSM using a two-dimensional time-stepping FEM
method. The major advantage of using a two-dimensional model is to reduce the computation time.
In this research, Ansoft® Maxwell 2D (Pennsylvania, PA, USA) software version 15 is used to simulate
Appl. Sci. 2017, 7, 845 8 of 20

3. Simulation and Experiment Process

3.1. Simulation
This study looks into the dynamics of LS-PMSM using a two-dimensional time-stepping FEM
method. The major advantage of using a two-dimensional model is to reduce the computation time.
Appl. Sci. 2017, 7, 845 8 of 20
In this research, Ansoft® Maxwell 2D (Pittsburgh, PA, USA) software version 15 is used to simulate the
healthy
the healthyandandfaulty condition
faulty of LS-PMSMs.
condition The
of LS-PMSMs. specification
The specificationofofthe
theLS-PMSM
LS-PMSMhas
hasbeen
beensummarized
summarized
in Table 2. Figure 7 illustrates the Maxwell 2D model for a LS-PMSM with meshed
in Table 2. Figure 7 illustrates the Maxwell 2D model for a LS-PMSM with meshed region. region.

Table 2. Motor
Table 2. Motor specifications.
specifications.

Parameters
Parameters Value
Value Parameters
Parameters Value
Value
Rated Power (HP) 1 Starting Torque (Nm) 2.3
Rated Power (HP) 1 Starting Torque (Nm) 2.3
Rated
Rated Voltage
Voltage (V)(V) 415
415 Number of
Number ofPoles
Poles 44
Rated Frequency
Rated Frequency (Hz) (Hz) 5050 Connection
Connection YY
Rated Torque
Rated Torque (Nm) (Nm) 4.8
4.8 Number of
Number of Stator
StatorSlots
Slots 24
24
Rated
Rated Speed
Speed (rpm)
(rpm) 1500
1500 Number of Rotor Slots
Number of Rotor Slots 16
16
Rated Current (A)
Rated Current (A) 1.3
1.3 Permanent
Permanent Magnet
Magnet N38H
N38H

Figure 7. Cross-section of LS-PMSM with mesh plotting.

plotting.

3.2. Experimental
3.2. Experimental and
and Signal
Signal Capturing
Capturing
An experimental
An experimental study
study was
was conducted
conducted to to validate
validate the
the effectiveness
effectiveness of of the
the proposed
proposed method
method
using the self-designed test rig shown in Figure 8a. It includes the LS-PMSM, sensor
using the self-designed test rig shown in Figure 8a. It includes the LS-PMSM, sensor and powder and powder
break. Magnetic
break. Magnetic powder
powder brake
brake offers
offers aa variable
variable mechanical
mechanical load
load byby adjusting
adjusting the
the voltage.
voltage. Two
Two one-hp,
one-hp,
four-pole LS-PMSMs
four-pole LS-PMSMs werewere considered
considered to to generate
generate data
data under
under four
four levels
levels of
of starting
starting load
load (0,
(0, 0.5,
0.5, 11 and
and
1.5 Nm). The first motor was considered as healthy and taken as a standard for evaluation
1.5 Nm). The first motor was considered as healthy and taken as a standard for evaluation with the with the
faulty motor.
faulty motor. The
The second
second motor
motor was
was faulty,
faulty,with
withone
oneBRB,
BRB,as asillustrated
illustratedininFigure
Figure8b.
8b.
 An experimental study was conducted to validate the effectiveness of the proposed method
using the self-designed test rig shown in Figure 8a. It includes the LS-PMSM, sensor and powder
break. Magnetic powder brake offers a variable mechanical load by adjusting the voltage. Two one-hp,
four-pole LS-PMSMs were considered to generate data under four levels of starting load (0, 0.5, 1 and
1.5Sci.
Appl. Nm).2017,The first motor was considered as healthy and taken as a standard for evaluation with the
7, 845 9 of 20
faulty motor. The second motor was faulty, with one BRB, as illustrated in Figure 8b.

Appl. Sci. 2017, 7, 845 9 of 20

A schematic diagram of the experimental test setup including the data acquisition technique and
computer interfacing is shown in Figure 9. The signal of the stator current was collected noninvasively
Figure8.8.(a)
(a)The
Theexperimental
experimental test rig and
and (b)
(b)the
therotor with
withone broken rotor bar (BRB).
while the systemFigure was on operation. Tworigcategories rotor
of signal one broken
exist: rotor
transient bar (BRB).
and stationary.
A transient signal is the signal obtained during motor start-up. It is selected as a feature source and
analyzed A schematic diagram of the
in the time-frequency experimental
domain. test setup
The sampling including
frequency is the datacomponent
a vital acquisitionin technique
capturingand
computer interfacing is shown in Figure 9. The signal of the stator current
of the transient signal and it is a significant point that should be taken into account. was collected noninvasively
while the system
According was on operation.
to Nyquist criterion, Two categories of of
implementation signal
thisexist:
method transient
needsand stationary.
a very A transient
high sampling
signal is the
frequency, signal
since obtained of
the majority during motor start-up.
the features It islow-frequency
exist in the selected as a feature
section source and analyzed
[33]. The sampling in
the time-frequency domain. The sampling frequency is a vital component in
frequency of 5 ks/s can provide a good resolution for the analysis of the transient signal here. Incapturing of the transient
each
signal
case, and it
40 tests is a performed;
were significant point
hence,that
320should
data setsbe taken
in totalinto account.
were acquired for each analysis.

Figure 9. Schematic diagram of the experimental test setup.

Figure 9. Schematic diagram of the experimental test setup.

4. Result and Discussion

According to Nyquist criterion, implementation of this method needs a very high sampling
frequency,
4.1. Performance since the majority
of LS-PMSM withofPresence
the features
of Faultexist in the low-frequency section [33]. The sampling
frequency of 5 ks/s can provide a good resolution for the analysis of the transient signal here. In each
In this
case, section,
40 tests werethe effects ofhence,
performed; BRB on320 thedata
performance ofwere
sets in total motor during for
acquired start-up
each are investigated.
analysis.
In the first step, the simulation result was used to analyze the performance of a healthy motor and a
motor with and
4. Result one broken
Discussion bar under different levels of load. In the second step, the experimental result
was used to confirm the effects of the fault on machine performance.
4.1.InPerformance
the simulationof LS-PMSM
study, FEMwithisPresence
appliedoftoFault
investigate the performance of LS-PMSM when any
BRB exists. The waveform of the magnetic field
In this section, the effects of BRB on the performancecontains comprehensive information
of motor during start-upabout the stator
are investigated.
andIn mechanical
the first step, parts of the motor.
the simulation Figure
result was10a shows
used the symmetrical
to analyze distribution
the performance of magnetic
of a healthy motorfluxand a
in motor
the healthy motor and Figure 10b presents the asymmetrical distribution of magnetic
with one broken bar under different levels of load. In the second step, the experimental result flux in a
motor with one BRB under 1.5 Nm starting torque. Comparing
was used to confirm the effects of the fault on machine performance. Figure 10a,b shows that the distribution
of the flux lines
In the around the
simulation broken
study, FEMbars differsto
is applied from healthythe
investigate bars and the density
performance increase when
of LS-PMSM of magnetic
any BRB
flux
exists. The waveform of the magnetic field contains comprehensive information about the stator of
in the rotor core around the broken bar, stator core, and the air gap. The concentration and
magnetic
mechanical flux parts
observedof thearound
motor.the broken
Figure 10a bar
showscreates an asymmetric
the symmetrical distribution
distribution of magnetic
of magnetic flux flux
in the
in healthy
this area. When
motor andaFigure
bar breaks, its current
10b presents is distributed
the asymmetrical in the adjacent
distribution bars,flux
of magnetic meaning thatwith
in a motor more one
current flows in them. The excess current causes saturation and generates more heat
BRB under 1.5 Nm starting torque. Comparing Figure 10a,b shows that the distribution of the flux lines in adjacent bars,
which results in an asymmetrical distribution of magnetic flux. The generated heat makes the situation
of the adjacent bars worse and then causes a problem for the permanent magnet. The residual flux
density and coercivity of permanent magnets reduce due to the proximity of the cages to the magnets
as a function of temperature.
According to Faraday’s law, an electromagnetic field will be induced in the cage because of the
Appl. Sci. 2017, 7, 845 10 of 20

around the broken bars differs from healthy bars and the density increase of magnetic flux in the rotor
core around the broken bar, stator core, and the air gap. The concentration of magnetic flux observed
around the broken bar creates an asymmetric distribution of magnetic flux in this area. When a bar breaks,
its current is distributed in the adjacent bars, meaning that more current flows in them. The excess current
causes saturation and generates more heat in adjacent bars, which results in an asymmetrical distribution
of magnetic flux. The generated heat makes the situation of the adjacent bars worse and then causes
a problem for the permanent magnet. The residual flux density and coercivity of permanent magnets
reduce due to the proximity of the cages to the magnets as a function of temperature.
According to Faraday’s law, an electromagnetic field will be induced in the cage because of the
magnetic field fluctuating. The electromagnetic field will then generate a current through the cage and
the situation becomes as a current carrying loop situated in a magnetic field. According to the Lorentz
law,
Appl. aSci.
magnetic
2017, 7, 845force is produced in the cage and causes the cage starts to rotate. Magnetic force 10 ofin
20
a healthy rotor is symmetrically distributed around the rotor. A magnetic asymmetry due to the BRB
Appl. Sci. 2017, 7, 845 10 of 20
introduces
The resultsan unbalanced
shown magnetic
in the Figure 11 force.
revealForthatboth the healthy of
the amplitude motor and motor
the magnetic withfor
force BRB,
the the magnetic
faulty motor
force
is The distributions
higher than for on the rotor
healthy bar at
motor. start-up
This (0 Nm) are
spontaneous computedofbythe
distribution theforce
FEMunavoidably
method. Theleadsresultsto
results shown in the Figure 11 reveal that the amplitude of the magnetic force for the faulty motor
shown
undue in thethan
Figure
mechanical
is higher
11healthy
reveal
forstress
thatbars,
in the theThis
motor.
amplitude of the
and spontaneous
it would magnetic
become moreforce for the faulty
vulnerable
distribution
motor iswearing
to additional
of the force unavoidably
higher than
leads to and
for healthy
anundue
eventual motor.
break. This spontaneous distribution of the force unavoidably leads to undue
mechanical stress in the bars, and it would become more vulnerable to additional wearing and mechanical
stress in the bars,
an eventual and it would become more vulnerable to additional wearing and an eventual break.
break.

Figure 10. The magnetic flux line in the (a) healthy motor and (b) faulty motor (one BRB).
Figure
Figure 10.10.
TheThe magneticflux
magnetic fluxline
linein
inthe
the(a)
(a) healthy
healthy motor
motor and
and (b)
(b)faulty
faultymotor
motor(one
(oneBRB).
BRB).

Figure
Figure
Figure 11.Magnetic
11.
11. Magneticforce
Magnetic forcein
force instarting
in starting torque
starting torque
torque in
inthe
in thehealthy
the healthyand
healthy andfaulty
and faultymotor.
faulty motor.
motor.

The squirrel-cage bars provide the accelerating torque which operates the rotor to a near-
The squirrel-cage
The squirrel-cage bars provide thetheaccelerating torque which operates the rotor to a near-
synchronous speed. In bars provide
addition to overcomingaccelerating
the appliedtorque which
load, the operates
accelerating the must
torque rotor to a
also
synchronous speed.
near-synchronous In addition
speed. Inmagnet to
addition overcoming
to overcoming the applied load,
the applied the accelerating
load,asthe torque
accelerating must
torque must also
overcome the generated braking torques, which is present a result of the permanent
overcome the generated magnet braking torques, which is present as a result of the permanent
magnet. When BRB occurs, the torque characteristics of the motor also change. Figure 12 illustrates
magnet. When BRB occurs, the torque characteristics of the motor also change. Figure 12 illustrates
the comparison of faulty and fault-free motors under a maximum starting torque. As is clear,
thethecomparison of faulty
motor with one and fault-free
BRB cannot motors
run at starting under
torque valueaofmaximum starting
“2.3 Nm”, which thetorque.
motor isAs is clear,
designed
thefor,
motor with one BRB cannot run at starting torque value of “2.3 Nm”, which the motor
and the starting torque value decreases to near 1.75 Nm. Accordingly, the value of the starting is designed
for,torque
and the starting torque
is decreased value
whenever decreases
there is a BRBto near 1.75 Nm.while
in LS-PMSM, Accordingly, the valuehas
this phenomenon of the
not starting
been
Appl. Sci. 2017, 7, 845 11 of 20

also overcome the generated magnet braking torques, which is present as a result of the permanent
magnet. When BRB occurs, the torque characteristics of the motor also change. Figure 12 illustrates the
comparison of faulty and fault-free motors under a maximum starting torque. As is clear, the motor
with one BRB cannot run at starting torque value of “2.3 Nm”, which the motor is designed for,
and the starting torque value decreases to near 1.75 Nm. Accordingly, the value of the starting torque
is decreased whenever there is a BRB in LS-PMSM, while this phenomenon has not been reported
for induction
Appl. Sci. 2017, 7,machines.
845 As the presence of BRB changes the torque characteristic of the LS-PMSM,
11 of 20
Appl. Sci. 2017, 7, 845
early detection of this fault is very important. 11 of 20

Figure 12. Comparison of the maximum starting torque [2.3 Nm] for (a) healthy and (b) faulty motor.
Figure 12. Comparison of the maximum starting torque [2.3 Nm] for (a) healthy and (b) faulty motor.
Figure 12. Comparison of the maximum starting torque [2.3 Nm] for (a) healthy and (b) faulty motor.
Another issue found during the simulation analysis is an increase of time duration for the
Another issue found during the simulation analysis is an increase of time duration for the
transient
Another section
issue when
found any bar is the broken. Figure 13 illustrates the startingoftime for both thefor healthy
transient section when any during
bar is broken. simulation
Figure 13analysis is an
illustrates theincrease
starting time time
for duration
both the healthy the
and faulty
transient motor
section whenunderany different
bar is levels Figure
broken. of load.13 As is clear the
illustrates in the figure,
starting timetheforstarting
both time
the trend
healthy
and faulty motor under different levels of load. As is clear in the figure, the starting time trend
increased
and based on
faultybased
motor increment in levels
the load. A recent study performed on athesquirrel-cage induction
increased onunder different
increment in the load. of Aload.
recentAsstudy
is clear in the figure,
performed starting time
on a squirrel-cage trend
induction
motor
increased revealed
based that
on the startup
increment in load
the is
load. not A important
recent in
study BRB fault
performed detection
on a based on
squirrel-cage the transient
induction
motor revealed that the startup load is not important in BRB fault detection based on the transient
analysis;
motor however,
revealed theystartup
thatthey
the mentioned
loadthatisthat
not this approach is especially
important suitable for applications with
analysis; however, mentioned this approachinisBRB fault detection
especially suitablebased on the transient
for applications with
heavy
analysis; startup
however, transients [34].
they mentioned Other thatresearchers
this approachalso did not consider the effects of starting load for
heavy startup transients [34]. Other researchers also is
didespecially suitable
not consider thefor applications
effects of starting withloadheavyfor
fault
startup detection
transients in induction machines based on the analysis of current in transient state [35,36].
fault detection in [34]. Othermachines
induction researchers also on
based didthe
notanalysis
considerofthe effectsinoftransient
current starting loadstatefor fault
[35,36].
However,
detection inthis research
induction indicatesbased
machines that the on starting
the load of
analysis effects in the
current in transient time
transient state should
[35,36]. be carefully
However,
However, this research indicates that the starting load effects in the transient time should be carefully
taken
this into account
research indicates for BRB fault detection in LS-PMSM.
taken into account for that
BRBthe starting
fault detection loadineffects in the transient time should be carefully taken into
LS-PMSM.
account for BRB fault detection in LS-PMSM.

Figure13.
Figure 13.Comparison
Comparisonofofthe
theresult
resultofofstarting
startingtime
timeunder
underdifferent
differentloads.
loads.
Figure 13. Comparison of the result of starting time under different loads.
ForLS-PMSM
For LS-PMSM that starts with a squirrel cage, the application of the transient current is even
For LS-PMSMthat thatstarts
startswith
witha squirrel
a squirrel cage, thethe
cage, application of the
application of transient current
the transient is even
current is more
even
more justified,
justified, due toduedue to the
thetocarryingcarrying of significant
of significant currents
currents in the in the squirrel cage during the start-up. This
more justified, the carrying of significant currents in squirrel cagecage
the squirrel during thethe
during start-up. This
start-up. is
This
is
why why the
thethe steady-state
steady-state signal
signal is impractical
is impractical for
for for
BRBBRB BRB fault
faultfault detection
detection in thisin this
motor. motor. The
The presence presence
of BRB of
is why steady-state signal is impractical detection in this motor. The presence of
BRB in
in LS-PMSM LS-PMSM can
can produce produce
changes changes in
in theinair the air gap flux and the current distribution between the
BRB in LS-PMSM can produce changes thegap
air flux
gap and
flux the
andcurrent distribution
the current between
distribution the bars
between the
bars during
during acceleration
acceleration from fromtozero
zero to rated
rated speed. speed.
This Thisindicated
part part indicated
that thatBRBs
the the BRBs slightly
slightly modified
modified the
bars during acceleration from zero to rated speed. This part indicated that the BRBs slightly modified
the performance
performance of theof the motor during startup and is critical in fault detection during transient time.
the performance of motor
the motorduring startup
during and and
startup is critical in fault
is critical detection
in fault during
detection transient
during time.time.
transient
The effect of BRB in starting torque is also surveyed experimentally. It was found that the motor
The effect of BRB in starting torque is also surveyed experimentally. It was found that the motor
with one BRB cannot run at a starting torque value of “2.3 Nm”, which the motor is designed for, and
with one BRB cannot run at a starting torque value of “2.3 Nm”, which the motor is designed for, and
the starting torque value decreases to near 1.70 Nm. The rate of torque variations for the faulty motor
the starting torque value decreases to near 1.70 Nm. The rate of torque variations for the faulty motor
is higher than for the healthy one, so a higher noise accompanied with lower performance are
is higher than for the healthy one, so a higher noise accompanied with lower performance are
expected from the faulty motor. The time duration for the transient section is also determined through
expected from the faulty motor. The time duration for the transient section is also determined through
Appl. Sci. 2017, 7, 845 12 of 20

The effect of BRB in starting torque is also surveyed experimentally. It was found that the motor
with one BRB cannot run at a starting torque value of “2.3 Nm”, which the motor is designed for,
and the starting torque value decreases to near 1.70 Nm. The rate of torque variations for the faulty
motor is higher than for the healthy one, so a higher noise accompanied with lower performance are
expected from the faulty motor. The time duration for the transient section is also determined through
experimental work. The stator current signal for the different load situations is illustrated in Figure 14
for healthy and faulty motors. It is obvious in this figure that instances of broken bars lengthen the
starting transient time of the motor. Figure 15 provides a comparison between the starting time for
both healthy and faulty motors in different levels of load. The results presented in Figure 15 are based
Appl. Sci. 2017, 7, 845 12 of 20
onAppl.
a mean of 40
Sci. 2017, samples for each condition.
7, 845 12 of 20

Figure 14. Experimental current signal in four load conditions for healthy and faulty motors.
Figure
Figure
(a) No 14. Experimental
Experimental
14.load; (b) 0.5 Nm; (c) current
current signal
1 Nm;signal
(d) in four
1.5in
Nm. four load
loadconditions
conditionsfor
forhealthy
healthyand
andfaulty
faultymotors.
motors.
(a)(a)
NoNo load;
load; (b)(b)
0.50.5 Nm;
Nm; (c)(c) 1 Nm;
1 Nm; (d)(d)
1.51.5 Nm.
Nm.

Figure
Figure 15.15. Comparison
Comparison of of
thethe experimental
experimental result
result of of starting
starting time
time under
under different
different loads,
loads, based
based onon
thethe
Figure
mean 15. of
value Comparison
40 samplesoffor
the experimental
each load. result of starting time under different loads, based on the
mean value of 40 samples for each load.
mean value of 40 samples for each load.
4.2. Fault Related Feature Evaluation
4.2. Fault Related Feature Evaluation
The proposed technique for the detection of BRB in LS-PMSM includes two steps: first,
The proposedpattern
the characteristic technique forwhich
of LSH, the detection
is shown of BRB in4, LS-PMSM
in Figure includes
is calculated, two steps:
and second, first,
the entropy
the characteristic
features patternInofthe
are computed. LSH,
firstwhich
stage,isknowing
shown inthe
Figure
value4,ofisthe
calculated, and
motor slip is second, the entropy
compulsory for LSH
features are computed. In the first stage, knowing the value of the motor slip is compulsory
calculation. Therefore, the speed of the tested motor was measured and the slip was calculated. Figure for LSH 16
Appl. Sci. 2017, 7, 845 13 of 20

4.2. Fault Related Feature Evaluation

The proposed technique for the detection of BRB in LS-PMSM includes two steps: first,
the characteristic pattern of LSH, which is shown in Figure 4, is calculated, and second, the entropy
features are computed. In the first stage, knowing the value of the motor slip is compulsory for
LSH calculation. Therefore, the speed of the tested motor was measured and the slip was calculated.
Figure 16 presents the behavior of LSH determined using Equation (2), where s and f are slip and
fundamental frequency, respectively. These values are obtained from measuring the motor speed for
both faulty and fault-free motors when there is no starting load and when there is a starting load equal
to 1.5 Nm. As is clear in this figure, the frequency of the LSH will change between the fundamental
frequency (s = 1), zero frequency (s = 1/2) and again near the fundamental frequency (s ≈ 0) when
the motor reaches a steady state. From this figure, it should be noted that there is a shift between the
faulty and fault-free motor under the same level of load that is also created by the effect of BRB.
Appl. Sci. 2017, 7, 845 13 of 20

Figure 16. The behavior of LSH from experimental data.

Figure 16. The behavior of LSH from experimental data.
The second stage is the quantitative approach, which begins with applying the wavelet
transforms to the startup current signal. By applying wavelet transform to the signal, sets of wavelet
The second stage is the quantitative approach, which begins with applying the wavelet transforms to
signals are obtained based on the approximation and detail coefficient. The method is to track the
the startup current signal. By applying wavelet transform to the signal, sets of wavelet signals are obtained
evolution of the LSH. As the LSH changes in the range of 50~0~80 Hz, its evolution is reflected by the
based on thesignals
wavelet approximation
and coversand the detail coefficient.
frequencies The method
lower than is to track
the fundamental the evolution
frequency of thecancels
and, hence, LSH. As the
LSHtheir
changes in the range of 50~0~80 Hz, its evolution is reflected by the
effects. The alternative variant proposed for detecting this evolution is to perform the signal wavelet signals and covers
the frequencies
decomposition lowerin athan
numberthe fundamental frequency
of levels that result in two and, hence, cancels
approximations andtheir
oneeffects. The alternative
detail signal and
variant proposed
covers for detectingfrequency
the aforementioned this evolution
range.isThe
to perform
number of the signal
levels decomposition
that in a number
lead to such a result with theirof levels
type of coefficients is marked in Figure 15. The next step, in this stage, involves
that result in two approximations and one detail signal and covers the aforementioned frequency range. entropy computation
The to quantify
number offailure
levelsinthat LS-PMSM.
lead to Two suchdifferent
a result features
with theirintroduced
type of in the previous
coefficients section are
is marked in then
Figure 15.
applied in these three frequency bands (A6, A7 and D7). After all data
The next step, in this stage, involves entropy computation to quantify failure in LS-PMSM. Two have been analysed, the result
different
is investigated with trend and boxplot graphs.
features introduced in the previous section are then applied in these three frequency bands (A6, A7 and
Figure 17 illustrates the trend and boxplot graphs for entropy features used as a fault feature in
D7). After all data have been analysed, the result is investigated with trend and boxplot graphs.
this study. A number of interesting conclusions can be drawn from the analysis of these figures.
Figure
The most17 illustrates
significant the the
is that trend andare
trends boxplot graphs
dissimilar for allfor
theentropy
conditions, features
and thusused
eachasindividual
a fault feature
in this
casestudy.
must be A studied
numberand of analyzed
interesting conclusionsThe
independently. cantrend
be drawn
graph from the analysis
for Shannon entropyofshows
thesean figures.
The overlap
most significant is that the trends are dissimilar for all the conditions,
between the healthy and faulty states at different levels of starting load for approximation and thus each individual
caseand
must be studied
detail and analyzed
of decomposition independently.
level seven (Figure 17b,c). TheThe trend
trendgraph
graphfor for Shannon entropyinshows
Shannon entropy the an
approximation
overlap between the of level six indicates
healthy that there
and faulty states is no overlapping
at different between
levels the healthy
of starting loadandforfaulty states
approximation
and at various
detail levels of starting level
of decomposition load. seven
However, by looking
(Figure 17b,c).atThe
the boxplot
trend graph graph,for
there is an overlap
Shannon entropy for in the
each condition in the same load (Figure 17a).The trend graph for
approximation of level six indicates that there is no overlapping between the healthy and faulty log-energy entropy shows an states
overlapping between the healthy situation and faulty state at various levels of starting load for an
at various levels of starting load. However, by looking at the boxplot graph, there is an overlap for each
approximation of level seven (Figure 17e). The trend graphs for log-energy entropy in the
condition in the same load (Figure 17a).The trend graph for log-energy entropy shows an overlapping
approximation of level six and detail of level seven indicate that there is no overlapping between the
between
healthytheandhealthy
faulty situation and faulty
states at various levelsstate at various
of starting levels of
load (Figure starting load for an approximation
17d,f).
Appl. Sci. 2017, 7, 845 14 of 20

of level seven (Figure 17e). The trend graphs for log-energy entropy in the approximation of level six
and detail of level seven indicate that there is no overlapping between the healthy and faulty states at
variousAppl.
levels of starting
Sci. 2017, 7, 845 load (Figure 17d,f). 14 of 20

Figure 17. Cont.

 Appl.
Appl. Sci.Sci. 2017,
2017, 7, 845
7, 845 1520
15 of of 20

Figure Trend
17.17.
Figure andand
Trend boxplot forfor
boxplot entropy features
entropy under
features different
under levels
different of starting
levels load.
of starting (a) (a)
load. Shannon
Shannon
entropy
entropy(A6); (b)(b)
(A6); Shannon
Shannonentropy
entropy(A7); (c) (c)
(A7); Shannon
Shannonentropy
entropy(D7); (d)(d)
(D7); log-energy
log-energyentropy
entropy (A6);
(A6);
(e) (e)
log-energy entropy
log-energy (A7);
entropy (f) log-energy
(A7); entropy
(f) log-energy (D7).
entropy (D7).
Appl. Sci. 2017, 7, 845 16 of 20

The evaluation of the statistical significance of differences (p-value ≤ 0.05) was verified with a
two-way ANOVA based on the motor condition, starting load condition, and their interaction. Tukey’s
honest significant differences method was also applied for multiple comparisons after ANOVA. Table 3
presents the results of the ANOVA models for the Shannon entropy feature in three different sub-band
frequencies based on the condition as mention above. As can be seen in this table, the p-value for the
features that were extracted from (A6) is more than 0.05 for the interaction condition states and also
for (D7) in both the motor condition and their interaction, which can be checked with a post-hoc test.
The performed ANOVA for Shannon entropy at (A7) reveals it can significantly distinguish different
conditions. Table 4 presents the results of the ANOVA models for the log-energy entropy feature in
three different sub-band frequencies based on similar conditions, as mentioned earlier. The performed
ANOVA for log-energy entropy features at different sub-band frequencies reveals its capability to
differentiate various states tested, as the p-value for all decomposition levels is significant.

Table 3. Analysis of variance for Shannon entropy features.

Source Sum of Squares df Mean Square F p-Value

Motor Condition 2.86 × 108 1 2.86 × 108 15.635 9.51 × 10−5
Load 5.35 × 109 3 1.78 × 109 97.572 1.45 × 10−44
A6 Interaction 1.66 × 107 3 5.53 × 106 0.302 0.824 *
Error 5.70 × 109 312 1.83 × 107
Shannon Entropy

Total 1.14 × 1010 319

Motor Condition 8.96 × 107 1 8.96 × 107 25.851 6.37 × 10−7
Load 1.87 × 108 3 6.22 × 107 17.941 9.11 × 10−11
A7 Interaction 3.96 × 107 3 1.32 × 107 3.808 1.05 × 10−2
Error 1.08 × 109 312 3.47 × 106
Total 1.40 × 109 319
Motor Condition 2.34 × 107 1 2.34 × 107 1.284 0.258 *
Load 3.07 × 109 3 1.02 × 109 56.233 4.4 × 10−29
D7 Interaction 1.02 × 108 3 3.41 × 107 1.874 0.134 *
Error 5.68 × 109 312 1.82 × 107
Total 8.87 × 109 319
* The p-value is greater than 0.05 and is not significant.

Table 4. Analysis of variance for log-energy entropy features.

Source Sum of Squares df Mean Square F p-Value

Motor Condition 2.43 × 108 1 2.43 × 108 304.907 4.17 × 10−48
Load 1.56 × 109 3 5.21 × 108 654.401 3.24 × 10−134
A6 Interaction 6.10 × 107 3 2.03 × 107 25.535 8.47 × 10−15
Log-energy Entropy

Error 2.48 × 108 312 7.96 × 105

Total 2.12 × 109 319
Motor Condition 2.06 × 107 1 2.06 × 107 5.854 1.61 × 10−2
Load 1.87 × 109 3 6.22 × 108 176.656 6.20 × 10−67
A7 Interaction 2.77 × 108 3 9.24 × 107 26.246 3.65 × 10−15
Error 1.10 × 109 312 3.52 × 106
Total 3.26 × 109 319
Motor Condition 1.72 × 108 1 1.72 × 108 228.217 4.44 × 10−39
Load 1.34 × 109 3 4.48 × 108 594.107 1.3 × 10−128
D7 Interaction 5.81 × 107 3 1.94 × 107 25.709 6.88 × 10−15
Error 2.35 × 108 312 7.54 × 105
Total 1.81 × 109 319

To carry out a statistical evaluation between the means for the features in various situations,
Tukey’s honest significant differences technique was used as a post-hoc test. Table 5 shows the
outcomes of the post-hoc test for two different features in three different sub-band frequencies.
The mean values were compared using Tukey’s honest significant differences test at p ≤ 0.05. As is
Appl. Sci. 2017, 7, 845 17 of 20

clear in Table 5, only the log-energy entropy feature is significant in (A6 and D7), and this feature is
not significant in a load of 1Nm for sub-band frequencies belonging to the (A7). The Shannon entropy
feature is also not significant in (A6 and D7) and also in (A7) at the loads (0 and 1 Nm).

Table 5. p-Value calculation from the post-hoc test procedure for features in different sub-band frequencies.

Shannon Entropy Log-Energy Entropy

Load (Nm)
A6 A7 D7 A6 A7 D7
0 0.448 * 1.00 * 0.58 * 3.5 × 10−7 4.2 × 10−7 3.0 × 10−7
0.5 0.153 * 0.40 * 0.84 * 1.2 × 10−7 6.4 × 10−5 9.4 × 10−6
1 0.911 * 2.2 × 10−3 0.939 * 6.0 × 10−8 1.00 * 5.5 × 10−4
1.5 0.494 * 8.3 × 10−4 1.00 * 6.0 × 1−8 1.3 × 10−6 6.0 × 10−8
* The p-value is greater than 0.05 and is not significant.

Table 6 presents the conclusion of the feature validation of the simulation and the experimental
results for two different features, using Coiflet mother wavelets. As is clear in this table, Shannon
entropy is not capable to detect the BRB in all frequency band ranges used. The log-energy entropy is
very significant in detecting BRB in the two frequency band ranges of 39.06–19.53 Hz, which belongs
to the decomposition of the original signal in the detail of level seven, and 39.06–0 Hz, which belongs
to decomposition of the original signal in the approximation of level six. However, log-energy entropy
is not significant for BRB detection in the frequency band range of 19.53–0 Hz, which belongs to the
decomposition of the original signal in the approximation of level seven. The results of the simulation
are also in accordance with the experimental results. Correspondingly, the most effective sub-band
frequency is the detail of level seven which includes the frequency band ranges of 39.06–19.53 Hz,
because this sub-band frequency is located in both (A6 and D7).

Table 6. Conclusion for feature validation in the simulation and experiment.

Simulation Experimental
Coefficientat Frequency
Decomposition Level Ranges (Hz) Shannon Log-Energy Shannon Log-Energy
Entropy Entropy Entropy Entropy
Approximation (6) [39.06–0] Na A Na A
Detail (7) [39.06–19.53] Na A Na A
Approximation (7) [19.53–0] Na Na Na Na
A: Applicable; Na: Not applicable.

5. Conclusions
The technique proposed here is well-suited for BRB detection in synchronous motors that
start with a squirrel cage. The reason is that these electrical machines carry significant currents
during the transient state, and it is difficult to diagnose the failure based on the monitoring of the
conventional steady-state current. According to the non-stationary characteristics of BRB faulty signals,
a fault-diagnosis method based on DWT is put forward in this paper. The main advantage of the
proposed DWT approach is that the feature of a specific fault mode is not at the same frequency band
as the fundamental, and hence is not affected by the main frequency, as shown in Figure 3. In this
research, the importance of starting load on broken bar detection in LS-PMSMs is also considered,
and the results show that the log-energy entropy feature is highly sensitive to the degree of load
complexity. Accordingly, the other advantage of this method is that fault detection and diagnosis can
be performed under different load conditions. Since the study of faults in LS-PMSM has just started,
it is of interest to propose further overall works according to the results achieved. Our future research
will be on the following topics: (1) to extend the detection method for the diagnosis of broken rotor
Appl. Sci. 2017, 7, 845 18 of 20

bars in LS-PMSM that use drive systems, and (2) the application of intelligent techniques to increase
the ability and accuracy of fault detection in the decision-making area.

Acknowledgments: The authors would like to express their gratitude to Ministry of High Education Malaysia for
financial support through grant number FRGS-5524356 and Universiti Putra Malaysia for the facilities provided
during this research work.
Author Contributions: Mohammad Rezazadeh Mehrjou designed and performed the experiments; analyzed
the data and prepare the manuscript, Norman Mariun is leader of the project and immensely in the preparation
of the manuscript, Norhisam Misron and Mohd Amran Mohd Radzi and Suleiman Musa revised and finalized
the manuscript.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Toliyat, H.A.; Kliman, G.B. Handbook of Electric Motors; CRC Press: Boca Raton, FL, USA, 2004.
2. WEG. WQuattro-Super Premium Efficiency Motor; WEG Industries: Jaraguá do Sul, Santa Catarina, Brazil, 2015.
3. Pyrhonen, J.; Jokinen, T.; Hrabovcova, V. Design of Rotating Electrical Machines; Wiley: Hoboken, NJ,
USA, 2009.
4. Mehrjou, M.R.; Mariun, N.; Marhaban, M.H.; Misron, N. Rotor fault condition monitoring techniques for
squirrel-cage induction machine—A review. Mech. Syst. Signal Process. 2011, 25, 2827–2848. [CrossRef]
5. Liu, Y.; Bazzi, A.M. A review and comparison of fault detection and diagnosis methods for squirrel—Cage
induction motors: State of the art. ISA Trans. 2017. [CrossRef] [PubMed]
6. Akçay, H.; Germen, E. Subspace-based identification of acoustic noise spectra in induction motors. IEEE Trans.
Energy Convers. 2015, 30, 32–40. [CrossRef]
7. Gyftakis, K.N.; Spyropoulos, D.V.; Kappatou, J.C.; Mitronikas, E.D. A novel approach for broken bar fault
diagnosis in induction motors through torque monitoring. IEEE Trans. Energy Convers. 2013, 28, 267–277.
[CrossRef]
8. Georgoulas, G.; Climente-Alarcon, V.; Antonino-Daviu, J.A.; Tsoumas, I.P.; Stylios, C.D.; Arkkio, A.;
Nikolakopoulos, G. The Use of a Multilabel Classification Framework for the Detection of Broken Bars
and Mixed Eccentricity Faults Based on the Start-Up Transient. IEEE Trans. Ind. Inform. 2017, 13, 625–634.
[CrossRef]
9. Mirzaeva, G.; Saad, K.I.; Jahromi, M.G. Comprehensive Diagnostics of Induction Motor Faults Based on
Measurement of Space and Time Dependencies of Air Gap Flux. IEEE Trans. Ind. Appl. 2017, 53, 2657–2666.
[CrossRef]
10. Gu, F.; Yesilyurt, I.; Li, Y.; Harris, G.; Ball, A. An Investigation of the Effects of Measurement Noise in the
Use of Instantaneous Angular Speed for Machine Diagnosis. Mech. Syst. Signal Process. 2006, 20, 1444–1460.
[CrossRef]
11. Drif, M.H.; Kim, H.; Kim, J.; Lee, S.B.; Cardoso, A.J.M. Active and Reactive Power Spectra-Based Detection
and Separation of Rotor Faults and Low-Frequency Load Torque Oscillations. IEEE Trans. Ind. Appl. 2017,
53, 2702–2710. [CrossRef]
12. Penrose, H.W. Estimating motor life using motor circuit analysis predictive measurements: Part 2.
In Proceedings of the IEEE International Symposium on Electrical Insulation, Indianapolis, IN, USA, 19–22
September 2004; pp. 15–17.
13. Picazo-Ródenas, M.J.; Antonino-Daviu, J.; Climente-Alarcon, V.; Royo-Pastor, R.; Mota-Villar, A. Combination
of Noninvasive Approaches for General Assessment of Induction Motors. IEEE Trans. Ind. Appl. 2015, 51,
2172–2180. [CrossRef]
14. Martinez, J.; Belahcen, A.; Muetze, A. Analysis of the Vibration Magnitude of an Induction Motor with
Different Numbers of Broken Bars. IEEE Trans. Ind. Appl. 2017, 53, 2711–2720. [CrossRef]
15. Hou, Z.; Huang, J.; Liu, H.; Ye, M.; Liu, Z.; Yang, J. Diagnosis of Broken Rotor Bar Fault in Open-and
Closed-Loop Controlled wye-Connected Induction Motors Using Zero-Sequence Voltage. IET Electr.
Power Appl. 2017, 11, 1214–1223. [CrossRef]
16. Mirafzal, B. Incipient Fault Diagnosis in Squirrel-Cage Induction Motors. Ph.D. Thesis, Marquette University,
Milwaukee, WI, USA, 2005.
Appl. Sci. 2017, 7, 845 19 of 20

17. Bangura, J.F.; Demerdash, N.A. Diagnosis and Characterization of Effects of Broken Bars and Connectors in
Squirrel-Cage Induction Motors by a Time-Stepping Coupled Finite Element-State Space Modeling Approach.
IEEE Trans. Energy Convers. 1999, 14, 1167–1176. [CrossRef]
18. Antonino-Daviu, J.A.; Climente-Alarcón, V.; Pons-Llinares, J.; Puche, R.; Pineda-Sánchez, M. Transient-Based
Analysis for The Detection of Broken Damper Bars in Synchronous Motors. Mech. Syst. Signal Process. 2013,
34, 367–377. [CrossRef]
19. Pineda-Sanchez, M.; Riera-Guasp, M.; Antonino-Daviu, J.A.; Roger-Folch, J.; Perez-Cruz, J.;
Puche-Panadero, R. Instantaneous Frequency of the Left Sideband Harmonic During the Start-Up Transient:
A New Method for Diagnosis of Broken Bars. IEEE Trans. Ind. Electron. 2009, 56, 4557–4570. [CrossRef]
20. Yan, R.; Gao, R.X.; Chen, X. Wavelets for fault diagnosis of rotary machines: A review with applications.
Signal Process. 2014, 96, 1–15. [CrossRef]
21. Glowacz, A. DC Motor Fault Analysis with the Use of Acoustic Signals, Coiflet Wavelet Transform, and
K-Nearest Neighbor Classifier. Arch. Acoust. 2015, 40, 321–327. [CrossRef]
22. Mallat, S.G. A theory for multiresolution signal decomposition: The wavelet representation. IEEE Trans.
Pattern Anal. Mach. Intell. 1989, 11, 674–693. [CrossRef]
23. Dash, R.N.; Subudhi, B.; Das, S. Induction motor stator inter- turn fault detection using wavelet transform
technique. In Proceedings of the International Conference on Industrial and Information Systems, Mangalore,
India, 29 July–1 August 2010; pp. 436–441.
24. Pilloni, A.; Pisano, A.; Riera-Guasp, M.; Puche-Panadero, R.; Pineda-Sanchez, M. Fault detection in induction
motors. In AC Electric Motors Control: Advanced Design Techniques and Applications, 1st ed.; Giri, F., Ed.;
John Wiley & Sons Ltd.: Oxford, UK, 2013; pp. 275–309. [CrossRef]
25. Polikar, R. Multi Resolution Analysis: The Discrete Wavelet Transform. In The Wavelet Tutorial Part IV; Rowan
University, College of Engineering Web Servers 10: Glassboro, NJ, USA, 1998.
26. Bae, H.; Kim, Y.T.; Lee, S.H.; Kim, S.; Lee, M.H. Fault Diagnostic of Induction Motors for Equipment
Reliability and Health Maintenance Based Upon Fourier and Wavelet Analysis. Artif. Life Robot. 2005, 9,
112–116. [CrossRef]
27. Sadeghian, A.; Zhongming, Y.; Wu, B. Online Detection of Broken Rotor Bars in Induction Motors by Wavelet
Packet Decomposition and Artificial Neural Networks. IEEE Trans. Instrum. Meas. 2009, 58, 2253–2263.
[CrossRef]
28. Romero-Troncoso, R.J.; Saucedo-Gallaga, R.; Cabal-Yepez, E.; Garcia-Perez, A.; Osornio-Rios, R.A.;
Alvarez-Salas, R.; Huber, N. FPGA-based online detection of multiple combined faults in induction motors
through information entropy and fuzzy inference. IEEE Trans. Ind. Electron. 2011, 58, 5263–5270. [CrossRef]
29. Hernandez-Vargas, M.; Cabal-Yepez, E.; Garcia-Perez, A.; Romero-Troncoso, R.J. Novel methodology for
broken-rotor-bar and bearing faults detection through SVD and information entropy. J. Sci. Ind. Res. 2012,
71, 589–593.
30. Cabal-Yepez, E.; Valtierra-Rodriguez, M.; Romero-Troncoso, R.J.; Garcia-Perez, A.; Osornio-Rios, R.A.;
Miranda-Vidales, H.; Alvarez-Salas, R. FPGA-based entropy neural processor for online detection of multiple
combined faults on induction motors. Mech. Syst. Signal Process. 2011, 30, 123–130. [CrossRef]
31. Pan, S.; Han, T.; Tan, A.C.; Lin, T.R. Fault diagnosis system of induction motors based on multiscale entropy
and support vector machine with mutual information algorithm. Shock Vib. 2016. [CrossRef]
32. Huo, Z.; Zhang, Y.; Shu, L. A Short Survey on Fault Diagnosis of Rotating Machinery Using Entropy
Techniques. In Proceedings of the 3rd EAI International Conference on Industrial Networks and Intelligent
Systems, Ho Chi Minh City, Vietnam, 4 September 2017.
33. Antonino-Daviu, J.A.; Riera-Guasp, M.; Roger-Folch, J.; Martínez-Giménez, F.; Peris, A. Application and
Optimization of the Discrete Wavelet Transform for the Detection of Broken Rotor Bars in Induction Machines.
Appl. Comput. Harmon. Anal. 2006, 22, 268–279. [CrossRef]
34. Riera-Guasp, M.; Antonino-Daviu, J.A.; Rusek, J.; Roger-Folch, J. Diagnosis of Rotor Asymmetries in
Induction Motors Based on the Transient Extraction of Fault Components Using Filtering Techniques.
Electr. Power Syst. Res. 2009, 79, 1181–1191. [CrossRef]
Appl. Sci. 2017, 7, 845 20 of 20

35. Tran, V.T.; AlThobiani, F.; Ball, A.; Choi, B.K. An Application to Transient Current Signal Based Induction
Motor Fault diagnosis of Fourier–Bessel expansion and Simplified Fuzzy ARTMAP. Expert Syst. Appl. 2013,
40, 5372–5384. [CrossRef]
36. Cabal-Yepez, E.; Romero-Troncoso, R.J.; Garcia-Perez, A.; Osornio-Rios, R.A. Single-Parameter Fault
Identification through Information Entropy Analysis at the Startup-Transient Current in Induction Motors.
Electr. Power Syst. Res. 2012, 89, 64–69. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).