N. BESSOUS1*, S. SBAA2, and A.

C
1
Department of electrical engineering, El-Oued Univ

2
BULLETIN OF THE POLISH ACADEMY OF SCIENCES Department of electrical engineering, Mohamed Khider
TECHNICAL SCIENCES, Vol. 67, No. 3, 2019
DOI: 10.24425/bpasts.2019.129655 Abstract. This paper presents the mechanical fault detection in squirrel cage induction
analyzed the rolling element bearing (REB) faults in SCIM. Rolling element bearing fau
catastrophic damage to rotating machinery. So, early detection of the REB faults in SC
mechanical vibrations for rotating electrical machines is the vibration analysis. Today, the
field who have the aim of replacing the vibration analysis technique. Among them, sta

Mechanical fault detection in rotating electrical machines subjects in the fault detection field. Motor current signature analysis (MCSA) has becom
generally based on fast Fourier transform (FFT) of the stator current signal. We detailed t
using MCSA-FFT and MCSA-DWT techniques spectrum. Another goal in this work is the use of the discrete wavelet transform (DWT) te
indicator based on MCSA-DWT technique has been developed in this study. This new ind
the quality form. The acquisition data is performed and a comparative study is carried
decision. The proposed subject is examined experimentally using 3 kW squirrel cage induc
N. BESSOUS 1*, S. SBAA2, and A.C. MEGHERBI 2
Key words: motor current signature analysis (MCSA); discrete wavelet transform (DW
1
Department of Electrical Engineering, El-Oued
currentUniversity,
spectrum. 39000, El-Oued
2
Department of Electrical Engineering, Mohamed Khider University, 07000 Biskra

1. Introduction T
Abstract. This paper presents mechanical fault detection in squirrel cage induction motors (SCIMs) by means of two recent techniques. More machine t
precisely, we have analyzed the rolling element bearing (REB) faults in SCIM. RollingLike mostbearing
element AC motors, the induction
faults constitute a majormotors
problem element
have
among
different faults which cause catastrophic damage to rotating machinery. Thusmany advantages
early detection of being
of REB faultsused by more
in SCIMs is ofthan importanc
90%importance.
crucial when
comparing
Vibration analysis is among the key concepts for mechanical vibrations of rotating it withmachines.
electrical other motor types.
Today, Their
there advantage
is massive lies
competition W
in: simple
between researchers in the diagnosis field. They all have as their aim to replace construction,
the vibration robust andAmong
analysis technique. mechanically avoid dam
strong,
them, stator current
analysis has become one of the most important subjects in the fault detection highfield.
starting
Motor torque,
currentgood speedanalysis
signature regulation,
(MCSA)operation the quali
in
has become
high
popular for detection and localization of numerous faults. It is generally based speed
on fast without
Fourier brushes,
transform (FFT)absence
of the of sparks
stator (operated
current monitorin
signal. We
have detailed the analysis by means of MCSA-FFT, which is based on the stator in hazardous conditions),
current spectrum. Another directly
goal in connected
this work is with
the use but also h
grid
of the
discrete wavelet transform (DWT) technique in order to detect REB faults. In power,
addition,squirrel
a new cage rotorbased
indicator of induction motor contains
on the MCSA-DWT technique M
windings
has been developed in this study. This new indicator has the advantage of expressing which
itself in the are permanently
quantity and quality short
form.circuited, different f
highlydata
The acquisition
efficient inwith
are presented and a comparative study is carried out between these recent techniques orderfull
to load,
ensureetc. Butdecision.
a final like anyThe other machine,
proposed broken ro
subject
is examined experimentally using a 3 kW squirrel cage induction motor test bed.SCIMs can affect multiple faults under different causes: faults [5-9
power imbalance, overvoltage, over-current, overload, A
Key words: motor current signature analysis (MCSA), discrete wavelet transform (DWT),faults,
construction rolling element bearing faults,aging
bad environment, rotor eccentricity,
of the acceleratin
stator current spectrum. machine, etc. According to the constituent elements of the technique
SCIM, we quote the percentage defects as shown in the [10-12]. I
following figure [1]: developm
1. Introduction numerical
150 Other faults 10% O
Like most AC motors, induction motors have many advantages of currently
being used in more than 90% when comparing it with other motor 100 Rotor faults 12% analysis w
types. Their advantage lies in: simple construction, robust and analyze t
50 Stator faults 37%
mechanically strong, high starting torque, good speed regulation, called: m
operation at high speed without brushes, absence of sparks (useful [15,16]. I
0 REB faults 41% processing
under hazardous conditions), direct connection to the grid power,
squirrel cage rotor of induction motor containing windings which in an early
Fig. 1. Percentage distribution of faults in SCIMs I
are permanently short circuited, high efficiency with full load, Fig. 1. Percentage distribution of faults in SCIMs
stator coil
etc. But like any other machine, SCIMs can cause multiple faults *e-mail: nbessous@yahoo.fr
under different causes: power imbalance, overvoltage, over-cur-
rent, overload, construction faults, bad environment, aging of the Many techniques are now used in order to detect different
machine, etc. As per the constituent elements of the SCIM, we faults such as: the eccentricity of the rotor, broken rotor bars,
quote the percentage defects, as shown in Fig. 1. short circuit in the stator and REB faults [5–9].
This classification depends on the power of the machine to Massive competition between researchers is accelerating 1to
be studied. The distribution of faults on the element bearings, lead to a simple, efficient and precise technique in the field of
the stator and the rotor show the overall importance of REB diagnosis of rotating machines [10–12]. In addition, several
faults [2­4]. works have focused their efforts on the development of a phys-
We say that monitoring is an essential step to avoid dam- ically and numerically suitable analytical model [13, 14].
age, to increase the life of the system, to ensure quality of the One of the simple techniques that are currently popular in
product, etc. A rotating machine monitoring is concerned not the industry is the vibration analysis, which is based on the
only with fault detection, but also with how these failures can vibration image observed in order to analyze its content. This
be detected. analysis technique is often called motor vibration signature
analysis (MVSA) [15, 16]. In fact, the goal of several vibra-
tion signal processing methods is to detect faults in rotating
*e-mail: nbessous@yahoo.fr machines at the earliest possible stage.
Manuscript submitted 2018-07-29, revised 2018-09-07, initially accepted On the other hand, another popular technique is based on the
for publication 2018-09-10, published in June 2019. stator coils. They produce a magnetic field that rotates around

Bull. Pol. Ac.: Tech. 67(3) 2019 571

 N. Bessous, S. Sbaa, and A.C. Megherbi

the outside of the motor. In addition, the rotor bars produce

another magnetic field that influences the stator winding. Air-
gap variation leads to a variation in magnetic field distribution
(such as e.g. rotor eccentricity).
As a result of the action-reaction phenomenon of the stator
and rotor parts, the EMF and current in the stator operate in
such a manner that a magnetic field created in the stator will
come to act in opposition to the stator magnetic field. Rotor
frequency will be linked in the stator with the slip. It can thus
be said that the speed of the rotor depends on the difference
between rotor and stator frequency, i.e. ( fs ¡ f r). Some har-
monics will be induced in both the stator and the rotor as the
rotor acts as the frequency converter in the magnitude ( fs ¡ fr
or fs + fr). Fig. 2. Experimental setup dedicated to REB faults
The stator current spectrum is often called motor current
signature analysis (MCSA). It is considered one of the many
techniques used in diagnosis of rotating machinery. The stator In this study, we used SCIM with the following charac-
current spectrum by means of the FFT tool is the goal of this teristics: 3 kW, p = 2 (number of pole pairs), fs = 50 Hz and
study. Our work is based on the REB faults detection by means 28 bars. Figure 2 shows the SCIM with some equipment applied
of the harmonic components verification in the stator current in order to take the measurements.
spectrum. We practically realized two REB fault types. The first is the
The second technique applied in this sphere is the dis- outer raceway fault (ORF) and the second is the inner raceway
crete wavelet transform (DWT). This technique is currently fault (IRF). Figure 3 shows photos of healthy and faulty element
used in many areas: filtering, electrical machine diagnosis, bearings.
etc. According to the DWT technique, we have defined a new
estimator, known as the mean square error (MSE), which
checks the existence of REB faults. This indicator has two
expressions: a quantitative and qualitative one. In addition,
another factor was used in this study which is based on the
DWT technique; it is the energy of every detail of this trans-
formation.
The purpose of this paper is the extraction of fault indicators
in the stator current signal. First, we did an experimental part
which ensures acquisition of the stator current signal. Exploita-
tion by two advanced techniques of signal processing was done Fig. 3. ORF and IRF
by means of analysis of the stator current signal attentively. In
addition, a comparative study applied led us to decide on the
best choice in an effective manner.
In this regard, the basic principles for REB faults detection 3. MCSA-FFT analysis
using the MCSA and DWT analysis of the stator current signal
are explained in the sections below. 3.1. Outer raceway fault detection by MCSA-FFT. Many
types of mechanical faults cause oscillations in the load torque
applied to the SCIM [17]. We will therefore study the peri-
2. Motor test bed odic variations effects of the load on the stator current signal
under an OR fault. Figure (3) presents a photograph of a bearing
The traditional method for measuring different quantities of defective at the outer raceway. This REB will be installed in the
the electric motor activity is based on an acquisition card and side load of the SCIM. It belongs to the 6206 series.
clearly defined measuring devices. But the process of extracting The stator current spectrum of the SCIM in absence and in
these magnitudes can be difficult to apply in more complicated the presence of a fault is very rich in harmonics. Indeed, the
work positions. frequencies presented in this spectrum are mainly the spectral
The method of measuring the quantities of electric motor ones related to the characteristic frequencies of REB faults,
activity requires installation of an adequate test bench. The test which are given by [16]:
bench consists of the motor tested, mounted onto a generator or
±
dynamometer. The motor tested is then connected to the load f charact–OR, IR, cage, ball = jν fs ± kfOR, IR, cage, ball j(1)
by a shaft.
This system provides data including current signal, voltage with, fOR, IR, cage, ball being the characteristic frequencies of vibra-
signal and other quantities. tion signal spectrum (MVSA-FFT).

572 Bull. Pol. Ac.: Tech. 67(3) 2019

 where, We s≈0),
will (a) analyze
: 0-150Hz, the(b) stator current
: 140-300 Hz, (c)spectrum
: 450-600ORFcontent of the
Hz, (d) : Around (d) Fig
-10 L-PSH
steady-state in both healthy and faulty states when the s≈0
steady-state 3f in+5f both healthyX:PSHs and faulty states
Healthy when the
fault
k is inany
the outer raceway
integer, ν is (ORF).
the order of the stator time fault-30 648.2 X: 748.1

Amplitude (dB)
in0 the outer s OR
raceway (ORF).
Y: -41.44 Y: -43.61
Figures (4) and (5) respectively represent
harmonics that are present in the power supply driving the the spectral
Figures
-50 9f -5f (4) and (5) respectively represent
f +5f
s OR the
ORFspectral (a)
0
content of the stator current
Mechanical fordetection
fault the twoin operations: at no-
rotating electrical machines using MCSA-FFT and MCSA-DWT techniques 7f -5f
motor (ν=1,3,5, etc.), fr is the mechanical rotor frequency content -20 ofs the ORstator f -fcurrent for the two s OR operations:
Healthy
U-PSH at no-
load (s = 0.004≈0) and at load (s = 0.04). s OR -20
9. harmonics due to the load-70 (s = 0.004≈0) and at load (s = 0.04).

Amplitude (dB)
andWe Nb is the number
clearly notice oftheballs equalof
presence
X: 599.6

Amplitude (dB)
Y: -85.45
-40 X: 679.8
f +f due to the
It has been statistically shown in [8] that the characteristic We
So,
-90 inclearly
this notice
case we the presence
write: Y: -95.89 of harmonics s OR -40
ORF.
So, in this case we  0.6 Nb  k  f r
f IRwrite: (3)
X: 17.7
X: 82.3

frequencies of the vibration signal can be approximated for ORF. -60

0 580
Y: -62.73
X: 36 Y: -67.68 X: 136
600 620 Y: -73.16
640 660 680 700 720 740 Y: -72.29
760 -60
where, f
most0 element bearings ball numbers
with f s  kfbetween
ORf -f
(4)
6 and 12 by: -10
f -f f ± L-PSH = jν fs ± kf(Hz) jORF
Frequency .(4) (d)
charact–OR–v, k OR
charact OR  ,k 0-80
-10 s r s r ORF (a)
Fig. 4. f -f
Stator current spectrum -f of different Healthy
f 648.2 frequency bands -80
3f +5f X: 748.1 (ORF,
kWe is will
any analyze
integer, theν stator
f -ffOR =is 0.4Nthe current k£spectrum
order frof
(2) the content
stator f time of the -10
-30 s rs X: s r ORF (a)

(dB) (dB)
b£ Healthy OR
+f We-100will (a)
s≈0), analyze
: 0-150Hz, the stator
(b) current
Y: -41.44
: 140-300 spectrum
Hz, (c) : 450-600 content
Hz, Y:(d)
Healthy of the
: Around
-43.61
steady-state
-30
harmonicsY:that X: 25.1 in both healthy
s OR
-45.12 are present in
X: 75
and faulty
the power supply states
? fdriving
when s OR
thethe f -f
Amplitude (dB)

0 s OR 50 75 PSHs f +5f 100 f +f 150 -100

-2f steady-stateX: 25.1in both healthyX:and faulty s states
OR when the sfault OR is

Amplitude
Y: -43.64 -30 Frequency (Hz)
fault in the outer fIR = 0.6N
raceway (ORF). b £ k£ fr(3)
s OR -50 Y: -43.64 ? f -2f
motor -50(ν=1,3,5, etc.), fr is the mechanical rotor frequency
Y: -45.12
found00in the outer raceway (ORF). s U-PSH
OR
Figures (4)X:Y: 40and (5) respectively represent theX:Y: 140 spectral 0

Amplitude
-50
-70 ORF (b)
and Nb iskthe
where is anynumber -70.6 of balls
integer, ν is the equal order9. of the stator time
-70.98 har- Figures 4 and X: 40 5, respectively, represent the spectral
X: 599.6
ORF
X: 140
Healthy con-(a)
content
-70
monics
of the stator current
that are present in the power supply driving
for the two operations:
X: 130
at no-
the motor
Y: -80.91
tent of thes stator
OR
Y: -70.6 13fs-5fOR X: 679.8
-20 9f -5fY: -85.45
current for Y:the 7f -5f
-95.89two soperations:
Y: -70.98
at no-load
X: 130Healthy -20
-70-20
-90 f -f OR

(dB)
load
So,(νin (s
= this = 0.004≈0)
1, 3,case we write: and at load (s = 0.04). s OR Y: -80.91

Amplitude (dB)
5, etc.), fr is the mechanical rotor frequency and Nb (s =-40 0.004 ¼ 0) and at load (s = 0.04).

(dB)
-90
Wenumber
clearlyofnotice the presence of harmonics due to the 580 600 620 640 660 680 700 720 3f -5f740 760 -40

Amplitude
-40
to9.f k 
balls equal X: 217.6

(3) is the f f  0.6 N  kf

f (3)
(4)
-90
We clearly notice the presence Y: -62.16 of (Hz)
Frequency harmonics due f +fto
s ORthe ORF.
s OR

Amplitude
ORF. 0
0 IR
charact OR  ,k b
Frequency (Hz)
s r 100
OR
0-60
X: 17.7
0 Y: -62.73
X: 82.3
100
-60
Fig. 4. Stator current L-PSH spectrum of differentY:frequency bands (ORF, X: 136 -60
where,
X: 36 -67.68
We -10will
0 analyze the stator
L-PSH current spectrum ORF content of the
(d) -10 Y: -73.16 Frequency (Hz) ORF (d)Y: -72.29
X: 282.2
a)-100 3f f -fin both healthy s≈0),
-80
b) 0-80 (a) : 0-150Hz, (b) : 140-300 Hz, (c) : 450-600Healthy Hz, (d) : Around
steady-state ORF f -fand faulty states
5f Healthy when the
(b) Y: -90.74
-80
time k is-10 3fr +5f
any s sinteger, is (ORF).
OR ν Healthy the r order ofs ORF
X: s648.2
the statorX: 748.1time
(a) -10
-30 3f 3fs+5fOR ORF X:PSHs 648.2
5f X: 748.1 (b)

(dB) (dB)
fault-30
(dB)(dB)

s
in the outer raceway Y: -41.44 11f -7f
Healthy s OR Y:=280Hz
-43.61 -100 s Healthy
Y: -41.44 s
11f -7f
Y: -43.61
=280Hz
harmonics that-8f are f -fpresent in the power f +5fsupplythe driving 140 160 180 200 220 f +5f240 s 260OR 280 300
s the
g the -30 7f 5f -6f
f s+f -100
Figures (4) and (5) respectively represent spectral -100
0

Amplitude
s OR s OR 7f -6f
Amplitude

-30
-50 X: 75 s OR 9f -8f OROR -30
-50 0 7fs-8fOR Frequency (Hz) s OR 5f -6f 150 1
(dB)

X: 25.1
sY: -43.64
OR 50
7f -6f 100
?s f -2f
OR Frequency (Hz) ORF s OR
ency motor -50 (ν=1,3,5,
content of the 9f
Y: -45.12
setc.),
OR fr is the
=180Hz formechanical
-7f current
stator thefstwo OR rotor s frequency
+2f operations: ORat no- 09f -5f s OR 9f -8f U-PSH (a)
Amplitude

U-PSH 9f -7ff -f=180Hz 7f -5f

f +2f s OR
-20 0 s OR Healthy 0

Amplitude
ORF
Amplitude

-50
-70 -50
-70 s OR ss OR OR (c)
loadN(sb is
and = 0.004≈0)
the number X: 40 and
X: 599.6
Y: -85.45 ofatballs
loadequal (s = 0.04).
9. X: 140
X: 599.6
7f
s
-10f
Y: -85.45
s
OR
OR 19f s
-5f OR ORF
Healthy (b)

(dB)
-70 Y: -70.6 X: 679.8 -20 X: 679.8
3f -8f
We-70 clearly
-90 X: 170.2notice the presence
X: 189.8
Y: -95.89 of X:
harmonics
229.9 due
X: 130
X: 270.1
Y:
to
-70.98
theX: 289.7
-40
-70
-90 -20 13f -5f
s OR
Y: -95.89 s OR 9f +6f
f +f
Healthy -20

(dB)
Y: -86.36 X: 189.8
So,
ORF. in this caseY: -90.83we write: Y: -89.23 Y: -80.91 Y: -87.69
Amplitude
X: 517.4 X: 229.9
X:s270.1
s r OR X: 289.7

Amplitude(dB)
Y: -91.82 X: 17.7X: 170.2

Amplitude (dB)
Y: -86.36
-90580 -40 Y: -90.83
Y: -62.73
Y: -47.07 X:Y:82.3
-89.23 Y: -87.69
600 620 640 660 680 700 720 740 760 580
-40
-60
Y: -91.82
600 620 X: 36640
X: 593.5
X: 136 660 680Y: -67.68
700 720 740 760 -40
(4) 160 f charact 180OR
-90
 220 
f s (Hz)kf 280 (4)300
-90
3f +5f3fs-5fOR Y: -72.29
Y: -57.87
Amplitude
Frequency 7f +6f Y: -73.16 Frequency
X: 493.6 X: 217.6 X:(Hz)
541.2
(3) 0 0
140  ,200
k 240OR 260 -60 s r Y: -66.83 Y: -62.16 Y: -66.93 s OR

f the We(3) 0 4. f -f
Fig.
will Stator
analyze
f
current
theIR
0.6
spectrum
IRstator current
f N
Frequency
-f of  k
different
(Hz)
spectrum
f 100
frequency bands (ORF, (3) 140
-80
Fig.-604.
160
Stator
180
current
200
X:
spectrum
512.8
of
220
different
240
frequency
260
bands
280
(ORF,
300 -60
ORFcontent of(a)the
-10 s r Frequency
s r b
b (Hz)rr ORF Y: -76.82
Frequency (Hz)
c) -10
s≈0), 0 (a) 0: 0-150Hz, (b) L-PSH
: 140-300 Hz, (c) : 450-600 Hz, (d) : Around (d) d) 0
s≈0),-80(a) : 0-150Hz, (b) : 140-300 Hz, (c) : 450-600 Hz, (d)
X: 582
: Around
where,
n the steady-state 0
-10 in fboths OR healthy
-f and ORFfaulty Healthy
states
Healthy ORF when f +f the (c) -100-80
(d) -100 L-PSH PSHs ORF Y: -89.04
(d)
X: 282.2

-30 -10 X:3f

-10
25.1 3f +5f ORFX: L-PSH 75 PSHs 5f11f s OR (b) -10 ORF (c)
Y: -90.74 -80
Amplitude (dB)

time fault-30 sf -6f

X: 648.2
Healthy 0 50 100
Healthy 150
in the s
outer s raceway (ORF). s s? f Healthy -2f Y: -43.61
X: 748.1
(dB)

Y: -45.12 OR Y: -43.64 Frequency (Hz) 11f

OR Healthy Y: -41.44
11f s-7f OR =280Hz -100 3f +5f
f -6f Healthy
gator k is
the time -50
-30 any integer, 3f +5f ν is the order
X: 648.2
of the stator time -30
-100
X:
X: 648.2
648.2 s X: 748.1
(dB)

-30 450 s OR 500 550 600

(dB)

ctral X: 748.1 X: 748.1

Amplitude (dB)

(dB)

s OR s OR 00 140 s s OR
OR 180 -100
-300
Figures 7f(4)-8f and (5) respectively represent the spectral
Y: -41.44f +5f Y: -41.44
Y:Frequency
-41.44 220 (Hz)
-30 160 200 240 260 Y: 280
Y: -43.61 300
(dB)

5f -6f 4
Amplitude

(dB) (dB)

Y: -43.61 -43.61
-50 s X:OR 7f -6f s OR ORF f -7f ORF (b)
riving
no- the
tency harmonics
content that are
-70.6 present
40
in
s the OR power supply f 9f
+5f -8fdriving
X: 140 s
the OR
(a) 09f -5f Frequency (Hz)
f +5f ORF
f -7f (a)
-20 of s the stator
-7f -fcurrent
=180Hz for the7f f two
9f -5f -5f operations: s atOR no-
Amplitude
Amplitude

Amplitude

Y:
-50 s 130 OR
Y: -70.98
U-PSH -50 11f7f+6f
ss
-5f OR
s rOR23fs-5fOR
-50 OR9f +2f OR s X:OR Healthy -20 0 s 21fOR -5f 13f -5f OR Healthy 0
s fOR
Amplitude

ss OR s OR
Amplitude

-70 -50 -20

-50 s OR fs-fOR 5f -8f Healthy ORF (d)
frequency motor(s(ν=1,3,5,
load -70 = 0.004≈0) X: etc.),
599.6
andfrratisload
s OR
the (s mechanical
= 0.04). rotor frequency Y: -80.91
U-PSH
s OR
s OR s ORF
U-PSH (c)
(dB)

-20 7f -10f 19f -5f 3f -10f Healthy

(dB)

Y: -85.45
o the and-90 WeN-70
-70
-70
clearly
-40 is the notice
number the
of
X: 599.6
presence
balls equal
X: 679.8
of
9. harmonics f +f due to the -70
-40
-40-20 9f -10f X: s
X: 599.6
599.6 OR X: 617.3 s OR
15f -f
3f -8f
s OR Healthy -20
(dB)

Y: -37.48
-90bb X: 17.7X: 170.2X: 489.9Y: -85.45 X: 189.8 Y: -95.89 X: 579.9 -70 s Y: Y:OR-85.45 s OR OR f 3f
+f -5f 7f -8f
Amplitude

X: 289.7 -85.45 X: 717.3

s OR X: 217.6
Amplitude

679.8 s
X: 229.9 X:
X: 679.8 9f +6f
s sOR ORX: 579.9 s OR

Amplitude (dB)
Amplitude

X: 679.8
ORF.-90
X: 270.1 Y: -90.61
(dB)

Y: -91.57 Y: -86.36
X:Y:82.3
-89.23 Y: -87.69 -40X: 17.7 X: 489.9
L-PSH Y: -62.16
X: 517.4Y:
Y: -95.89
Y: -47.1
s r
0 Y: -62.73
-90 Y: -90.83 Y: -95.89 100 Y: -91.82 -90
-60 Y: -62.73 Y: -91.57 X: 641.2-95.89 X: 82.3 Y: -90.61
-40
Amplitude

X: 136
-60 -40
So, in this case we write:
X: 36
-90580 600 620 640 660
Y: -73.16 Frequency (Hz)
680 Y: 700
-67.68 720 740 760
Y: -72.29 -60
-90 X: 612.7
X: 36 Y: -47.07
Y: -59.18 Y: -67.68 U-PSH X:
X: 136
593.5
Amplitude

(4) Y: -73.16 Y:X:

-72.29
741.1
450 580
00 140 600
 180 620 500 Frequency 640 220660 240
(Hz) 550
680 260 700 720 600 760 580
740 -60 7f600 Y: -67.74620
+6f X: 493.6 640 660 680 X: 693.4
X: 541.2 700 3f 720+5f
Y: -57.87
740 Y: 760
-f 
-69.93
f  f  kf Y: -71.91 X: 712.7 X: 282.2
160 200 Frequency (Hz) 280 300 -80 X: 593.5 s r s OR -60
-80 f -f
(4) -80450
-60 500 (Hz) 550 600
Y: -66.83 Y: -66.93
(a) (4) -10 Fig. 4.
3f Stator current ORFspectrum fFrequency
of Frequency
different
s (Hz) 5f (Hz)
frequency
OR bands (b)
(ORF, Y: -57.87 Frequency
X: 665
X: 512.8
Frequency (Hz)
Y: -77.53 Y: -90.74
f the
X:
-10 s s r charact
charact  OR
OR  
 ,,kk
s r s ORF (a) X: 622 Y:
Y: -76.82 -83.07
s≈0), 0 (a) : 0-150Hz, Healthy
(b) : 140-300 Hz, (c) -80 Y:
of :different
450-600 Hz, (d) : Around Y: -66.49
11f frequency
-7f =280Hz Fig. 4. Stator current spectrum
Healthy Hz,ofc)220 different frequency bands (ORF,
X: 582
X: 721.9
nenttheof the Fig. 4. Fig.
Stator4. fStator
current current
spectrum spectrum
of different frequencys OR bands bands
(ORF, (ORF,
s -100
0): a) 0‒150 Hz, b) 140‒300 450‒600 -80
OR We-30 will
-100 analyze
7f25.1-8f
-30-10 0 X:s≈0),
the
-f stator current
X: 75 PSHs ORF spectrum content f of
+f the
(c) ¼ -100 -80
140 160 180 200 240Hz, d) 260Around 280PSHs300
Y: -89.04
Y: -76.03
(dB)

s OR 50 100 5f s -6f
OR 150 s≈0), (a) : 0-150Hz, (b) : 140-300 Hz, (c) : 450-600 Hz, (d) : Around
(a) : 0-150Hz,7f(b) : 140-300 Hz, (c) : 450-600 Hz, (d) : Around
Amplitude(dB)

s OR -6fFrequency (Hz) s OR 0 50 100 150

when the steady-state Y: -45.12inf -6f
9f -7f
both=180Hz healthy ORand
sY: -43.64 Healthyfaulty 9fstates
fPSHs
+2f
11f
s?s-8ffsOR-2fwhen the
OR
-100
580 600 620
Frequency
640Frequency 660 (Hz)
PSHs
(Hz)
680 700 720 740 760 -100
0 s OR -100
0
ctral fault -50 in0 the outer s raceway (ORF). 0 450 500 Frequency (Hz) 550 ORF 5
Amplitude

OR s OR
-50-30
a) b) 600
(dB) (dB)

ORF (b) Frequency (Hz) (c)

X: 40 ORF
X: 140 (a) 7f -10f 19fs-5fOR ORF (b)
te no-
spectral-70Figures
-20 9f
s 0(4)
-5f
OR andf -f
Y: -70.6 (5)sX: 189.8
13f -5f respectively 7f represent
-5f f
Healthy
Y:
-7f
the spectral
-70.98
s OR 0
Fig.
-20 0 5. Stator
s current
OR spectrum of different frequency bands
Healthy (ORF,
Amplitude

-20
-70-50 OR s OR X: 130Healthy -20 13f -5f 3f -8f 23f+6f -5f ORF Healthy Fig
s OR ORF X: 289.7
(a) s=0.04), 21f
(a)s:-5f 0-150Hz, s (b)OR 5f:517.4
140-300 11f
sHz,+6fOR(c)r : 450-600 Hz, (d) ORF : Around
(a)(d)
content ofX:Y: 170.2 the9fstator current
Y: -86.36 for the two operations: at no-
X: 229.9
ns: at no- X: 270.1 Y: -80.91
9f -5f OR -8f 9f s OR
(dB)

-5f Y: -87.69 7fs -5f

(dB) (dB)

Y: -89.23
7f Y:-5f X: s r s=0.0
(dB)

-90.83 ss OR s OR Healthy Healthy

-40 -20 s OR f -f -91.82
Healthy -20-20 OR f -fX: 617.3Y: -47.07 PSHs ss 3f OR -10f
o the load -90 (s = 0.004≈0)
-40
-90-70 and ats load X: 217.6(s = 0.04).
OR
s OR
f 3f+f -5f
-40
-40 9f -10f s OR
s OR
OR
s OR X: 593.5
Amplitude

ORX: 493.6 Y: -37.48 X: 217.615fsX: -f 541.2

Amplitude

s sOR ORX: 579.9 3f -5f Y: -57.87

Amplitude (dB)
Amplitude

s OR 7fs-8fOR
Amplitude (dB)

(dB)
Amplitude

X: 17.7 X: 489.9 Y: -62.16 7f +6fs OR 3fX: +5f

717.3
due to the -90 We
140
-60
0
-60
clearly
Y: 160
-40
-62.73 notice
180
Y: -91.57 the
X: 36
200 presence 220 of X:harmonics
240 82.3
100
260 due
280 to
f +f X:
the
300
Y: -90.61
136 -40
-60-40
-60
s r Y: -66.83L-PSH Y: -62.16 Y: -66.93 s
Y: -47.1
f +fOR
Frequency (Hz)Y: -67.68 X: 512.8 X: 641.2
ss OR
Amplitude

ORF.0 s ORY: -72.29 OR

Amplitude

X: 17.7 Y: -73.16 Frequency (Hz) X:

X: 17.7
17.7
X: 612.7
Y: -76.82Y: -59.18
X:
U-PSH
Y: -62.73 X: 82.3 X: 282.2 Y:
Y: -62.73
-62.73 X: 82.3
82.3 X: 582 X: 741.1
-80 0 450 -60 500X: 36 550 600
X: 136 -80-60
-60 Y: -67.74 X: 36 X: 36
X: 693.4 Y: -67.68 X:
X: 136
136 X: 282.2
(b) -10-80 -73.16ORF (Hz)
Y: -67.68 Y: -90.74
(c) Y: -72.29 -80 X: 593.5 Y:Y:-71.91
-67.68X: 712.7 Y: -89.04 Y: -69.93
ORF Y:Frequency Y:Y:
Y:
Y: -73.16 -72.29
(a) -10 3f 5f11f (b) Y: -57.87
-73.16 X: 665 Y: -77.53
Y: -72.29
-90.74
Hz 0 s
-100 f -f fs-6fOR Healthyf -f Healthy s s X: 622 Y: -83.07
11f -7f =280Hz -80
-100 -80
140 -80
Y: -66.49
6f (a) -10 -100
-30 ss rr160 180 200ss rr 220 240ORF s 260OR 280(a) 300 -100 450
X: 721.9
180 500200Frequency 220 (Hz)240550 260 600
(dB)

OR 140 160 280 300

OR -30 0 7f -8f 50 Frequency (Hz) 100 5f -6f 150 Y: -76.03
(dB)(dB)

f +f (3) 0 X:
s OR f -f f IR
7f 0.6
s
-6f N
Frequency
OR b  k 
(Hz) f r
Healthy
9f -8f f -7f
f s OR
+f (3) -100
-1000
Frequency (Hz)
s OR c) -100 0 09f -7fss OR s OR s ss OROR d)
Amplitude

-50 0 X: OR
-30 =180Hz X: X: 75 f +2f OR 150 000580 21f -5f600 620 50640 660 680 23f 100700
-5f 720 740 150 760
25.1 75
(dB)
Amplitude

-50
25.1 50 ? f100 ORF
-2f 11f +6f
OR where, Y:
Y:-10-45.12
7f
-45.12 s OR
-10f 19f s
Y:
Y: -43.64
-5f-43.64 Frequency
L-PSH
OR
s OR (Hz)
sORF ORF (c)(b) (d)
OR
s Healthy
OR 3 -10 s OR 5f -8fFrequency
L-PSH s OR
Frequency s (Hz) r(Hz) ORF s OR
ORF ORF (c)
(d)
(d)
-20 s OR -200 7f -10f X: 617.319fs-5fOR 3f -10f Healthy Healthy 3
Amplitude

-70-20 0 13f -5f 3f -8f Healthy

Healthy
Amplitude

-50 Fig.
9f 5. s
-10fStator OR Healthy
+5f current
Y: -37.48spectrum of different frequency bands (ORF,
X: 289.7 s OR -20 s OR
-70 40 3f +5fX: 189.8 X:s 648.2OR 9f +6f 3f 15f -f
3f -8f ORF (b)
(dB)

ator time k is any integer,

X:X: 40
ν is the order of the stator time
ORF (b) X: 648.2 7f -8f
(dB)

X:
X: 140 X: 579.9 s OR
-30 X:489.9
-30
Amplitude (dB)

X: Y:
140-87.69 140 289.7
X: 748.1 s OR X: 748.1
(dB) (dB)

(dB)

s OR X: 517.4 s r s OR s OR X: 717.3
X: 170.2
Y:Y: -70.6
X: 229.9
Y: -41.44 X: 270.1 s=0.04), (a) : 0-150Hz, (b)X::517.4 140-300
Y: -41.44 Hz, (c) : 450-600 9f +6f Hz, (d)s: Around
OR
(dB)(dB)

Y: -86.36 Y:
Y: -70.98
Y: -90.61
Y: -70.98 -40 -20 Y:-91.57
-70.6 Y: -47.07
13f -5f Y: -89.23 X:
X: 130
-70.98 -87.69
Y: -43.61
Healthy -40
-20 13f -5f X: 641.2
L-PSH Y: -47.1
s r Healthy
Y: -43.61
-90-40 Y:
-70
-90.83 Y: -91.82 130 X: 593.5
Y: -47.07 PSHs
riving the harmonics that are present sin the
OR power supply driving
-5f Y: -57.87the -40 ss OR OR
Amplitude

-90 f +5fY: 3f
Y: -80.91 f +5f U-PSH
Amplitude

91 -80.91
Amplitude

Amplitude

X: 217.6 Y: -59.18 X: 593.5

7f-50 ORs OR -50
(dB)

+6f X: 493.6 X: 541.2 3f

s +5f X: 612.7 s OR
Amplitude

X: 741.1
300 450
-60(ν=1,3,5,s r Y: -66.83 500
Y: -62.16
Y: -66.93 550 s OR 600 -60 7f +6f X: 693.4 Y: -57.87
frequency motor etc.), is
512.8the mechanical rotor frequency 3f +5f U-PSH Y: -69.93
Y: -67.74 X: 493.6
180 frX: -40 X: 541.2
-60
140 -40 160 200 Frequency 220 (Hz) 240 260 280 U-PSH300 Y: -71.91 X: 712.7
-90 3f -5f
X: 593.5 s
-60 r Y: -66.83 Y: -66.93 s OR
3f -5f
Amplitude

X:X:665
Amplitude

Amplitude

Y: -76.82
Frequency (Hz)
X: 217.6 Y: -57.87 X: 512.8X: 217.6
217.6 Y: -77.53
ss OR
-70 s OR -70 X: 622 OR
(c)
and 0N 0b is -60
-80
-80 the number of balls equal
X: 599.6
Y: -85.45
Y: -62.169. 100
X: 582
Y: -89.04
X: 282.2 -80
-60
X: 599.6
Y: -85.45 Y: -66.49
Y:
Y: -83.07
Y: -62.16
Y: -76.82 -62.16
X: 582
X: 721.9
(b) Frequency (Hz) X: 679.8 Y: -90.74
-80
X: 679.8
-10 -90 ORF Y: -95.89 (c) -90 Y: -95.89 Y: -76.03
Y: -89.04
Hz So, -100in
0 this case we write:
-100 -80 160 f -6f 180ORF Healthy 11f
s
X: 282.2 -100
-80580 600 620 640 660 680 700 720 740Y:
X:
X: 282.2
282.2
760
6f (b) -10 140 4503f 580 s OR 600 500200 640 220660 5f240
620 550
680 260
700 280
720 (b) 740600
300 Y: -90.74
760-100580 600 620 640 660 680 700 720 740 Y: -90.74
-90.74
760
Frequency (Hz)
OR -30 ss 
f
Frequency (Hz)ss   f  kf 450 500 Frequency (Hz) 550 600
(dB) (dB)

R=280Hz (4) Healthy Frequency (Hz)

OR11fss-7fOR =280Hz (4) Frequency (Hz)
00 -100 OR  ,k s ORf -7f -100
Fig.
R
-30 21f140
7f -5f charact
-8f 160 current 180
5f -8f 11f +6f22023fs-5f
200 s of rdifferent240 s260
5f OR -6f
ORF 280 300Fig. 0 5. Stator
140 160 current 180 spectrum 200 of 220 different240 frequency 260 bands 280 (ORF,300
Amplitude
(dB)

5f -6f -50 Fig.s s 4.

OR ORStator 7f -6fspectrum OR ORF
frequency bands(c)(d) (ORF, 4. Stator current spectrum of11fdifferent frequency bands (ORF,
ent sof OR the
WeFig. will
-205. analyze
Stator s 7f current
OR-10f the stator
spectrum
19f ss current
-5f
ss OR of
OR
OR different spectrum
Frequency3f
frequency
(Hz)
9f
-10f : 450-600 content
-8f bands of
s(ORF,
s the
OR
OR s = 0.04): a)
s=0.04),
3 s≈0),
0‒150
21f
(a) -5f
: Hz,
0-150Hz,(b) 140‒300
(b) :
5f -8f Hz,
Frequency
140-300 (c)
Hz, 450‒600
+6f(Hz)
(c) : 23fHz,
450-600
s
-5f (d)
OR Hz, Around
(d) :
ORF PSHs
Around(d)
-20 9fs≈0), 0 9f
s(a) : OR
-7f0-150Hz,
X: 617.3
=180Hz (b) :OR
140-300 3ffss-8f
+2fHz,s (c)OR ss OR OR HealthyHz,Healthy
(d) : Around s OR
0 (a) : 0-150Hz, (b) : 140-300 s OR PSHsHz,s(c) :r 450-600 Hz, (d) : Around
Amplitude

when
X: 289.7 thesteady-state -10f in both healthy and faulty states when the
Amplitude

-50 s OR 15f -f OR -20 3f -10f Healthy

(dB)

s OR Y: -37.48 OR
9.9
-70 s OR s OR s PSHs OR 9f +6f
X: 717.3 7fORF -8f X: 617.3
PSHs s OR ORF (c)
Amplitude(dB)

Y: -87.69 X: 517.4 Y:s-47.1 r s OR (c) 9f -10f -10f Y: -37.4819fs-5f15f -f

s 7fOR
0.61
-40
(dB) (dB)

fault-40 in the-20outer X: 489.9

raceway
7f
Y: -91.57 s
-10f Y:(ORF).
L-PSH
OR X:-47.07
19f -5fOR
641.2 s
X: 579.9
Healthy
Y:593.5
-90.61 -20
-40 ss OR
OR s OR ORs OR
X: 717.3
Y: -47.1 Healthy s
7f -8f
OR
-70
Bull. Pol.X: Ac.: Tech. 67(3) 2019 3f -8f U-PSH X:
L-PSH 3f -8f 573
Amplitude

X: 189.8 Y: -59.18
e spectral X: 289.7
Figures
-90 0(4)
X: 612.7
and (5) respectively
X: 189.8
X: represent
X: 229.9
s OR the spectral
Y: -57.87 X: 289.7
X:741.1
289.7 0 X: 641.2 ss OR OR 9f +6f
Amplitude

170.2 229.9 X:
-60 Y: 7f X: +6f
170.2 X: 493.6 Y: -86.36 X: X:
541.2
693.4 X:3f
X: 270.1
270.1+5f
9f +6f U-PSH
(dB)

(dB)

600
300 Y: -87.69 -60 Y: -67.74
s -90.83r Y: -66.83
Y: -86.36
X: 517.4
Y:Y:
Y:
Y:
-66.93
-89.23
-89.23X: 712.7
Y: s
Y: -91.82 OR
s r ORF
Y:
Y: -87.69
Y:-69.93
-87.69 X: 612.7 X: 517.4
Y: -59.18
X: 517.4 ss rr ORF
ns: at no- content X: 593.5Y: -90.83
of-40the9fstator current X: 665for
-71.91
the two550 (a) -60 (a)
7foperations: atHealthy
no-
-91.82 X: 741.1
-90450 -5f X: 512.8 -40 9f -5f Y: -47.07
-47.07 X: 693.4
s OR X:500 600
Y: -47.07 Y:
Y: -57.87 622Y:f -76.82
Y: -5f
-77.53
X: 593.5 X: 593.5 s
Y: -67.74
OR Y:7f -5f X: 712.7 X:
Healthy 593.5Y: -69.93
X: 593.5
-20 -f
Frequency (Hz) -20 f -f -71.91
emplitude

Y: -83.07 s OR X: 582 s OR
mplitude

mplitude

280
load
300 140
(s = 0.004≈0)
-80
-80 160 7fs+6f 180
and 493.6
Y: X:
at
-66.49s
200
load220
OR (s = 0.04). 240
X: 541.2 3f-89.04
X:+5f
721.9
Y: -57.87
7f +6f
Y: -57.87
ss rr
X: s
493.6 OR
X: 493.6X: 622
X: 665
Y: -83.07
X: 541.2
X: 541.2
Y: -77.53
3f +5f
ss OR
Y: -57.87
Y: -57.87

Y: -66.93 260 280 OR 300

e (dB)

(dB)

(c) -60 r Y: -66.83

Y:
sY: -76.03 -80
-60 Y:
Y: -66.83
-66.83 Y:
Y: -66.93
-66.93 OR
due to the We clearly notice the presence of harmonics due to the
Y: -66.49
-40 Frequency
X: 512.8 (Hz)
-40
X:
X: 512.8
512.8 X: 721.9
-100 Y: -76.82 f +f Y:
Y: -76.82
-76.82 f +f Y: -76.03
 N. Bessous, S. Sbaa, and A.C. Megherbi
First, and from figure (4), we notice the characteristic Secondly, and in the presence of a slip value (at the
First, and from figure (4), we notice the characteristic Secondly, and in the presence of a slip value (at the
frequencies of the mixed eccentricity fault around the load operation), it is clear in figure (5) that the ORF has
frequencies of the mixed eccentricity fault around the load operation), it is clear in figure (5) that the ORF has
fundamental:
First, First,and fs-f
andfrom r=25.1
from Fig. Hz
figure and
4, (4),
we wefs+f
notice notice =75
rthe Hz.characteristic
the In thefrequen-
characteristic band made its the
Secondly,
we notice mark on the
and
frequencies in the stator
because current
presenceof theof spectrum.
mixeda slip Around
value
eccentricity the
(atthat
fundamental: fs-f r=25.1 Hz and fs+fr =75 Hz. In the band made its mark on the stator current spectrum. Around the
[0- 150
frequencies Hz] of
ofof the the figure,
mixed we can notice precisely
around the fundamental frequency, we notice the frequencies because
[0- cies 150ofHz] the mixed the figure, eccentricity
eccentricity faultcan
we around fault
the fundamental
notice precisely fre-
the load
have the operation),
fundamental following it is clear
values:
frequency, fswe ¡ in figure
fnotice
r = 26 Hzthe(5)and that
fs + fthe
frequencies ORF
r = 74.1 has
Hz.
because
presence of
quency: fs ¡ ffsr-f=
fundamental: the peaks
=25.1 Hzaround
andfs +fsf+fthe fundamental,
=75 Hz. In the whose
band of the
made mixed markeccentricity that have the fs-fr=26 the
values:frequencies Hz
of the its
toon thethe stator thatcurrent spectrum.
values: Around
25.1 Hz and r = 75 Hz. In the [0–150 Hz] According formula of the characteristic
presence of the peaks around the fundamental, whose
r r mixed± eccentricity have the fs-fr=26 Hz
frequencies
[0- band
150 ofHz] thearefigure,
of in agreement
the we can notice
figure, we with precisely
can the formula
noticetheprecisely
presenceof ofthethe and
of ORF fs+frfcharact–OR–v,
fundamental =74.1 Hz.
frequency,k = jν fs we ±0.4(k)Nnoticeb £the fr j,frequencies
we have found what
because
frequencies are in agreement with the formula of the and fs+fr =74.1 Hz.
characteristic
presencepeaks around of frequencies
the thepeaks
fundamental of the
around equation
frequency,
the (5). frequencies
whose
fundamental, whose are According
follows:
of the mixed theeccentricity
formula of that the characteristic
have the values: frequencies
f -f =26 Hzof
characteristic
in agreement
frequencies of theofequation (5). According the formula of the characteristic frequencies s r of
frequencies arewith in fthe formula
agreement  f s  kf r the
with characteristic
the formula frequencies
of the
(5) and f +f 
ORF s f charact
r –=74.1 Hz.   f 0.4( k ) N  f , we found:
of equation frequencies
characteristic (5). f Mix ecc
Mix ecc  f s  kf r
of the equation (5). (5) ORF
According
fcharact–OR–9,5
fcharact  ,k 
OR  ,k =
the
ORformula
j9fof
fs ss¡the0.4(5)N
0.4( k )bN£ bfr
characteristic
b
j =f rr18frequencies
Hz found:
, we of
The values can be written as follows:
The values can be written as follows:

fcharact –  9 f  0.4(5)
0.4(5)
7f N
0.4(5)N  f  18 82HzHz

fcharact f Mix ecc
f0.4
f s figure N  jfwef s  kf r
k40 fr jHz (5) ORF   f
fcharactfcharact

OR  9,5  9,k f=
s 
s j f s 0.4(
¡ Nbk) Nf rb r 18
b £ r f j = f r Hz
, we found:
1,1  frs  s b
b= ±notice (5) ORcharact–OR–7,5
9,5OR
Thef 
First,
values
charact
ORand
OR 1,1
from
 f 
can bes written as 0.4
Mix–eccN (4),
 f
b follows:r  40 Hz
the characteristic Secondly, –
and in the presence of a slip value (at the
0.4N £ fr=f36.4 HzHz
7,5  9 0.4(5) 18
 fcharact–OR–1,1 it f=
7 fs ¡ in bN 82
frequencies of the mixed eccentricity fault around the
The values can be written as follows:
fcharact
load  operation),
f charact OR 9,5 7 fissss clear
0.4(5) figure
Nbbb  fr(5) r  that 82Hz
Hz
the ORF has
fcharact

 f   0.4 Nb  f  140 40 Hz
HzHz. In the band
fcharact OR –1,1 fs-ffrss=25.1
fundamental:  0.4Hz Nb and frr fs+f140 r =75 made itsf +mark on the stator current spectrum. Around the
  OR  7,5 r
 OR  1,1 Hz charact–OR–1,1 = fs + 0.4Nb £ fr = 136.4 Hz.
[0- 150 Hz] of the =figure,
fcharact–OR–1,1 fs ¡ 0.4N web £can fr =notice
40 Hz precisely the 
f charact
fundamental  f7s 
frequency,
7,5  f s 0.4  0.4(5)
Nb  N
we f r b36.4
notice thef r  Hz 82Hz because
frequencies

f f 0.4 0.4NbN 140 f  OR 1,1  f  0.4 N  f  36.4 Hz
fcharact
presence ORof+1,1the b  ffrr 
(2) Hz
130fundamental,
Hz
 f ss speaks around
 fbr£the whose of the Themixed eccentricity that have the values: fs-fr=26fre- Hz
charact  OR 1,1 s b r

fcharact
charact OR
f
1,2 0.4
= Nfs b+ (2)
0.4N f130 = Hz Hz
140 ORF causes components to appear in the higher
frequencies are in agreement with the formula of the
 OR  1,2 and  f +f =74.1 Hz.

fcharact  fs   0.4 Nspectrum. 36.4
f r  136.4 Hz
b Hzshown in fig-
charact–OR–1,1 r
quency of
ORthe1,1 stator f s current We have
s r

fcharact
characteristic 0.4
f s orders
– frequencies
=N fs b¡(2)  130
 fbrequation
of0.4N
the fr Hz
= (5).

fcharact
According theformula 0.4 ofNstator
the r  136.4
fcurrent
b  characteristic Hzfrequencies Hz]of
At the OR f1,2higher of the harmonics,
(2)£ 130we Hz.have seen ure (5)
The the
 OR spectrum
ORF
1,1
causes of the
components to in
appearthe [580‒760
in the higher
At the higher orders
charact–OR–1,2
of the harmonics, we have seen
the existence many fseries Mix ecc f s  kf r
of characteristic frequencies(5) of ORF
band. The
fcharact
frequency

ORFOR
fcharact
Figutre
OR of
6causes
1,1 the
shows 
f ,sstator
k  current
components
0.4 fN
spectrals b 0.4( k)to
analysis Nbappear
136.4
f rspectrum. of thefHz in
wethe
,current higher
found:
rWe have shown
in the
the existence At the many orders
higher series ofofthe characteristic
harmonics, frequencies
we have seen of
the frequency
[1000‒1500 ofHz]the stator
band; current
the spectrum.
difference between We have
healthy shown
and
REB
TheAt faults
valuesthe higheralong
can bethe the frequency
orders
written of
as the bands.
follows: harmonics, we have seen inf figure (5) the
REB
This
faults
existence
successful
along
of many frequency
series
comparison of bands.
characteristic
of any frequencies
experimental of REB
results in
faulty The
figure
stateORF
(5)
is  spectrum
9,5clearly9 f svisible.
causes
the spectrum 0.4(5)
components of the
of theNb stator
stator  current
tof rappear 18Hz
current inin thehigher
inthe
band
the band
the  existence
fcharact many series
0.4 Nbof characteristic Hz frequencies
 40experimental of charact
[580-760Hz].
OR
Figure (6) shows a spectral analysis of the
This
with faultssuccessful
the OR 1,1 
along thefcomparison
characteristic sfrequency
frequency
 off r any
bands. formulas of the OR fault
results frequency of the
[580-760Hz]. Figure stator(6)current showsspectrum.
a spectralWe have shown
analysis of the
REB faults along the frequency bands. current
inf figure in
with the
allows
This
This
 successful
characteristic
us successful
to ensure
frequency
comparison
comparisonthe effectiveness of formulas
any experimental
of any experimental
of the results
OR fault
of the diagnostic
results
with current (5) the
in7,5the 7band
 spectrum f  0.4(5)
band
[1000-1500Hz];
of theN stator
[1000-1500Hz];
(a)
 f r current 82theHzindifference
the the band
difference
between
charact OR
[580-760Hz]. healthy ands faulty state bis clearly appeared.
f
allows
method
with
the
charact
us
the for

to
characteristic 
characteristic
OR  1,1 f
ensure s  0.4
the
frequency
a fault infrequency N b  f
effectiveness
formulas
the element r  140
formulas
of Hzof
the the
OR
bearings.
diagnostic
fault
of theThe
allows
OR tablefault
us between0 healthy and faulty state is clearly appeared. of the
Figure (6) shows a spectral analysis
current in Absence
the band
method
below
allows
to ensure for the
represents
uselement
a effectiveness
to ensure
fault in the ofelement
some the comparison
the diagnostic
effectiveness
bearings.
values method The
of thebetween
for table
a fault
diagnostic the
0
fcharact
-20  f s ofofthe
Absence 0.4
the N[1000-1500Hz];
 fr 
characteristic
characteristic 36.4Hz
frequenies of the
frequenies
theORF difference
of the ORF
(a)
below
f in therepresents
 f  some
bearings.
0.4 N comparison
The
(2) Table
f  1130 values
represents
Hz between
some the
compari- between
-20
 OR healthy and faulty state is clearly appeared. (a)
1,1 b
(dB)

theory
method and the
for abetween experimental
faultexperimental
in calculation in no-load
the the element bearings.in The tablein
charact  OR  1,2 s b r -400
(dB) (dB)

theory son values and the theory andcalculation

experimental no-load
calculation
OR 1,1  f s  0.4 Nb  f r  136.4 Hz

operation
below (s≈0).
represents
operationsome comparison values between the fcharact
-40 Absence of the characteristic frequenies of the ORF
Amplitude

(a)
no-load
operation (s≈0). (s ¼ 0). -20
-60
Amplitude

-60
theory At the andhigher the orders experimental of the harmonics,calculationweinhave no-load
seen The ORF causes components to appear in the higher
-40
-80
operation
the existence (s≈0). many series Table of 1
characteristic frequencies of -80
Amplitude

Table
Table 1 1 frequency
-60
-100 of the stator current spectrum. We have shown
REBTheoreticalfaults
Theoretical
and practical
along andthe practical
values of bands.
frequency values
the harmonics (ORF, s = 0)
of
Theoretical and practical values of the harmonics (ORF, s = 0) harmonics (ORF, s = 0) -100
in -80
figure
1000 1050 (5) the1100spectrum
1150 1200of 1250 the stator
1300 1350 current 1400in 1450 the band 1500
ThisFormula successful comparison Theoretical Tableof1 any Practical experimental
Amplitude results 1000 1050 1100 1150 1200
[580-760Hz].
Frequency
Frequency
1250 (Hz)
1300 1350 1400 1450 1500
Figure (6) shows a spectral analysis of the
(Hz)
Formula
Theoretical Formula
and Theoretical
practical valuesTheoretical
of Practical
the harmonics Practical Amplitude
Amplitude
(ORF, -1000
with the
± characteristic frequency
Values formulas
Values values [dB]OR0)fault
of the s = 0
current inFrequenies
the caused by ORF
1100 band by[1000-1500Hz]; 1300 1350 the 1400 difference

fcharact f s  k  fOR Values values Values [dB] 1000 1050 1150
caused1200
(b)

fallows f
OR
  f  k 
us s to ensure f
jν f ± k£ f [Hz]
the j [Hz]
-20 Frequenies (b)1250 (Hz)
ORF 1450 (b) 1500
[Hz]effectiveness [Hz] of the diagnostic
= s -20 Frequency
between healthy and faulty state is clearly appeared.
(dB)

charact ORcharact–OR OR
[ Hz] [ Hz]
Formula OR
Theoretical Practical Amplitude[dB] 0
(dB) (dB)

-40
method for a fault in the element bearings. The table X: 1385 X: 1485
X: 1285
-400
fcharact OR  f s j fks¡fORfOR j Values Values [dB] Frequenies caused by ORF
 X: 1117 -62.35Y: -59.17
Amplitude

X: 1485
X: 1185 Y: -63.68
X: 1285 Y:
X: 1385 (b)
040 040.9 – 70.6 -20
-60 Absence X:of
Y: the
-67.73
1117 characteristic frequenies
X: 1361 of the ORF Y: -59.17
below represents some[Hz] comparison values between the (a)
Amplitude

X: 1461
40 [Hz]
40 -70.6 -20
-60
X: 1017 Y: -67.73
Y:
X: -72.2
X:
11851261 Y: -63.68 Y: -62.35

f s  fOR
X: 1085 Y:
X: -78.29
1361 Y:
X: -77.96
1461
Y: -82.9
X: 1017 X: 1161 Y: -81.22
Y: -72.2
X: 1261
40 40 -70.6
(dB)

f s  andfORj fs +
Y: -87.07
theory thefOR jexperimental140calculation in no-load -40 X: 1085 Y: -78.29 Y: -77.96
-80 Y: -82.9 Y: -91.97
X: 1161 Y: -81.22 X: 1385 X: 1485
140.9 – 70.98 -40
-80 Y: -87.07
X: 1117
X: 1285
Y: -62.35Y: -59.17
Amplitude

Y: -91.97 X: 1309
Y: -63.68
X: 1185
operation (s≈0). -60 Y: -67.73
Amplitude

-100 Y: -72.2
X: 1261 X:Y:
X: -79.3
1309
1361 X: 1461
X: 1017
40 40 -70.6 -60
ffss  ffOR
Y: -77.96

j fs ¡ 2 fOR j 140 130 140 -70.98 X: 1085 Y:Y: -79.3

-78.29
-100 Y: -82.9 X: 1161 Y: -81.22
140 130.9-70.98 – 80.91 1000 1050 1100 1150 Y: -91.97 1200 1250 1300 1350 1400 1450 1500
Y: -87.07
140 -80
f s  fOR OR -80 Frequency
1000 1050 1100 1150 1200 1250 1300 X: 1309 (Hz)
Frequency (Hz) Y: -79.3
1350 1400 1450 1500
130Table 1230 130 229.9-80.91 – 89.23 -100
s  2 f OR
fTheoretical j fs +and2 fOR j 130 130 -80.91 Fig.
-100 6. Stator current spectrum of the frequency bands 1000-1500Hz
ffss 2ffOR practical140 140
values of the harmonics -70.98
(ORF, s = 0) Fig. 6. Stator
1050 current spectrum
OR 1000
1000 1050 1100
1100 1150
(s=0.04),
1150 (a)
1200 : of the frequency
1250
Healthy,
Frequency
1200
1300
1250 (Hz)(b) : ORF
1300
bands
1350 1000-1500Hz
1400
1350 1400 1450 1500
1450 1500
230 290 229.9 -89.23 (s=0.04), (a) : Healthy, (b) : ORF
j 5 fs ¡ 6 fOR jTheoretical 130 289.7 – 87.70 Frequency (Hz)
f ss  22
ffFormula
 2 fffOR
OR
130
230 Practical
229.9 -80.91
Amplitude
-89.23 Fig.0 6. Stator current spectrum of the frequency bands 1000-1500Hz
We notice the appearance of: ORF thefrequency
characteristic

fcharact OR  f js 7
s
 fks¡fOR6 fOR j Values
OR
290
Values [dB]
190 289.7 189.8 -87.7 – 86.36 Fig. We
-206. Stator Frequenies
notice
current the
caused
(s=0.04), by ORF
(a) : of
appearance
spectrum Healthy, (b)
1000–1500 of Hz the characteristic bands(b)
5ff s26ffOR [Hz]
230 [Hz]
229.9 -89.23 frequencies of the (s = ORF 0.04): a) which
Healthy, located
b) ORF in the frequency
Amplitude (dB)

290 289.7 -87.7 frequencies of the ORF which located in the frequency
5 fs s  6 fOR -40
range of 1000-1500 Hz, which haveX: the
1285 following values:
X: 1385 X: 1485
j 9ORfs ¡ 7 fOR j 190 180 180.9 – 91.28
rangeWe of notice
1000-1500 the Hz, appearance
which have ofthe the characteristic
following X: 1461values:
X: 1117 Y: -62.35Y: -59.17
Y: -63.68
189.8 -86.36 X: 1185
75 ff s  6 f
Y: -67.73
290 289.7 -87.7 -60
frequencies

fcharact of the ORF which
 -87.0711 f sX:11610.4(10) located in the
Nb  f r  712.7 Hz Y: frequency
X: 1361
6 6f ffOR
Y: -72.2
X: 1261
190
40 189.8
40 -86.36
-70.6 X: 1017
7 f sss  
fcharact
X: 1085
11Hz, 0.4(10)
f sY:-91.97 Nbthe
Y: -78.29
 712.7
ffollowing
-77.96
Hz
the appearance X: 1309
Y: -81.22
OR
j 9OR
fs ¡ 8 fOR j 270 270.1 – 91.82 Y:OR 11,10 Y:
-82.9
OR
180 180 -91.28
range
-80 of
We notice
OR 1000-1500
11,10 which have
of characteristic r ORF values:
frequencies,
9
7 fff ss  67 ffOR 190
180 189.8
490 180
-86.36
-91.28 which are located in23
thef frequency range of f1000–1500 Hz
Hzand
Y: -79.3
9  7 fjOR fs ¡ 6 f j 140 140 489.9-70.98 – 91.57 f-100

 11
23,51100 23  0.4(10)
0.4(5) N N 
b  ff r1350
 712.7Hz
r  717.3
f ss  fOROR OR 270 270.1 have

fcharact
charact 
theOR
1000 OR
 23,5
11,10
following
1050 s
values:
1150 0.4(5)
f s 1200 1250N1300
b
b
r  717.3 Hz
1400 1450 1500
180 579.9-91.82 Frequency (Hz)
99 f  87 ffjOR
9fffsss 28ffOR fs ¡ 7 fOR j 180
270
130 580 270.1
130
-91.28
-91.82
-80.91 – 90.61
OR 
fFig.
 6. 23,5 23
7= 0.4(5)
f fs 0.4(8)
s ¡ = 717.3
NbNb bf£r ffr rj 741.1
– Stator current spectrum of the frequency bands 1000-1500Hz
Hz
HzHz
s OR
490 600 489.9 599.6-91.57
fcharact–OR–11,10
fcharact 7 f ssj 11f
charact OR 7,8  (s=0.04),
OR  7,8  0.4(8)0.4(10)N
Nb (b)f r: ORF
(a) : Healthy,  741.1 712.7
Hz
9fffs s 6
68ffjfOR
3 fs + 5 fOR j 270
490
230
270.1
489.9
229.9
-91.82
-91.57
-89.23
– 85.45
f ss  2 fOR OR  f
fcharact

–
21fjs23f
 7=the 0.4(5)N
0.4(8) NN
0.4(3)
fs appearance f rfrfjr = 717.3
741.1 Hz Hz
Hz
1309.2
21,3  bb the
OR s¡ b£
 charact–OR–23,5
j fs ¡ 7 fOR j 580 680 579.9 679.8-90.61 – 85.89 WeOR
fcharact notice
7,8  21 f  0.4(3) N of
 f  characteristic
1309.2 Hz
ffs 
f 7
 6 fffOR
7
490
580
290
489.9
579.9
289.7
-91.57
-90.61
-87.7 frequencies – OR  21,3
of the ORF
= j 7f
s
which b £
located
b r
in the
Hz frequency
5 fs s  6 fOR f s ¡ 0.4(8)N fr j = 741.1
s OR
 charact–OR–7,8
3ff s Secondly,
OR
75ffOR and in the 600 599.6 -85.45 fcharact
range of  21
1000-1500 fs 
Hz, 0.4(3)Nhave
f s0.4(14)
which N  fthe 1309.2
f rfollowing
1159.6 HzHz
values:
3 fss  5 fOR
580
600
190
579.9
presence 599.6
189.8
-90.61
-85.45
of a slip value (at the load
-86.36 f 
+
 OR  21,3
1,14  f  0.4(14) N  f  1159.6 Hz
f OR 1,14
charact
b b r
= s 21 f + 0.4(3)N
b £ fr = 1309.2 Hz
j f s 0.4(10) Nb r j f  712.7 Hz
7 f s  6 fOR
11,10  11

OR
operation), it is clear in680 Fig. 5 that the ORF has made its mark on fcharact
charact–OR–21,3
 OtherORsideband frequencies
3the 7
ffs stator5fffOR 600
680
679.8
599.6
679.8
-85.89
-85.45
-85.89 f charact
Other
+ORsideband s0.4(14) N(SBF)
1,14  f s frequencies
b caused
b  f r caused
(SBF)
by Hz
r 1159.6 this fault
by this fault
f  7
9 fs s  7 fOR
s current
OR spectrum.
180 Around the
180 fundamental
-91.28frequency, can befcharact–OR–1,14
presented = 0.4(14)Nbtable:
fs +following
in jthe £ fr j = 1159.6 Hz.
can be presented in
23 the following
0.4(5) table:
fcharact  f  N r  717.3
b  fcaused Hz fault
OR
f 7f 680 679.8 -85.89 Other ORsideband
 23,5 frequencies
s (SBF) by this
270 270.1 -91.82
9 fs s  8 fOR can be presented in the following table:
574 OR
fcharact OR 7,8  7 f s  0.4(8) NBull.
b  f r  741.1Hz
Pol. Ac.: Tech. 67(3) 2019
490 489.9 -91.57
f s  6 fOR
580 579.9 -90.61

fcharact OR  21,3 21 f s  0.4(3) Nb  f r  1309.2Hz
f s  7 fOR
 36.4 36 -73.16 always present in agreement with the formula (5), among
f s  fOR Table 2 In fthis first step, we will analyze the stator current in
them, s-fr = 25.1 Hz and fs+fr = 75 Hz.
Theoretical and practical values of the harmonics (ORF, s = 0.04) the 0case when the slip is close to zero (at no-load
136.4 136 -72.29
fFormula
s  f OR Theoretical Practical Amplitude
3f -f
operation).s IR
f -f
s IR
IRF f +f (a)
-20 Healthy s IR
electrical machines Figure (7) shows the presence of harmonics at the

 fOR Mechanical
13ORfs  f5s fkOR
fcharact Values fault detection
218[Hz] 217.6 in rotating-62.16
Values [dB] using MCSA-FFT
X: 25.1 and
X: 75 MCSA-DWT techniques

Amplitude (dB)
[Hz] Y: -45.98 Y: -44.28 21f -9f
frequency
-40 15.4 Hz, 84.6 Hz, 115.45fHz, -f 159.6
s IR Hz and X: 84.6 s IR
X: 184.6

3 f s  5 fOR 282 282.2 -90.74 184.6 Hz around the fundamental.

X: 15.4
-60Y: -68.3 X: 115.4 Moreover, the series
Y: -63.26
X: 159.6
of
Y: -66.42
Other sideband frequencies
18 (SBF) caused by this
17.7 fault are
-62.73 In this first
harmonics is step,
verified we will by analyze
Y: -74
the equation the statorbelow current
Y: -76.79 in the
on the
9 f s  5 fin
presented OR the Table 2: case
-80 when the slip is close to zero (at no-load operation).
7 f s  10 fOR 514 512.8 -
Table 2 illustrated
In this band of [0-1000]
first step, we willHz. analyze the stator current in
Theoretical and practical values of the harmonics (ORF, 76.82→517.8
s = 0.04) the Figure 7 shows
-100case when f the slip
the presence is ofclose
harmonicszero at the (at
frequency of
7 f s  5 fOR 82 Table 2 82.3 -82.3
-2sfs 15.4 Hz, 84.6 Hz,

115.4 
Hz,
 ,k 159.6 Hz
 f s andtokf184.6 IR Hz
no-load
around (6)
the
Formula 518Table
Theoretical 2 Practical
517.4 Amplitude
-47.07 operation).
In0 this first step, we will analyze
charact  IR
the stator current 200 in
19 f s  5 fOR
Theoretical and practical values of harmonics (ORF, s =(SBF)
0.04) fundamental
The value.50Moreover,
frequencies due tothe100
theseries mixed
150
of harmonics
eccentricityis verified
Theoretical
fcharact   f  kand practical
 f values
Values [Hz]of the harmonics
Values (ORF, s [dB]
= 0.04) the Figure
case (7)
when showsthe theFrequency
slip presence
is close (Hz) of harmonics
to zero (at at are
no-load the
by the
always0 equation
present below on the illustrated theband of [0–1000] Hz.in
15.4 in agreement with formula (5), among
OR s OR [Hz]
36.4Table 2 Practical
36 -73.16 In
In this
frequency
operation). this first step,
Hz, we
84.6will Hz, analyze
115.4 theIRFstator
Hz, 159.6 current
Hz and
f ss ±Formula
3 Formula
Theoretical 8fOR fORand practical Theoretical
541.2
Theoretical
values of the
541.2
Table 2 Practical
harmonics
Amplitude
-66.93
(ORF, s [dB]
(SBF)
Amplitude
= 0.04) them, f -f =
first
25.1
step,
Hz
we
and
will
f
23f
+f
analyze
-5f
= 75 Hz.
the stator current (b) in
f
Theoretical
 fs  kand
fcharact ORcharact–OR Valuesvalues
[Hz] values
Values
 fOR practical values of the harmonics (ORF, s = 0.04)
= the
184.6
-20 case
the Figure
case Hz
s r when
around
(7)f shows
when ± the
the slip
fundamental.
the =is
slip s iss
presence
r
IR
close
close to
Moreover,
of
to zero
harmonics
Healthy
zero (at
the
(at no-load
series
at
no-load of
the
0
operation).
harmonics Hz,fs+2f
is charact–IR–v,
verified IRby
jν f ±equation
the s115.4
k fIR j .(6)
IRF Hz, below on and the
136.4
641.2 [Hz]
136
641.2 -72.29
-37.48 (SBF) frequency f -2f15.4 84.6 k
f -fHz, 159.6 Hz

(dB)
5 ff
9 Formula
=
sss
j
Formula

ν f8 f f
5 OROR
s ± k£ f 18[Hz]
Theoretical
OR OR j Theoretical
17.7
Practical
[Hz]
Practical
-62.73
Amplitude
[dB]
Amplitude operation).
-40 3f -f s IR s IR f +f (a)
Values [Hz] Values [dB] 184.6Figure
illustrated s IR (7)
band shows
of [0-1000] the presence
Hz. of harmonics at the
TheHz around thedue fundamental. Moreover, theareseries of
 s IR
 OR  f s  k  fOR
ffcharact -20
Figure (7) shows mixedHealthy
k f fORj Values X: 75 the presence of harmonics at the
[Hz] Values [dB]

Amplitude
frequencies to the eccentricity always
charact ORj9 fs f5s 5 218
718 [Hz]
217.6
717.3 -62.16
-47.1 (SBF) frequency
-60 15.4 Hz, 84.6 Hz, 115.4
X: 394.5 Hz, 159.6 Hz and
13
23 f  f 018.4 17.7 –62.73 harmonics is verified by the equation below on the
X: 25.1
¡ [Hz]  Y: -44.28
15.4 fHz, Y:with   115.4
f s(5), 5fkf (6)

(dB)
18 17.7 -62.73 frequency
present inX:agreement 84.6 Hz,
formula fsHz,
¡IRfr 159.6
21f
= 25.1-9f HzHz and and
9 f s  5 f OR
7 f ss  5 fOR
OR
82 82.3 -82.3 -40
184.6 Hz
Y:
219.1
around
-45.98
the
X: 319.1
fundamental. X: 328.8 Y: -78.5 -f
s IR s IR
X: 494.5

illustrated band ofcharact

[0-1000] Y: -88.04 Moreover, the series of
them. Hz.
-81.9
IR  ,k X: 84.6 Y: -82.94
OR
184.6
fs + Hz
-80X:fr15.4 75around
Y: -85.7
Hz among the fundamental. Moreover, the series of
X: 184.6

Amplitude
7 fs ¡ 5 fOR j 1082.4
282 82.3
282.2 –82.3
-90.74 =
339fffssj 714 712.7 -77.53=717.3- Y: -63.26
10 5 ff 18 17.7 -62.73 The
harmonics
-60Y: -68.3 frequencies
is verified due by to thethe mixed eccentricity
equation below on the
X: 115.4
are
Y: -66.42
5 f 
  f  kf
X: 159.6
 f OROR 18
82 17.7
82.3 -62.73
2sfs
-82.3 harmonics is verified by theY: -74 equation below on the
(6)
79 f ss jf 5¡fOR
Y: -76.79
-100
always present incharact
agreement withs the formula (5), among
ORfOR j 36.4
1036.4 36
1136.4 -73.16
–73.16 illustrated
illustrated band
band of of [0-1000]
[0-1000] Hz.
 IR  ,k
Hz.
IR
f
7f sf ss14  s f OR
10OR 514
1159.6 512.8
1161 -
-91.97 -80
ffOROR
76.82→517.8
them,The
200 fs-ffrequencies
r = 25.1 Hz
250 
ff charactand
due
300
 f s+f
to(a)
r =the
350
 75 Hz.
mixed
ff s(Hz)
 kf400 eccentricity
450 are
500
(6)
7
7 ff s j fs5 5+ffOR fOR j
82
82
1136.4
82.3
82.3
1136.4
-82.3
-82.3
–72.29 -100
always0
0 3f present

incharact  IR
agreement
Frequency

withs the  kf IR (6)
IRF formula (5), among
136.4 136 -72.29
-2sfs  ,k
IR
21fffssss ff3OR
36.4
1309.2 36
1309 -73.16
-79.3 -f  IR  f,k-f
OR fOROR 518 517.4 -47.07 (SBF) them,The 0 fss-ffrequencies
IR= 25.1 50 Hz due
and fs to
+fIR IRF
=the
100 75 mixed
Hz. 150
f +f
eccentricity (a)
are
200
(c)
19 fj 13 s fs5 ¡f5ORfOR j 1218.4 1217.6 –62.16 The frequencies
-20 r dueFrequency
s tor the mixed Healthy eccentricity
(Hz)
s IR
are
13 5 218 217.6 -62.16 always
-20
0 present in
L-PSH agreement
X: 75
with
Healthy the formula (5), among
ffffsss study fOR has noticed
X: 25.1
36.4
136.4 36
136 -73.16
-72.29 always present in agreement
 f f -f with the IRF5fformula 21f(5),
-9f among

(dB)
Our f 8f¡OR f 36.4 the series
36 of additional
-73.16 -400 3f -f -5f Y: -45.98 Y: -44.28
-f

(dB) (dB)
3 f ssj 3 s f OR
OR
OR
5which
fOR j have 541.2
1282.4 541.2
1282.2 -66.93
–90.74(SBF) them, s-f IR
f21f
-40 fs-fsr = IR
them, = 25.1
25.1 Hz Hz and and ffs23f
s +fIR = 84.675 Hz. sIRF
+fX:rU-PSH
-5f
= Hz. IR
75 Healthy
sf +fIR
17f
s s-5f
IRIR (b)
(a)
frequencies severe amplitudes along -90.74
harmonic -20 rs -63.26
IR X: 184.6
Amplitude
0X: s r
15.4 s Y:
282 282.2 -20 Healthy X: 159.6 Y: -66.42
13 3 fffsthese  5 ffOR 136.4
218 136
217.6 -72.29
-62.16 -60 Amplitude X: 75 X: 115.4

fs 8¡5ffOR
0Y: -68.3
ffjsss7s
X: 25.1
order; 10 frequencies
ORf j are due to mixed
136.4
641.2
1514.4
136 eccentricity.
641.2
1512.8
-72.29
-37.48
–76.82 (SBF)
→ 517.8 -40 3f -f Y: -45.98 Y:
f +2f fs-fIR
-44.28 f -f Y: -74 IRF
IRF5fs-fIR 21fsY: f-9f
+f (a)
(dB)
-76.79
sf-f-2f
5
In this study, it has f
OR OR
OR
been shown that the spectral
-60
-20
-80
-20
3f
-40 s s IR IR
IR
X: 548.5
s IR s IRX: 84.6 Healthy fs+fIRIR (a)
s IR X: 184.6
514 512.8 – -
2sf X: 648.5 Healthy
Amplitude

7
13 f
3 ffsss   10 55ffOR f 218
282 217.6
282.2 -62.16
-90.74 s
X: 15.4
X: 25.1Y: -55.72
X: 75 Y: -63.26
Amplitude

X:-60.17
115.4 Y: -66.42
-60Y: -68.3 X: 513.2Y: Y:
(dB)

analysis
13 method OR
of the218 stator current 217.6has given -62.16
adequate X: 75
21f X:-9f
23 fjf19 f 55¡ff5OR
159.6
76.82→517.8
s 
-80 -45.98 Y: -44.28
X: 25.1 X: 713.9
718 717.3 -47.1 (SBF) -40 X: 394.55fs-fIR
(dB)

-60 Y: -45.98 Y: -44.28

Y: -74
21f -9f
sY: IR-93.07
-76.79
OR -100
-40 Y: -94.19 5f -f s IR Y:
results fors ORF. s OR fOR j We noticed 1518.4 the 1517.4
efficiency, –47.07
even, -2sfaround
(SBF) X: 219.1
X: 319.1 X: 84.6
328.8 Y: -78.5 s IR
X: 494.5
X: 184.6
Amplitude

s
282
514 282.2
512.8 -90.74
-(SBF) -80X: 15.4 Y: -81.9 X:84.6
X:
Y: -63.26 Y: -82.94
7 3 f  10 5 ff X: 184.6
Amplitude

X: 15.4 Y: -85.7 X: 115.4 Y: -66.42

518
282 517.4
282.2 -47.07
-90.74 -60
-80Y:
-100 0-68.3 50 100
Y:-63.26
Y: -88.04 150 X: 159.6 Y: -66.42 200
19
33ffsfsjs310
fundamental ss 55ffOR ORfrequency,714 whose frequencies
712.7 are very clear
-77.53=717.3-
76.82→517.8 -60500
Y: -68.3 550 600 X:
Y: 115.4
-74 650 700 Y: 750
fs ¡ fOR
X: 159.6
OR 1541.2 1541.2 –66.93 (SBF) Frequency Y: -74 (Hz) -76.79
8ORfOR j 2sf -100 Frequency (Hz) Y: -76.79
for 7the f induction
 10 f motor. 514 512.8 -2sf
- ss -800
-100
-80
541.2
514
518 results 512.8 541.2 (SBF)
-66.93-(SBF)
73f ffsss
Finally,
19 j 5f10 85¡ffffOR
sthe
OR
fexperimental 1159.6 517.4
1161 -47.07
obtained 76.82→517.8 -91.97
confirm the Fig.0 7. Stator current 50 spectrum 23fof-5f different
100 IR (Hz) frequency
IRF 150 bands (IRF, (b)200
s s 14
8OR
OR OR j 1641.2 1641.2 –37.48
76.82→517.8 (SBF) -20200
-100 250 300 Frequency (b)
s 350 400
Healthy 450 500
OR -2sf s=0.004), (a) : 0-200Hz, (b) :
Frequency200-500 (Hz) Hz, (c) : 500-760 Hz
accuracy with the literature 641.2 results. 641.2 -37.48
s
-2sf(SBF) -1000
f +2f
(dB)

5 ffjs23  8s53¡fffOR 518

541.2 517.4
541.2 -47.07
-66.93
s
(SBF) 0 fs-2fIR
19 3 f f8 5 fOR j 1718.4
1309.2 1717.3
1309 –47.1
-79.3 (SBF) -400 50 s IR 23f -5f 100 IRF 150 (b)200
21
19 f ss  5 fOR
s s  OR OR
OR
518 517.4 -47.07 (SBF) 0
The detection operation
-20
50 s 100
Frequency
Frequency
IRF
byIR (Hz)
the FFT
(Hz)
150
Healthyof the stator
200
(c)
(dB)Amplitude

3.2 Inner raceway f jhas fault 718 detection1712.7by MCSA-DWT.

717.3 -47.1 (SBF) In 0
-20
current made L-PSH
it fspossible
Healthy
IR 23f -5fto Y: carry IRFout acceptable
3 10 1714.4 –77.53 = 717.3 -60 +2f
23 f
j f 5 f 541.2
641.2 541.2
641.2 -66.93
-37.48 (SBF) 0
(dB)

Our ¡
study noticed
3 8 bearingthe series of additional
5 f -2f
3 ffcases, ffOR
X: 394.5
ss 
s
ORORelement 541.2 541.2 -66.93 (SBF) X: 494.5(b)
most the faults are manifested by:
X: 319.1
(dB)

-40 21fs X:
 8 – 2sf monitoring
-20
-5fIR
219.1
of Y: -81.9 23fsX:
IR the fault indicators
sY: -85.7 s
-5f
U-PSH IR -78.5 IRF
328.8
IR in the IRF.
Healthy 17f -5f Y: -82.94(b)
frequencies s OR
which havecontour
severe of amplitudes along harmonic s -40
-80
-20 Y: -88.04 s IR
Amplitude

noises,
3 f  ripples
10 f on the 714 the surface
712.7 of the(SBF)
-77.53=717.3- balls, Moreover, f -2f based on f +2f the statorX: 394.5
Healthy
current spectrum and
Amplitude

641.2
718 641.2
717.3 -37.48
-47.1 (SBF) -60
23
order; 5 +8
ffsfsjthese
sfs
fffOR
85inner frequencies are due to mixed
641.2 641.2 eccentricity.
-37.48 2sf(SBF) fsX:+2f IR
(dB)

-40 fs-2f
etc. 5 The s 
14 OR
ORf j 1159.6 1161.4
OR OR raceway fault (IRF) that cause many
–91.97 s -100
-40
-60
according
IR
s X: 219.1
Y:to
-85.7equation
IR s 319.1IR
(6);
X: 548.5 Y: -81.9 we X: 328.8 Y: -78.5
give some examples
X: 494.5
of
Y: -82.94 the
In this study, it has been shown that the spectral
Amplitude

problems in: electromagnetic

1159.6 torque,
1161 rotational -91.97 speed, -80 Y:X:-88.04
648.5
f 14 +ffOR
Amplitude

Y: -55.72
718
714 717.3
712.7 -47.1 (SBF)
-77.53=717.3- -60200
characteristic frequencies of IRF. Table (3) summarizes
323
23fssjff21
analysis s  10 s5
fmethod
5 f 3 fOR j of the718 1309.2
stator 1309.4
current717.3has isgiven –79.3
-47.1adequate
(SBF) -60
-80
250 300
X: 319.1
350
Y: -60.17X: 394.5 400 450 X:X:713.9
500
rotor, stator
s
OR
ORcurrent, etc. The objective the detection
2sf s -100 compares
and
X:
Y:
X:513.2
219.1
X:-94.19
219.1
the
X: 319.1
Y: -81.9
Frequency
X: 328.8 Y:
theoretical
(Hz)
X: 394.5
-78.5
Y: -78.5
and experimental
494.5
X:-93.07
Y:Y: 494.5
-82.94
results for ORF. Wetransform
noticed the efficiency, even, around
Y: -85.7
1309.2 1309 -79.3 -800 Y: -81.9 X: 328.8
Y: -88.04 Y: -82.94
IRF 21
3 f by
f f  fast

1410 3 fff Fourier 714
1159.6
714
(FFT) of the stator
712.7
1161
712.7
current
-77.53=717.3-
-91.97
-77.53=717.3-
-80
frequencies
-100200
Y: -85.7
in
250case of300 IRF.
Y: -88.04
IRF
350 400 450 500
(c)
fundamental
3 f Our
signal. s
s  10
s s
study
ORORfrequency, whose frequencies
fOR has noticed the series of additional frequencies
OR
are very2sfs clear -100
-20
500 550 600Frequency Healthy
650
(Hz) 700 750
2sfs -1000 L-PSH Frequency (Hz)
forwhich
the
Ourinduction
The havestudy
different severe motor.
hasmode noticed
1159.6
1309.2
amplitudes the 1161
operations
along series
used
1309
the toofvalidate
harmonic additional
-91.97
-79.3
order; the
these
14
(dB)

21 ff s f s 14  3 f
f f 1159.6 1161 -91.97 20021fs-5fIR
Fig.
2007. Stator
250 300
250current spectrum of 350
U-PSH
IRF
different 400 450 -5f
17f 500
Finally,
frequencies
diagnostic the OR ORexperimental
which
procedure have severeresults obtained
amplitudes along confirm
harmonic the -40 300 Frequency (c) 350 (Hz) frequency
400 bands
450s (IRF, IR (c) 500
frequencies
s OR are due to are: mixedateccentricity.
no-load operation with a -200 s=0.004), (a)L-PSH : 0-200Hz, Frequency Healthy
(b) : 200-500 (Hz)Hz, (c) : 500-760 Hz
(dB)Amplitude

accuracy
order;
healthy
21 Our f
In theseand

this with
study
3 f thehas
frequencies
faulty
study, literature
element
it has are
noticed
1309.2
been results.
due to mixed
the
bearing,
shown series
1309
at full
that eccentricity.
the of
load additional
spectral-79.3analysis
operation 0
(dB)

1309.2 1309 -79.3 -60 21fs-5fIR IRF

with
21 In
frequencies
method
f s  3 f OR
a healthy
sthis of theandstudy,
which
OR it
have
stator has been
severe
defective
current has shownbearing. resultsspectral
amplitudes
element
given adequatethat
along the harmonic
for ORF. -40
X: 548.5 U-PSH
IRF 17f -5f (c)
s IR (c)
The
-20 detection L-PSH operation
Y: -55.72 by
-60.17 the FFT of the stator
X: 648.5
Healthy
Amplitude

-20 Healthy
3.2 Inner raceway fault detection by MCSA-DWT. In
Y:
order;
We Our
analysis
Our these
noticed study
method
study has
of the
frequencies
the efficiency,
has noticed
stator
are
even
noticed the
duecurrent
around
the series
to mixed has
fundamental
series of
of additional
given
eccentricity. adequate
frequency,
additional -80 21f X:-5f
current
513.2
made
L-PSH
it possible U-PSH to carry out 17f
X: 713.9
acceptable
Y:-5f
(dB)

-93.07
-60
-40 21fss-5fIR
Y: -94.19
most
frequencies
resultsIn this
whose
frequencies cases,for ORF.
frequencies the
which
study, element
which haveWehave
itarehas bearing
severe
noticed
very been
severe the
clear faults
amplitudes
shown
amplitudes are
efficiency,
thatmanifested
along
for the induction even,
along harmonic
around
theharmonic
spectral
motor. by: monitoring
-40 IR
of the
X: 548.5
fault indicators
U-PSH
X: 648.5
in the IRF.
17f s-5fIR
s IR
Amplitude

Y: -55.72
-100
noises,
order;
fundamental ripples
these on
frequencies
frequency,the contour
are
whosedue of
to the surface
mixed
frequencies of
eccentricity.
are the
very balls,
clear Y: -60.17
Amplitude

analysis Finally, method the of the

experimental
order; these frequencies are due to mixed eccentricity. stator current
results has
obtained given
confirm adequate
compat- -80
-60
500 X: 513.2 550
Moreover, based on the 600
stator (Hz) 650
current spectrum700 X: 713.9 750
Y: -93.07 and
Frequency
etc. The inner raceway fault (IRF) that cause many
Y: -94.19
for In
the
results
ibility
In this this
induction
for with study,
ORF. literature
study, it
motor.
We has
it noticed been
results.
has been shown
the shown that
efficiency,
that even,the spectral
around
the spectral
-60
according to equation
X:
X:
548.5
548.5
Y: -55.72
(6); weY: -60.17
X:
X: give some examples
648.5
648.5 of the
problems
analysis methodin: electromagnetic
of the stator torque,
current hasrotational
given speed,
adequate -100
Fig.
-805007. X: Stator
513.2 current spectrum ofY:different
Y: -55.72
frequency bands (IRF,
Finally,
fundamental the experimental
frequency,
analysis method of the stator current has given adequate whoseresults obtained
frequencies areconfirm
very the
clear characteristic
-80 s=0.004), X:
Y: 513.2
-60.17
-94.19 frequencies of IRF. Table (3) summarizes
550 600 650 700 Y:
X: 713.9
-93.07 750
X: 713.9
(a) : 0-200Hz, (b) : 200-500
Frequency (Hz) Hz, (c) : 500-760 Hz
rotor,
results
accuracy
for3.2.
the stator
for
Inner with
induction ORF. current,
the
raceway We etc.
noticed
literature
motor. fault The theobjective
results.
detection efficiency,
by is
MCSA-DWT. the
even, detection
around
In most and
Y: -94.19 Y: -93.07
results for ORF. We noticed the efficiency, even, around 5 -100 compares the theoretical and experimental
IRFcases,
Finally, by
fundamental fast
the the Fourier
frequency,
element
experimental
fundamental frequency, whose frequencies are very cleartransform
bearing whose
faults (FFT)
results of
frequencies
are obtainedthe
manifested stator
are very
by:
confirm current
noises, clear
rip-
the Fig.
-100500
frequencies
7. Stator current
in 550
case of
spectrum
IRF.600 of different
650 frequency 700bands (IRF, 750
The
500 detection
s=0.004), (a) 550 operation
: 0-200Hz, 600 Frequency
(b) : by (Hz)
200-500 theHz,FFT
650
(c) : of theHzstator
700
500-760 750
signal.
for
3.2 the
ples
accuracy
for theInner oninduction
the
with
induction raceway themotor.
contour fault
of detection
the surface
literature
motor. byballs,
results.of the MCSA-DWT.
etc. Meanwhile, In Frequency (Hz)
current madecurrent
Fig. 7. Stator it possible to carry
spectrum of different out bands
frequency acceptable
(IRF,
the
most The
Finally, inner
cases, different
the raceway
the mode
experimental
fault (IRF)
element
Finally, the experimental results obtained confirm the operations
results
causes
bearing used
obtained
many
faults to
problems
are validate
confirm
manifested the
in: electro-
by: Fig. 7. Stator current spectrum of different frequency bands (IRF,
monitoring
The of
s=0.004), the
detection
(a) : fault
0-200Hz,indicators
operation
(b) : by in
200-500 the
the
Hz, IRF.
FFT
(c) : of the
500-760 Hz stator
3.2magnetic
diagnostic
accuracy
noises, Inner rippleswith torque,
procedure
raceway the rotational
on literature
thefault
contour speed,
are:detection
at of rotor
no-load
results. the and
surface
by stator
operation
MCSA-DWT. of current,
thewith etc.
balls,
Ina Fig. 7.s=0.004),
Stator current spectrum
(a) : 0-200Hz, (b) of different
: 200-500 Hz,frequency bands
(c) : 500-760 Hz (IRF,
accuracy with the literature results. Moreover,
current based
made
s = 0.004): on
a)it0‒200 the b)stator
possible
Hz, current
to carry
200‒500 Hz, c) outspectrum
500‒760 Hz and
acceptable
The
healthy
etc. The objective
and inner faultyis the detection
element
raceway
most cases, the element bearing faults are manifested by: of IRF
bearing,
fault (IRF)by
at fast
full
that Fourier
load
cause transform
operationmany The
The detection
according
monitoring toofequation operation
the fault
detection (6); we by
indicators
operation give
by inthe
theFFT
some
the of
of the
examples
IRF.
FFT the ofstator
the
stator
3.2(FFT)
with
problems
noises, a healthy
Inner of in:
ripples the stator
raceway and current
defective
fault
electromagnetic
on the signal.
element
detection
contour thebearing.
by
oftorque, MCSA-DWT.
rotational
surface of thespeed, In
balls,
3.2 Inner The
raceway
different modes
fault of
detection
operation
by
used
MCSA-DWT.
to validate the
In
diag-
current
Moreover,
current mademade
characteristic based it
it onpossible
frequencies
possible to
the ofstator
IRF. carry
current
to carry out
Table out
(3) acceptable
summarizes
spectrum and
acceptable
most
rotor,
etc.
most The cases,
stator
cases, inner the element
current,
raceway
the element etc. bearing
The
fault faults
objective
(IRF)
bearingoperation are is
that
faults arewith manifested
the
cause
manifesteddetectionmanyby:
by: monitoring of the fault indicators in the IRF.
nostic procedure are: no-load a healthy and and compares
according
monitoring to equationthe
of theoperation
The detection theoretical
(6); we
fault indicators give
by FFT of inthe and
some
thestator experimental
IRF.current madethe
examples of it
noises,
IRF
problems
noises, by ripples fast
ripples in:Fourier on
on the
the contour
transform
electromagnetic
contour of
of the
(FFT)
torque,
the surface
of the of
stator
rotational
surface of the
the balls,
current
speed,
balls, Moreover, based on the stator current spectrum and
faulty element bearing and full load operation with a healthy frequencies
characteristic in
possible to carry
Moreover, case
basedouton of
frequencies IRF.
acceptableof IRF.
the stator currentof(3)
Table
monitoring summarizes
fault indicators
spectrum and
etc.
signal.
rotor, The stator inner raceway
current, etc. Thefault (IRF)
(IRF) that
fault objective is thecause detectionmany according to
etc.
and
problems
IRF The
The
by
defective
fast
inner
in:
different
raceway
element bearing.
electromagnetic
Fourier mode operations
transform torque,
(FFT) used
of
that cause
rotational
theto validate
stator
many
speed,
current the
and to equation
compares
in the IRF.
according equationthe (6); we
we give
give some
(6);theoretical andexamples
some examples of
of the
experimentalthe
problems in: electromagnetic torque, rotational speed, characteristic
frequencies in frequencies
case of IRF. of IRF. Table
characteristic frequencies of IRF. Table (3) summarizes (3) summarizes
rotor, stator current, etc.
rotor, stator current, etc. The objective is the detectiona
diagnostic
signal. procedure are: Theat objective
no-load is the
operation detection
with and
IRF
healthy by fast
and Fourier
faulty mode transform
element (FFT)
bearing, of the stator current and compares
compares the the theoretical
theoretical and and experimental
experimental
IRF The
Bull. byPol. different
fast Ac.: Fourier transform
Tech. 67(3) 2019 (FFT)at
operations offull
used thetoload
stator operation
validate current the 5 frequencies in case of
frequencies in case of IRF. IRF. 575
signal.
with
diagnostic
signal. a healthy procedureand defective
are: element
at no-load bearing.
operation with a
healthyThe
The and different
different faulty modemode
element operations
bearing, at
operations used
used to
fulltoload validate
operation
validate the
the
diagnostic
with
diagnostic a healthy procedure
procedureand defective are:
are: at no-load
element
at no-load operation
operation with
bearing. with aa
healthy
healthy and and faulty faulty element
element bearing,
bearing, at at full
full load
load operation
operation
 f s  f IR
Formula Theoretical Practical Amplitude -40 5f -f f +f
s IR

(d(
Values Values [dB] -40 s IR
 IR  f s  k  f IR

fcharact X: 493.7

Amplitude
160.14 159.6 -76.79 X: 121.5
21 f s  9 f IR
X: 21.5 X: 78.6 Y: -65.56
X: 178.6
[Hz] [Hz]

Amplitude
-60Y: -64.53 Y: -66.2 X: 394.4 Y: -64.31 Y: -65.54
-60 X: 307.2 Y: -81.18
X: 207.3 X: 262.8 Y: -86.05
218.92 219.1 -85.7 -80
f 2f 15.54 15.4 N. Bessous,
-68.3 S. Sbaa, and -80
A.C. Megherbi Y: -94.03 Y: -91.45

3sf s  f IRIR -100

318.92 319.1 -81.9 -100
f s  2 f IR 200 250 300 350 400 450 500
115.54 115.4 -74 accord- a) 00
5Moreover,
f s  f IR based on the
515.54 stator
Table 3 current
513.2 spectrum and
-94.13
50 Frequency 100 (Hz)
Frequency (Hz)
150 200
ing13 f 
tosequation
Theoretical f and practical
IR (6), we values Table
give some 3
of theexamples
harmonicsof the scharacteristic
(IRF, = 0.004) 0 0 f -f IRF (a)
Table 3 harmonics -20 00 3fs-n 3frb-ffr L-PSH fss-fIR IRF IRF (c)
(a)
Theoretical and practical 84.46
values of the 84.6and (IRF, s =-63.26
0.004) HealthyIRF
f
frequencies
Theoretical  f of IRF.
IR and practical
Table
715.54 3
values summarizes
Tableof 3
the 713.2
harmonics compares
(IRF, s = theoreti-
-93.07
0.004) -20
-20 0 3f s-fIR 23f -11f f -fIR IRF Healthy
(b)
(a)
17 Formula
s
f s experimental
f IR Theoretical Practical Amplitude s IR s IR s
f -f
X: 617.6 IR Healthy
5f -f IRF Healthy f +f

(dB)
cal and
Theoretical and practical frequencies
values of the in the
harmonics case of
(IRF, IRF.
s = 0.004) -20
-20 3f -f Healthy f s+fIR (a)
 Formula Theoretical
Values Practical
Values Amplitude
[dB] -40 s IR Y: -39 s IR s IR
3f -f f +2f 5fs-fIRHealthy

(dB)
X: 517.6 X: 717.6
fcharact
 Formula
  f  k  f IR Theoretical 184.46 184.6 -66.42 -20X: 21.5f -2f sX: 78.6IR 5f -f X: 121.5 U-PSH fs+fIR
fsIRIR ffssIR k  to [Hz] we canPractical [Hz]that theAmplitude -40
-40 sIRIR

(dB)
Values Values [dB] s Y: -48.53 Y: -48.67

(dB)
X: 178.6
fcharact
According
f IR this table, say frequency

Amplitude
Formula Theoretical Practical Amplitude -40
-40 Y: 21.5
-64.53 f +4f 5f s -f IR Y: -64.31 fs+fIR X:

(dB)
fcharact  f k f Values
[Hz] Values
[Hz] [dB] -60 X: s IR Y:
X: -66.2
78.6 X: 121.5 Y: 178.6
-65.54

Amplitude
  

Amplitude
s IR
are Table
Values likely3 toValues [dB] on -40 s IR 3fs+4fIR X: 493.7
Y: -64.31
bands 
fcharact that
 IR
  f sswe  k  analyzed
f IRIR bring information -60X:
Y: 21.5
-64.53 X: 121.5 Y: 178.6
-65.54

Amplitude
Y: 78.6
-66.2
[Hz] [Hz] -60 X: X:

Amplitude
160.14 159.6 -76.79
21 f
 IR
Theoretical  9 f and practical [Hz]values of harmonics [Hz] (IRF, s = 0.004) -60 X: 21.5
Y: -64.53 X:X: 641.4
Y: 78.6
-66.2 X: 665.2 X:
Y: 121.5
-64.31 Y: -65.56
X: 178.6
Y: -65.54

Amplitude
-60
the existence of the IR15.54
s IR fault. 15.4 -68.3 -80
-60
Y: -64.53 X: 565.3 Y: -56.04
Y: -66.2 Y:
X: -80.89
394.4 Y: -64.31 Y: -65.54
3 f s  f IR 15.54 15.4 -68.3 -80
-80 Y: -86.91
X: 262.8
X: 307.2 Y: -81.18
Slip 3plays f s 2 ffFormula an important role in219.1
15.54 Theoretical
218.92 the calculation
15.4Practical -85.7
of
Amplitude -80
-80 Y: -94.03
X: 207.3 Y: -86.05

3f s±f
Y: -91.45
IR -68.3 -100
s  f IR
harmonics; it
IR
depends, systematically,
15.54 values 15.4byvalues the variation
-68.3 of -80
-100
-100
3fcharact–IR
f s  f IR= jν fs ± k£115.54 fIRj 115.4 -74 -100
the 5 load.
f f s ffIR
2Figure (8) 115.54
shows the
318.92 [Hz]spectrum
319.1 [Hz]
115.4 of the-81.9 [dB]
current
-74
-100 0
-100 500 550 50
50
600
Frequency
100
100 650 (Hz)
150
700
200
750
5 sf s  f IRIR 115.54 115.4 -74 200
0
250 300 Frequency 350 (Hz) 400
150
450
200
500
under 5 ffull  jload
f3 f operation,
f j 115.54 whose015.54 slip is
115.4 equal
015.4 to 0.048.
–
-74 68.30It 00 50 Frequency
Frequency
100 (Hz)
(Hz)
150 200
5ffsssthat  ffIRIR s ¡ IR 84.46
515.54 84.6
513.2 -63.26
-94.13 b) 0 0 50 Frequency 100 (Hz) 150
IRF (b)200
shows 13f sf sf IRf IR around
IR the fundamental
84.46 frequency,
84.6 there
-63.26 are Fig. 00 8. Stator current
23f -11f
s-11fspectrum
IR Frequency (Hz)
of different frequency IRF bands (IRF, (b)
84.46 in 84.6 -63.26 -20 3f -n f 23f Healthy
other f sparticular  fjIR5 fs ¡ fIRharmonics j 184.46
84.46 the 184.6
115.54 spectrum
84.6
115.4 which
-63.26
-66.42 can
– 74.30 -200 s=0.048), fs -2frb (a) r 23f sL-PSH
-11fIR (b)
: 0-200Hz, f +2f : 200-500 Hz, (c) IRF: IRF
500-750 Hz (c)
Healthy (b)
ff 
 ff (b)

(dB)
715.54 713.2 -93.07 -20 s-2fIR 23fs-11fIR f s+2fIR IRF
Healthy
confirm17f sf s
s thef ffIRexistence
IR of IRF. The calculated values of
184.46 184.6 -66.42 -20
-40 f s X: 617.6
IR s Healthy

(dB)
IR s IR IR
j f j 184.46 084.46 184.6 084.6 –
-66.4263.26 -20 f -2f f
Y: -39 +2f Healthy
¡ -40According IR to theoretical fs+2fcalculations obtained in the

(dB)
these fharmonics  f IRfs IRare: 184.46
s X: 517.6 X: 717.6

(dB)
fs-2f IR X: 493.7
U-PSH Y:Y:-65.56

Amplitude
160.14 184.6
159.6 -66.42
-76.79 -40
21f fss s 9f IR s Y: -48.53 -48.67
-40

(dB)
 According IRf IR+to this 160.14
table, we can say that the frequency -60 IR f +4f say s that IR X: 493.7

Amplitude
21 f s IR93,1 fsIR f3IR jf s 160.14 159.6 -76.79 table
-40 below, we can the
X: 394.4frequencies X: obtained
f charact 0.6are (1)  Nto f r 184.6 21.48 Hz
Y: -65.56
bring  information s IR

Amplitude
jwe 184.46 – 66.42 -60 493.7

Amplitude
X: 307.2
bands b 159.6 -76.79 3fs+4fIR X:
21 fthat analyzed likely on Y: 394.4
-81.18
s  9 f IR experimentally demonstrate -86.05the existence of the IRF.
X: Y: 493.7
-60 -65.56

Amplitude
X: 262.8
160.14
218.92 159.6
219.1 -76.79
-85.7 -60
-80 X: 207.3 Y: 307.2
X:
21 9
X: 641.4 X:
Y: 665.2
f
the existence f  2 f f of the IR fault. -60 Y: 207.3
X: -94.03 X: 565.3
Y:
X: -91.45
262.8 Y: 307.2
-86.05
X: -81.18
394.4 Y: -65.56
s s j 21IR fIR
f s  2 f IRs 5 IRf j 218.92
¡ 9 f 218.92 160.14 219.1 159.6 -85.7
– 76.79 These
-80
-80 harmonics
Y: 207.3
-94.03 Y: -86.91 Y:
X: are
-91.45
262.8 more
X: Y:
or
-56.04
less
Y:
X: -80.89
Y: 394.4
-81.18
readable given the
f charact 0.6 (1) N f 121.48 Hz
X: 307.2
    X: Y: -86.05
219.1 -85.7 -80 Y: -81.18
Slip f splays IR25,1
 f IRan important s 218.92
318.92 role in
b the
219.1
319.1 r calculation-85.7
-81.9 of -100
magnitude
-80 X:
Y: 207.3
of realized
-94.03 X:
Y: 262.8
-91.45
Y: -91.45fault (a small hole).
Y: -86.05

harmonics; fs   2j ffsitIR¡depends,
2 fIR j 318.92 218.92 319.1
systematically, by219.1 the variation – 85.70
-81.9 of
-100
-100
Y: -94.03

f  2 f 318.92 319.1 -81.9

-100200
-100200
250 300 350 400 450 500
the
 s
fcharact f s IR2Figure
load. f7,1
f s fs 2j fffsIR
IR
 7 f
(8)  0.6
shows
318.92  (1)
the N
+ 2 fIR j s 515.54 318.92b513.2r 319.1 -94.13

spectrum
319.1f  78.52
of the Hz
current
-81.9 500
250
550
300 Frequency
300600 Table350
350 (Hz) 400
4 (Hz) 650 400
450
700450
500
750
under
13 515.54whose slip 513.2
– 81.90
-94.13 0200 250 Frequency
Frequency (Hz)
500
13 full f s  load IR
f IR operation, 515.54
is
513.2
equal to 0.048.
-94.13
It 200 3f -n and
Theoretical
0
c) -200 3fs-nrbfr s rb
250
f practical
r
300
L-PSH Frequency
values of350
Frequency (Hz)
(Hz) 400
the harmonics IRF (IRF, 450
s = 0.048) (c)
500
13 f  f IR
shows
f charact that
17 fssIRj1,1f IR
13 13 around
s fIRs j 0.6
f ¡ f the fundamental
515.54
715.54 (1)  Nb 713.2
515.54 frequency,
f r  178.52-93.07
513.2
713.2 513.2 there
Hz
-94.13
–
-93.07 are
94.13 Fig. 8.3fStator -n f current
L-PSH
spectrum
IRF
PracticalHealthy
X: 617.6 of different frequency bands (IRF, (c)
715.54 -200 Formula Theoretical
L-PSH IRF
Healthy
Amplitude (c)
other 17 particular
f s  fIR harmonics715.54 in the 713.2 spectrum which -93.07 can  s=0.048),
3fs-n rbfr
(a) : 0-200Hz, Y: 617.6
X: -39
(b) : 200-500 Hz, (c)
IRF: U-PSH
500-750 Hz717.6
(dB)

s
X: 517.6
rb r L-PSH
Values Values [dB]X:
(c)
17 fs  j 17f IRfexistence
IR
715.54 713.2 713.2 -93.07 – 93.07 f
-20
-40   f  k  f Y: -39 Healthy
tofIRthis
Y: -48.53 X: 617.6 Y: 717.6
-48.67
(dB)

confirm According the s¡ j 715.54

table,
of IRF. we can
Thesay that the frequency
calculated values of -20charact  IR X: s 517.6 IR
f +4f[Hz] [Hz] Healthy U-PSH X:
X:
17 f  f -40 Y: 517.6
-48.53 X:
Y: 617.6
-39 Y: 717.6
-48.67
(dB)

X: s IR
Amplitude

bands
these According s
that we analyzed
harmonics
IR to this table, we can say that the frequency
are: are we likely tosaybring information on -40According f +4f Y: -39 calculations
-48.53to theoretical U-PSH X:
3fsobtained
+4f in the
(dB)

X:
Y: 517.6 Y: 717.6
-48.67
-60 fs+4fIR U-PSH
Amplitude

bands According
that we to
analyzed this table,
are likelycan to bringthat the frequency
information on -40
-60 below,X: 565.3
Y: -48.53
s IR X: 641.4 X: 665.2 3f +4fIR Y: -48.67
According to this table, we can say that the frequency table we can say that the-80.89frequencies IR obtained
Amplitude

the existence f +4f

we of 3the IR fault. s

f charact fthis 0.6
table, we(1)  Nsay f r the
b bring 21.48
 information Hz bands 21.5 3f +4fIR -64.53
Y: 641.4
-56.04X:
Y:
bands that
According ofanalyzed
to are likely
can to bring
that frequency on s 21.48
IR X: 665.2
Amplitude

-60
-803 f  f Y:
the
bands
Slip existence
that
plays
 IR 3,1 wean the
analyzed s IR
important fault.
are likely
role to
in the information
calculation on
of experimentally
-60
-80 s IRY:
-86.91
X: 565.3
-86.91 demonstrate the
Y: 641.4
X: -56.04X:
existence
Y: 665.2
-80.89 3fss+4fof
IR the IRF.
thethat
Slip existence
we havean
plays ofanalyzed
the IR fault.
important are likelyrole to generate
in the information
calculation onofthe X: 565.3 X:
Y: 641.4
-56.04X:
Y: 665.2
-80.89
the existence
harmonics; itan of
depends,the IR fault.
systematically, by the variation of of These
-80
-100 harmonics X:
Y: 565.3
-86.91
are more or less
Y: -56.04Y: -80.89
readable given
(is notthe
Slip
f  plays of
existence
harmonics; it  the 5
depends, important
IR f fault.
 0.6  role
(1) 
systematically, N in  thef
by  calculation
the121.48
variation Hz of -80
-100 5 f  f
Y: -86.91 121.48 121.5 -64.31
Slip
the load.
charact plays
Slip
 IR  Figure
5,1
plays ananimportant (8) shows
s
important role
role in calculation
thein the
spectrum
b the calculation
r of variation
the of
current
of harmonics; magnitude s of
IR realized fault (a small hole). alone, with750
harmonics;
the load. it depends,
Figure (8) shows systematically,
the spectrum by the of variation
the0.048. of
current
-100500 550 600 650 700
harmonics;
under
it load. full
depends, loadit depends,
operation,
systematically, systematically,
whose
onthe slip
the spectrumisby
variation of the
equal to
the of
It -100500 550 Frequency (Hz)
600 650 healthy state750
700
the
fcharact Figure 7 (8) f s theshows
0.6 (1)  of theload.
78.52
Fig. 8
current Frequency (Hz)
under
the
shows
showsload. fullthat
IRthe load
Figure
7,1  operation,
around
spectrum (8) of
shows whosethe N
fundamental
thewhose
current
slip isf requal
frequency,
spectrum
bunder fullofload
to Hz
0.048.
thetherecurrentIt
are
operation,
500
5008. Stator 550
550 600
600 Table 4 (Hz)
Frequency
650
650 frequency
700= -70.08) 750
700 750
under full load operation,
shows that around the fundamental frequency, there are slip is equal to 0.048. It Fig. current spectrum 78.52
Frequency of different
of the(Hz) 78.6 bands (IRF,
(IRF, s-66.2
under
other full load operation,
particular harmonics whose
in theslipspectrum
is equal to 0.048.can It Fig.fs=0.048), f IR (a)current
s8.Stator
Theoretical and practical values
spectrum
(b) :of200-500
different harmonics
frequency bandsHz(IRF, = 0.048)
shows
other that around
particular the fundamental
harmonics in the frequency,which
spectrum there can
which are Fig.s=0.048),
8. Stator(a)
: 0-200Hz,
current spectrum
(b) :of200-500
different
Hz, (c) : 500-750
frequency bandsHz(IRF,
f 
shows
confirm
other
that
 IRthe
charactparticular 
aroundf
1,1 existence 0.6
the
sharmonics (1)  N
fundamental
of IRF.
TableThe f
4b spectrum
in the r  178.52
frequency,
calculated Hz
there
which
are
valuescanof Fig.
Fig. 8.Stator
8.Formula
s=0.048), Stator
: 0-200Hz,
current
current
(a) spectrum
Theoretical
spectrum
: 0-200Hz,
178.52 :of200-500
(b) of different
Hz,
different
(c) : 500-750
frequency
Practical
Hz,
178.6frequency bands (IRF,
Amplitude
bands
(c) : 500-750 Hz
65.54 (IRF,
confirm the existence
otherTheoretical
particular of values
harmonics IRF. The
in of
the calculated
spectrum values
which of
can  fs=0.048),
According
ss= ff IR k (a)fto
0.048): theoretical
a): 0-200Hz,
Values
0‒200 Hz, calculations
(b)b): 200-500
200‒500 Hz, c) :obtained
Hz,(c)
Values 500-750
500‒750 Hzin the
[dB]
these
confirm harmonics are:
and
the existence practical harmonics (IRF, s = 0.048) fcharact 
these
 harmonics
confirm are: of
the existence
IRF. The calculated values of
of IRF. The calculated values of tableAccording
 IR
below, we
s to
IR theoretical calculations obtained in the
[Hz]
can207.04
say that [Hz]
the207.3
frequencies obtained
f charact 3are:
f s  0.6Theoretical
 (1)  N  f r  21.48 Hz According to theoretical calculations obtained in the
3,1 
these
 harmonics
 IR Formula table below, we
f s  2 f IR todemonstrate can say that the frequencies -94.03
obtained
f charact
these harmonics
 IR 3,1  f s  0.6  (1)  Nbb Practical
3are: f r  21.48 Amplitude
Hz According
experimentally
table
theoretical calculations
can say thatthe
below, wedemonstrate
experimentally thethe
obtained
existence of the
frequencies
existence
inIRF.
the
obtained
of given
the IRF.
f  ±
 3 f s fIR0.6 (1)  Nb values
j values fr  21.48Hz table
These below, we can
harmonics are say
more that
or the
lessfrequencies
readable obtained
the
ffcharact  IR3,1 jν fs35± fk£
 f charact–IR
charact

=
 IR 3,1
 0.6
fs  0.6 (1)
(1) N

[Hz] Nb  ffr 
 [Hz] 21.48Hz
121.48 Hz
[dB] experimentally
These  f IR demonstrate
3 f sharmonics
experimentally
21.48
are fault the less
more(a or
demonstrate
existence
21.5 of-64.53
readable
the existence
the IRF.
of given
the IRF.the
f 
charact  IR 5,1
 5 f  0.6  (1)  N  f  121.48Hz
s b r magnitude
These of realized
harmonics are more small
or hole).
less readable given the
magnitude of realized fault (a small hole).
j  5 j f  0.6 (1)  N  f  121.48

f charact  IR 5,1
 IR 35,1
fs ¡ fIR
s
 21.48 b
Hz
–64.53
r
21.5 These
whose
magnitude
harmonics
slip is
of equal are
to
realized
moreIt or
0.048.
121.48
fault (a
less
shows
small
readable
that
121.5
hole).around
5 f  fofIR realized fault (a small hole). alone, with
given
the
-64.31 notthe
(isfunda-
 57 ff   0.6  (1)  ff  121.48

ff 0.6 N
(1)  N  Hz
charact s b r

charact  IR 5,1 s 78.52Hzb r magnitude
mental sfrequency, there areTable
other4 particular harmonics in the
f 
charact  IR 7,1
 7 f  0.6
s
j  7j f  0.6 121.48
 IR 57,1
fs ¡ fIR  (1)  N  f  78.52 Hz
b r
121.5
–64.31 (if not Table 4 harmonics (IRF, healthy state
f 
charact s
(1)  N  f  78.52 Hzwith
b r Theoretical
spectrum which and practical values
can confirm of the
the
Table 4existence of IRF. = 0.048)
sThe calcu-
ff

charact

charact
 IR 7,1
 IR 7,1  7f f 0.60.6
s
s  (1)  N  f  78.52
 (1)  Nb  f  178.52 Hz
alone,
b
Hz
b
r
r
Theoretical and practical values
lated values
Theoretical of
and these harmonics
practical
of the
Table
values of 4
are:
the
harmonics (IRF,=s-70.08)
harmonics (IRF, s
= 0.048)
= 0.048)
Formula Theoretical
78.52 Practical
78.6 Amplitude
-66.2
f 
charact  IR 1,1
 f  0.6  (1)  Nb  f  178.52
s
Hz state
healthy r
f s  f IRand practical
Theoretical
 Formula
values of thePractical
Theoretical harmonics (IRF, s = 0.048)
Amplitude
f 
charact  IR 1,1
 f  0.6  (1)  Nb  f  178.52
s
= Hz
r
–70.08) 6 fcharact –  f  k  f
Formula
Values
Theoretical
IR = j 3Values
¡ 0.6£(1)£N
Values
Practical
[dB]
Amplitude
b £ fr j = 21.48 Hz
f 
charact  IR 1,1
charact  IR 1,1  f  0.6  (1)  Nb  f  178.52Hz
s
s
r
r
 f IR
fcharact
Formula  f s
 k
charact–IR–3,1  f fs[Hz]
Theoretical
178.52
Values
Values
[Hz]
Practical
178.6
Values
[dB]
Amplitude
65.54
[dB]
 fsIR– fsIR
  IR s IR
j fs ¡ fIR j 078.52 078.6 –66.20 fcharact  k  f IR [Hz]
Values
[Hz]
Values [dB]
IR  f s  k  f IR = j 5 fs[Hz]
fcharactfcharact–IR–5,1 ¡ 0.6£(1)£N[Hz] b £ fr j = 121.48 Hz
[Hz] [Hz]
207.04
21.48 207.3
21.5 -94.03
-64.53
j fs + fIR j 178.52 178.6 –65.54 3f sf–s2 fIRIR 21.48 21.5 -64.53 = j 7 fs ¡ 0.6£(1)£Nb £ fr j = 78.52 Hz
3ffcharact–IR–5,1
f 21.48 21.5 -64.53
j fs ¡ 2 fIR j 207.04 207.3 –94.03 3 f+ss  f IR 21.48
121.48 21.5
121.5 -64.53
-64.31 (is not
53ffcharact–IR–1,1
s  f IR
IR = j f 0.6£(1)£N
121.48
+ 121.5
£ f j = 178.52
alone,Hz.
-64.31 (is not
j fs + 2 fIR j 307.04 307.2 –86.05 5 f s  f IR s IR
121.48
s
121.5
b
-64.31
r
alone,(is not
with
5 f s  f IR 121.48 121.5 healthy
-64.31
with
state
(is not
5 f 
According
s f IR to theoretical calculations obtained
alone,
healthy with
state
= -70.08)
alone, in
with the
j 23 fs ¡ 11 fIR j 263.73 262.8 –91.45 healthy state
= -70.08)
78.52 78.6 -66.2state
healthy
Tablef4,  f
we can say that the
78.52 frequencies
78.6 obtained =experimen-
-70.08)
= -66.2
tally ffdemonstrate
s  f IR
s IR -70.08)
j fs + 4 fIR j 564.08 565.3 –86.91 the existence
78.52 of the IRF. These
78.6 harmonics
-66.2
 f 78.52
178.52 78.6
178.6 -66.2
65.54
664.08 665.2 –80.40
f
are more s  f IR
s orIR less readable given the magnitude
178.52 178.6 of realized
65.54 fault
j 3 fs + 4 fIR j f s hole).
6 (a small f 178.52 178.6 65.54
f  f IR 178.52
207.04 178.6
207.3 65.54
-94.03
ffsss 2ffIR 207.04 207.3 -94.03
f s  2 fIRIR 207.04 207.3 -94.03
f s  2 f IR 207.04 207.3 -94.03
576 f s  2 f IRIR Bull. Pol. Ac.: Tech. 67(3) 2019
 664.08 665.2 -80.4 sa
3 f s  4 f IR
We used in this specialized study the spectral analysis
method f s of2 fthe stator307.04 307.2
current (MCSA-FFT) -86.05
in order to The theorem of Shannon is applied in order to determine
IR Level : j=1of a1 : 0- fsa/4
Mechanical fault detection
detect the ORF and the IRF in REB of the SCIM. The in rotating electrical the
machines frequency
using bands
MCSA-FFT and approximations
MCSA-DWT techniques and details.
d1 : fThe
sa/4-fs/2

23 f  11 f 263.73 262.8 -91.45 figure 10 shows the levels of each band.

results thats have IR been obtained are sufficiently acceptable
for the analysis of the two faults in the REB (ORF and
In this specialized study, 564.08 we have 565.3
used the spectral analysis
-86.91 Original Signal
IRF). fThe signatures are clear along the stator current
method s  of4the
f IR stator current (MCSA-FFT) in order to detect s(t)
spectrum, and especially around the fundamental Level : j=2 a2 : 0- fsa/8 d2fsa: fsa/8-fs/4
the ORF and IRF in REB of the SCIM.
664.08 665.2The results that have
-80.4
3 f s  4 f IR
frequency.
been obtained are sufficiently acceptable for analysis of the
two We
faults in the
used in REB (ORF and IRF).
this specialized studyThe thesignatures
spectral analysisare clear
4.along
MCSA-DWT
method theofstator statoranalysis
the current spectrum,
current and especiallyinaround
(MCSA-FFT) order theto Fig. 10. Decomposition scheme of frequency bands
fundamental frequency. Level : j=1 a1 : 0- fsa/4 d1 : fsa/4-fs/2
detect the ORF and the IRF in REB of the SCIM. The
The signature of a system is
results that have been obtained are sufficiently acceptable the recorded trace as a Higher level decompositions can be obtained in by basing
function of time of
for the analysis of thethese twocharacteristics.
faults in the REB This(ORF signature
and on theorem of Shannon. The frequency band of each level
must
4.
IRF). beThe
MCSA-DWT processed
signatures in order
analysis
are clear to extract
along the the characteristics
stator current
j returns  f 
that make it possible
spectrum, and especially to define around the situation the of the system Level
fundamental : j=2 a2to: 0- fsa 0/8 saj  for approximation d2 : fsa/8-fs/4 bands, and
atThe
time t.
signature
frequency. of a system is the recorded trace as a function of  2 
time of these techniques
Various characteristics. haveThisbeen signature must beusing
exploited processedthe  f sa f 
in order to extract the characteristics that make it possible to   jsa1  for detail bands. We can stop the operation
acoustic analysis,
4. MCSA-DWT analysis the electromagnetic field, the analysis of 2
j
2 10. Decomposition scheme of frequency bands
define the situation Fig.10.
the motor current, of thethe system voltage,
induced at time t. the instantaneous
of
Fig.
decomposition
Decomposition scheme of frequency bands
by the following number:
power,Variousthe techniques
vibration, have
etc. been exploited,
The signature of a system is the recorded trace as a using acoustic anal-
ysis, thethe electromagnetic field, analysis of in
theThis
motor current, Higher level decompositions can  be obtained f  in by basing
Among
function ofmost timerecent
of these techniques used
characteristics. fault detection
signature on theorem of Shannon. The logfsa sa band
frequency
  of each level
isinduced
must bevoltage,
the discrete processed instantaneous
wavelet in transform
order to power, vibration,
(DWT).
extract etc.
theIncharacteristics
addition, this log  f  fs  
thatDiscrete
techniquemake is itwavelet
very transform
possible to define
efficient (DWT) is among of
the situation
and greatly thethemost
enhances recent
system the j returns to  0N
 Nf   int
LL for approximation
sa s
log  2 (7)  bands, and(7)
LL = j int 
techniques
at timeof
quality used
t. our in fault
work because detection . In addition,
the current sensors thisare technique
easy to  2  
log(2) 
is very efficient
Various and and
techniques greatly enhances the quality of our work 
 
implement do nothave requirebeen expensive
exploited using additionalthe  f sa f 
 jsa1  for detail bands. We can stop the operation
because the current sensors are easy
acoustic analysis, the electromagnetic field, the analysis of to implement and do not where,
transducers. j
 2 f 2is the  sampling frequency, and f is the fundamental
require
theInmotor expensive
current,additional
the willinduced transducers.
voltage, the instantaneous where, sa s
In
this
this
section,
section, we
we
will apply
apply
the
the
DWT
DWT
technique
technique
by
by
exploit- of decomposition
frequency. by the following number:
power, thethe
exploiting vibration, etc.
stator current signal. So, the MCSA-DWT fsa is the sampling frequency, and fs is the fundamental
ing
Amongthe stator
the mostcurrent
recentsignal. So,
techniques the MCSA-DWT
used in fault technique
detection [18] proposed a differentnumber
frequency.  f sa of   levels for adequate
technique
is used in isorder
usedtoinanalyze
order totheanalyze REB the REB
faults. faults.
Discrete wavelet analysis; they used the following  logformula:
 
is the discretewavelet wavelettransform
transform is(DWT). In addition, this   s f
The discrete
transform is isconsidered to be among considered among the N  int (7)
technique
best signal very efficient
processing techniques. and the best signal
greatly enhances processingthe LL  log  2  
techniques. f
quality
DWT of our
allows workus because the current sensors
to decompose themeans are easy
signal bytoa [18] proposed a different log  number sa  levels for adequate
of
DWT allows us to decompose the signal by of succes-  fs 
implement and do not require expensive additional analysis; they N used int the following+formula: 1 or 2(8)
successive
sive transition from a high-pass
transition from a high-pass filter (HP) and low pass filter filter (HP) and low where, LL =
transducers. log(2)
pass filter
(LP).InAfter (LP). After this operation, we get two   f  
this this operation,
section, we will we get applytwothe coefficients:
DWT technique the approxi- by  log 
sa

coefficients:
mate a and the the approximate
detail d . a1 is the i and the detail
aapproximate shape dofi. a1
the is the fsa is the sampling frequency,
origi-
exploiting the stator current signal. So, the MCSA-DWT
i i
 N int   s  fs is1 the
and
f
or
fundamental
2this stopping. (8)
approximate
nal signal without shape of and
noise, thed1original
is the signalshape
detailed without of the noise,
signal [19] has
frequency. explained the
LL reasons for
 log  2   the choice of
technique is used in order to analyze the REB faults.
and
Thed1
that is the wavelet
influences
discrete detailed
the original shapesignal
transform of(noise).
the signal thatamong
Successive
is considered influences
operation
the For fsa = 12.8 k Hz, we find  the advisable  number of decom-
onof original signal
this decomposition (noise). The
is represented successive operation
in the following of
diagram:this positions:  
best signal processing techniques.
decomposition
DWT allows may us represent in the following
to decompose the signal diagram: by a [18]
[19]proposed
has been aexplaineddifferent number
the
12.8£10 reasons
3
of of levels for adequate
the choice of this
log
successive transition from a high-pass filter (HP) and low analysis;
stopping. they used the following50 formula:
a2 NLs = int + 1 = 9 levels.
pass filter (LP). After this operation, we
2 get two For fsa=12.8 kHz; we log(2) find
 f sa the  advisable number of
coefficients: the approximateaa1 i and the detail di. a1 is the
LP
decompositions:  log  
  fs  
approximate shapeLP of the original signal without noise,
2   N12.8
LL  10 int  
3  1 or 2 (8)
Signal s(t)
and d1 is the detailed shape of theHP signal that influences
We summarize  log the frequency
  logbands

 2  in the Table 5.
2   50 
on original
Sampling rate fsasignal (noise). The successive operation d2 of this N Ls int      1 9 levels 

log  2   Table 5
decomposition may HP represent 2 in the following diagram:
[19] has been explained the reasons of the choice of this
d1  Details frequency  bands ( fsa = 12.8 kHz)
stopping. 
Fig. a2 We fsummarize
For Frequencythe frequency
bands of decomposition
sa=12.8 kHz; we find the advisable number of
bands inlevels this (Hz)
table.
Fig. 9. DWT decomposition
9. DWT decomposition process
LP process 2
a1 decompositions: d1 3200–6400
2  3 
LP  12.8 d10 1600–3200
Signal s(t)  log  2
 
The Shannon theorem is applied in order
HP to determine the fre-
2   50 d3    1 9 levels 800–1600
Sampling rate fsa d2  N Ls int   
quency bands of approximations
HP 2
and details. Figure 10 shows 
log  2
d4  400–800
the levels of each band. d1  
Higher level decompositions can be obtained based on the 5 d 200–400
Fig. 9. DWT decomposition process We summarize the frequency bands in this table.
Shannon theorem. The frequency band of each level j returns to 7 d 6 100–200
f f f d7 50–100
0 → saj for approximation bands, and saj → jsa¡ 1 for detail
2 2 2 d8 25–50
bands. We can stop the process of decomposition by using the
d9 12.5–25
following number:

Bull. Pol. Ac.: Tech. 67(3) 2019 7 577

 d9 12.5 - 25 multi-decomposition under ORF for different conditions.
The calculation of the relative energy associated with each
The fault indicator can be found in any level. The level of decomposition allows us to differentiate also the
frequency bands (BF) of each di are considered an healthy
N. Bessous, S. Sbaa, and and faulty IM. Figures (11) and (14) show the
A.C. Megherbi
orientation to the important harmonic values. energy variation of both motors (healthy and faulty state,
In practice, the choice of the wavelet is not at no-load and at load operation).
crucial. Indeed one seeks
The fault indicator a wavelet
can be which
found at any offers
level. a good
The frequency 10
compromise between the temporal and frequency
bands (FB) of each di are considered an orientation to the
resolution,
importantsuch a choice
harmonic depends often on the objective of
values.

is(t)
0
the proposed subject.
In practice, the choice ofAdditional
the waveletcriteria suchIndeed
is not crucial. as
regularity,
one seekssymmetry, a rapid
a wavelet which decrease
offers a good at infinity may
compromise be
between -10
necessary.
a temporal and frequency resolution, and such a choice often 1 2 3 4 5 6 7 8 9 10
We choice the Daubechies mother (Db44) for all 10
depends on the objective of the proposed subject. Additional
analyzes
criteria concerning this application
such as regularity, symmetry andpartrapid
[19].
decrease at infin-

is(t)
ity may be necessary. 0
4.1. REB faults
We have detection
chosen by Energy-DWT.
the Daubechies The(Db44)
mother wavelet energyfor
variation of the
all analyzes details this
concerning has application
very goodpart
indicator
[19]. for the -10
1 2 3 4 5 6 7 8 9 10
SCIM faults. Our study have based on the energy analysis 0.5
of 4.1.
eachREB faults
detail. Wedetection by Energy-DWT.
have verified the detail The energy
which hasvari-
a
ationinfluence
strong of the details hasstator
on the a verycurrent
good indicator
signal. for SCIM faults.
0 0.18

a9
TheOur study of
energy hasthe
based on dj
detail theisenergy analysis
calculated by: of each detail. We 0.16
0.14
have verified the detail which N has a strong
2 influence on the 3 4 5 6 7
stator current signal. E j   d j ( n) (9) -0.5
0 2 4 6 8 10
The energy of detail d j isn calculated
1 by: 0.5
0.02
N is the total number of samples in the signal and j is 0
N -0.02
the level of detail. 2
E j = ∑ j dj (n)j .(9) 0
6.5 7 7.5 8 8.5
d9
We can calculate the differentn =1
energies from d1 to d9. We
have made a comparison between the two cases: healthy
and faulty. Any
N is the change
total numberindicates the existence
of samples of and
in the signal the REB
j is the -0.5
0 2 4 6 8 10
level This
fault. of detail.
analysis led us to select the interesting 5
We can
frequency calculate the different energies from d1 to d9. We
band.
have made a comparison between the two cases: healthy and 0
d8

faulty.
4.2. REBAny change
faults indicatesby
detection theMSE-DWT.
existence of the REB fault.
Among the
This analysis
methods led us to select
of distinguishing the frequency
between band
signals that analysis
is the is of inter- -5
byest
thetomean
us. square error estimator 'MSE'. This estimator
10
0 2 4 6 8 10

can be used as a quality indicator to give a percentage of

4.2. REB faults
resemblance detection
between by MSE-DWT.
X1 signal and another Among the methods
X2 signal. Our
of distinguishing between signals is the analysis by the mean 0
d7

new approach is based on a study of the correspondence

squarebetween
degree error estimator
details(MSE).
in order This
to estimator
determine cana be usedand
well as a
quality indicator to give a percentage of resemblance between -10
accurate calculated value (not per vision). 0 2 4 6 8 10
Time (s)
the X1 signal and another X2 signal. Our new approach is based Time (s)
The coefficient „MSEs‟ of resemblance for two signals Fig. 11. Multi-level decomposition of the stator current at no-load
on a study of the correspondence degree between details in
operation: Healthy (in blue), ORF (in blue)
and Xto2 determine
X1order (in each position which
a correcti)and havecalculated
accurate point numbers n
value (not Fig. 11. Multi-level decomposition of stator current at no-load opera-
per vision). tion: Healthy (in blue), ORF (in red)
is calculated
The MSE bycoefficient
the following formula: for two signals X and
of resemblance 1
X2 (in each position i) which have point numbers n is calculated
by the following formula: 5
x 10
n 2 15 4
∑ i =1(X1, i ¡ X2, i) Healthy
MSE = .(10) ORF
n
8 x 10
4
3
3
This indicator was used for details di to determine the fre- 10
Energy (J)

quency band on the one hand, and for the determination of the
Energy (J)

2
dominant detail in the new signal. Our fault analysis is based 2
on the MSE for detail signals or approximations. The resem-
blance of the signals is checked for an MSE value that tends 5 1

towards zero. 1
0
1 2 3 4 5 6
4.3. Outer raceway fault detection by DWT. Figures 11‒14
represent the detail and approximation signals (d7, d8, d9 and 0 0
1 2 3 4 5 6 7 8 9
a9) obtained by means of multi-decomposition under ORF for Levels
Levels of
of details
details
different conditions. The calculation of the relative energy asso- Fig. 12. Energy evolution at no-load operation (ORF)
ciated with each level of decomposition allows us to differenti- Fig. 12. Energy evolution at no-load operation (ORF)
10
obta
578 0 Bull. Pol. Ac.: Tech. 67(3) 2019 amp
is(t)

sens
-10 amp
0 2 4 6 8 10
10
that
 10 Compare the detail and approximation signals
obtained for machines with defects, we see that the
0 amplitude of the coefficients d7, d8 and d5 is more
is(t) Mechanical fault detection in rotating electrical machines using MCSA-FFT and MCSA-DWT techniques
sensitive compared to the others. This variation of the
-10 amplitude in the signals d5, d7 and d8 is due to the effect
0 2 4 6 8 10
10
that the corresponding frequency bands are affected by the
of coefficients d7, d8 and d5 is more sensitive as compared to
defect.
the others. This variation of the amplitude in signals d5, d7 and
Now, we make the fault detection by the
0 d8 is due to the fact that the corresponding frequency bands are
is(t)

observation and/or the comparison between the

affected by the defect.
decomposition levels. These details contain much
Now, we turn to the fault detection by means of observation
-10 information about the state of the machine to diagnose.
0 2 4 6 8 10 and/or comparison between decomposition levels. These details
0.5 Our proposal focuses on the calculation of
contain a massive amount of information about the statethe MSEofvalue.
the
In addition,
machine to be it aims to avoid
diagnosed. any opposition
Our proposal focuses onor calculation
overlapping
0 0.16 ofofthehuman opinions
MSE value. on theit aims
In addition, visualto evolution of signals.
avoid any opposition
a9

0.14
0.12 or overlapping of human opinions on the visual evolutioninofthe
Tables below show the calculation of MSE values
-0.5
4 4.5 5 5.5 6 presence
signals. Theoftables
the ORF.
below show the calculation of MSE values
0 2 4 6 8 10
0.5
in the presence of ORF.
0.04
0.02 Table 6
-0.020
-0.04
4 4.5 5 5.5 6 Table 6
0 MSE values at different decomposition levels (ORF,
d9

MSE values at different decomposition levels (ORF, 0£TN)

0×TN)
-0.5
MSE(di,
MSE (di,a9)
a9) MSE
MSE Value
value Ressemblance
Resemblance
0 2 4 6 8 10 -05
MSE-d1
MSE-d1 3.4938×10
03.4938£10 ‒05 Almost
Almost present
present
10
MSE-d2 5.5600×10-05 Almost present
MSE-d2 05.5600£10 ‒05 Almost present
0 MSE-d3 0.0028 Less absence
d8

MSE-d3 00.0028 Less absence

MSE-d4 0.0077 Less absence
-10 MSE-d4 00.0077 Less absence
0 2 4 6 8 10 MSE-d5 0.2329 Absence
10 MSE-d5 00.23290.0056 Absence
MSE-d6 Less absence

0 MSE-d6
MSE-d7 00.005624.9483 Less Great
absence
absence
d7

MSE-d8
MSE-d7 24.948319.2467 GreatGreat absence
absence
-10
0 2 4 6 8 10 MSE-d9
MSE-d8 19.24670.0035 GreatLess absence
absence
Time (s)
Time (s) MSE-a9 0.0068 Less absence
Fig. 13. Multi-level decomposition of the stator current at load operation: MSE-d9 00.0035 Less absence
Fig. 13. Multi-levelHealthy (in blue),ofORF
decomposition (incurrent
stator blue) at load operation:
Healthy (in blue), ORF (in red) MSE-a9 00.0068 Less absence
Table 7
MSE values at different decomposition levels (ORF,
6
TN×3/4)
Table 7
x 10 MSE values
MSE (di, at
a9)different decomposition
MSE value levels (ORF, TN£3/4)
Ressemblance
4
Healthy MSE-d1
MSE (di, a9) 3.2868×10
MSE
-05
value Almost present
Resemblance
ORF
3 x 10
4 MSE-d1 03.2868£10 ‒05 Almost present
2.5
‒05
MSE-d2 05.8192£10 Almost present
Energy (J)

2
2 MSE-d3 00.0057 Less absence
1.5

1
MSE-d4 00.0203 Absence

1 9 MSE-d5 00.1741 Absence

0.5

0 MSE-d6 00.0066 Less absence

1 2 3 4 5 6
0 MSE-d7 24.0351 Great absence
9 1 2 3 4 5 6 7 8 9
Levels
Levels of
of details
details MSE-d8 51.1002 Great absence
Fig. 14. Energy evolution at load operation (ORF)
Fig. 14. Energy evolution at load operation (ORF) MSE-d9 00 Present
Compare the detail and approximation signals MSE-a9 00 Present
obtained for machines with defects, we see that the
ate also between the healthy and faulty IM. Figures 11 and 14
amplitude of the coefficients d7, d8 and d5 is more
show the energy variation of both motors (healthy and faulty The representation of the mean square error evolution
sensitive compared to the others. This variation of the
state, at no-load and at load operation). for the two types of operations is illustrated in Fig. 15. From
amplitude in the signals
When comparing the d5, d7 and
detail and d8 is due to thesignals
approximation effect Table 7, we see that the detail d9 and the approximation a9 have
10
that the corresponding frequency bands are affected by
obtained for machines with defects, we see that the amplitude the an excellent similarity (MSE = 0). This value of the MSE = 0
defect.
Now, we make the fault detection by the
observation and/or
Bull. Pol. Ac.: Tech. the comparison between the
67(3) 2019 579
decomposition levels. These details contain much
10 information about the state of the machine to diagnose.
Our proposal focuses on the calculation of the MSE value.
In addition, it aims to avoid any opposition or overlapping
 MSE-d4 0.0203 correspond
Absence to the confirms
correspond a decrease
frequencies
tocorrespond
the frequencies
around ofobserved
to thearound
frequencies
50Hz. inaround
of 50Hz. detailofsignals
50Hz.d5 and d8.
Calculating the energy stored in each level
-d7
-d9MSE-d7
MSE-d9 24.0351
MSE-d7
0 24.0351
MSE-d5
MSE-d9 0 Great
24.0351
absence
0 Great
0.1741
Present only
absence Great
Present Absence (19).
onlyThis
absence
Present (19).confirms
variation
This
onlyvariation
(19).
leadsThis
us
leads
tovariation
a decrease detect
us to detect
the
observedleads
IRin
the
fault
usdetail
to
IRand
detect
fault and
the IR
signals fault
d5In and and
d8.
MSE-d5 0.1741 Absenceto
10
specify
to
10
specify
the But, 10 we
frequency
the
to specifyhave
frequency
band.
the seen
band.
The an
frequency
d7
The increase
and d7
band.
d8and in
bands
The
d8 d7
bands
d7 (17).
and d8 load
bands
-d8
-a9MSE-d8
MSE-a9 51.1002
MSE-d8
0 51.1002
MSE-a9 0 Great
51.1002
absence
0 Great
Present absence Less
Present Great absence
Present But, we have
operation, the seen an increase in d7 (17).level
In d8load
MSE-d6 0.0066 absence
correspondcorrespond
to the frequencies
tocorrespond toenergy
the frequencies
around of increase
thearound
frequencies
50Hz. was
of 50Hz. found
around at detail
of 50Hz.
MSE-d6 MSE-d9 0.0066 Present Less absence operation,
only 10(19).
the
This variation leads us to detect the IR fault andd8
energy increase was found at detail level

is(t)

is(t)

is(t)
-d9MSE-d9
he representation 0
The representation
The
MSE-d7 0
of representation
theofmean 0
the24.0351
mean
square Present
of square
the
errormean
error
Great Present
0
square 0
10error 0
N. absence
10
Bessous, S. Sbaa, and A.C. Megherbi
for
-a9 theMSE-d7
ution for
MSE-a9two
evolution
thetypes
two0 of
types 24.0351
foroperations
MSE-a9
MSE-d8 the
0 oftwo
operations
types
isPresent
illustrated Great
0 ofisPresent
51.1002 operations absence
illustrated
in Great illustrated in only
isinPresent
absence to (19).
specifyThis
thevariation
frequency leads us to
band. detect
The d7 andthe IR
d8 fault
bandsand
021and to specify the frequency band. The d7 and d89 bands
error 32 correspond 54 2to the 65 3frequencies 875around986 of10950Hz.
).
re From
(15). table
From
figure (7),
table
(15).we -10 -10 -10
of(7),
From
see we that
table
seethethat
(7),
detail
the
we d9detail
see and
that
d9error
the
anddetail
0 1 d9

is(t)

is(t)

is(t)
he MSE-d8
representation
The representation
The 51.1002
representation
the of mean
the mean
square Great absence 4031
MSE-d9 0 of square
the
error mean square
Present correspond to the
764
frequencies around of
7 108
50Hz.
10
oximation
for approximation
utiontheMSE-d9
for
twothe a9 approximation
indicates
evolution
thetypes have
two of a9
that
types have
an
there
foroperations
theoftwoexcellent
an
will a9be
operationsexcellent
havesimilarity
difficultyan insimilarity
excellent
distinguishing10 similarity
between
10 1010
MSE-a9 0 types
is illustrated
ofisoperations
0 that
illustrated
inPresentisinillustrated in
Presentthat 10 -10
SE=0).
).reThis
From
(15).value
This(MSE=0).
the of
table
From
figurevaluethewe
following
(7),
table
(15). of
MSE=0
This
the
(7),two
From
see wevalue
MSE=0
indicates
signals:
that
table
see of
thethatthe
indicates
(7),
detail
the
weMSE=0
d9 there
detail
see that
and
that indicates
d9 there
and-10
the detail -10there
d9
1 21and 32 43 1 54 2 65 3 76 4 87 5 98 6 109 7 108 9 10
aximationMSE-a9
bedifficulty
a the – d9-healthy
will
difficulty
in
be VS
distinguishing 0
d9-with
a inrepresentation
difficulty
distinguishing ORF;
inbetween
distinguishing
between the Present
the between 0 the

is(t)
0
approximation The
a9 approximation
have
a9 have
an excellent
ana9 excellentof
have the
similarity mean
an similarity
excellent square error
100 similarity

is(t)

is(t)

is(t)
10 100
owing
two –signals:
a9-healthy VS a9-with ORF.
Thissignals:
two following two signals:

is(t)
evolution thefor the two types ofthat
operations is illustrated in 0
E=0). valueThe
This(MSE=0).
of representation
value ofMSE=0
Thisthe value
MSE=0 of the
indicates the mean
indicates
MSE=0there square
thatindicates
there error
that there
hy-healthy
VS d9-with
VS
-
a figured9-healthy
d9-with
ORF;
(15). ORF;
aFromVS d9-with
tableof(7), ORF;we see that thethe detail the 32 -10
avolution
bedifficultywill
difficulty
for thein
be distinguishing
two indifficulty
distinguishing
types inbetween
distinguishing
operations betweenthe
is illustrated 0 1 d9
between
-10 in02and
-10 -10

is(t)

is(t)

is(t)
-healthy
y VS a9-with
VS
- a9-healthy
a9-with
ORF. ORF.
VS a9-with ORF. 1 40131 5242 6353 7464 8575 9686 797
10 88
10 99 1010
owing
two signals:
two
gure (15). following
signals:
the From two
approximation
table signals:
(7), a9
we have
see an the
that excellent
detail d9
0.5similarity
and 0.5 -10 10 0.5
a) At 0% of the load 1 2 3 4 5 6 7 8 9 10
-healthy
he y VS d9-with VS
- d9-healthy
(MSE=0). d9-with
ORF; ThisORF;
VS d9-with
a9value of theORF;
anMSE=0 indicates -10 that-10 there 10 -10
25 approximation 25 have excellent similarity
1 21
-healthy
y VS a9-with
MSE=0). VS
- a9-healthy
will a9-with
ThisORF.
be a ORF.
value VS a9-withinORF.
difficulty
of0%the MSE=0 distinguishing
indicates that between the0.1632 0.16 43 1 54 2 65 3 76 4 87 5 98 6 1097 108 9 10
0.50there

is(t)
0.50 0.14 0 0 0.16

a9

a9

a9
a): At(a): 0% At
of the
0% load
of(a):
the At
load of the load 0.140.5 0.14
will
20 befollowing
a 2520 two signals:
difficulty in distinguishing between the 0.12 00.12 0.12

is(t)
25
- d9-healthy
ollowing two signals: VS d9-with ORF; -0.5
4
-10 4.5
4 4.55 45.5
5 4.5
5.5
6 56 5.5 6
0 0 -0.5 0 00.162 0.16-0.5 00 4 2 0.16

a9

a9

a9
1 3 4 5 6 7 8
10 8 9 1010
0.2- At
a9-healthy 0.2VS a9-with ORF.
): 2 4 2 6 6 4 8 8 6 10
15 .2 At(a):
0%
d9-healthy 20
20 VS
of the
0% load
15
d9-with ORF;
of(a):
the At
load 0% of the load 0.14 0.14 0.14
0.5 0.12 -100.12 0.12
MSE MSE

0.5 0.5
0.5
4 1 4.5 4 2 4.5 5 3 45.5 5 4 4.5 5.565 56 6 5.57 68 9 10
a9-healthy
10
.1 0.1
VS
2510
a9-with
0.1
ORF. -0.5 -0.5 -0.5
15 15 0 0 2 0.5 2 0 4 4 2 6 6 4 8 8 6 10 10 8 10
.2 0.2 0.2
0.50 0.50 00 0.16
d9

d9

d9
0.5

a9
(a): At 0% of the load 0.01 0.01
0 -0.01 0 0.01
0.14-0.010
0 25 0
5 20 5 0 -0.01
-0.02 -0.02 0.12-0.02
10
.1 0 10.1 02 1310 24 0.1
35 0 46 1 5 2 6 3 4 5 6 5 0.16
5.5
5
0 -0.5 4 5.5
6 4.56.5
56 55.5
6.5 6
5.5 6.5
6

a9
-0.5 -0.5
(a): At 0% of the load 00 00 -0.52 0 0 4 0.14 4 0.01
d9

d9

d9
0 20 0 2
0.01 0.01 2 6 6 4 8 8 6 10 10 8 10
0 00 0.12 2 0 4 6 8 10
05 1
2 03 2 15 5 1 4 0.2 0 2
3 4
5 3
5
6 4
6
7 5
7
8 6
89 79 5 8 5 -0.01 -0.02
9-0.02 -0.01 -0.01
-0.02
0.55 5
MSE

0 1 02 13 24 Details 35 0 46 1Details
5 2 6 3 4 Details 5 6 5.5
5 4 5.5 6 4.56.5
5
6 55.5
6.5 5.5
6 6.56
Details -0.5 -0.5 -0.5 -0.5
60 02 0 4 24 2 6 64 4 8 8 66 10 10 88 1010
0 15 0.21060
0 0.1 0 0 2
MSE

b):21 At(b):
75%32 Atof75%
the
431 load
of
(b):the
542 Atload
75%653 of the 764 load875 869 79 50 8 50 9 0.5 050
d8

d8

d8
50 b)Details
At 75% of the loadDetails
d9
50 Details 0.010
-0.01
60 10 0.2 0.1 560
0.2
0
0.2
0 1 2 3 4 5 6
-0.02
-5 -5 -5 5 5.5 6 6.5
b): At(b):
40 75%Atof75% the load
40 of
(b):
theAtload 75% of the load 0-0.5
d9

00 00 2000 4
d8

d8

d8

2 4 22 6
0.01 6 44 8 86 6 10 108 8 1010
50 5 0
0 050 10 10 10 -0.01
-0.02
MSE MSE

30 1 1 2
030 2 3 34 54 6 5 6 7 8 9 5
.2
0.1 0.2
0.1 0.2
0.1 Details 5 5.5 6 6.5
40 40 -5 -5 -0.5 -5
6020 00 00 2 02 00 4 10 8 101042 2 6 64 4 8 8 66 10
20 0
d7

d7

d7

30 1 230 (b): 3 At 75% 4 of the 5 load 6 7 810 5 0

10 910
d8

.10
10 0.10 5010 0.10 Details
-10 -10 -10
20 060 1 02 120 3 24 350 461 5 2 6 3 4 5 6 00 00 2 200 4 4 2 6 6 4 8 8 6 10 10 8 10
d7

d7

d7

0.2 Time (s)Time (s) Time (s)

0 (b):40At0 75% of the load 0 -50
d8

1
2 2
3 13
4 245 356 467 578 689
79 Fig. 16.8 Fig. 9 Multi-level 2 4 6 8 10
100 50 0 10 0 -10Multi-level
16. decomposition
Fig. decomposition
16. Multi-level
of the stator
of
decomposition
the
current
stator at
current
no-load
of the
at stator
no-loadcurrent at no-load
13 24 Details
350 461Details
5 2 6 3 4 Details -10 -10
10
MSE

0 1 02 30 5 6 0 0 operation:
2 2 0 4 Healthy
operation:
Healthy 4operation:
(in blue),2(inIRF
6 blue),
(in 6 4 (in
Healthy
blue)
IRF 8(in
blue) 8 IRF
blue), 6 10(in blue)
10 8 10
0.2 MSE
0 40 Fig. 15. Fig. 0 15. 0.1MSE evolution
evolution Fig.
for ORF 15. for
MSE ORF
evolution for ORF
Time6 (s)Time (s) Time (s)
6 -5 6
21 32 4
1
3 5
2
4 6
35 7
46 8
5
7 7
9 Fig.2.59
6
8 8 xFig.
10 9 x 10 0
2.5
x 102 4stator at 6 stator current8at no-load 10
20 DetailsDetails Details 16. Multi-level
16. Multi-level 0Fig.2.5
decomposition
decomposition
16. Multi-level
of the stator
of
decomposition
thecurrent current
no-load
of the
at no-load
d7

Healthy
10 Healthy Healthy
MSE

r Inner
raceway
30 4.4.
raceway
fault
Innerfault
detection
raceway
detection
byfault Details
DWT.
bydetection
DWT.
The The by DWT. The operation: operation:
Healthy Healthy
(in blue),
operation:
(inIRF
blue),
(in
Healthy
blue)
IRF (in(inblue)
blue), IRF (in blue)
Fig. 15. Fig.
MSE15.
0.110 evolution
0MSE evolution
for
Fig.ORF
15. for
MSE ORFevolution for ORF 10 2 xIRF 10 -10 IRF 2 x 10 IRF
6 6 6
action
of information
of extraction
information
from of the
from
information
stator
the current
stator
fromcurrent
inthe
steady
stator
in steady
current in 2 xsteady
0 1 Fig. 2 15.3 MSE4 evolution
5 6 for ORF 2.5 2.5 4 4
0 2.5 24 4 6 8 10
20generally x 10 0x 10 Healthy x 10
d7

ernerally
is
Inner done
racewaystate
4.4.
raceway
0 in
is
done
fault load
generally
Inner in
faultoperation.
load
detection
raceway done
operation.
detection
by in
This
fault load
DWT.
bychoice
This
operation.
choice
detection
DWT.
Theis byThis
is
The DWT. choice 3 is Healthy
The 3 3 Healthy Time (s)
Time (s)
the1 of 2 difficulty
3 of the 4 difficulty
5 no-load6 no-load
7 8 9 IRF Fig. IRF16. IRF
(J)

(J)

(J)

essary
because
action
of 10 because
ofof
necessary
information
extractiondifficulty
information
0
the
from because
ofthe
fromencountered
information
stator encountered
the current
statorfrominin
current
the
Details encountered
inin
steady
stator currentinin2no-load
steady
1.5
steady
1.524 1.52 Multi-level decomposition of the stator current at no-load
-10
Energie Energie

Energie Energie

Energie Energie

eration
inisorderindone
operation
to
order
4.4.0extract
to
1 extract
in
2theorder
3information
the4to information
extract
5 6concerningthe concerning
information isTheconcerning
4
2 x 10 2 x 104
2 x 10
operation: Healthy
nerallygenerally
state in
is
done
Innerload
generally
inoperation.
racewayload done
operation.
fault in
This loadchoice
detectionThis
operation.
bychoice
isDWT.This choice3 is 3Fig.
extraction 0 16. Multi-level32 decomposition4 (in blue), IRF (in blue)
of(s)the 6stator current8 at no-load 10
Fig. 15. MSE evolution for ORF 6 Time
SCIM
faults. faults.
the
0 because SCIM faults. 1 1 1
1.5x 10operation: Healthy (in blue), IRF (in red)
(J)

(J)

(J)

essary
because 1 of ofthe
information
necessary
2 ofdifficulty
thebecause from
3 difficulty 4 ofthe
encounteredthe stator
encountered current
5difficulty
in no-load
6 in inno-load
7a steady
encountered 8 state is
9 gen-
in1.5no-load
11.5 1 2.5 1
he
in
ation wavelet
The
orderin4.4. wavelet
erally
operation
to
order transform
done
extract
to The transform
wavelet
during
extract
inthe order application
theload
information application
transform
Details in
operation.
toinformation
extract concerningstator
application
in stator
This choice
the concerning
information in stator
isconcerning
necessary2 2 Fig. 16. Multi-level
2 decomposition
Healthy of the stator current at no-load
Inner raceway fault detection by DWT. 0.5 The 0.5 0.5 operation: Healthy (in blue), IRF (in blue)
signal
ent
SCIM signal
is
faults. faults.current
based
because
the SCIMis signal
basedon
ofoffaults.
the the
is
on
difficulty based
multi-resolution
the multi-resolution
encountered on the
during multi-resolution
no-load operation
1 0 1 1 IRF
omposition
ition extraction
of the decomposition
signal. Fig.
of transform
the signal. information
15. MSE evolution
ofapplication from forthe
ORF stator current in steady 11
0
x 10 x 10
0
2 1 1 3 2 42 3 6 5 1 14 6 25 36 4 5 6
when attempting tothe signal.
extract the in
information concerning in SCIM
4
he wavelet
The state wavelet
is The
generallytransform
wavelet
done application
in transform
load stator
application
in stator
operation. This choice stator
is 2.5
igures Figures
(16), (17),
(16), (18)
Figures
(17), and (18)
(16),
(19) and (17),
show
(19) (18)
clearly
show and clearly
the
(19) show
the 0
clearly the0 0 3
ent
ignalInner
.4. signal faults.
iscurrent
based
is because
raceway signal
basedon fault theon
is thebased
multi-resolution
the
detection multi-resolution
on encountered
the DWT.
by 0.5 1 0.521
multi-resolution
The 0 32 43 1 0 54Healthy
0.5 2 65 3 76 4 87 5 98 6 97 8 9
Energie (J)

of
utionmany necessary
of evolution
many
coefficients
The coefficients
of many
wavelet for ofdifferent
transform
difficulty
coefficients
for different
conditions.
applicationfor
conditions.
different
in the stator
in no-load
conditions.
0
current1
1.5
2 1 3 2 4Levels 3 5 of 14Levels
details
6 25 of details
IRF 36 4Levels 5 of details
6
omposition
tion of
xtraction thedecomposition
signal.
of
operation the signal.
ofisinformation
in order of the
from
to signal.
the
extract stator
the current
information inconcerning
steady 2 2
of
effect
iguresthe of
IRF
Figures
(16),The
the
signal effect
IRF
(17),clear
(16), isof
is(18) for
based
Figures
(17), clear
the
andsome
(18) for
IRF
on(16),the
(19) coefficient
andsome
is
(17),
show clear
(19) coefficient
(18) forsignals
multi-resolution
clearly
show andsome signals
clearly
the(19)coefficient
decomposition
show
the 0signals
clearly of the
Fig. the
17.
0 Fig.
Energy
17. evolution
Energy0 Fig.
evolution
at 4no-load
17. Energy
at no-load
operation
evolution
operation
(IRF)
at no-load
(IRF) operation (IRF)
tate is the generally
SCIM done
faults. in load operation. This choice 1is 21 32 143 31x 1054 2 65 3 76 4 87 5 98 6 97 8 9
of
ution
many signal.
of evolution
many
coefficients
coefficients
of manyfor differentfor
coefficients
different
conditions.
forconditions.
different conditions. Levels of
1 Levels
details
of detailsLevels of details
Energie (J)

ecessary because The of wavelet

the difficulty transform encounteredapplication in in
no-load stator 1.5
of
effect
the ofIRF theisFigures
The effect
IRF
clearisof 16‒19
for
clear
the
some clearly
for
IRF show
coefficient
some
is clear the
coefficientevolution
forsignals
somesignals of multiple
coefficient coef-
signals
Fig. 17. Fig.
Energy
17. evolution
Energy
0.5 2evolution
Fig.
at no-load
17. Energy
at no-load
operation
evolution
operation
(IRF)at no-load
(IRF) operation (IRF)
peration current
in order
ficients signal
for to is conditions.
extract
different based on The the
the information effect multi-resolution
concerning
of the IRF is clear 0
1 2 3 4 5 6
he SCIM decomposition
faults.
for some coefficient of the signals
signal. and is not clear for others. d7 and 1
Figures (16),
d8 arewavelet
substantially (17), (18)
affected andpresence
byapplication
the (19) show ofinIRclearly
fault. the 0 1
The transform stator 1 2 3 4 5 6 7 8 9
urrent evolutionsignal ofis many
Calculating thecoefficients
based energy on stored 10 level
thefor inmulti-resolution
different
each 10 confirms 10 0.5
conditions. Levels of details
0
The a decrease
effect ofobserved
the IRF inis detailclear signals
for some d5 coefficient
and d8. Butsignals we also Fig.1 17. 2Energy 3 evolution
4 5at no-load
6 operation (IRF)
ecomposition of the signal.
observe an increase in d7 (17). During load operation, the
Figures (16), (17), (18) and (19) show clearly the 10 10 10 0
energy increase was found at detail level d8 only (19). This 1 2 3 4 5 6 7 8 9
volution variation
of many coefficients
leads for IRdifferent
us to detect the fault and conditions.
to specify the fre- Levels of details
Levels of details
he effectquency
of theband.
IRF The
is clear
d7 and d8 bands coefficient
for some correspond tosignals
the frequen- Fig. 17. Energy evolution at no-load operation (IRF)
cies around 50 Hz. Fig. 17. Energy evolution at no-load operation (ORF)
10
580 Bull. Pol. Ac.: Tech. 67(3) 2019

10
 2 4 2 0 4 6 4 2 6 8 6 4 8MSE-d5 10 8 6 10
10 8 MSE-d5MSE-d5 10 0.1668 0.1668
MSE-d5 0.1668 Absence 0.1668
AbsenceAbsence Absence
10 MSE-d4MSE-d4MSE-d4 0.0229 0.0229
MSE-d4 0.0229 Absence 0.0229
AbsenceAbsence Absence
-10 MSE-d6MSE-d6MSE-d6 0.0050 0.0050
MSE-d6 0.0050Less absence
Less
0.0050
absence
Less absence Less absence
2 4 2 0 4 6 4 2 6 8 6 4 8 10 8 6 10 10 8 MSE-d5MSE-d5
MSE-d5 10 0.1668 0.1668
MSE-d5 0.1668 Absence 0.1668
AbsenceAbsence Absence
d7 0
10 MSE-d7MSE-d7MSE-d7 21.114921.1149
MSE-d721.1149 Great absence
Great
21.1149
absence
Great absence Great absence
MSE-d6MSE-d6MSE-d6 0.0050 0.0050
MSE-d6 0.0050Less absence
Less
0.0050
absence
Less absence Less absence
Mechanical fault detection in rotating electrical
MSE-d8MSE-d8 machines
55.4360using
MSE-d8 MCSA-FFT
55.4360
MSE-d855.4360 and
Great MCSA-DWT
absence
Great
55.4360
absence techniques
Great absence Great absence
-10
0 MSE-d7MSE-d7MSE-d7 21.114921.1149
MSE-d721.1149 Great absence
Great
21.1149
absence
Great absence Great absence
d7

2 4 2 0 4 6 4 2 6 8 6 4 8 10 8 6 10 10 8 MSE-d9MSE-d9
MSE-d9 10 0.0103 0.0103
MSE-d9 0.0103 Absence 0.0103
AbsenceAbsence Absence
Time (s)Time (s)Time (s) Time (s)
MSE-d8MSE-d8MSE-d8 55.436055.4360
MSE-d855.4360 Great absence
Great
55.4360
absence
Great absence Great absence
evel
omposition
Multi-level
decomposition
Fig. of
decomposition
18. theMulti-level
stator
of thecurrent
stator
of
decomposition
theat
current
stator
load operation:
at
current
load
of the
operation:
at stator
load operation:
currentMSE-a9
at loadMSE-a9
operation:
MSE-a90.0511 0.0511
MSE-a9
a)anAt0.0511 Absence0.0511
AbsenceAbsence Absence
althy2 Healthy
-10
(in4blue),
10
2 Healthy
0(inIRF
4blue),6(inblue)
(in 4blue),
IRF 2(in
6 blue)
Healthy
IRF 8 6(inblue)
4blue),
8 10 8 (in6 blue)
10 10 8 MSE-d9MSE-d9
MSE-d9 10 0.0103 gives
0.0103
MSE-d9
illustrative
no-load
0.0103 Absence
summary
(0%)
0.0103
of the MSE
AbsenceAbsence
evolution as a
Absence
Time (s)Time (s)Time (s) (in TimeIRF(s)
6 6 10 10 10 function
10 of each detail.
10
evel
omposition
Multi-level
decomposition
Fig.of xthe10
decomposition
18. Multi-level
stator
of thecurrent
stator
of
decomposition
theat
current
stator
load operation:
at
current
load
of the
operation:
at stator
load operation:
currentMSE-a9
at load operation:
(a):MSE-a9
At(a): MSE-a9
no-load 0.0511
At(a):
no-load
(0%) 0.0511
MSE-a9
At no-load
(0%) 0.0511
(a):(0%) Absence
At no-load 0.0511
Absence
(0%) Absence Absence
4Healthy Table 8
is(t)

althyHealthy
(in blue),(in 0IRF
blue),
(in(inblue)
IRF
blue),
(in
Healthy
blue)
IRF (in(inblue)
blue), IRF (in blue)
lthyHealthy
6 Healthy 6 Healthy 108 10
0.2
8 10
0.2
8
0.2 MSE
10 values
8
0.2 at different decomposition levels (IRF, 0×TN)
10 x 10 (a): At(a):
no-load
At(a):
no-load
(0%)
At no-load
(0%) (a):(0%)
At
IRF IRF 4-10 IRF MSE (di, a9)no-load MSE
(0%)Value Ressemblance
lthy
10
4
Healthy
x 10 3
4 101
Healthy 2
x 10
4
Healthy
3 4 5 6 7
86 886 9 86 10 gives
6
8 an illustrative summary of the MSE evolution as a
0.2 0.2 0.2 0.2
MSE-d1 2.7053×10 -05
Almost present

MSE MSE

MSE MSE

MSE MSE

MSE MSE
2.5
IRF IRF 10 2.5 IRF 0.1 0.1 0.1 function 0.1of each detail.
(J) (J)

Table 8
-05
is(t) is(t)

10 2 x 10 3
4 4 0 2 x 104 64 64 64 4 MSE-d2
6 6.8581×10 Almost present
2.5
2 0 2.5 MSE MSE-d3values at different0.0029
decomposition levels (IRF, 0×TN)
Energy

1.5 1.5
0.1 0.1 0.1 0.1 Less absence
2 -10 2 42 420 420 0 4 MSE (di,
2 0 a9) MSE Value Ressemblance
1 1 2 3 4 5 6 7 8 0 19 0 2 10 1 3 0 2 4 1 3 5 2MSE-d4
4 6 3 05 4 16 5 2 6 3 0.00734 5 6 Less absence
Energy

1 2 MSE-d1 2.7053×10-05 Almost present

1.5 10 1.5
-10 20 2008 200 0
2
1 0 0.5 2 4 6 1 2
10 30 2 10 03 04
55 2MSE-d5 7 0.1511 79 Absence
0.5 1 1 2 1 3240 2 4 1 3 1
3
4 4 632
MSE-d25
605 4 16 3
5
6
75 2 6 36.8581×10
4
6
8 4 5 7
895-05 6 8
69 Almost 8
present 9
1 0.5 1 DetailsDetailsDetailsDetails
Details
0 MSE-d6 0.0088 Almost present
a9 is(t)

0 0 0 60
0 60
0 60
0 60 MSE-d3 0.0029 Less absence
2 1 4 3 215 4 3 0.5
3 0.5 6 51 4 62 5 36 4 5 6 1 (b):2
1At(b):3
21Atof 4
32Atload 5
431 load 6
542Atload 7
653 of the
8
764 load 8759 869 79 8 9
75% (b):
75%
the ofDetails
75%
the of
(b):
b) the
At 75%
MSE-d7
Details 75%
of the load
Details 0.5414
Details Great absence
0 0 50 79 50 8 50 9 50 MSE-d4 0.0073 Less absence
0234 1345 0 24
56 3567 4678 5789 689 60 60 60 60
3 2 1 4 3 2 -10
5 4 03 6 51 4 62 52 3 6 4 4 5 6 MSE-d8
MSE-d5 8.5167 Great absence
Levels Levels
of details Levels
of details of detailsLevels of details 6
40
0.28 At(b):
(b):
40 75%
0.2 Atof
40
10
(b):
75%
the
0.2 At
load
of75%
the load
40 of
(b): 75% of the0.1511
theAtload
0.2 load Absence
0 0.5 50 50 50 50 -04
ergy 234 19. evolution
19. Fig.
evolution
Energy 1345at loadFig.
-0.5
Energy 2operation
45at
6 19.
evolution 35at
67(IRF)
loadEnergy
operation 4678(IRF) 5at789(IRF)
loadevolution
operation 689
load operation79(IRF) 8 9 MSE-d9
MSE-d6 5.7579×10
0.0088 Almost
Almost present
present
0 2 4 6 8 10
MSE MSE

MSE MSE

MSE MSE

MSE MSE
Levels Levels
of details Levels
of details of detailsLevels of details 30 0.2
30 0.2
30 0.2 0.2
30 MSE-a9 0.0034 Less absence
enalysis
by
MSE theThe
is
by
MSEconsidered
the
analysis
0.5 isMSE
considered
is
byasconsidered
the
a good
0.02 MSE
as aindicator
good
isasconsidered
aindicator
good indicator as40a good 40indicator
0.1 40
0.1 0.1 40 MSE-d70.1 0.5414 Great absence
ergy
19. Fig.
evolution
Energy at0 loadFig.
19. evolution
a9

Energy evolution
operation
at 19.
load Energy
operation
at
-0.020 (IRF)
loadevolution
operation
(IRF) at(IRF)
load operation (IRF)
l-defined
agives
well-defined
which
a well-defined
value gives
value
for athe well-defined
value
for
resemblance
thefor resemblance
thevalue
resemblance
or fororthe30 20resemblance
or 30 20
30 or
20 20 MSE-d8
30 Table 9
8.5167 Great absence
4 4.5 5 5.5
enalysis
by
MSE the
ocessed
tween
the The
is
by
MSEconsidered
processed
not
the the
analysis
0is
signals.MSE
between considered
processed is
by
signals.
The as
the considered
the
a good
signals.
MSEThe MSE
as
processed aindicator
allows
MSE good
Theisasconsidered
aindicator
allowsgood
signals.
MSE
us to indicator
allows
usThe
to usa 6to
asMSE good0.1indicator
allows 0.1 to 0.1
us MSE0.1values at different decomposition levels (IRF,
d9

-04
-0.5 10 10
0 10
0 0 10 MSE-d9 0 5.7579×10 Almost present
l-defined
agives
well-defined
which
linformation
entthe athewell-defined
value
allinformation
represent 0gives
value
for
information
of the athe
allof well-defined
thevalue
for
2resemblance
the
resemblance
the forresemblance
information
of the
resemblance
the value
resemblance
or
resemblance
4by of
a by for
theora6the20resemblance
ora 2080 by
resemblance
by 120 0ora2 101 3 0 2 4 1203 5 2 4 6 3 05 4 16 5 2 6 3 4TN×3/4) 5 6
0.5 signals. MSE-a9 0.0034 Less absence
ocessed
tween
the
utable
and processed
not
computablethesignals.
visible
single
and between
processed
visible
value.
and The
computablethe
visiblesignals.
MSE
value.
This The
processed
allows
MSE
0interest
value.
and
This The allows
signals.
MSE
usthe
visible
interest
This
of to allows
usThe
interest
value.
of toThis
the MSE
ofus0 to
10the
allows
interest us0theto
0812 of10
0 0
-0.5 0.02
-0.02 1 10 0 123 10 0 234 101345MSE (di,
02456 a9) 3567 MSE
4678
Value
5789 689
Ressemblance
79 8 9
linformation
ent 0 2 4 6
wn s the
valueinallthe
shown information
represent
is
MSEthe
shown
in information
twotheoftables
value the
inallthe
two isoftables
the
resemblance
theinformation
below.
shown
two oftables
resemblance
the below.
below.
Figure
in resemblance
4 the4.5
by of
Figure
two
(20) atables
by
the a resemblance
5 Figure
(20) by
5.5below.
(20) a 0 by
6 Figure (20)
1 0 a2 1 3 0 2 4Details
1 3 5 2 4Details
6 3 05 4 16 5 2 6 3
MSE-d1 Details 2.9296×10
Table
4Details5-059 6
Almost present
100
utable
and
computable
visible
single
and visible
value.
and
computable
visible
value.
This interest
value.
and
Thisvisible
interest
This
of theinterest
value.
of theThis of 0theinterest
0 of 0the 0 MSE values at different decomposition levels (IRF,
d9

1 1
2 2
3
1 3
4
2 4
3
5
1 MSE-d2 5
4
2
6 6
7
3
5 7
8
4
6
8.3884×10 8
9
7
5 -05 9
8
6 9
7
Almost 8
present 9
wn
value
s shown
in the
isMSE
shown
in
twothe
value
tables
in
twothe
is tables
below.
shown
two tables
below.
Figure
in thebelow.
Figure
two
(20)tables
Figure
(20)below. (20) Figure (20) Fig. 20. Fig.
MSE 20.
evolution
Fig.
MSE
Details 20.
evolution
MSE
for IRF
Details evolution
Fig.
for IRF
Details 20. MSE
for IRF N×3/4) for IRF
Tevolution
Details
0 Details MSE (di, a9) MSE Value Ressemblance
d8

-0.5 MSE-d3 0.0075 Less absence

0 2 4 6 8 10
Fig. 20. Fig.
MSE20.
evolution
Fig.
MSE20.evolution
for
MSE IRF
evolution
Fig.
for IRF
20. for
MSE IRF
evolution for
for IRF
MSE-d4 Fig. 20. MSE evolution IRF
-05
10 MSE-d1 2.9296×10
0.0229 Almost present
Absence
-10 MSE-d2 8.3884×10 -05
Almost present
0 2 4 6 8 10 MSE-d5 0.1668 Absence
10
0
d8

MSE-d3 0.0075 Less absence

The analysis
MSE-d6 by MSE is 0.0050considered a good indicator which
Less absence
11 11 11 11 MSE-d4
gives a well-defined value 0.0229
for the resemblance orAbsence
lack thereof
-100 MSE-d7 21.1149 Great absence
d7

0 2 4 6 8 10 betweenMSE-d5
the processed signals. MSE allows us toAbsence
0.1668 represent all
10 11 11 11 11 MSE-d8 on the resemblance
the information 55.4360 by a single computable
Great absenceand
-10 MSE-d6 0.0050 Less absence
0 2 4
Time (s)
6 8 10 visible value.
MSE-d9This aspect of the MSE value is shown
0.0103 Absencein the
0 Table 8 MSE-d7
and 9. Figure 20 21.1149
gives an illustrative Great absence
summary
Absenceof the
d7

Fig. 18. Multi-level decompositionTime

of the(s)
stator current at load operation: MSE-a9 0.0511
Healthy (in blue), IRF (in blue) MSE evolution
MSE-d8 as a function of each detail. Great absence
55.4360
-1018. Multi-level
Fig. 6 10
0x 10 2 decomposition
4 of the stator
6 current at
8 load opera-10 MSE-d9
4 tion: Healthy (inTime (s)IRF (in red)
blue), (a): At no-load 0.0103
(0%)
Table 8
Absence
Fig. 18. Multi-levelHealthy
decomposition of the stator current at load operation: 8
MSEMSE-a9
values 0.0511 Absence
0.2 at different decomposition levels (IRF, 0£TN)
Healthy (in blue), IRF (in blue)
6 IRF 10
34 x 10 x 104 MSE
6 (di,
(a):a9)
At no-load MSE
(0%) Value Resemblance
MSE

2.5
Healthy 0.1
8 MSE-d1
0.2 2.7053£10 ‒05 Almost present
Energy (J)

2 IRF 4
‒05
2
3 x 10
4 6 MSE-d2 6.8581£10 Almost present
1.5
MSE

2.5 2 0
0.1 0 1
MSE-d3 2 0.0029
3 4 5 6 Less absence
Energy (J)

12 4
2
1 0 MSE-d4 0.0073 Less absence
0.5
1.5 1 2 3 4 5 6 7 8 9
2 0 Details
MSE-d5 0 1 2 3
0.1511 4 5 6 Absence
0
1 60
1 2 3 4 5 6
1 (b):2At 75%
0 MSE-d6 of the
3 0.0088
load 5 6Almost7present8
0 0.5
501 4 9
1 2 3 4 5 6 7 8 9 Details
0 Levels of details 60 MSE-d7
0.2 0.5414 Great absence
1 2 3 4 5 6 40 (b): At 75% of the load
0 Fig. 19. Energy evolution at load operation (IRF) 50 MSE-d8 8.5167 Great absence
1 4 5 2 6 7 3
8 9
MSE

30
The analysis by the MSE is
Levelsconsidered
of details as a good indicator
Levels of details
0.2
0.1
40 MSE-d9 5.7579£10 ‒04 Almost present
which gives Fig. a
19.well-defined
Energy evolutionvalue
at loadfor the resemblance
operation (IRF) or 20
30 MSE-a9 0.0034 Less absence
MSE

not between the

Fig. 19. processed
Energy signals.
evolution at loadThe MSE
operation
The analysis by the MSE is considered as a good indicator allows
(IRF) us to 10 0.1
0
which gives a well-defined value for the resemblance ora
represent all the information of the resemblance by 20 0 1 2 3 4 5 6
single
not computable
between and visible
the processed value.
signals. TheThis
MSE interest
allowsofusthe
to 0
Bull. Pol. Ac.: Tech. 67(3) 2019 101 02 3 4 5 6 7 8 581 9
MSE valueallis the
represent shown in the twooftables
information the below.
resemblanceFigureby(20)a 0 1 2 3 4Details
5 6
single computable and visible value. This interest of the 0
1 2 3 20. MSE
4 evolution
5
Fig. for6 IRF 7 8 9
MSE value is shown in the two tables below. Figure (20) Details
 N. Bessous, S. Sbaa, and A.C. Megherbi

Table 9 [3] L. Saidi, J.B. Ali, E. Bechhoefer, and M. Benbouzid, “Wind tur-
MSE values at different decomposition levels (IRF, TN£3/4) bine high-speed shaft bearings health prognosis through a spectral
Kurtosis-derived indices and SVR”, Applied Acoustics 120, 1‒8,
MSE (di, a9) MSE Value Resemblance (2017).
MSE-d1 02.9296£10 ‒05
Almost present [4] F.B. Abid, S. Zgarni, and A. Braham, “Distinct Bearing Faults
Detection in Induction Motor by a Hybrid Optimized SWPT and
‒05
MSE-d2 08.3884£10 Almost present aiNet-DAG SVM”, IEEE Trans. En.Conv. (2018).
[5] A. Głowacz and Z. Głowacz, “Recognition of rotor damages in
MSE-d3 00.0075 Less absence
a DC motor using acoustic signals”, Bull. Pol. Ac.: Tech 65(2),
MSE-d4 00.0229 Absence 187‒194 (2017).
[6] B. Rachid, A. Hafaifa, and M. Boumehraz, “Vibrations Detection
MSE-d5 00.1668 Absence in Industrial Pumps based on Spectral Analysis to Increase Their
MSE-d6 00.0050 Less absence Efficiency”, Manag. Syst. .Prod. Eng. 21(1), 55‒61 (2016).
[7] R. Dash and B. Subudhi, “Stator inter-turn fault detection of an
MSE-d7 21.1149 Great absence induction motor using neuro-fuzzy techniques”, Archives of Con-
trol Sciences 20(3), 363‒376 (2010).
MSE-d8 55.4360 Great absence
[8] C.T. Kowalski, and M. Kaminski, “Rotor fault detector of the
MSE-d9 00.0103 Absence converter-fed induction motor based on RBF neural network”,
Bull. Pol. Ac.: Tech 62(1), 69‒76 (2014).
MSE-a9 00.0511 Absence [9] R. Valles-Novo, J. de Jesus Rangel-Magdaleno, J.M. Ramirez-
Cortes, H. Peregrina-Barreto, and R. Morales-Caporal, “Empir-
ical mode decomposition analysis for broken-bar detection on
5. Conclusion squirrel cage induction motors”, IEEE Trans. Instr. Meas. 64(5),
1118‒1128 (2015).
The observation of stator current signal is a widely used [10] A.M. Júnior, V.V. Silva, L.M. Baccarini, and L.F. Mendes, “The
technique. In this study, the motor current signature analysis design of multiple linear regression models using a genetic algo-
technique gave good indicators of REB faults. For this rea- rithm to diagnose initial short-circuit faults in 3-phase induction
motors”, Applied Soft Computing, 63, 50‒58 (2018).
son, MCSA has been preferred until now for fault detection in
[11] G. Singh, and V.N.A. Naikan, “Detection of half broken rotor bar
rotating electrical machines. However, MCSA-FFT does not fault in VFD driven induction motor drive using motor square
allow for time-frequency representation as it cannot study the current MUSIC analysis”, Mechanical Systems and Signal Pro-
phenomena of the transient regime or the non-stationary signals cessing 110, 333‒348 (2018).
caused by a variable load. [12] A. Mejia-Barron, M. Valtierra-Rodriguez, D. Granados-Lieber-
MCSA-DWT gave precise results on the state of the SCIM. man, J.C. Olivares-Galvan, and R. Escarela-Perez, “The appli-
A comparison between the decomposition levels (di or ai) has cation of EMD-based methods for diagnosis of winding faults in
a transformer using transient and steady state currents”, Measure-
given us more information about the existence of the fault.
ment 117, 371‒379 (2018).
A new indicator was introduced in this study; it is the mean [13] N. Bessous, S.E. Zouzou, and A. Chemsa, “A new analytical mod-
square error (MSE) which represents a sound contribution for el dedicated to diagnose the rolling bearing damage in induction
the final decision in our analysis. motors-simulation and experimental investigation,” in Control En-
Because of the disadvantages of FFT analysis, many import- gineering and Information Technology (CEIT) IEEE, 1‒9, (2016).
ant pieces of information in the non-stationary regime can be [14] M. Ojaghi and R. Akhondi, “Modeling Induction Motors Under
exploited more flexibly by means of the wavelet transform Mixed Radial-Axial Asymmetry of the Air Gap Produced by Oil-
Whirl Fault in a Sleeve Bearing”, IEEE Trans. Mag. (99) (2018).
(WT) or other techniques.
[15] C. Wang, X. Bao, S. Xu, Y. Zhou, W. Xu, and Y. Chen, “Anal-
ysis of Vibration and Noise for Different Skewed Slot-Type
Acknowledgements. This research was supported by the Elec- Squirrel-Cage Induction Motors”, IEEE Trans. Mag. 53(11), 1‒6
trical Engineering Laboratory of Biskra (LGEB), University of (2017).
Mohamed Khider in Biskra, BP 145 RP, 07000 Biskra, Algeria. [16] C. Kumar, G. Krishnan, and S. Sarangi, “Experimental investi-
We thank our colleagues from LGEB who have provided insight gation on misalignment fault detection in induction motors using
and expertise that greatly assisted the research, although they current and vibration signature analysis”, in Futuristic Trends
on Computational Analysis and Knowledge Management IEEE,
might not agree with all of the interpretations/conclusions of
61‒66 (2015).
this paper. [17] A. Ibrahim, “Contribution au diagnostic de machines élec-
tromécaniques: Exploitation des signaux électriques et de la vi-
tesse instantanée”, Doctoral dissertation, Université Jean Mon-
References net-Saint-Etienne, (2009).
[18] J.A. Antonino-Daviu, M. Riera-Guasp, J.R. Folch, and M.P.M.
[1] W.T. Thomson, “A review of on-line condition monitoring tech- Palomares, “Validation of a new method for the diagnosis of ro-
niques for three-phase squirrel-cage induction motors – past, tor bar failures via wavelet transform in industrial induction ma-
present and future”, in Proc. IEEE Int. Symp. SDEMPED 1, 3–18 chines”, IEEE Trans. Ind. App. 42(4), 990‒996 (2006).
(1999). [19] N. Bessous, S.E. Zouzou, W. Bentrah, S. Sbaa, and M. Sahraoui,
[2] R.K. Patel and V.K. Giri, “Condition monitoring of induction mo- “Diagnosis of bearing defects in induction motors using discrete
tor bearing based on bearing damage index”, Archives of Electri- wavelet transform”, International Journal of System Assurance
cal Engineering, 66(1), 105‒119 (2017). Engineering and Management 9(2), 335‒343 (2018).

582 Bull. Pol. Ac.: Tech. 67(3) 2019