Document 0

Neural Approach for Fault Classification in Induction
Motors by Using Motor Current and Voltage

A Thesis
Submitted in the partial fulfillment of the requirements
For the Degree of
Master of Engineering
In
Electrical Power System
By
Praful Dewanand Thamke
Roll No. 1327
Under the guidance of

Dr. Mrs. A. U. Jawadekar
DEPARTMENT OF ELECTRICAL ENGINEERING

SHRI SANT GAJANAN MAHARAJ COLLEGE OF ENGINEERING
SHEGAON 444 203 M.S (INDIA)
Winter 2017
SHEGAON 444 203 M.S (INDIA)
CERTIFICATE
Certified that the contents of thesis entitled “Neural Approach for Fault
Classification in Induction Motors by Using Motor Current and Voltage”
is a bonafied work carried out under my supervision by Praful Dewanand
Thamke in partial fulfillment of the requirements for the degree of Master of
Engineering in Electrical Power System.
Date: Dr.Mrs.A.U.Jawadekar
Advisor
Approved
(Dr.S.R.Paraskar) (Dr.S.B.Somani)
Head
Principal
Electrical Engg. Department
______________________________________________________________
E-mail.principal@ssgmce.ac.in Phone No.07265-252116, 252216
hod_electrical@rediffmail.com Fax No. 07265-252346, 253602

SHEGAON 44 203 (INDIA)
CERTIFICATE OF APPROVAL
The foregoing dissertation entitled, “Neural Approach for Fault

Classification in Induction Motors by Using Motor Current and Voltage”
is hereby approved as a creditable study of an Engineering Student
(Electrical Power System) carried out and presented by Praful Dewanand
Thamke in satisfactory manner to warrant its acceptance as a pre-requisite of
his Master of Engineering degree in Electrical Power System.
Internal Examiner External Examiner

ACKNOWLEDGEMENT
I first pray to our divine source of inspiration “Shri Sant Gajanan

Maharaj” whose blessings are always with us.
I would like to express my sincere thanks for to my project guide
Dr.Mrs.A.U.Jawadekar and for his valuable guidance, constant
encouragement and kind co-operation throughout the period of work which
has been instrumental in the success of project.
I would like to thank Shri Sant Gajanan Maharaj College of
Engineering, Shegaon for providing a course where I could learn and
broaden my engineering perspective. I express my extreme gratefulness to
our honorable principal Dr. S. B. Somani providing me a healthy academic
ambience. I would also like to thank head of the department of electrical
engineering Dr.S.R.Parasakar for his valuable guidance.
Finally, I sincerely thank to all those who have helped me in the
endeavor of completing the project. My deepest thanks to my parents who
have continuously encouraged me to fulfill the goal successfully. Last but not
the least I am thankful to all my friends who have been contributed through
advice and timely support for the completion of this project work.
Place: Shegaon Praful Dewanand Thamke

Date: ME (EPS) Final Year
ABSTRACT
Induction motor is playing a key role as prime-mover or energy

conversion device in many engineering areas. Fault diagnosis of such
machines is gaining importance in industry because of the need to increase
reliability and to decrease possible loss of production due to machine
breakdown. The concept of magnetic field pendulous oscillation is the robust
technique is presented for the diagnosis of the various fault occurred internally
to the induction motor and also the classification is done with the help of ANN.
In this, the current signals and the voltage signals are used as an input to
compute and subsequently observed the magnetic field orientation. To
validate this purpose, the experimental investigation were performed and the
result of this test is given here for several types of fault on 2-hp, 3-phase
squirrel cage Induction motor.
i
Chapter 1 Introduction
INTRODUCTION
1.1 Background
Induction motors are most widely used electrical machines in industry
because of their low cost, reasonably small size, ruggedness, low
maintenance, and operation with an easily available power supply. More than
80% of electromechanical conversion in industrial drives belongs to the
Induction Motor. Furthermore, the total number of operating electrical
machines in the world was around 16.1 billion in 2011, with a growth rate of
about 50% in the last five years [1]. Accordingly, in many industrial processes,
the induction motor plays a key role as a prime mover. Although induction
motors and specifically squirrel-cage induction motors have a rugged
structure, as with all energy conversion systems, their reliability is not
absolute. Since in many applications induction motors are the main prime
movers in such industrial processes. Sudden failure of such motor may be
very damaging or catastrophic in an industrial system, in which electrical
motor is the prime mover. Consequently, failure motors reduces the
productivity in industry. Hence the frequent maintenance schedule is very
important to prevent this failure. Generally mechanical faults, for e.g. Rotor cut
bar, stator interturn faults, bearing faults etc. may often have negligible effects
on the machine performance, however such faults may rapidly lead to
substantial interturn faults and subsequently catastrophic failures. Such failure
may cause damages the motor insulation which leads to generates heat in the
defective region which causes the fault to rapidly progress to more severe
forms, which may damage stator winding and core of the motor. Earlystage
fault detection will enable orderly process shutdown, thereby avoiding
expensive repairs and minimizing lost production time. Accordingly, it is
worthwhile to detect these faults at an early stage to prevent further damage
to the machine and involved systems [2].
Accordingly, an on-line fault diagnostic system becomes a valuable tool
to increase industrial efficiency and reliability. Thus, on-line condition
monitoring and diagnostics are becoming very important issues in electric
machine protection systems since they greatly improve reliability, availability
and maintainability in a wide range of applications.
SSGMCE, Shegaon Page 1

1.2 Condition Monitoring

An on-line condition monitoring of three-phase machines is generally
provided by some combination of electrical, mechanical and thermal
monitoring techniques. Even though thermal and vibration monitoring devices
and conventional electrical relays are used to monitor the machine’s
performance, they are installed only on expensive or “sensitive” machines,
where the cost of the whole system can justify their installation. Meanwhile,
sensitivities of the thermal and mechanical sensors are affected by aging and
ambient conditions, which may lead to situations of failure of monitoring of
such conditions. These forms of condition monitoring are also limited in their
ability to detect incipient faults such as broken rotor bar faults and the inter-
turn stator insulation failures, particularly in their earliest stages. Modern
electrical monitoring techniques have mainly concentrated on motor current
spectral analysis techniques for broken bar faults and the negative sequence
component concepts for detecting inter-turn stator insulation failures. These
indices have served as the foundations for other techniques which make use
of statistical correlation, pattern recognition, and AI techniques in order to
improve the reliability of the diagnostic outcomes.
It should be noted that even modern on-line condition monitoring or
fault diagnosis is not a new development. It has been a relatively “hot
research area” in many technical conferences and there are many technical
papers in the literature concerning this topic, some of which will be reviewed
in the literature search section of this chapter.
However, this research area still constitutes an open problem because
these methods are sensitive to the mechanical load variations, motor
manufacturing imperfections, operating ambient (environmental) conditions
and the complications in motor current and voltage signals associated with
PWM-based drives. These aspects could render an on-line condition
monitoring unreliable under some operational conditions. Moreover, as will be
delineated throughout this dissertation the degree of fault severity cannot be
evaluated using conventional fault diagnostic indices. Accordingly, in dealing
with the two electrical types of motor component failure, subject of this
dissertation, this research will concentrate on finding effective means of

(1) Detection of broken-bar faults, when a motor is energized by Three

phase balance power supply under different load levels, and
(2) Detection of incipient inter-turn faults in the presence of a degree of
motor manufacturing imperfection, such as the case particularly in
random wound motors, operating under various load levels.
1.3 Aim Of The Work

The main aims of this work are to classify the different types fault that
is to classify healthy, rotor broken-bar condition and stator inter-turn short
circuit condition occurred in an Induction motor under full-load condition by
using following techniques
 By using magnetic field oriented Pendulous Oscillation phenomenon
 By peak values of Negative sequence current and peak values of
swing angle obtained for each cycle.
 With the help of Multi-layer feed-forward artificial neural network.

Chapter 2 Literature Review
2. LITERATURE REVIEW
2.1 Introduction
Condition monitoring is defined as the continuous evaluation of health of
the plant and equipment throughout its service life. It is important to be able to
detect the faults while they are still developing. This is called incipient failure
detection. The incipient detection of motor failure also provides a safe
operating environment. It is becoming increasingly important to use
comprehensive condition monitoring scheme for continuous assessment of
the electrical condition of the electrical machines. By using the condition
monitoring, it is possible to adequate warning of imminent failure. In addition,
it is also possible to schedule future preventive maintenance schedule.
Condition monitoring allows the machine operator to have the necessary
spare part before the machine is stripped down, thereby reducing outage
time. Therefore effective fault classification of electric machine is critical in
improving the reliability, safety and productivity.
In this chapter, the literature on fault classification is reviewed. This
review covers some important and most useful techniques such as condition
monitoring, fault classification and diagnosis, Artificial Neural Network etc. In
addition, this review also covers the major development in this field from early
research to most recent.
GB.Kliman proposed tum-to-tum fault detector technique that has a

sensitivity approaching one part in a thousand. The fault indication is fully
developed in two cycles of the fundamental making it suitable as a protective
function on all sizes of motor. The detector software is self-adapting to the
motor and instrumentation characteristics needing no detailed data on the
motor other than can be obtained by itself in a short learning period while the
motor operates normally plus a normal start. The method demonstrated
should open the door for the integration of tum fault protection as a standard
part of any motor protection package [4].
Sahraoui, Ghoggal, Zouzou, Aboubou and Razik have proposed a new

mathematical model of the induction motor operating under stator inter-turns

short circuits. The model is based on the multiplied coupled circuit approach.
The inductances calculation is performed an extension in 2- D of the modified
winding function approach (EMWFA), which was able to take into account the
space harmonics in addition to the effects of rotor bar skewing and to the
linear rise of MMF across the slots. From the results it is shown that the inter-
turn short circuit gives rise to some spectral components which appear in the
current line spectrum [5].
Bachir, Tnani, Trigeassou and Champenois have suggested a new model

of squirrel cage induction motors under stator and rotor faults. First, they
study an original model that takes into account the effects of inter-turn faults
resulting in the shorting of one or more circuits of stator phase winding. Then,
they propose a new faulty model dedicated to broken rotor bars detection.
The corresponding diagnosis procedure based on parameter estimation of the
stator and rotor faulty model is proposed. The estimation technique is
performed by taking into account prior information available on the safe
system operating in nominal conditions. In this paper the output error (OE)
identification technique is used to estimate the parameters [6].
Behrooz Mirafzal developed a new technique based on the “swing-like”

pendulous oscillation of the rotor magnetic field orientation for rotor broken-
bar fault detection. In this, he developed an associated variance index derived
from the pendulous swing angle to distinguish a rotor broken-bar fault from an
inter-turn stator-winding short circuit when such faults occur concurrently.
Mainly, this paper has shown that as the number of broken bars increases,
the range of the “swing-like” pendulous oscillation of the rotor magnetic field
orientation will increase in a monotonically unambiguous manner [7].
Behrooz Mirafzal again proposed a robust interturn fault diagnostic method

based on the pendulous oscillation concept and examined for the case
studies of the 2-hp and the 5-hp induction motors. The results shows, the
strength and fidelity of this method, even in the presence of a degree of
machine construction imperfections. It was also shown that the swing angle
magnitude is a function of the ratio between the short circuit circulating current
and the line (phase) current. Moreover, an interturn fault can be detected if

the circulating current increases slightly beyond the phase current level even
before the occurrence of higher levels of damaging circulating currents.
Hence, the swing angle index enables one to detect stator interturn faults at
an early stage to prevent further damage to the machine and involved
systems [2].
Vilas N. Ghate and Sanjay V. Dudul proposed a new approach to intelligent

fault detection and classification of three-phase induction motor based on
RBF-MLP cascade neural network. In this paper, Simple statistical
characteristics, such as standard deviation, kurtosis, energy, entropy,
variance, etc., are extracted to derive rich faulty information from stator
current [8].
Nejjari and Benbouzid have used the Park‟s vector patterns for detecting
different types of supply faults, such as voltage imbalance and single phasing.
In addition a neural network based back propagation algorithm is used to
obtain the machine condition by testing the shape of the Park‟s vector
patterns. Two neural network based approach have been used, these are
classical and decentralized. The generality of the proposed methodology has
been experimentally tested and the authors claim that the results provide a
satisfactory level of accuracy [9].
Bouzid has suggested a neural network approach for the detection and
location automatically of an inter-turn short circuit fault in the stator windings
of an induction motor. In this paper they have used a feed forward multi-layer
perceptron neural network which is trained by the back propagation
technique. The phase shift between the phase voltage and line current of an
induction motor is used as the input to the neural network. The desired output
is set to either „one‟ or „zero‟. If a short circuit is detected and located on one
of the three phases, the corresponding neural network output is set to „one‟;
otherwise, it is „zero‟ [10].

Chapter 3: Induction Motor
3 INDUCTION MOTOR
3.1 Introduction
An induction motor comprises a magnetic circuit interlinking two electric
circuits which are placed on the two main parts of the machine:
(i) The stationary part called the stator and
(ii) The rotating part called the rotor.
Power is transferred from one part to the other by electromagnetic
induction. For this induction machine is referred as an electromechanical
energy conversion device which converts electrical energy into mechanical
energy. Rotor is supported on bearings at each end.
Generally, both the stator and rotor consist of two circuits: (a) an
electric circuit to carry current and normally made of insulated copper or
insulated aluminium and (b) a magnetic circuit, shown in Figure 3.1, to carry
the magnetic flux made of laminated magnetic material normally steel. The
different components of an induction motor are shown in figure 3.2.
Figure 3.1: Magnetic circuit of stator and rotor of an induction motor

Figure 3.2: An induction motor
3.2 Construction
A 3-phase induction motor has two main parts (I) stator and (II) rotor.
The rotor is separated from the stator by a small air-gap which ranges from
0.4 mm to 4mm, depending on the power of the motor
3.2.1 Stator
It consists of a steel frame which encloses a hollow, cylindrical core
made up of thin laminations of silicon steel to reduce hysteresis and eddy
current losses. A number of evenly spaced slots are provided on the inner
periphery of the laminations shows in figure 3.4. When 3-phase supply is
given to the stator winding, a rotating magnetic field of constant magnitude is
produced.
Figure 3.3: Stator of an Induction motor.

3.2.2 Rotor
The rotor mounted on a shaft is a hollow laminated core having slots
on its outer periphery. The winding placed in these slots (called rotor winding)
may be one of the following two types.
1 Squirrel cage type

It consists of a laminated cylindrical core having parallel slots on its
outer periphery. One copper or aluminum bar is placed in each slot. All these
bars are joined at each end by metal rings called end rings shows in figure
3.5. The entire construction (bars and end rings) resembles a squirrel cage
and hence the name. The rotor is not connected electrically to the supply but
has current induced in it by transformer action from the stator. Those
induction motors which employ squirrel cage rotor are called squirrel cage
induction motors. Most of 3 phase induction motors use squirrel cage rotor as
it has a remarkably simple and robust construction enabling it to operate in
the most adverse circumstances. However, it suffers from the disadvantage
of a low starting torque. It is because the rotor bars are permanently short
circuited and it is not possible to add any external resistance to the rotor
circuit to have a large starting torque.
Figure 3.4: Squirrel cage rotor

2 Wound type
It consists of a laminated cylindrical core and carries a 3 phase
winding, similar to the one on the stator shows in figure 3.6 the rotor winding
is uniformly distributed in the slots and is usually star- connected. The open
ends of the rotor winding are brought out and joined to three insulated slip
rings mounted on the rotor shaft with one brush resting on each slip ring.
Figure 3.5: wound Rotor
Besides the above two main parts, an induction motor consists some
other parts which are named as follows:
(i) End flanges: There are two end flanges which are used to support
the two bearings on both the ends of the motor.
(ii) Bearings: There are two set of bearings which are placed at both
the ends of the rotor and are used to support the rotating shaft.
(iii) Shaft: It is made of steel and is used to transmit generated torque
to the load.
(iv) Cooling fan: It is normally located at the opposite end of the load
side, called non-driving end of the motor, for forced cooling of the
both stator and rotor.
(v) Terminal box: It is on top or either side of the outer cylindrical
frame of stator to receive the external electrical connections.

3.3 Operation
When the stator winding of an induction motor is connected to a three-

phase supply, a uniform rotating magnetic field is produced therein [3],
which induces e.m.f. in the rotor which is free to rotate coaxially with the
stator core with the help of ball bearings. Rotor being short circuited, either
through the end rings or an external resistance, currents are produced due
to this induced e.m.f. This current interacts with the rotating magnetic field
to develop a torque on the rotor in the direction of the rotating magnetic
field. As the rotor is free to rotate, the torque will cause it to move round in
the direction of the stator field. This makes a three-phase induction motor
as self-starting.
In transforming this electrical energy into mechanical energy, in an

induction motor some losses occur which are as follows:
 Friction and windage losses, 5–15 %

 Iron or core losses, 15–25 %
 Stator losses, 25–40 %
 Rotor losses, 15–25 %
 Stray load losses, 10–20 %.
Full-load motor efficiency varies from about 85 to 97 %.
Induction motors are simpler, cheaper, and efficient. Among them
squirrel cage induction motor is more rugged and work more efficiently
compared to wound-rotor induction motor. If supply voltage and frequency are
constant, then a squirrel-cage induction motor runs at a constant speed which
makes it suitable for use in constant speed drive. Several standard designs of
squirrel-cage induction motors are available in the market to fulfil the
requirements of different starting and running conditions of various industrial
applications. These are classified [16] as class A, class B, class C, and class
D. In Table 3.1, a comparison of different classes of squirrel-cage induction
motors is presented.

Table 3.1: Various classes of squirrel-cage induction motor
Class A Class B Class C Class D

Properties Normal starting Normal starting High starting High starting
torque, high torque, torque and torque, low
starting current low starting low starting starting
current and high
and low current current
operating slip
operating slip and low
operating slip
Uses Fan, pump For constant Compressor, For driving
load etc. speed conveyors, intermittent load,
where torque is drive such as crashers etc. e.g. punch press
low pump,
etc.
at start blower
3.4 Faults: Causes and Effects

Induction motors are rugged, low cost, low maintenance, reasonably
small sized, reasonably high efficient, and operating with an easily available
power supply. They are reliable in operations but are subject to different types
of undesirable faults. From the study of construction and operation of an
induction motor, it reveals that the most vulnerable parts for fault in the
induction motor are bearing, stator winding, rotor bar, and shaft. Besides due
to non-uniformity of the air gap between stator-inner surface and rotor-outer
surface motor, faults also occur.
In general, induction motor faults can be categorized into electrical and
mechanical faults

Figure 3.6: Types of fault in Induction motor.
Electrical faults/asymmetries can be further categorized into rotor and

stator faults. All these possible faults in induction motors and their associated
subsets are summarized as depicted in the block diagram schematic of Figure
3.6. The statistical report of the Motor Reliability Working Group of the IEEE-
IAS, which surveyed 1141 motors, [3] and the study carried out by the
Electrical Power Research Institute (EPRI), which surveyed 6312 motors, [4]
are summarized here in Table 3.2.
Table 3.2: Percentage failure of Induction motor.
IEEE-IAS survey EPRI Survey

Major Components
Percentage of Failures Percentage of Failures
Bearing Faults 42% 41%
Winding faults 28% 36%
Rotor Faults 8% 9%
Others 22% 14%
As one can observe from this table, bearing and winding related
failures are the dominant trouble areas. Moreover, the percentage of winding
related and rotor related failures are around 40 percent. This dissertation is

specifically dealing with only these two electrical types of component failure,
namely broken bar and inter-turn faults.
The majority of electric machine component failures can be related to
the three main components of the machine: the stator, the rotor, and the
bearings. Since these components account for over 80 percent of all electric
machine failures, most condition monitoring schemes have concentrated on
detecting faults in one of the three components. Again, this dissertation is
dealing with two types of stator and rotor faults mainly inter-turn shorts in
stator windings and broken bar faults in squirrel-cages motor.
3.4.1 Stator Faults
Stator faults may be divided into two types these are as follows
1. Stator winding related faults
2. Stator core related faults
One of the weakest components in an electric machine, both
mechanically and electrically, is the winding insulation. Stator faults are
caused by partial stator winding insulation failures shown in figure3.7.
Figure 3.7: damaged stator winding insulation
Such partial stator winding insulation failures may in turn be caused by

one or more of the following causes: frequent machine overloading, coil
vibration, frequent motor starts and stops, transient voltage stress, and PWM
inverter induced surges, particularly in the presence of a substantial cable

length between a motor and its drive, ambient stresses, and aging of the
stator winding insulation [3, 4]. Inter-turn faults lead to generated heat in the
defective region of a winding which causes the fault to rapidly progress to
more severe forms such as phase-to-phase and phase-to-ground faults.
3.4.2 Rotor related Faults
Although the squirrel-cage rotor is rugged, bar defects do occur,

particularly in larger machines. These defects come from two sources. The
first is associated with the high temperatures and the large centrifugal forces
developed during transient operations such as start-ups. This may be
aggravated by frequent start-stop duty cycles. Meanwhile, defective casting or
poor end-ring joints formed during manufacturing are the second source.
Once the initial defect occurs, localized overheating may develop in the cage,
and hence propagation of the fault is continued by multiple start-ups as well
as load fluctuations which produce high centrifugal forces. The condition is
further accentuated by the heating and cooling cycles of the rotor associated
with high frequency duty cycles of starts and stops. The rotor with broken bar
is shown in figure 3.8.
Figure 3.8: Broken rotor bar

Chapter 4: Artificial Neural Network
4 ARTIFICIAL NEURAL NETWORK.

4.1 Introduction
Artificial Neural Network (ANN) is an efficient computing system whose
central theme is borrowed from the analogy of biological neural networks.
ANNs are also named as “artificial neural systems,” or “parallel distributed
processing systems,” or “connectionist systems.” ANN acquires a large
collection of units that are interconnected in some pattern to allow
communication between the units. These units, also referred to as nodes or
neurons, are simple processors which operate in parallel. Every neuron is
connected with other neuron through a connection link. Each connection link
is associated with a weight that has information about the input signal. This is
the most useful information for neurons to solve a particular problem because
the weight usually excites or inhibits the signal that is being communicated.
Each neuron has an internal state, which is called an activation signal. Output
signals, which are produced after combining the input signals and activation
rule, may be sent to other units.
4.2 Biological Neuron

Basically, a biological neuron receives inputs from other sources,
combines them in some way, performs a generally nonlinear operation on the
result, and then outputs the final result. Figure 4.1 shows the relationship of
these four parts.
Figure 4.1: Biological neuron

Within humans there are many variations on basic type of neuron, yet,
all biological neurons have the same four basic components. They are known
by their biological names – cell body (soma), dendrites, axon, and synapses.
 Cell body (Soma): The body of neuron cell contains the nucleus and
carries out biochemical transformation necessary to the life of neurons.
 Dendrite: Each neuron has fine, hair like tubular structures (extensions)
around it. They branch out into tree around the cell body. They accept
incoming signals.
 Axon: It is a long, thin, tubular structure which works like a transmission
line.
 Synapse: Neurons are connected to one another in complex spatial
arrangement. When axon reaches its final destination it branches again
called as terminal arborization. At the end of axon are highly complex and
specialized structures called synapses. Connection between two neurons
takes place at these synapses.
Dendrites receive the input through the synapses of other neurons. The
soma processes these incoming signals over time and converts that
processed value into an output, which is sent out to other neurons through the
axon and the synapses.
4.3 An Artificial Neuron

The artificial neuron simulates four basic functions of a biological
neuron. Figure 4.2 shows basic representation of an artificial neuron.
Figure 4.2: A basic artificial neuron.

Figure 4.2 shows various inputs to the network are represented by the
mathematical symbol, x(n). Each of these inputs is multiplied by a connection
weight. The weights are represented by w(n). In the simplest case, these
products are summed, fed to a transfer function (activation function) to
generate a result, and this result is sent as output. This is also possible with
other network structures, which utilize different summing functions as well as
different transfer functions.
Some applications like recognition of text, identification of speech,
image deciphering of scenes etc. require binary answers. These applications
may utilize the 77 binary properties of ORing and ANDing of inputs along with
summing operations. Such functions can be built into the summation and
transfer functions of a network.
Seven major components make up an artificial neuron. These
components are valid whether the neuron is used for input, output, or is in the
hidden layers.
Component 1: Weighting Factors

A neuron usually receives many simultaneous inputs. Each input has
its own relative weight, which gives the input the impact that it needs on the
processing element's summation function. Some inputs are made more
important than others to have a greater effect on the processing element as
they combine to produce a neural response. Weights are adaptive coefficients
that determine the intensity of the input signal as registered by the artificial
neuron.
They are a measure of an input's connection strength. These strengths can be
modified in response to various training sets and according to a network's
specific topology or its learning rules.
Component 2: Summation Function

Inputs and corresponding weights are vectors which can be
represented as (i1, i2 . . . in) and (w1, w2 . . . wn). The total input signal is the
dot product of these two vectors. The result; (i1 * w1) + (i2 * w2) +…….. + (in *
wn); is a single number.

The summation function can be more complex than just weight sum of
products. The input and weighting coefficients can be combined in many
different ways before passing on to the transfer function. In addition to
summing, the summation function can select the minimum, maximum,
majority, product or several normalizing algorithms. The specific algorithm for
combining neural inputs is determined by the chosen network architecture and
paradigm. Some summation functions have an additional „activation function‟
applied to the result before it is passed on to the transfer function for the
purpose of allowing the summation output to vary with respect to time.
Component 3: Transfer Function

The result of the summation function is transformed to a working output
through an algorithmic process known as the transfer function. In the transfer
function the summation can be compared with some threshold to determine
the neural output. If the sum is greater than the threshold value, the
processing element generates a signal and if it is less than the threshold, no
signal (or some inhibitory signal) is generated. Both types of response are
significant. The threshold, or transfer function, is generally non-linear. Linear
functions are limited because the output is simply proportional to the input.
The step type of transfer function would output zero and one, one and
minus one, or other numeric combinations. Another type, the „threshold‟ or
ramping function, can mirror the input within a given range and still act as a
step function outside that range. It is a linear function that is clipped to
minimum and maximum values, making it non-linear. Another option is a „S‟
curve, which approaches a minimum and maximum value at the asymptotes.
It is called a sigmoid when it ranges between 0 and 1, and a hyperbolic
tangent when it ranges between -1 and 1. Both the function and its derivatives
are continuous.
Component 4: Scaling and Limiting

After the transfer function, the result can pass through additional
processes, which scale and limit. This scaling simply multiplies a scale factor
times the transfer value and then adds an offset. Limiting is the mechanism
which insures that the scaled result does not exceed an upper, or lower

bound. This limiting is in addition to the hard limits that the original transfer
function may have performed.
Component 5: Output Function (Competition)

Each processing element is allowed one output signal, which it may
give to hundreds of other neurons. Normally, the output is directly equivalent
to the transfer function's result. Some network topologies modify the transfer
result to incorporate competition among neighboring processing elements.
Neurons are allowed to compete with each other inhibiting processing
elements unless they have great strength. Competition can occur at one or
both levels. First, competition determines which artificial neuron will be active
or provides an output. Second, competitive inputs help determine which
processing element will participate in the learning or adaptation process.
Component 6: Error Function and Back-Propagated Value

In most learning networks the difference between the current output
and the desired output is calculated as an error which is then transformed by
the error function to match particular network architecture. Most basic
architectures use this error directly but some square the error while retaining
its sign, some cube the error, other paradigms modify the error to fit their
specific purposes. The error is propagated backwards to a previous layer.
This back-propagated value can be either the error, the error scaled in some
manner (often by the derivative of the transfer function) or some other desired
output depending on the network type. Normally, this back-propagated value,
after being scaled by the learning function, is multiplied against each of the
incoming connection weights to modify them before the next learning cycle.
Component 7: Learning Function

Its purpose is to modify the weights on the inputs of each processing
element according to some neural based algorithm.

4.4 An Artificial Neural Network
Figure 4.3: An artificial neural network
Figure 4.3 shows an artificial neural network. Inputs enter into the
processing element from the upper left. The first step is to multiply each of
these inputs by their respective weighting factor [w(n)]. These modified inputs
are then fed into the summing function, which usually sums these products;
however, many different types of operations can be selected. These
operations can produce a number of different values, which are then
propagated forward; values such as the average, the largest, the smallest, the
OR values, the AND values, etc. Other types of summing functions can also
be created and sometimes they may be further complicated by the addition of
an activation function which enables the summing function to operate in a
time sensitive way.
The output of the summing function is then sent into a transfer function,
which turns this number into a real output (a 0 or a 1, -1 or +1 or some other
number) via some algorithm. The transfer function can also scale the output
or control its value via thresholds. This output is then sent to other processing
elements or an outside connection, as dictated by the structure of the
network.

4.5 Transfer (Activation) Functions

The transfer function for neural networks must be differential and
therefore continuous to enable correcting error. Derivative of the transfer
function is required for computation of local gradient. One such example of a
suitable transfer function is the sigmoid function. The sigmoid function is a S-
shaped graph. It is one of the most common forms of transfer function used in
construction of ANNs. It is defined as a strictly increasing function.
Mathematically its derivative is always positive. It exhibits a graceful balance
between linear and nonlinear behavior. One example of it is a logistic function
represented by the equation
This function has certain characteristic. At extremes of (v): (v) is flat

and ‟(v) is very small. At midrange of (v): ‟(v) is maximum as seen in
Figure 4.4.
Figure 4.4: Logistic transfer function

Several other transfer functions can also be employed as shown in Figure 4.5.

Figure 4.5: Transfer functions with different characteristic constant values
4.6 Types of Artificial Neural Networks

4.6.1 Single Layer Feed Forward Network
A neural network in which the input layer of source nodes projects into
an output layer of neurons but not vice-versa is known as single feed-forward
or acyclic network. In single layer network, „single layer‟ refers to the output
layer of computation nodes as shown in Figure 4.6
Figure 4.6: A Single layer feed-forward network

4.6.2 Multilayer Feed Forward Network

This type of network shown in Figure 4.7 consists of one or more
hidden layers, whose computation nodes are called hidden neurons or hidden
units. The function of hidden neurons is to interact between the external input
and network output in some useful manner and to extract higher order
statistics. The source nodes in input layer of network supply the input signal to
neurons in the second layer (1st hidden layer). The output signals of 2nd layer
are used as inputs to the third layer and so on. The set of output signals of the
neurons in the output layer of network constitutes the overall response of
network to the activation pattern supplied by source nodes in the input first
layer.
Figure 4.7: A multilayer feed forward network
Short characterization of feed-forward networks:

1. Typically, activation is fed forward from input to output through „hidden
layers‟, though much other architecture exists.
2. Mathematically, they implement static input-output mappings.
3. most popular supervised training algorithm: backpropagation algorithm
4. Have proven useful in many practical applications as approximators of
nonlinear functions and as pattern classificatory.
4.6.3 Recurrent Network

A feed forward neural network having one or more hidden layers with
at least one feedback loop is known as recurrent network as shown in Figure
4.8. The feedback may be a self-feedback, i.e., where output of neuron is fed

back to its own input. Sometimes, feedback loops involve the use of unit delay
elements, which results in nonlinear dynamic behavior, assuming that neural
network contains nonlinear units.
Figure 4.8: A recurrent network

There are various other types of networks like; delta-bar-delta,
Hopfield, vector quantization, counter propagation, probabilistic, Hamming,
Boltzman, bidirectional associative memory, spacio-temporal pattern, adaptive
resonance, self-organizing map, recirculation etc.
A recurrent neural network has (at least one) cyclic path of synaptic
connections. Basic characteristics:
1. All biological neural networks are recurrent
2. Mathematically, they implement dynamical systems
3. Several types of training algorithms are known, no clear winner
4. Theoretical and practical difficulties by and large have prevented
practical applications so far.
4.7 Training of Artificial Neural Networks

Once a network has been structured for a particular application, it is
ready for training. At the beginning, the initial weights are chosen randomly
and then the training or learning begins. There are two approaches to training;
supervised and unsupervised.

4.7.1 Supervised Training

In supervised training, both the inputs and the outputs are provided.
The network then processes the inputs and compares its resulting outputs
against the desired outputs. Errors are then propagated back through the
system, causing the system to adjust the weights, which control the network.
This process occurs over and over as the weights are continually tweaked.
The set of data, which enables the training, is called the "training set." During
the training of a network, the same set of data is processed many times, as
the connection weights are ever refined.
Sometimes a network may never learn. This could be because the
input data does not contain the specific information from which the desired
output is derived.
Networks also don't converge if there is not enough data to enable
complete learning. Ideally, there should be enough data so that part of the
data can be held back as a test. Many layered networks with multiple nodes
are capable of memorizing data. To monitor the network to determine if the
system is simply memorizing its data in some non-significant way, supervised
training needs to hold back a set of data to be used to test the system after it
has undergone its training.
If a network simply can't solve the problem, the designer then has to
review the input and outputs, the number of layers, the number of elements
per layer, the connections between the layers, the summation, transfer, and
training functions, and even the initial weights themselves. Another part of the
designer's creativity governs the rules of training. There are many laws
(algorithms) used to implement the adaptive feedback required to adjust the
weights during training. The most common technique is known as back-
propagation.
The training is not just a technique, but a conscious analysis, to insure
that the network is not over trained. Initially, an artificial neural network
configures itself with the general statistical trends of the data. Later, it
continues to „learn‟ about other aspects of the data, which may be spurious
from a general viewpoint.
When finally the system has been correctly trained and no further
learning is needed, the weights can, if desired, be „frozen‟. In some systems,

this finalized network is then turned into hardware so that it can be fast. Other
systems don't lock themselves in but continue to learn while in production use.
4.7.2 Unsupervised Or Adaptive Training

The other type is the unsupervised training (learning). In this type, the
network is provided with inputs but not with desired outputs. The system itself
must then decide what features it will use to group the input data. This is often
referred to as self-organization or adaption. These networks use no external
influences to adjust their weights. Instead, they internally monitor their
performance. These networks look for regularities or trends in the input
signals, and makes adaptations according to the function of the network. Even
without being told whether it's right or wrong, the network still must have some
information about how to organize itself. This information is built into the
network topology and learning rules. An unsupervised learning algorithm
might emphasize cooperation among clusters of processing elements. In such
a scheme, the clusters would work together. If some external input activated
any node in the cluster, the cluster's activity as a whole could be increased.
Likewise, if external input to nodes in the cluster was decreased, that could
have an inhibitory effect on the entire cluster.
Competition between processing elements could also form a basis for
learning. Training of competitive clusters could amplify the responses of
specific groups to specific stimuli. As such, it would associate those groups
with each other and with a specific appropriate response. Normally, when
competition for learning is in effect, only the weights belonging to the winning
processing element will be updated. Presently, the unsupervised learning is
not well understood and there continues to be a lot of research in this aspect.
4.8 Learning Rates

The rate at which ANNs learn depends upon several controllable
factors. A slower rate means more time to spend in producing an adequately
trained system.
With faster learning rates, however, the network may not be able to
make the fine discriminations that are possible with a system learning slowly.

Most learning functions have some provision for a learning rate

(learning constant). Usually this term is positive and between 0 and 1. If the
learning rate is greater than 1, it is easy for the learning algorithm to
overshoot in correcting the weights, and the network will oscillate. Small
values of the learning rate will not correct the current error as quickly, but if
small steps are taken in correcting errors, there is a good chance of arriving at
the best minimum convergence.
4.9 Learning Laws (Algorithms)

Many learning laws are in common use. Most of them are some sort
of variation of the best known and oldest „Hebb's Rule‟.
 Hebb's Rule: This was introduced by Donald Hebb in „Organization of
Behaviour‟. The basic rule is: If a neuron receives an input from
another neuron and if both are highly active (same sign), the weight
between the two neurons should be strengthened.
 Hopfield Law: If the desired output and the input are both active and
both inactive, increment the connection weight by the learning rate,
otherwise decrement the weight by the learning rate.
 The Delta Rule: This rule is based on the simple idea of continuously
modifying the strengths of the input connections to reduce the
difference (the delta) between the desired output value and the actual
output of a processing element.
 The Gradient Descent Rule: This is similar to Delta Rule in that, the
derivative of the transfer function is still used to modify the delta error
before it is applied to the connection weights. However, an additional
proportional constant tied to the learning rate is appended to the final
modifying factor acting upon the weight.
 Kohonen's Law: In this, the processing elements compete for the
opportunity to learn or update their weights. The element with largest
output is declared the winner and has the capability of inhibiting its
competitors as well as exciting its neighbours. Only the winner is
permitted an output and only the winner plus its neighbours are allowed
to adjust their connection weights.

4.10 Backpropagation For Feed-Forward Networks

The backpropagation (BP) algorithm is the most commonly used
training method for feed forward networks. Consider a multi-layer perceptron
with „k‟ hidden layers. Together with the layer of input units and the layer of
output units this gives k+2 layers of units altogether, which are numbered by
0, 1, 2..., k+1. Let the number of input units be K, output units be L and of
units in hidden layer m be Nm.
The weight of jth unit in layer m and the ith unit in layer m+1 is denoted
by wijm. The activation of the ith unit in layer m is xim (for m = 0 this is an input
value, for m = k+1 an output value). The training data for a feed-forward
network training task consist of T input-output (vector-valued) data pairs
u(n)=(x10(n),...,xk0(n))t, d(n)= (d1k+1(n),...,dLk+1(n))t,
Where „n‟ denotes training instance.
The activation of non-input units is computed according to
Xim+1(n)= ∑ .
Presented with training input u(t), the previous update equation is used
to compute activations of units in subsequent hidden layers, until a network
t
response is obtained in the output layer. The
objective of training is to find a set of network weights such that the summed
squared error ∑ ‖ ‖2 ∑ is minimized. This is
done by incrementally changing the weights along the direction of the error
gradient with respect to weights ∑ using a (small) learning
rate :
This is the formula used in batch learning mode, where new weights
are computed after presenting all training samples. One such pass through all
samples is called an epoch. Before the first epoch, weights are initialized,
typically to small random numbers. A variant is incremental learning, where
weights are changed after presentation of individual training samples:
The subtask in this method is the computation of the error gradients

The backpropagation algorithm is a scheme to perform these

computations. The procedure for one epoch of batch processing is given
below. Input: current weights , training samples.
Output: new weights.
1. For each sample n, compute activations of internal and output units
(forward pass).
2. Compute, by proceeding backward through m = k+1, k, ..., 1, for each
unit the error propagation term
For the output layer and
∑ ⌋
For the hidden layers, where
is the internal state (or potential) of unit . This is the error backpropagation
pass. Mathematically, the error propagation term represents the error
gradient with respect to the potential of the unit .
3. Adjust the connection weights according to
After every such epoch, compute the error. Stop when the error falls
below a predetermined threshold or when the change in error falls below
another predetermined threshold or when the number of epochs exceeds a
predetermined maximal number of epochs. Many (order of thousands in
nontrivial tasks) such epochs may be required until a sufficiently small error is

achieved. One epoch requires O(T,M) multiplications and additions, where M

is the total number of network connections.
The basic gradient descent approach (and its backpropagation

algorithm implementation) is notorious for slow convergence, because the
learning rate g must be typically chosen small to avoid instability. Another
approach to achieve faster convergence is to use second-order gradient
descent techniques, which exploit curvature of the gradient but have epoch
complexity O(T M2).

Chapter 5: Space Vector
5 SPACE VECTOR
5.1 Introduction
Space vector is a transformation for analyzing three-phase electric
systems. The term “space” originally stands for the two-dimensional complex
plane, in which the three-phase quantities are transformed. In electric
machines it is combined with the magnetic field [13]. In order to understand
the concept of space vector, it is helpful to start from observing the structure
of an electric machine that is driven by a three-phase current.
An electric machine is usually composed of a stator and a rotor.
Usually a stator is the outer part of the machine with windings, which are
connected to the three-phase grid to generate a rotating magnetic field.
Different rotors are available in different types of machines. However,
the structure of the stator is almost the same. In a three phase machine with
one pole pair, three copper windings are embedded in the slots of the stator
iron structure. Figure 5.1 shows one distributed winding of the stator with a
simplified drawing.
Figure 5.1: One of the distributed windings in a machine stator
In a distributed winding, the copper wire turns are not uniformly

distributed along the stator circumstance. This arrangement is used for
generating a magnetic field with sinusoidal distribution along the stator inner

circumstance. For a three phase machine with one pole pair, the simplified
winding structure is shown in figure 5.2
Figure 5.2: Windings in an electric machine
Three windings are organized with 120° angle difference in the physical
space between each other. On the right side of the figure they are illustrated
in a simplified way, where a winding is represented with an inductor.
The arrowed lines attached to the windings define the directions of the
currents in the windings in a form of vector. The three current currents are
named as ia, ib and ic. The vector magnitudes are equal to the current
magnitudes, respectively.
It must be noticed that the vector directions are not the directions of the
physical currents flowing in the windings. The vectors defined with directions
are only used for analysis.
At a certain time point, e.g., the currents in the three windings have the
directions and values as shown in Figure 4.3; they could be summarized
together using vector addition. The result is ia, ib and ic as shown in the figure
5.3.

Figure 5.3: Addition of the current vectors
5.2 Definition
Referring to Figure 5.3, the sum of the current vectors as
Isum= + + = + + (1)
Using the Euler’s formula,
√ √
I sum= (2)
The transformation of space vector is directly derived from the sum of

the vectors. Based on the equation above, we put a coefficient 2/3 to I sum, in
order to keep constant magnitude of the vectors during the transformation.
The space vector transformation is thus defined as,
⃗s= I sum= (ia+αib+α2ic)=is (3)
Where, and =
And ⃗s is the magnitude and θ is the angle of the space vector. Instead
of the coefficient, 2/3, other coefficients could also be used, e.g.√ , for
constant power transformation. In this exercise we always use coefficient, 2/3,

for constant magnitude transformation.

The above equation represents a transformation from a quantity in

three-phase system to a vector in a complex plane. Principally, the space
vector is a pure analytical quantity and independent of any physical system.
Actually, any three-phase system can be described with space vectors.
However, one can combine this complex plane with the machine model
for analysis. For that, the plane will be fixed on the stator, with the real axis
coinciding with the u-direction of the stator. It must be noted that the complex
plane is attached to the magnetic field model, not to the physical rotor.
5.3 Graphic illustration of the rotation

The currents defined in (3) can also be illustrated graphically as
following, where the initial phase angle is zero.
Figure 5.4: Phase currents in the machine windings

The current space vector is a rotating vector. This is illustrated with figures
5.5.

Figure 5.5: Graphic show of the rotating current space vector
In this chapter, different techniques have been exploited by using the

space vector form of stator phase voltage and current to classify the rotor
broken bar fault and stator inter-turn short circuit fault. These techniques have
gained popularity over other techniques due to their generalization capability,
which means that they are able to perform satisfactorily even for unseen fault.

Chapter 6: Methodology
6. METHODOLOGY
6.1 Introduction
When the induction motor is in the faulty condition such as if the motor
rotor bar get damage or inter-turn short circuit occurring in one or more
phases in the winding then such types of induction motor is represented as an
unsymmetrical three phase system. The analysis of such system can be
achieved by using symmetrical component or space vector form namely
positive, negative and zero sequence components. In this chapter, different
techniques have been exploited with the help of space vector form of stator
phase voltage and current to classify the rotor broken bar fault and stator
inter-turn short circuit fault. These techniques have gained popularity over
other techniques due to their generalization capability, which means that they
are able to perform satisfactorily even for unseen fault.
In this dissertation three types of techniques were used to classify the
different types of faults occurred in the induction motor which are based on
the space vector form of stator current and voltage signal. These are:
1. Fault classification using Pendulous oscillation.
2. Fault classification by using Negative Sequence current and swing
angle.
3. Fault classification by using Artificial neural network
6.2 Fault classification using Pendulous oscillation

6.2.1 Phenomenon
In healthy condition, Stator of Induction motor can be represented as a
symmetrical three-phase circuit having equal in magnitude with the phase
displacement of 120° with each other. In these conditions, the magnetic fields
of an induction machine rotate at synchronous speed. However, any
asymmetry occurred in the stator windings or rotor bars, which may be due to
manufacturing imperfections or machine failure or by misalignment and mainly
to incorrect assembly, disturbs the air-gap magnetic field causing the air-gap
magnetic field to oscillate around its original synchronously rotating axis.
Although, this oscillation may exist even for a healthy machine due to the
machine structural imperfections, this oscillation will be significant and

detectable in a case of internal failure in induction machines such as broken

bar and inter-turn faults. This oscillation is known as the pendulous oscillation.
It has been shown that the range of this pendulous oscillation progressively
increases with the increase in the number of broken bars [7]. It will be shown
here that the range of this pendulous oscillation also progressively increases
in proportionality with the circulating loop current magnitude in the shorted
coil. However, the pendulous oscillation caused by inter-turn shorts is different
in its nature from the pendulous oscillation caused by broken bar faults [2].
6.2.2 Motor Fault classification Using Pendulous Oscillation

Phenomenon
When the Induction motor is in the faulty condition such as if the

interturn short circuit occurring in one or more of the phases in the stator
winding then such types of induction motor is represented as an
unsymmetrical three-phase system. It is well known in power system, the
analysis of such asymmetrical system can be achieved by using symmetrical
component or space vector namely Positive, Negative and Zero sequence
components. And most of the investigation for online fault diagnosis of
Induction Motor are done by calculating the Negative Sequence current and
Negative and Zero sequence Impedance [11] [12].
The different shape of the pendulous oscillation phenomenon is

obtained from the different types of faults. The shape of the pendulous
oscillation describe the types of fault occurred in the machine. This pendulous
phenomenon can be achieved from the angular phase shift between the
space vector of stator currents and voltages are plotted in a polar coordinated
in such a way that the radius indicates the absolute value of the real part of
the space vector of the stator current. The space vector can be achieved by
the calculating negative sequence current and voltage of the measured
quantities of phase current and voltage respectively. The space vector voltage
⃗⃗⃗ s and current s can be expressed from (3) as follow:

⃗⃗⃗ s(t)= (va(t)+αvb(t)+α2vc(t)) (4)
s(t) = (ia(t)+αib(t)+α2ic(t)) (5)
Where, α=exp(j2π/3) is the space vector transformation operator.
And va(t), vb(t), vc(t) and ia(t), ib(t), ic(t) are phase voltages and phase
current of phase a, b, c respectively. After defining the space vector of stator
voltage and current, the angular phase shift called swing angle and the
absolute value of the real part of the space vector of the stator current at any
point (r, δ) in the polar plot is determined by following expression.
r(t)=abs(Re( s(t) )) (6)
δ(t)= s(t) ⃗⃗⃗ s(t) (7)
The flowchart for fault classification by using pendulous oscillation

phenomenon is shown in figure 6.1.

Figure 6.1: Flowchart for fault classification by using pendulous oscillation
As the simulation results for each condition generates a different shape

of the trace of the ⃗⃗⃗⃗⃗ =r δ vector, which can be used for fault classification
purposes. The differences in these shapes can be explained based on the
physics of each of the above-mentioned conditions.

6.3 Fault classification by using Negative sequence current

and swing angle.
Symmetrical (unfaulted) motors powered by symmetrical multiphase

voltage sources will have no negative sequence currents flowing in the leads.
A turn-to-turn or rotor fault will break that symmetry and give rise to a negative
sequence current which may then be used as a measure of fault severity [4].
As the severity increases, the amplitude of the negative sequence
currents also increases. For different types of fault the amplitude of negative
sequence current is also different depends on the severity. Therefore the
peak values of an on-line trace of the absolute values of real part negative
sequence component of the stator currents ( r ), which were obtained for each
cycle, for different condition are used to classify the motor weather it is
healthy, broken rotor bar or stator inter-turn short circuit condition in a
graphical manner.
Similarly the swing angle is the peak to peak value of δ is known as swing
angle ‘ and it is expressed as
= max - min (8)
Machine failure disturbs the air gap magnetic fields which is oscillates
around its originally synchronous axis. This oscillation can be defined by
measuring the phase displacement between the stator voltage and current
which is denoted by δ. The peak to peak values of δ is known as swing angle
express in (5).
For classification of different condition of motor such as healthy, rotor

broken bar and stator inter-turn short circuit condition, the peak quantity of r
and obtained for each cycle is used. The chart to obtain the peak quantity
of r and form each cycle for the stream of the 50 cycles are shown in
Figure 4.7

Figure 4.7: Flow Chart to define peak quantity of r and

6.4 Fault Classification by Artificial Neural Network.
Multilayer perceptron is one of the most popular neural network

structure were used for the classification purpose. Perceptron is the basic
operational unit of artificial neural networks. It employs supervised learning
rule and is able to classify the data into different classes. This network with a
simple architecture may be used for both modeling and classification tasks.
The feed-forward architecture is used to interconnect the layers from the input
layer towards the output layer. The multilayer feed-forward architecture is
shown in figure 4.9.
Figure 4.9: Multilayer feed-forward ANN Network
The number of neurons in the input and output layers is governed by the
number of inputs and outputs of the pattern to be recognized. However, the
number of neurons in the middle layer can be selected depending upon the
applications. Input patterns are exposed to the network whose output is
compared with the target values to calculate the error which is corrected in the
next pass by adjusting the synaptic weights to get desired output.
In the proposed work, a three-layer feed-forward neural network is

selected for fault diagnosis of an induction motor as this problem of fault
diagnosis is likely a highly complex nonlinear mapping problem because both
the inputs and outputs are multiple variables without clear linear relationships.

Chapter 7: Experimental Setup and Data Generation
7 EXPERIMENTAL SETUP AND DATA

GENERATION
7.1 Introduction
Stator winding faults are one of the most important causes of faults in
induction motors. Such faults are caused by several types of stress such as
thermal, mechanical, electrical and environmental acting on the insulation
system. All these stresses interact with each other in such a way that to
degrade in the insulation system. Different types of stator faults [1] can
develop under such stresses. From which inter-turn short circuit fault is one of
the most common types of stator fault [3-4]. In most cases, the inter-turn short
circuit fault progresses to a coil to coil, phase to phase, or phase to ground
fault, causing the final breakdown of the motor.
The squirrel cage of an induction motor consists of rotor bars and end
rings. If one or more of the bars is partially cracked or completely broken, then
the motor is said to have broken bar fault. It has been observed that in
squirrel-cage induction motor rotor asymmetry occurs mainly due to
manufacturing defect, such as during the brazing process non-uniform
metallurgical stresses may occur in cage assembly which led to failure during
rotation of the rotor. Also heavy end rings of rotor result in large centrifugal
forces which may cause extra stresses on the rotor bars
This chapter presents an experimental set up for capturing the
induction motor phase voltage and current under healthy, rotor broken bar
and stator inter-turn short circuit fault conditions. For this experimental set up,
we have used a 2-hp squirrel cage induction motor. The circuit diagram of
experimental setup is shown in figure 7.1. And Table 7.1 shows the required
equipment to perform experiment for capturing the instantaneous values of
Induction motor phase voltage and current signal.

Figure 7.1: Circuit diagram of experimental setup
Table 7.1: Equipment Required
Sr. No. Name of Equipment specification

1 3-Ph Induction Motor 2HP/1.5kW, 415v, 3.80Amp, 1380 RPM
2 DC-Generator (as load) 2.237kw, 230v, 9.7Amp, 1500RPM
Input range :0.01Amp-5Amp AC,
3 Current transformer
Output:400mV/A
4 Potential Transformer 230V/6V, 200mA, step-down transformer
16-bit USB 2.0-based , Input range of ± 10V,
5 Adlink DAQ USB-1900
Up to 250 kS/s analog input
6 Dimmerstate 3-phase, voltage range (0-600V)
7 Resistor Bank 5kw, 250V, for single phase
2nd generation, i-5, windows 8. with
8 PC
MATLAB R2010a software

7.2 Experimental setup

The test were performed at various situation such as for healthy,
broken rotor bar and stator inter-turn short circuit condition. The experimental
setup to obtain the stator voltage and current signal is shown in figure 7.2.
Figure 7.2: Experimental setup
1. 3-phase Induction Motor

2. D.C. Generator
3. Current Transformer
4. Potential transformer
5. Adlink DAQ USB-1902
6. Resistor Bank
7. Computer with interfacing with DAQ
1. Induction Motor
The experimental set up is developed by reconfiguring of an existing
induction motor rated at 2-hp. It has 24 coils, 36 slots in all. Each phase
comprises 8 coils carries 300 turns. A phase has been tapped where each
tapping is made after 10 turns, near to the star point (neutral).
In order to emulate stator inter-turn short-circuit faults, the motor has a
phase winding that was prepared with taps for the purpose of “experimental

mimicking” of incipient inter-turn faults. Therefore one of the three phases has
been tapped where each tapping is made after every 10 turns near to the star
point (neutral).The tapings are drawn from the coils where each group
comprises of approximately 70 to 80 turns. The arrangement for tapings is
Figure 7.3: Taping Arrangements

The number of taps to be soldered in the windings is restricted by the amount
of space available inside the motor housing. These taps are specially added
at the motor terminal of one of the phases since the stator faults are likely to
occur closest to the terminal end of the windings to limit the short-circuit loop
current. The design characteristics of this reconfigurable induction motor are
given in Table 7.2.
Table 7.2: 2-HP Induction Motor Characteristics

Motor Parameter Specification
Power (HP) 2 HP
Voltage 440 Volt
Current 3.80 Amp
Speed 1350 RPM
No. of Pole 4 pole
No. of Coil per phase 8 Coils

No. of turns per phase 300 turns

No. of turns per Coil 37 turns
No. of stator slot 36 slots
2. D.C. Generator
Device which converts mechanical energy to electrical energy is known

as a Generator. And DC generators field coils are self-excited generators
which get started with the initial current in the field coils. When generator is
switched off, a small magnetism is developed in rotor iron which induced
electromotive force in the armature due to which current is produced in the
field windings. Initially, weak magnetic field creates less current in the coil, but
to sustain self-excitation, the additional magnetic flux increases the
electromotive force in the rotor, due to which voltage keep on increasing until
the machine takes the full load. The small amount of magnetism is present in
the rotor iron. This residual magnetic field of the main poles, induced
electromotive force in the stator coils, which produces initial current in the field
windings. Due to flow of small current in the coil, an increase in magnetic field
occurs. As a result, voltage output increases, in turn, increases the field
current. This process continues as long as the electromotive force in the
armature is more than the voltage drop in the field winding. But, after at
certain level, field poles get saturated and at that point electric equilibrium is
reached and no further increase in armature emf and increase in current. The
resistance of the field winding has certain fixed value, at which self-excitation
can be achieved. This resistance value may vary according to electric
parameters of the generator.
In this experiment, self-excited dc generator was used as a load of an

Induction Motor. The motor was loaded at no load to full load conditions. For
loading an induction machine from no load to full load, the dc generator
loaded by switching the resistor bank. The specification of dc generator is
shown in table 7.3

Table 7.3: Self Excited DC Generator Characteristics

Motor Parameter Specification
Power (kw) 2.2 kw
Voltage 230 Volt
Excited voltage 220 Volt
Current 9.7 Amp
Speed 1500 RPM
No. of Pole 4 pole
3. Current Transformer
A CT is an instrument transformer in which the secondary current is
substantially proportional to primary current and differs in phase from it by
ideally zero degree. The secondary AC is commonly used in two ways:
 To measure the size of the current in the primary circuit
 To supply electronic equipment this would otherwise be damaged by
very large currents
Current transformers allow delicate equipment to be protected from
significant currents while simultaneously allowing these pieces of equipment
to be powered by those currents. This happens because current transformers
produce a secondary current via electrical induction.
To capture the motor current, Fluke i-5s current transformer known as
current clamper were used shown in Figure 7.4. The specification of fluke i-5s
current clamper is shown Table 7.4.
Figure 7.4: FLUKE AC current clamp

Table 7.4: Specification of FLUKE i-5s AC current clamp
Measurement Range 0.01A-5A

Output 400 mV/A
0.01A to 1 A 1% + 5 mA
Accuracy (48 Hz to 65 Hz)
1 A to 5 A 1%
Working Voltage 600 V ac rms
Duty Cycle 0.01 to 6 A continuous
4. Potential Transformer
A voltage transformer or potential transformer is just like a general
purpose step down transformer shown in figure 7.5. Primary of this
transformer is connected across the phase and ground. Just like the
transformer used for stepping down purpose, potential transformer i.e. PT has
lower turns winding at its secondary. The system voltage is applied across the
terminals of primary winding of that transformer, and then proportionate
secondary voltage appears across the secondary terminals of the PT.
In this experimental setup, 230v/6v, 200mA potential transformer was
used for capturing the phase voltage of an induction motor.
Figure 7.5: Potential transformer
5. Adlink DAQ USB-1902

The current and voltage signals are captured at sampling frequency
of 1 KHz with the help of ADLINK DAQ Which is shown in Figure 7.6.

Figure 7.6: ADLINK DAQ USB-1902
.
The USB-1902 Series of 16-bit can import maximum input samples of
250 kS/s. The series also delivers 2-CH, 16-bit analog output capable of up to
1 MS/s update and programmable function I/O. The software programmable
function I/O supports a variety of functions including TTL digital I/O, general-
purpose timer/counter, and PWM output. The USB-1902Series’ analog input,
analog output, and function I/O are capable of functioning simultaneously at
full speed. The data is captured with the help of ADLINK DAQ compatible with
MATLAB. The ADLINK DAQ has 10 input ports and 10 output ports. Each port
has maximum voltage rating of 10 volts.
7.3 Data Generation

Many tests were performed on 2-hp Squirrel cage induction motor in
the laboratory under full load condition on various condition of motor such as
healthy, broken rotor bar and inter-turn short circuit. Under normal operation
and balanced conditions, phase voltages and currents equal in magnitude and
shifted by 1200 electrical due to the induction motor parameters. However,
under faulty operation, the phase currents are unequal and also their phase
shifts. The different phase currents of the induction motor under both before
and after the inter-turn short circuit fault have been investigated through
simulation study using MATLAB software.

7.3.1 Healthy Condition

2 H.P motor is fed by three-phase balanced power supply. The
motor is varied from no load to full load. In this case, the magnetic field of an
induction motor is rotate at synchronous speed. There is no disturbance
occurred in the air gap magnetic field so the motor run at its rated speed with
small oscillation which is due to manufacturing imperfection only. In this case,
Stator phase voltage and current signals are captured for full load conditions
at a sampling frequency of 1 kHz are shown in figure 7.7.
Figure 7.7: voltage and current signal under healthy condition

7.3.2 Broken Rotor bar Fault

The induction motor under test has 32 rotor bars, to carry out the rotor
broken bar test, five rotor bars are broken on both sides of end rings and
stator current signals are captured at full load conditions. Figure 7.8 shows
the stator phase voltage and current signals from the motor terminals at full
load during broken bar fault conditions.
Figure 7.8: voltage and current signal under broken rotor bar condition.
7.3.3 Stator inter-turn short circuit

This fault is due to failure of insulation of the stator winding. It is mainly
termed as inter-turn short-circuit fault. Different types of stator winding faults
are
(i) Short circuit between two turns of same phase called turn-to-turn fault
(ii) Short circuit between two coils of same phase called coil to coil fault,
(iii) Short circuit between turns of two phases called phase to phase fault,
(iv) Short circuit between turns of all three phases

(v) Short circuit between winding conductors and the stator core called coil
to ground fault,
(vi) Open-circuit fault when winding gets break.
Different types of stator winding faults are shown in Figure 7.9. Short-
circuit winding fault shows up when total or a partial of the stator windings get
shorted. Open-circuit fault shows up when total or a partial of the stator
windings get disconnected and no current flows in that phase/line
Figure 7.9: Star-connected stator showing different types of winding

fault
In this test, stator windings of induction motor were modified to have

several accessible tapings that can be used to introduce short circuits. For
this experimentation phase A is tapped, where each tapping is made after 10
turns. Different experimentations were conducted with 20 turns short circuited
in phase A of motor and for different loading conditions. Phase voltage and
stator current signals captured are shown in figure 7.10 for 20 turn’s stator
short circuit condition.

Figure 7.10: voltage and current signal under stator inter-turn short
circuit condition.
In this way, experimentally, the induction motor phase voltage and
current are captured under healthy, broken rotor bar and stator inter-turn short
circuit condition at full load.
As seen from figure 7.7, 7.8 and 7.10, it may observed that the
variation in voltage signal for healthy, rotor broken bar and rotor short circuit
condition is negligible small but in the current signal, there is a variation for
healthy, broken rotor bar and stator inter-turn short circuit condition.
These captured signals of induction motor stator phase voltage and
current then further used for fault classification purpose.

Chapter 5 Result and discussion
8 RESULTS AND DISCUSSION

In this dissertation the tests were performed at various situations such
as for healthy, broken rotor bar and stator inter-turn short circuit condition and
the proposed methods were applied to the obtained signal of induction motor
stator current and voltage for the classification of healthy, broken rotor bar
and stator inter-turn short circuit condition. The results obtained are:
8.1 Pendulous Oscillation Phenomenon

As representative examples of the many tests performed on 2-hp
Squirrel cage induction motor in the laboratory, the pendulous oscillation of
the 2-hp induction motor in polar co-ordinate plot are shown below
respectively for healthy, with five broken rotor bar and stator inter-turn short
circuit with 20 turn condition.
8.1.1 For Healthy Condition

A three phase Induction motor which is energized by the balance three
phase power supplies. As the motor works on the healthy condition, there will
be no oscillation occurred in the system. The polar plot for healthy condition is
Figure 8.1: pendulous oscillation in healthy condition

In healthy condition, the phase shift between the terminal voltage and current
is very small nearly unity. Therefore, the polar plot of (r, δ) is a straight line.
8.1.2 For Broken Rotor Bar

In this test, the three phase induction motor with broken rotor bar
energized by a three phase balance power supply. The induction motor has
32 rotor bars, to carry out the rotor broken bar test; five rotor bars are broken
on both sides of end rings. Figure 8.2 shows the polar plot for the rotor broken
bar at full-load condition.
Figure 8.2: pendulous oscillation for Broken rotor bar
Here, for any angle r and δ quantities changing from zero to its maximum
values [1]. Therefore, in this case, polar plot of (r,δ) has a filled petal shape
8.1.3 Stator Inter-turn Short-Circuit

An Induction motor with stator inter-turn fault is energized by a three
phase balance power supply. For inter-turn fault test, the stator windings of
induction motor were modified to have several accessible tapings that can be
used to introduce short circuits. For this experimentation phase A is tapped,
where each tapping is made after 10 turns. Different experimentations were
conducted with 20 turns short circuited in phase A of motor and for different

loading conditions. So the air gap magnetic field gets oscillated due to the
short circuited winding only. So that the air gap magnetic field can be resolved
into two components in which rotate at the synchronous speed but in opposite
direction. And also, the resultant magnetic field can be resolved into a main
rotation at synchronous speed and an oscillation around the main rotation,
where the speed of this oscillation is twice the synchronous speed [1]. The
speed of the variation of r and δ quantities is same so that at any angle of δ,
the r quantity will not changes from zero to its maximum values because r and
δ are varying with time simultaneously at the same rate. Therefore in this
case, the polar plot of (r, δ) has unfilled petal shape in other word constitutes
only the outer boundary of the petal shape. Figure 8.3 shows the experimental
result of polar plot at full-load for 20-turn short circuit condition.
Figure 8.3: pendulous oscillation with stator inter-turn short circuit
It is observed that, practically it is not possible to get an exact straight (radial)

line on polar coordination of pendulous oscillation on healthy condition due to
minor fluctuation are presents and also this is due to construction
imperfections of 2-hp induction motor. It is also observed that a slight degree
of curvature occurred in the plot this is due to the presence of internal stator
winding resistance of an induction motor.

8.2 Fault Classification using Negative Sequence current

and swing angle
In this method, the peak swing angle and peak values of negative
sequence current both obtained for each cycle were used. The swing angle,
Δδ, is obtained by calculating the maximum spread of Δδ = δmax − δmin in a
polar plot diagram. Thus, for each power cycle, a single value for the swing
angle, Δδ, is obtained. The results of a peak values calculated swing angle for
different condition in an Induction motor are shown in Figure 8.4, 8.5 and 8.6
for healthy, broken rotor bar and stator inter-turn short circuit condition
respectively and figure 8.7 shows the comparative plots for healthy, rotor
broken bar and stator inter-turn short circuit condition.
Figure 8.4: Δδ for Healthy

Figure 8.5: Δδ for rotor broken bar
Figure 8.6: Δδ for Stator Inter-turn short-circuit
Figure 8.7: fault classification using Δδ
Similarly, at the same conditions, the peak values of an on-line trace of the
negative sequence component of the line currents, which were obtained for
each cycle, for different condition in an Induction motor are shown in Figure

8.8, 8.9 and 8.10. And by comparing all the plots the classification is done
shown in figure 8.11
Figure 8.8: In(max) for Healthy
Figure 8.9: In(max) for Broken rotor bar

Figure 8.10: In(max) for Stator Inter-turn short circuit
Figure 8.11: fault classification using In(max)
Hear the classification of healthy condition and broken rotor bar is quite
difficult with swing angle curve method because of the closeness or
“bunching-up” nature of the traces. This is because of the fact that the
rewound 2-hp motor had an inherent manufacturing degree of unbalance due
to the manufacturing (or construction) imperfection resulting from the random
winding installation process and the layout of the winding taps. But with the
help of peak values of Negative sequence current, there are no difficulties to
classify the different types of condition. So that for the classification purpose,

by using peak value of Negative sequence current per cycle has better
accuracy than swing angle curve.
8.3 Artificial Neural Network

A multilayer neural network includes an input layer, an output layer,
and one or more hidden layers. Each layer may include many neurons. A
neuron in each layer of the network is connected to all the nodes or neurons
in the previous. Multilayer perceptron (MLPs) are layered feed forward
networks typically trained with static back propagation. It consists of one input
layer, one hidden layer and one output layer. Randomized data is fed to
neural network and number of processing elements in hidden layer are varied
and decided by trial and error method.
The directly measured signals are not suitable for detection, since large
number of samples is burden for processing and computation. So for fault
characterization suitable features are extracted and selected such that they
preserve the critical information for decision making. In present study motor
line current has been analysed by using modern data reduction technique.
The success of FFANN to distinguish healthy and faulty condition of motor
depends upon pre-processing of input data. The input is such that it contains
the concentrated information, consequently the results are accurate.
In this dissertation, three-layer FFANN is used and trained with a
supervised learning algorithm called back propagation. FFANN consists of
one input layer, one hidden layer and one output layer. The input layer
consists of 2 processing elements. The output layer consists of three
processing elements representing healthy condition, rotor bar and inter-turn
fault.
For generalization, randomized data are fed to the network, Tanigmoid
transfer functions is used for training the network and percentage accuracy of
classification is obtained. For this transfer functions, the network is trained
with maximum Iteration of 500, training data assumed is 60%, testing data is
40%, step size=0.8 and momentum of 0.7. The minimum squared error is
considered as a discriminating factor for the classification of different
conditions of the motor. With these assumptions the percentage accuracy of
classification for induction motor under healthy, broken rotor bar and inter-turn

fault conditions with respect to the number of processing elements in the

hidden layer is obtained.
The fault classification results for three phase induction motor under
full-load condition are shown in table 8.1 and figure 8.12.
Table 8.1: Variation Of Hidden Neuron And Percentage Accuracy
Percentage accuracy of classification

No. of processing
Elements Broken Rotor
Healthy Inter-turn short-circuit
bar
1 00% 100% 100%
2 100% 00% 100%
3 100% 100% 100%
Figure 8.12: percentage accuracy with variation of neurons
It is seen that the Tansigmoid transfer function with three processing

element is capable of classification of different faults in an induction motor
with 100% accuracy.

Chapter 6 Conclusion And Future Scope
9 CONCLUSIONS AND FUTURE SCOPE
9.1 Conclusion
The common types of faults in induction motor are studied and various types
of current based condition monitoring and fault detection techniques are
reviewed. A robust and intelligent fault detection and classification of three-
phase induction motor based on pendulous oscillation concept and by using
Artificial Neural network has been introduced and examined with the 2-hp,
three phase Induction motor. The experimental results showed the strength
and fidelity of this method, even in the presence of a degree of machine
construction imperfections. The pendulous oscillation techniques gives the
accurate classification of healthy, rotor broken bar and stator interturn short
circuit condition of motor which shown that the negative sequence of stator
current is the responsible one to detect rotor broken bar fault and stator
interturn faults of machine which further used in ANN for classification
purpose. And also observed that Feed-forward ANN with the supervised
learning algorithm and Tansigmoid transfer function and with three processing
elements in the hidden layer is the best network to classify multiple faults in
an induction motor with 100% accuracy.
9.2 Future Scope

In the present work, several soft computing techniques have been
presented for the detection and location of stator inter-turn short circuit fault in
the stator winding of an induction motor.
 Pendulous oscillation phenomenon is successfully used to detect rotor
broken bar and stator inter-turn short circuit fault together with identification
the severity of this fault of an induction motor. The same approach can be
extended to identify the other faults such as bearing fault, and eccentricity
related fault of an induction motor.
 As the induction motor has the symmetrical stator winding at healthy
condition and fault classification techniques totally depends on disturbance of
symmetry of an induction motor. So proposed method can be extended to

Chapter 6 Conclusion And Future Scope
identify the faults at all three symmetrical winding motors such as three phase
synchronous motor and also three phase generator.
These techniques can also be extended for identification of synchronism
test in generator with the grid supply.
 Neural network ensembles are receiving increasing attention in the
field of research such as control, fault diagnosis, decision making,
identification, robotics etc. due to their learning and generalization abilities,
nonlinear mapping, and parallelism of computation. However, for solving fault
detection problems the neural networks may get stuck on a local minimum of
the error surface, and the network convergence rate is generally slow. A
suitable approach for overcoming these disadvantages is the use of wavelet
functions in the network structure. Wavelet function is a waveform that has
limited duration and an average value of zero. A wavelet neural network has a
nonlinear regression structure that uses localized basis functions in the
hidden layer to achieve the input-output mapping.
The integration of the localization properties of wavelets and the learning
abilities of neural network results in the advantages of wavelet neural network
over neural network for the detection and location of an inter-turn short circuit
fault in the stator winding of an induction motor.

References
REFERENCES
[1] J.Stack, T.G.Habetler, and R.G.Harley, ―Fault classification and fault
signature production for rolling element bearings in electric machines,‖
IEEE Transactions on Industry Applications, vol. 40, no. 3, pp. 735–
739, May/June 2004.
[2] B Mirafzal,R. J. Povinelli,A.O. Demerdash , ―Interturn Fault Diagnosis
in Induction Motors Using the Pendulous Oscillation Phenomenon‖
IEEE transactions on energy conversion, vol. 21, no. 4, december
2006.
[3] Singh GK, Al Kazzaz, ―Induction machine drive condition monitoring
and diagnostic research—a survey‖. Electr Power Syst Res 64(2):145–
158, 2003.
[4] G.B.Kliman, W.J.Premerlani, R.A.Koeg1 & D.Hoevveler, ―A New
Approach to On-Line Turn Fault Detection in AIC Motors‖, 0-7803-
3544-919$65 .00 0 1996 IEEE, pp-687-693.
[5] M. Sahraoui, A. Ghoggal, S. E. Zouzou A. Aboubou, H. Razik
―Modelling and Detection of Inter-Turn Short Circuits in Stator Windings
of Induction Motor,‖ IEEE Trans. Energy Convers. vol.EC 1, issue 1,
pp. 4981-4986, 2006.
[6] S. Bachir, S. Tnani, J. C. Trigeassou, and G. Champenois, ―Diagnosis
by Parameter Estimation of Stator and Rotor Faults Occurring in
Induction Machines,‖ IEEE Trans. Ind. Electron., vol. 53, issue 3, pp.
963–973, 2006.
[7] B.Mirafzal and N. A. O. Demerdash, ―Inductionmachine broken-bar
fault diagnosis using the rotor magnetic field space vector orientation,‖
IEEE Trans. Ind. Appl., vol. 40, no. 2, pp. 534–542, Mar./Apr. 2004.
[8] V. N. Ghate and S. V. Dudul ―Cascade Neural-Network-Based Fault
Classifier for Three-Phase Induction Motor,‖ IEEE transactions on
industrial electronics, vol. 58, no. 5, may 2011, pp. 1555-1562.
[9] H. Nejjari, M. H. Benbouzid, ―Monitoring and Diagnosis of Induction
Motors Electrical Faults using a Current Park’s Vector Pattern Learning
Approach,‖ IEEE Trans. Ind. Appl., vol. 36, no. 3, 2000.
SSGMCE, Shegaon. Page 67

References
[10] M. B. K. Bouzid, G. Champenois, N. M. Bellaaj, L. Signac, and K.

Jelassi, ―An Effective Neural Approach for the Automatic Location of
Stator Inter-Turn Faults in Induction Motor,‖ IEEE Trans. Ind. Electron.,
vol. 55, no. 12, pp. 4277-4289, 2008.
[11] J. L. Kohler, J. Sottile, and F. C. Trutt, ―Alternative for assessing the
electrical integrity of induction motors,‖ IEEE Trans. Ind. vol. 28, no. 5,
pp. 1109–1117, Sep./Oct. 1992.
[12] S. B. Lee, R. M. Tallam, and T. G. Habetler, ―A robust, on-line turn-fault
detection technique for induction machines based on monitoring the
sequence component impedance matrix,‖ IEEE Trans. Power
Electron., vol. 18, no. 3, pp. 865–872, May 2003.
[13] J. M. Aller, A. Bueno and T. Paga, ―Power System Analysis Using
Space–Vector Transformation‖, IEEE Transactions On Power Systems,
Vol. 17, No. 4, November 2002,pp.957-965.
[14] Howard Demuth and Mark Beale, ―Neural Network Toolbox‖, For Use
with MATLAB , The Mathworks, user’s guide version 4.
[15] Christopher M. Bishap, ―Neural Network for Pattern Recognition‖,
Department of computer science, university of Toronto, clarendon
press.OXFORD, 1995, pp. 1-6,98-100,116-161.
SSGMCE, Shegaon. Page 68

Document 0

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Document 0

Uploaded by

Copyright:

Available Formats

Neural Approach for Fault Classification in Induction

Motors by Using Motor Current and Voltage

Under the guidance of

DEPARTMENT OF ELECTRICAL ENGINEERING

E-mail.principal@ssgmce.ac.in Phone No.07265-252116, 252216

hod_electrical@rediffmail.com Fax No. 07265-252346, 253602

The foregoing dissertation entitled, “Neural Approach for Fault

Internal Examiner External Examiner

I first pray to our divine source of inspiration “Shri Sant Gajanan

Place: Shegaon Praful Dewanand Thamke

Induction motor is playing a key role as prime-mover or energy

SSGMCE, Shegaon Page 1

1.2 Condition Monitoring

SSGMCE, Shegaon Page 2

(1) Detection of broken-bar faults, when a motor is energized by Three

1.3 Aim Of The Work

SSGMCE, Shegaon Page 3

GB.Kliman proposed tum-to-tum fault detector technique that has a

Sahraoui, Ghoggal, Zouzou, Aboubou and Razik have proposed a new

SSGMCE, Shegaon Page 4

Bachir, Tnani, Trigeassou and Champenois have suggested a new model

Behrooz Mirafzal developed a new technique based on the “swing-like”

Behrooz Mirafzal again proposed a robust interturn fault diagnostic method

SSGMCE, Shegaon Page 5

Vilas N. Ghate and Sanjay V. Dudul proposed a new approach to intelligent

SSGMCE, Shegaon Page 6

Figure 3.1: Magnetic circuit of stator and rotor of an induction motor

SSGMCE, Shegaon Page 7

Figure 3.2: An induction motor

Figure 3.3: Stator of an Induction motor.

SSGMCE, Shegaon Page 8

1 Squirrel cage type

Figure 3.4: Squirrel cage rotor

SSGMCE, Shegaon Page 9

Figure 3.5: wound Rotor

SSGMCE, Shegaon Page 10

When the stator winding of an induction motor is connected to a three-

In transforming this electrical energy into mechanical energy, in an

 Friction and windage losses, 5–15 %

SSGMCE, Shegaon Page 11

Table 3.1: Various classes of squirrel-cage induction motor

Class A Class B Class C Class D

3.4 Faults: Causes and Effects

SSGMCE, Shegaon Page 12

Figure 3.6: Types of fault in Induction motor.

Electrical faults/asymmetries can be further categorized into rotor and

Table 3.2: Percentage failure of Induction motor.

IEEE-IAS survey EPRI Survey

SSGMCE, Shegaon Page 13

3.4.1 Stator Faults

Figure 3.7: damaged stator winding insulation

Such partial stator winding insulation failures may in turn be caused by

SSGMCE, Shegaon Page 14

3.4.2 Rotor related Faults

Although the squirrel-cage rotor is rugged, bar defects do occur,

Figure 3.8: Broken rotor bar

SSGMCE, Shegaon Page 15

4 ARTIFICIAL NEURAL NETWORK.

4.2 Biological Neuron

Figure 4.1: Biological neuron

SSGMCE, Shegaon Page 16

4.3 An Artificial Neuron

Figure 4.2: A basic artificial neuron.

SSGMCE, Shegaon Page 17