Speed Estimation of Induction Motor Using Artificial Neural Network and Implementation in MATLAB

Generated by Foxit PDF Creator © Foxit Software
http://www.foxitsoftware.com For evaluation only.

1
Abstract:
Motors play an important role in daily lifelike in industrial manufacturing and in many other
applications. Induction motors are robust, simple, small in size, low in cost, almost maintenance-
free and possess a wide range of speeds compared to DC motors. However, complexity of signal
processing and poor precision limits its usage. Speed estimation in an induction motor is very
difficult because of its non-linear dynamic nature. Filed Oriented Control or Vector Control
developed by Blaschke(1972) and Vas (1900) respectively is used in motor control. But it gives
incorrect measurement of flux at low speed and lacks robustness. It has high drive cost, low
reliability and noise immunity. Also these methods employ speed sensors. However, the
algorithm of vector control theory requires manipulation of the electric parameters of the motor
so that the governing equations in rectangular coordinates can be developed, prior knowledge of
the state equations is necessary when the estimation theory is used to estimate the speed
precisely. However, the values of the electric parameters may deviate from the designated values
due to changes in the working environment, temperature, speed, external load and noise.
The speed estimation technique employed here is dependent on expressions obtained from the
induction motor dynamic equations. The equations have singularity therefore direct speed
estimation cannot be employed. Two ANNs are used here to recover the speed from these two
equations. The two equations are then combined and singularities are removed. This method is
robust and is easily implementable using commercially available ANN tools.
1
2
LIST OF FIGURES
FIGURE…………………………………………………………………………… PAGE NO.
FIG 1.1 Induction Motor……………………………………………………………….09
FIG 1.2 Coventional Per Phase equivalent circuit…………………………………......12
FIG 1.3 Dynamic equivalent circuit on Stationary Frame…………………………… 14
FIG 1.4 Dynamic Equivalent Circuit on Arbitrary Frame …………………………....15
FIG 1.5 d-q Axis Equation of Induction Motor………………………………………..17
FIG 2.1 Numerator Curve……………………………………………………………...20
FIG 2.2 Denominator Curve…………………………………………………………...20
FIG 2.3 Continuity……………………………………………………………………..21
FIG 2.4 Block Diagram of ANN……………………………………………………….23
FIG 3.1 Neuron Model…………………………………………………………………25
FIG 3.2 MLP…………………………………………………………………………...26
2
3
Table of Contents
CERTIFICATE………………………………………………………02
ACKNOWLEDGEMENT…………………………………………..03
ABSTRACT………………………………………………………….04
LIST OF FIGURES…………………………………………………05
CHAPTERS
1. INDUCTION MOTOR
1.1 Introduction 08
1.1.1 Induction Motor 08
1.1.2 Principle 08
1.1.3 Construction 08
1.2 Speed of Induction Motor 09
1.3 Speed Control of Induction Motor 09
1.4 Dynamic Nature of Induction Motor 11
1.4.1 Per phase equivalent Circuit 12
1.4.2 Stationary Frame Circuit 14
1.4.3 Arbitrary Frame Circuit 15
1.5 d-q Axis Equation 17

1.6 Characteristic Equation 17
2. SPEED FUNCTION 19
2.1 Speed expression 19
2.2 Method of Singular Point 21
2.2.1 Non-Singularity 21
2.2.2 Continuity 21
2.2.3 Square integrable 22
3
4
3. ARTIFICIAL NEURAL NETWORK 24
3.1 Neural Network 24

3.2 Definitions 25
3.3 Function Approximation 27
3.4 Data 27
3.5 Training 30
4. CONCLUSION 49
5. REFERENCES 50
6. APPENDIX 51
A. MATLAB Command 51
B. Plots 52
4
5
CHAPTER 1
INDUCTION MOTOR
1.1 Introduction
1.1.1 Induction Motor
Induction motor is the most popular type of a.c. motor. It is very commonly used for industrial
drives since it is cheap, robust, efficient and reliable. It has good speed regulation and high
starting torque. It has a reasonable overload capacity. Along with variable frequency AC
inverters, induction motors are used in many adjustable speed applications which do not require
fast dynamic response.
1.1.2 Principle:
It works on the principle of electromagnetic induction. A rotating magnetic field is produced

when a 3- phase supply is connected to the 3- phase winding of the stator.
1.1.3 Construction:
A three phase induction motor consists of mainly two parts:
1. Stator
2. Rotor
The stator is the stationary part and the rotor is the rotating part. The stator is built up of high-
grade alloy steel laminations to reduce eddy current losses.
The rotor is also built up of thin laminations of the same material as stator. The laminated
cylindrical core is mounted directly on the shaft or a spider carried by the shaft.
There are two types of induction motor rotors:
1. Squirrel cage rotor
2. Wound rotor
5
6
FIG 1.1
1.2 Speed of induction motor:
Induction motor speed is given by following formula:
Where,
v = speed of rotor
f = frequency of rotor
And, n = number of poles
1.3 Speed control of induction motor:
The main method employed for speed control of induction motor are as follows:
1. Pole changing methods

2. Stator voltage control
3. Supply frequency control
4. Rotor resistance control
5. Slip energy recovery
6
7
1. Pole changing methods:
The number of stator poles can be changed by
﴾a﴿multiple stator windings,
﴾b﴿methods of consequent poles
﴾c﴿pole-amplitude modulation ﴾PAM﴿.
The methods of speed control by pole changing is suitable for cage motors only because the
cage rotor automatically develops number of poles equal to the poles of the rotor winding.
2. Stator voltage control:
The speed of a 3-phase induction motor can be varied by varying the supply voltage .Torque
developed in induction motor is proportional to square of the supply voltage. Speed control
is obtained by varying the supplying voltage until the torque required by the load is
developed at the desired speed.
3. Variable-frequency control:
The synchronous speed of an induction motor is given by:
Ns=120f/P
The synchronous speed and, therefore, the speed of the induction motor can be controlled
by varying the supply frequency.
4. Rotor resistance control:
The speed of wound induction motor can be controlled by connecting external resistance in the
rotor circuit through slip rings. This method is not applicable to cage motors.
7
8
5. Supply energy frequency:
In the rotor resistance control,the slip power in rotor circuit is wasted as I2 R Loss during the
low speed operation. The efficiency of the drive system by this Method of speed control is,
therefore, reduced. The slip power from the rotor Circuit can be recovered and fed back to
the a.c. source so as to utilize it outside the motor. Thus, the overall efficiency of the drive
system can be increased. This method of speed control is used in large power applications
where variation of speed over a wide range involves a large amount of slip power.
1.4 Dynamic model of induction motor:
The concept of vector control has opened up a new possibility that induction motors can be
controlled to achieve dynamic performance as good as that of DC or brushless DC motors. In
order to understand an analyze vector control , the dynamic model of induction motor is
necessary . it has been found that the dynamic model equations developed on a rotating reference
frame is easier to describe the characteristic of induction motors. It is the objective of the project
to derive and explain induction motor model in relatively simple terms by using the concept of
space vectors and d-q variables .when we choose a synchronous reference frame in which rotor
flux lies on the d-axis ,dynamic equations of induction motor is simplified and analogous to a
DC motor.
1.4.1 CONVENTIONAL PER PHASE EQUIVALENT CIRCUIT :
8
9
FIG 1.2
The inductance of rotor and stator is given by:
Inductance of rotor circuit –
Lr = Llr + Lm
Where,
Inductance of stator circuit –
Ls = Lls + Lm,
Where,
Lr = rotor inductance,
Llr = rotor leakage inductance,
Ls = stator inductance,
Lls = stator leakage inductance,
Lm = magnetizing inductance of motor.
If the excitation frequency injected into the stator is ωe and
the actual speed converted into electrical frequency unit is ωo, slip s is defined by,
s = ﴾ ωe – ωo﴿
ωe
9
10
= ω r/ ω e
Where ωr is called the slip frequency, which is the frequency of actual rotor current.
The power consumption in stator circuit is given by,
Is2Rs
The power consumption in rotor circuit and load (output) is given by ,
Ir2Rs/s
Torque produced is given by:
T =Ir2Rr (P/2)(1-s)/s e =Ir2Rr(P/2 e)
Where P is the no. of poles, although the per-phase equivalent circuit is useful in analyzing and
predicting steady-state performance, it is not applicable to explain dynamic performance of the
induction motor. In the next section, we will develop dynamic model of induction motors in
general frame work and introduce several equivalent circuits as special cases.
With space vector notation, voltage equations on the stator and rotor circuits of induction motors
are,
Vs s = Rs Is s + p s s (1)
Vr’ = Rr’ Ir’ + p r’ = 0 (2)
It is very convenient to transform actual rotor variables (Vr’, Ir’,r’) from eqs.2 . on a rotor
reference frame into a new variables ( Vr s, Ir s, r s) on a stator reference frame as in the
derivation of conventional steady-state equivalent circuit.
Vs s = Rs Is s + p s s (3)
0 = Rr Ir s + (p - jωo) r s (4)
Where ωo = p θo, is the speed of the motor in electrical frequency unit and
s s = Ls Is s + Lm Ir s (5)
r s = Lm Is s + Lr Ir s (6)
1.4.2 DYNAMIC EQUIVALENT CIRCUIT ON STATIONARY REFERENCE FRAME:
10
11
FIG 1.3
EQUATIONS:
It is very convenient to transform actual rotor variables (Vr’, Ir’, λr’) from . on a rotor reference
frame into a new variables ( Vr s, Ir s, λr s) on a stator reference frame as in the derivation of
conventional steady-state equivalent circuit. Let the stator to rotor winding turn ratio be n and the
angular position of the rotor be θ, and define
Ir s = (1/n) exp (j θ ) Ir’,
λr s = n exp (j θ ) λr’ (7)
Also, by defining referred rotor impedances as Rr = n2 Rr’, etc., we have
Vs s = Rs Is s + p λs s (8)
0 = Rr Ir s + (p - jωo) λr s (9)
Where ωo = p θo, is the speed of the motor in electrical frequency unit and
λs s = Ls Is s + Lm Ir s (10)
λr s = Lm Is s + Lr Ir s (11)
The above 4 equations (8-11) constitute a dynamic model of the induction motor on a stationary
(stator) reference frame in space vector form. These model equations may be simplified by
eliminating flux linkages as
Vs s = (Rs + Ls p) Is s + Lm p Ir s (12)
0 = (Rr + Lr (p - jωo)) Irs + Lm (p - jωo) Is s. (13)
From (12-13) , The dynamic equivalent circuit model on a stationary reference frame can be
drawn as in Fig.1.2 For steady-state operation with excitation frequency ωe, p in may be
replaced by jωe and after some algebraic manipulation, we get
11
12
Vs s = (Rs + jωe Ls ) Is s + Lm p Ir s (14)
0 = (Rr/ s + jωe Lr) Ir s + jωe Lm Is s. (15)
which exactly describes the conventional steady-state equivalent circuit of Fig. 1.1.
1.4.3 DYNAMIC EQUIVALENT CIRCUIT ON ARBITRARY REFERENCE
FRAME:
FIG 1.4
Equations:
Now, the previous procedure can be generalized so that the dynamic model is described on an
arbitrary reference frame rotating at a speed ωa, where is a special case with ωa,= 0 To do that,
define the new space vector on the arbitrary frame as
Y a = exp(- j a ) Y s (16)
and reconstruct all the model equations in terms of the new space vectors. In the arbitrary
reference frame, Eqs are modified to
Vs a = (Rs + Ls p) Is a + Lm p Ir a + jωa s a (17)
12
13
0 = (Rr + Lr p) Ir a + Lm p Is a + j (ωa - ωo) r a, (18)
with new flux linkage equations defined by,
s a = Ls Is a + Lm Ir a (19)
r a = Lm Is a + Lr Ir a (20)
As before, by substituting Eqs. into Eqs, we have
Vsa = (Rs + Ls (p + jωa)) Is a + Lm (p + jωa ) Ir a (21)
0 = (Rr + Lr (p + jωa - jωo)) Ir a + Lm (p + jωa - jωo) Is a (22)
where eliminated flux linkage variables are eliminated.
The generalized equivalent circuit on a arbitrarily rotating frame based on Eq. is shown in ..
Now, depending on a specific choice of ωa, many forms of dynamic equivalent circuit can be
established. Among them, the synchronous frame form can be obtained by choosing ωa = ωe.
This form is very useful in describing the concept of vector control of induction motors as well
as of PM synchronous motors because at this rotating frame, space vector is not rotating, but
fixed and have a constant magnitude in steady-state. Since space vectors in the synchronous
frame will frequently be used, they are denoted without any superscript indicating the type of
frame.
Another possible reference frame used in vector control is the rotor reference frame by choosing
ωc = ωo which is , in fact, the reverse step of . with n =1.
1.5 d-q axis equations of induction motor:
13
14
FIG 1.5
In many cases, analysis of induction motors with space vector model is complicated due to the
the fact that we have to deal with variables of complex numbers. For any space vector Y, define
two real quantities Sq and Sd as,
S = Sq - j S d .
In other words, Sq = Re (S) and Sd = - Im (S) illustrates the relationship between d-q axis and
complex plane on a rotating frame with respect to stationary a-b-c frame. Note that d- and q-axes
are defined on a rotating reference frame at the speed of ωa = p a with respect to fixed a-b-c
frame.
Fig. 3.1 Definition of d-axis and q-axis on an arbitrary reference frame with the above definition
can be translated into the following 4 equations of real variables expressed in a matrix form.
1.6 Characteristic equation:
The d-q axis dynamic equations for the squirrel cage induction
motor are given by [l]
14
15
V = AI (1)
where,
In the above equations, subscript s denotes stator quantities,
r denotes rotor quantities, q and d refer to the quadrature and direct axis quantities respectively
and L, is the magnetizing inductance.
CHAPTER-2
15
16
SPEED FUNCTION
2.1 Speed Expression:
If the stator voltages and stator currents are known along with the machine parameters. We have
only three unknowns r , ird and irq . We can thus solve for r (speed of induction motor) in
terms of stator quantities only. First , we obtain the rotor currents as function of stator quantities
and r ,from the first two rows of characteristic equation . since the rotor currents are not
accessible in a squirrel cage induction motor . the expressions for ird and irq are obtained as:
ird= 1/Lm [(Vsd-Rsisd)dt- Lsisd]
irq=1/Lm[(Vsq-Rsisq)dt-Lsisq]
We can substitute ird and irq in the last two rows of characteristic equation and obtain the
equations of rotor speed r as:
r = -[2disd/dt – RrLsisd + RrVxddt + LrVxd] /[2isq+LrVxqdt]
Where,
2=Lm2-LrLs,
Vxd=Vsd-Rsisd,
Vxq=Vsq-Rsisq
The speed can be recover from this equation directly but due to singularities in this function it is
difficult to calculate speed for regular induction motor operation.
16
17
FIG 2.1
FIG 2.2
It can be seen from the above figures of numerator and denominator functions that both
waveforms are in phase, resulting in simultaneous zero-crossings, and hence singular points.
Hence we cannot obtain the speed of induction motor directly.
So for calculating the speed of induction motor we use artificial neural network.
17
18
2.1 METHOD OF SINGULAR POINTS :
One of the necessary conditions for an ANN to approximate a function is that the function
should be square integrable, non-linear, singular and continuous.
2.1.1 Non singularity:
A square matrix is nonsingular if Ax = 0n implies x = 0n. Otherwise it is a singular matrix
Properties of singular matrix:
1.A n×n is nonsingular if and only if r(A) = n.
2.A is nonsingular if and only if A has a linear inverse A−1.
2.1.2 Continuity:
FIG 2.3
A continuous function is a function for which, intuitively, small changes in the input result in
small changes in the output continuity of a function in the following intuitive terms: an
infinitesimal change in the independent variable corresponds to an infinitesimal change of the
18
19
dependent variable .We call a function continuous, if, and only if, it is continuous at every point
of its domain. More generally, we say that a function is continuous on some subset of its domain
if it is continuous at every point of that subset.
2.1.3 Square integrable:
A real or complex-valued function of a real or complex variable is square-integrable on an

interval if the integral of the square of its a absolute value, over that interval, is finite.
The basic idea in this method is to partition the main speed function having singularities into
smaller function which do not have any singularities , and to train small ANN’s to identify these
smaller functions. The desired output can be obtained from the outputs of these ANN’s by
avoiding the singular points of the main functions. In this case one of the simplest ways to
partition the functions is to consider their numerators and denominators separately.
N1= - [2disd/dt-RrLsisd+RrVxddt+LrVxd]
D1= 2isq+LrVxqdt
ANN’s can be trained to approximate N1 and D1. The output of these ANN’s can then be passed
through a filter which performs the required division at points where both the numerators and
denominators are non-zero.
Inputs given to the numerator ANN are:
Isd,disd/dt,vsd,vsddt and isddt
Input given to thedenominator ANN are:
vsq,vsqdt and isqdt
19
20
Block diagram of ANN
FIG 2.4
CHAPTER-3
20
21
NEURAL NETWORK
3.1 Neural Network
An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by

the way biological nervous systems, such as the brain, process information. The key element of
this paradigm is the novel structure of the information processing system. It is composed of a
large number of highly interconnected processing elements (neurons) working in unison to solve
specific problems. ANNs, like people, learn by example. An ANN is configured for a specific
application, such as pattern recognition or data classification, through a learning process.
Learning in biological systems involves adjustments to the synaptic connections that exist
between the neurons. This is true of ANNs as well.
Use
Neural networks, with their remarkable ability to derive meaning from complicated or imprecise
data, can be used to extract patterns and detect trends that are too complex to be noticed by either
humans or other computer techniques. A trained neural network can be thought of as an "expert"
in the category of information it has been given to analyse. This expert can then be used to
provide projections given new situations of interest and answer "what if" questions.
Other advantages include:
1. Adaptive learning: An ability to learn how to do tasks based on the data given for training
or initial experience.
2. Self-Organisation: An ANN can create its own organisation or representation of the
information it receives during learning time.
3. Real Time Operation: ANN computations may be carried out in parallel, and special
hardware devices are being designed and manufactured which take advantage of this
capability.
4. Fault Tolerance via Redundant Information Coding: Partial destruction of a network
leads to the corresponding degradation of performance. However, some network
capabilities may be retained even with major network damage.
21
22
FIG 3.1
3.2 Definitions:
Feed forward network:
Feed-forward networks have the following characteristics:

1. Perceptrons are arranged in layers, with the first layer taking in inputs and the last layer
producing outputs. The middle layers have no connection with the external world, and hence
are called hidden layers.
2. Each perceptron in one layer is connected to every perceptron on the next layer. Hence
information is constantly "fed forward" from one layer to the next., and this explains why
these networks are called feed-forward networks.
3. There is no connection among perceptrons in the same layer.
Feedback Network:
By using loops in the network, Feedback networks transfer signals in both directions. Feedback
networks are powerful and complex. Feedback networks state is changing dynamically until they
reach an equilibrium point. Until the input changes, they remain at the equilibrium point.
Feedback architectures are called as interactive or recurrent
Back propagation:
It is a supervised learning method, and is an implementation of the Delta rule. It requires a

teacher that knows, or can calculate, the desired output for any given input. It is most useful for
feed-forward networks (networks that have no feedback, or simply, that have no connections that
22
23
loop). The term is an abbreviation for "backwards propagation of errors". Back propagation
requires that the activation function used by the artificial neurons (or "nodes") is differentiable.
FIG 3.2
Supervised learning:
Supervised learning incorporates an external teacher, so that each output unit is told what its
desired response to input signals ought to be. During the learning process global information
may be required. Paradigms of supervised learning include error-correction learning,
reinforcement learning and stochastic learning.
An important issue concerning supervised learning is the problem of error convergence, i.e. the
minimisation of error between the desired and computed unit values. The aim is to determine a
set of weights which minimizes the error.
Neuralwares Predict:
NeuralWorks Predict is an integrated, state-of-the-art tool for rapidly creating and deploying
prediction and classification applications. Predict combines neural network technology with
23
24
genetic algorithms, statistics, and fuzzy logic to automatically find optimal or near-optimal
solutions for a wide range of problems. Predict incorporates years of modeling and analysis
experience gained from working with customers faced with a wide variety of analysis and
interpretationproblems.
Predict requires no prior knowledge of neural networks. With only minimal user involvement it
addresses all the issues associated with building robust models from available empirical data.
Predict analyzes input data to identify appropriate transforms, partitions the input data into
training and test sets, selects relevant input variables, and then constructs, trains, and optimizes a
neural network tailored to the problem. For advanced users, Predict also offers direct access to
all key training and network parameters.
3.3 FUNCTION APPROXIMATION
When input data originates from a function with real-valued outputs over a continuous range, the
neural network is said to perform a traditional function approximation. An example of an
approximation problem could be one where the temperature of an object is to be determined
from secondary measurements, such as emission of radiation. Another more trivial example
could be to estimate shoe size based on a person’s height. These two examples involve models
with one input and one output. A more advanced model of the second example might use gender
as a second input in order to derive a more accurate estimate of the shoe size.
3.4 DATA FOR NEURAL NETWORK TRAINING:
Induction motor parameters used in simulation work:
Parameters Symbol Value

Stator resistance Rs 0.49
Rotor resistance Rr 0.45
Stator inductance Ls 0.0388mH
Rotor inductance Lr 0.0354mH
Magnetizing inductance Lm 0.0354mH
24
25
Numerator
X1 X2 X3 X4 X5 N1
9.54 7.8 4.9 6.8 5.1 -2.111
8.9 8.7 4.79 6.34 5.31 -1.87
9.3 8.4 2.51 3.23 7.13 0.02
9.1 8.2 1.52 2.45 7.81 0.562
10.1 9.3 2.34 1.9 8.12 0.84
10.5 9.4 1.21 0.78 9.89 1.7761
9.5 8.9 2.4 1.75 8.3 0.961
8.4 8.6 1.59 2.34 7.9 0.6287
9.4 10.1 4.5 6.7 4.98 -2.0787
8.3 10.5 4.4 6.54 4.95 -2
8.7 10.8 4.35 6.48 5.1 -1.94
8.5 10.7 3.4 6.3 5.15 -1.82
9.6 8.5 2.9 2.41 7.9 0.54
10.4 9.2 2.5 1.8 8.5 0.97
10.9 9.8 2.4 0.6 10.15 1.87
11.1 10.3 2.1 0.45 13.5 2.695
11.5 10.2 2.12 0.4 15.1 3.0669
11.8 10.7 1.9 0.35 15.9 3.2732
11.3 10.4 1.8 0.32 16.8 3.4882
11.4 10.6 1.6 0.3 17.5 3.6583
11.7 9.6 2.1 0.4 14.5 2.93
12.3 11.2 2.5 0.5 13.2 2.5905
12.5 11.3 2.4 2.1 10.5 1.28
12.8 11.6 2.2 2.8 9.9 0.8401
12.6 11.4 4.35 6.4 5.2 -1.889
12.4 11.7 4.45 6.6 5.1 -2.005
12.9 11.5 4.5 6.8 4.9 -2.1413
13.2 12.1 4.9 7.1 3.9 -2.51
13.4 12.5 5.2 7.5 3.5 -2.789
13.6 13.8 5.35 7.9 3.2 -3.0403
X1=disd/dt ,X2=isd, X3=Vsd, X4=Vsddt, X5=isddt
25
26
Y1 Y2 Y3 D1
80.9 46.27 700 -0.0105
70.8 69.13 610 -0.00813
90.4 70 138 0.000072
99.3 113 120 0.00193
120 130 108 0.00275
130.1 204 98 0.00553
140.3 127 100 0.00278
120.25 93 91 0.001722
135.55 95 490 -0.00513
112.9 100 460 -0.00444
115.7 98 425 -0.00391
132.7 97.5 390 -0.0033
122.5 88 130 0.000892
134.45 102 125 0.00144
145.85 131.1 116 0.00263
129.89 153 110 0.00354
124.6 160 108 0.0038
127.7 161 116 0.003719
154.85 154 112 0.00354
149.65 168 133 0.00365
148.43 149 140 0.00285
152.35 156 175 0.00251
145.76 101 137 0.001219
155.91 84 128 0.000785
143.87 137 390 -0.00189
156.24 136 394 -0.00201
165.34 134 402 -0.00214
130.5 130 410 -0.00251
135.9 125 416 -0.00279
141.95 120 420 -3.04403
Y1=disq/dt, Y2=Vsqdt, Y3=isqdt
26
27
3.5 TRAINING STEPS FOR NEURAL NETWORK:
27
28
28
29
29
30
30
31
31
32
32
33
33
34
34
35
35
36
36
37
37
38
38
39
39
40
40
41
41
42
42
43
43
44
44
45
45
46
CONCLUSION:
This project outlines technique for speed estimation of induction motor using artificial neural
networks. The dynamic model of induction motor is considered and expression for rotor speed is
obtained . The expression obtained have singularities thus ANN’s cannot be used to obtain the
speed directly from speed function. A method is used in this project in which two ANN’s are
trained to approximate the numerator and the denominator functions in the speed expression. By
training such ANN’s and using a filter to avoid singular points , the speed can be recovered .
REFERENCES:
1. Electrical Machinary by A.E. Fitzgerald.

2. Electric Machines by Ashfaq Hussain.
3. Control System Engineering by I. J. Nagrath and M. Gopal
46
47
4. “Dynamic model of induction motors for vector control” Dal Y. Ohm, Drivetech Inc.,
Blacksburg,Virginia.
5. “Speed Estimation Of Induction Motor Using Artificial Neural Networks” by Prashant
Mehrotra, John E. Quaicoe and R. Venkatesan.
6. “Motor Speed Identification Via Neural Network” by L.Ben Brahim.
7. “Neural Network Documentation” by Wolframesearch.
8. www.google.com
9. www.wikipedia.com
10. Neuralware predict
APPENDIX:
A. MATLAB COMMANDS:
>> y1= [80.9 70.8 90.4 99.3 120 130.1 140.3 120.25 135.55 112.9 115.7 132.7
122.5 134.45 145.85 129.89 124.6 127.7 154.85 149.65 148.43 152.35 145.76 155.91
143.87 156.24 165.34 130.5 135.9 141.95];
47
48
>> y2= [46.27 69.13 70 113 130 204 127 93 95 100 98 97.5
88 102 131.1 153 160 161 154 168 149 156 101 84
137 136 134 130 125 120];
>> y3= [700 610 138 120 108 98 100 91 490 460 425 390
130 125 116 110 108 116 112 133 140 175 137 128
390 394 402 410 416 420];
>> d1= -0.12*(10^-9)*y1+0.0354*(10^-3)*y2-0.01734*(10^-3)*y3;
>> plot(d1);
>> x1=[9.54 8.9 9.3 9.1 10.1 10.5 9.5 8.4 9.4 8.3 8.7 8.5
9.6 10.4 10.9 11.1 11.5 11.8 11.3 11.4 11.7 12.3 12.5 12.8
12.6 12.4 12.9 13.2 13.4 13.6];
>> x2=[7.8 8.7 8.4 8.2 9.3 9.4 8.9 8.6 10.1 10.5 10.8 10.7
8.5 9.2 9.8 10.3 10.2 10.7 10.4 10.6 9.6 11.2 11.3 11.6
11.4 11.7 11.5 12.1 12.5 13.8];
>> x3=[4.9 4.79 2.51 1.52 2.34 1.21 2.4 1.59 4.5 4.4 4.35 3.4
2.9 2.5 2.4 2.1 2.12 1.9 1.8 1.6 2.1 2.5 2.4 2.2
4.35 4.45 4.5 4.9 5.2 5.35];
>> x4=[6.8 6.34 3.23 2.45 1.9 0.78 1.75 2.34 6.7 6.54 6.48 6.3
2.41 1.8 0.6 0.45 0.4 0.35 0.32 0.3 0.4 0.5 2.1 2.8
6.4 6.6 6.8 7.1 7.5 7.9];
>> x5=[5.1 5.31 7.13 7.81 8.12 9.89 8.3 7.9 4.98 4.95 5.1 5.15
7.9 8.5 10.15 13.5 15.1 15.9 16.8 17.5 14.5 13.2 10.5 9.9
5.2 5.1 4.9 3.9 3.5 3.2
>> n1=-(-0.12*10^(-9)*x1-0.03474*10^(-3)*x2+0.0354*x3+0.45*x4-0.22*x5);
>> wr=n1./d1;
>> plot(wr)
B. PLOTS:
48
49
Numerator plot
49
50
Denominator plot
50
51
Speed plot
51
52
52

Speed Estimation of Induction Motor Using Artificial Neural Network and Implementation in MATLAB

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Speed Estimation of Induction Motor Using Artificial Neural Network and Implementation in MATLAB

Uploaded by

Copyright:

Available Formats

Generated by Foxit PDF Creator © Foxit Software

http://www.foxitsoftware.com For evaluation only.

FIGURE…………………………………………………………………………… PAGE NO.

FIG 1.1 Induction Motor……………………………………………………………….09

FIG 1.2 Coventional Per Phase equivalent circuit…………………………………......12

FIG 1.3 Dynamic equivalent circuit on Stationary Frame…………………………… 14

FIG 1.4 Dynamic Equivalent Circuit on Arbitrary Frame …………………………....15

FIG 1.5 d-q Axis Equation of Induction Motor………………………………………..17

FIG 2.1 Numerator Curve……………………………………………………………...20

FIG 2.2 Denominator Curve…………………………………………………………...20

FIG 2.3 Continuity……………………………………………………………………..21

FIG 2.4 Block Diagram of ANN……………………………………………………….23

FIG 3.1 Neuron Model…………………………………………………………………25

FIG 3.2 MLP…………………………………………………………………………...26

1.1.1 Induction Motor 08

1.2 Speed of Induction Motor 09

1.3 Speed Control of Induction Motor 09

1.4 Dynamic Nature of Induction Motor 11

1.4.1 Per phase equivalent Circuit 12

1.4.2 Stationary Frame Circuit 14

1.4.3 Arbitrary Frame Circuit 15

1.5 d-q Axis Equation 17

2.1 Speed expression 19

2.2 Method of Singular Point 21

2.2.3 Square integrable 22

3. ARTIFICIAL NEURAL NETWORK 24

3.1 Neural Network 24

3.3 Function Approximation 27

1.1.1 Induction Motor

It works on the principle of electromagnetic induction. A rotating magnetic field is produced

A three phase induction motor consists of mainly two parts:

There are two types of induction motor rotors:

1. Squirrel cage rotor

1.2 Speed of induction motor:

Induction motor speed is given by following formula:

And, n = number of poles

1.3 Speed control of induction motor:

1. Pole changing methods

1. Pole changing methods:

The number of stator poles can be changed by

﴾a﴿multiple stator windings,

﴾b﴿methods of consequent poles

﴾c﴿pole-amplitude modulation ﴾PAM﴿.

2. Stator voltage control:

The synchronous speed of an induction motor is given by:

4. Rotor resistance control:

5. Supply energy frequency:

1.4 Dynamic model of induction motor:

1.4.1 CONVENTIONAL PER PHASE EQUIVALENT CIRCUIT :

The inductance of rotor and stator is given by:

Inductance of rotor circuit –

Inductance of stator circuit –

Llr = rotor leakage inductance,

Lls = stator leakage inductance,

Lm = magnetizing inductance of motor.

If the excitation frequency injected into the stator is ωe and

The power consumption in stator circuit is given by,

The power consumption in rotor circuit and load (output) is given by ,

Torque produced is given by:

T =Ir2Rr (P/2)(1-s)/s e =Ir2Rr(P/2 e)

Vr’ = Rr’ Ir’ + p r’ = 0 (2)

1.4.2 DYNAMIC EQUIVALENT CIRCUIT ON STATIONARY REFERENCE FRAME:

>> d1= -0.12(10^-9)y1+0.0354(10^-3)y2-0.01734(10^-3)y3;