Graduate Theses, Dissertations, and Problem Reports

2015

Parameter identification of induction motor

Yousef Abdel-Rehim

Follow this and additional works at: https://researchrepository.wvu.edu/etd

Recommended Citation
Abdel-Rehim, Yousef, "Parameter identification of induction motor" (2015). Graduate Theses, Dissertations, and Problem Reports. 5018.
https://researchrepository.wvu.edu/etd/5018

This Thesis is brought to you for free and open access by The Research Repository @ WVU. It has been accepted for inclusion in Graduate Theses,
Dissertations, and Problem Reports by an authorized administrator of The Research Repository @ WVU. For more information, please contact
[email protected].
 PARAMETER IDENTIFICATION OF INDUCTION MOTOR

by

Yousef Abdel-Rehim

Thesis submitted to the

College of Engineering and Mineral Resources
at West Virginia University
In partial fulfillment of the requirements
for the degree of

Masters of Science
in
Electrical Engineering

Prof. Muhammad A. Choudhry, PhD, Chairperson

Prof. Hany H. Ammar, PhD
Prof. Natalia A. Schmid, PhD

Lane Department of Computer Science and Electrical Engineering

Morgantown, West Virginia

2015

Keywords: Induction motor, genetic algorithm, particle swarm optimization, parameter

identification, nonlinear modeling, LabVIEW software

Copyright 2015 Yousef Abdel-Rehim

 Abstract

PARAMETER IDENTIFICATION OF INDUCTION MOTOR

by
Yousef Abdel-Rehim

Numerous recent techniques of induction motor parameters calculating are hard to be done
and expensive. Accurate calculations of the parameters of these motors would allow savings
in different prospective like energy and cost. The major problem in calculating induction
motor parameters is that it’s hard to measure output power precisely and without harm during
the operation of the machines. It will be better to find other way to find out the output power
with certain amount of inputs like input voltage and current.

Particle swarm optimization (PSO) and genetic algorithms (GAs) are often used to
estimate quantities from limited information. They belong to a class of “weak” search
procedures, that is, they do not provide the “best” solution, but one close to it. It is a
randomized process in which follows the principles of evolution.

In this thesis genetic algorithm and partial swarm optimization are used to identify
induction motor parameters. The inputs used to estimate electrical and mechanical
parameters are measured stator voltages and currents. The estimated parameters compare
well with the actual parameters. Data Acquisition (DAQ) is used to obtain these variable with
the help of LABVIEW software. The induction motor used is a 7.5-hp with a constant
frequency and in free acceleration. IEEE standard test of 7.5-hp induction motor is used to
compare with performance of the simulated and measured data obtained. According to the
output results, method of optimizing induction machine can be used in different models of
induction motor

ii
 Acknowledgments

I would like to take this opportunity to express my sincere gratitude to my advisor Dr.
Muhammad A. Choudhry whose guidance and encouragement was instrumental in the
successful completion of my work.

I would also like to thank my committee members Dr. Hany H. Ammar and Dr. Natalia
A. Schmid for supportive advice and discussion. Furthermore, I am very thankful to
professors and staff at Lane Department of Computer Science and Electrical Engineering
(LCSEE) for making my work such an enjoyable and memorable experience.

Special thanks goes to my family for their inspiration and encouragement during this
work. None of my accomplishments would have been possible without their endless love.

iii
Contents
Abstract ................................................................................................................ ii
Acknowledgments .............................................................................................. iii
List of Figures ...................................................................................................... 6
List of Tables ....................................................................................................... 8
Chapter 1: Introduction ........................................................................................ 9
1.1 Thesis Outline ................................................................................... 9
1.2 Literature Survey............................................................................. 10
Chapter 2: Induction motor ............................................................................... 11
2.1 Principle of operation of induction motor: ..................................... 11
2.2 Types of tests used .......................................................................... 12
2.2.1 Locked rotor test:…………………………………………….12
2.3 Iron core loss Effect 𝑹𝒄: ................ Error! Bookmark not defined.
2.4 System Model of Induction Motor: ................................................ 14
2.4.1 Objective…………………………………………………….18
Chapter 3: Particle Swarm optimization ........................................................... 19
3.1 Theory ............................................................................................. 19
3.2 Parameters identification ................................................................ 20
3.3 PSO steps of operation:................................................................... 20
3.4 Advantage and disadvantage of PSO: ............................................ 21
Chapter 4: Genetic Algorithm ........................................................................... 22
4.1 Mutation and Crossover ........................................................................ 23
4.1.1 Crossover………………………………………………………23
4.2.2 Mutation:………………………………………………………27
4.3 Fitness Function: ................................................................................... 29
Chapter 5: Data Acquisition System ................................................................. 30
5.1 Data Acquisition System (DAQ) .................................................... 30
5.2 Lab VIEW Software ....................................................................... 31
5.2.1 Front Panel User Interface…………………………………...31
5.2.2 Block Diagram………………………………………………31
 5.2.3 Lab VIEW Features………………………………………….31
Chapter 6: Case Study and Results ................................................................... 35
6.1 Case Study ............................................................................................. 35
6.2 Experimental Test Procedure ................................................................ 35
6.2.1 Calculations:……………………………………………………36
6.4 Results .................................................................................................... 38
6.4 Results of 7.5-hp Motor ........................................................................ 43
6.5 Comparison between GA and PSO Algorithms .................................. 46
Chapter 7 Conclusion and future work ............................................................. 47
Appendix A: Hardware & Software.................................................................. 48
Appendix B: Code ............................................................................................. 51
GA Code ...................................................................................................... 51
FLUX………………………………………………………………..55
PSO Code..................................................................................................... 57
References .......................................................................................................... 62

5
List of Figures

Figure 2.1: Equivalent model for Induction machine ............................................................ 12

Figure 2.2 effect with and without 𝑅𝑐 .................................................................................... 14
Figure 2.3 𝑅𝑐 Effect................................................................................................................ 14
Figure 2.4 dqo equivalent circuit for induction motor ........................................................... 16
Figure 5: One point Crossover ............................................................................................... 23
Figure 4.6: One point crossover for offspring ....................................................................... 23
Figure 4.7: Two Point Crossover ........................................................................................... 24
Figure 4. 8: Continuous two point crossover for offspring ................................................... 24
Figure 4.9: Uniform crossover ............................................................................................... 25
Figure 4.10: Continuous Uniform Crossover ........................................................................ 25
Figure 4.11: Least popular crossover .................................................................................... 26
Figure 4.12: Arithmetic Crossover ......................................................................................... 27
Figure 4.13: Single Point Crossover and Mutation ............................................................... 27
Figure 4.14: Multi-point Crossover and Mutation ................................................................. 28
Figure 4.15: Example for fitness function ............................................................................. 29
Figure 5.16: Data acquisition block diagram for a three-phase system ................................ 30
Figure 5.17: GUI in LabVIEW .............................................................................................. 32
Figure 5.18: Three phase voltage and current Output from LabVIEW for 7.5-hp machine 33
Figure 5.19: Rotor speed data from LabVIEW for 7.5-hp machine ..................................... 34
Figure 6.20: Rotor Speed for 3-hp machine........................................................................... 40
Figure 6.21: Speed vs Torque characteristics for 3-hp machine ........................................... 40
Figure 6.22: Stator current Ias, for 3-hp machine .................................................................. 41
Figure 6.23: Stator current Ias, for 2250-hp machine ............................................................ 41
Figure 6.24: Torque, Te for 2250-hp machine....................................................................... 42
Figure 6.25: Rotor Speed vs Torque, for 2250-hp machine .................................................. 42
Figure 6. 26: Rotor Speed, for 2250-hp machine .................................................................. 43
Figure 6.27 Simulated and measured stator currents of 7.5-hp induction motor.................. 44
Figure 6.28. GAs estimated and no-load and blocked-rotor test data estimated stator currents
of 7.5-hp induction motor .......................................................................................... 45

6
Figure 6. 29. Estimated and measured rotor speeds of 7.5-hp induction motor. .................. 45
Figure A. 30: Name plate for 7.5-hp Induction machine....................................................... 48
Figure A.31: Physical Connection for Transducers ............................................................... 49
Figure A.32: Voltage Transducer ........................................................................................... 50
Figure A.33: Current Transducer ........................................................................................... 50

7
List of Tables

Table 1: Induction Machine Model Parameters ..................................................................... 15

Table 2: No load, blocked rotor and DC resistance test results............................................. 35
Table 3: Standard test parameters for 7.5-hp induction motor. ............................................. 37
Table 4: Test parameters for 3-hp induction motor (J = 0.089kgm 2 ) .................................. 38
Table 5: Test parameters for 50-hp induction motor (J = 1.662kgm 2) ................................ 38
Table 6: Test parameters for 2250-hp induction motor (J =63.87kgm 2 ) ............................. 39
Table 7 Results for 7.5-hp machine (J = 0.5kgm 2) .............................................................. 44

8
Chapter 1: Introduction

In the United States, electric motors use most of the electricity produced. Most of these
motors are induction motors because induction motors are efficient. In order to increase the
efficiency of induction motor, it has to be upgraded with some features. Information, like an
accurate motor parameters and machine model, must be known in order to improve its
function. In case of inaccurate parameters, the predicted output will be inaccurate. Examples
of outputs are Torque and Flux.
Knowing that the induction motor is nonlinear, variables of the rotor are unknown which
in turn makes most of the parameters unknown too. Therefore, in order to get all of these
parameter, we have to get it through another way by using known values and control it as
well. On doing this, we can control other parameters that are unknown. These known
parameters used are input voltage of stator, current and speed of motor.
Great methods are created to generate the best values for the unknown parameter through
using Genetic algorithm (GA) and Particle optimization Swarm (PSO) and different methods
of estimation. In this thesis, two optimization methods will be used, Genetic Algorithm (GA)
and Particle Swarm Optimization. Identification of induction motor parameters will be
performed by using DAQ card which will be connected to a personal computer (PC) to
observe and calculate the unknown parameters.
.

1.1 Thesis Outline

This thesis is designed to identify the parameters of induction motor by different methods.
Chapter 2 will present what is induction motor and nonlinear induction motor model and
how to solve it through Simulink/Matlab.
Chapter 3 will introduce PSO and how it will be used to estimate parameters of the induction
motor.
Chapter 4 will introduce GA and how it will be used to estimate parameters of the induction
motor.
Chapter 5 will show the results and case study done by the use of Data Acquisition card and
LabVIEW to know the output parameters from induction motor.
And finally, Chapter 6 will show the implementation of PSO and GA to optimize parameters
of induction motor, estimated results and evaluation.

9
1.2 Literature Survey
To be able to identify the unknown parameters of induction motor, two main tests will be
done on the induction motor according to IEEE test procedure. These tests are no load test
and block rotor test [36].
In [3], a balanced three phase induction motor was used as a model. From this model,
parameters were estimated. Some measurements were expected to be able to identify
procedures like stator current and stator voltage.
In early 90’s, two scientists, Richard R. Bishop and Gill G.Richards, introduced GA to
identify the parameter of induction motor by using steady state mode. They used some known
parameter like stator resistance and rotor reactance from the induction motor user manual to
find other parameters. For these tests they used a three phase induction motor with 0.5 hp.
These tests measure the motor terminal impedance Zm and Torque Tm at a rated voltage.
These scientists succeeded to prove that GA can be used to identify unknown parameters
[31].
[47] Stated that by using steady state models, motor parameter can be identified. Fitness
function was introduced while trying to do identify the parameter. Fitness function is defined
as the least error between the torques created and input torque. Here came the big advantage
of GA which was used to get this least error .In this situation the parameters of the motor
were assumed to be constant. This aimed to use the parameters where precision wasn’t
necessary. This method showed that performance of genetic algorithm was affected by those
constant value. By applying the parameter as constant on 5-hp, a huge error (20%) was
noticed. In case of deep bar model, it reduced from 3% to 6%. They concluded that accurate
values must be used to get accurate outs with low errors.
[17] and [18] talked about creating a solution to identify parameter of induction motor by
using qdo axis for modeling of induction motor. The induction motor that was used was 1kw.
Genetic algorithm was applied to identify the parameters. The genetic algorithm was
compared with a technique called analytical least square to identify induction motor
parameter. It was noticed that quality of the parameters depend on accuracy of equipment
used in the test for the motor.
[48] Demonstrated the use of genetic algorithm to identify induction motor by adding 4
different levels of noise. The mathematical model used was Kron’s voltage equation. To
check the efficacy of Genetic algorithm, another method, simple Random search (SRS), was
used under the same condition used in genetic algorithm. It was noticed that the values of
Genetic algorithm parameter where more accurate and efficient than SRS.
The fast increase in calculating power by using personal computers allowed researchers to
implement several optimization algorithms and verify their efficiency. Researchers
developed many algorithms that mimic natural phenomena. Examples of these algorithms
include the Simulated Annealing Genetics Algorithms (GA), Ant Colony Optimization
Algorithms. As genetic algorithm was spring an estimation method called Particle Swarm
optimization (PSO) was created in 1995 by Dr. Eberhart and Dr. Kennedy based on the
animal society which didn’t have leader like fish schooling and birds flocking [37].

10
Chapter 2: Induction motor

This thesis will talk about parameter identification of induction motor using genetic algorithm and
particle swarm optimization. Before we get involved more in the complicated detail, we have to talk
more about some fundamentals to understand the complicated parts in this thesis.
First we have to know what the difference between motors and generators is and how they operate
differently, and what is needed to able to perform the thesis in the right way.
Motor is a device used to convert electricity to mechanical movement. Most mechanical motions
happen due to interaction between electric motor’s magnetic field and winding current to create force
of motion and generator is the vice versa.
In our research we will do our experiment on motor more specifically on induction motor which
is an AC electric motor. When current enters the rotor, it produces torque due to interaction between
current and magnetic field created by stator winding.

2.1 Principle of operation of induction motor:

In induction motor, when supplying an AC power to the motor's stator , a magnetic field will be
created. The rotor speed will not be the same as field created from the stator. So, the rotor of the
induction motor will have a lower speed than stator magnetic field. That will lead of a creation
current in the rotor which will affect the secondary winding of the motor. Currents will be created
in the rotor winding when the magnetic flux induced. The currents created in the rotor will lead to
creation of magnetic field in the rotor which will react against the stator field .

Induction motors are the most used in the industry as they are not difficult to build. They also have
great advantage as they are reliable and have self-starting capability. To be able to control induction
motor, some of the parameters should be known. These known parameters must be very accurate.
Inaccuracy may lead to worsening of the motor performance.

11
 Figure 2.1: Equivalent model for Induction machine

It was found that the easiest way to get parameters of induction motor is to execute locked rotor
test and no-load test. The idea of using linear parameter estimation technique is to find out ‘𝑅1 ’
which is stator resistance, ‘𝑋1 ’ which is stator leakage-reactance, ‘𝑋𝑚 ’ magnetizing reactance, ’ 𝑋′2’
which is rotor self-reactance and ‘𝑅′2 ’ rotor resistance.

2.2 Types of tests used

- Locked Rotor Test
- No Load Test

2.2.1 Locked rotor test:

Locked rotor test is also known as blocked rotor test and stalled toque test. In this test, voltage
applied can’t be rated voltage, as the rated voltage is 220. This voltage is very high in case of locked
rotor test, as when this voltage is applied to the motor while it is locked, a huge current will pass in
the stator winding which will lead to overheat and damage the motor. Therefore, the applied voltage
must be low and test time should be short time.

In block rotor test, the rotor will be locked to avoid its motion, leading to production of slip factor
s=1 that will make 𝑅′2 /s=𝑅′2 . This will make 𝑅′2 become much less resistance than 𝑅𝑐 and therefore,
𝑅𝑐 and 𝑋𝑚 parts of the circuit will be removed or ignored due to their huge resistance and reactance.

No load Test:
No load test is also known as open circuit test. A balanced rated voltage will be applied to the
stator winding for a No load test. The motor speed will be almost at synchronous speed. Synchronous

12
speed will lead to s= very small value, and therefore according to figure, 𝑅′2 /s= Very big value ,
removal of the part which include 𝑋′2 and 𝑅′2 will occur as their value is huge. In the end, all current
will pass only in 𝑅𝑐 and 𝑋𝑚

From both test locked rotor test and no load test certain parameter can be calculated like ‘𝑅1 ’,
‘𝑋1’, ‘𝑋𝑚 ’,’ 𝑋′2’ and ‘𝑅′2 ’ which are stator resistance , stator leakage-reactance , magnetizing
reactance , rotor leakage -reactance and rotor resistance respectively

For identifying induction motor parameters, Particle Swarm Optimization (PSO) and Genetic
Algorithm (GA) are used [5], [6], [37] .The aim of using those two algorithms to get electrical and
mechanical parameters of the nonlinear motor model. To identify the parameters 3, 50 and 2250-hp
models are used. In lab experiment the model used is 7.5-hp induction motor.

2.3 Iron core loss effect∶

1. 𝑅𝑐 is added to account for hysteresis and eddy current losses

2. Iron core losses causes a reduction in performance.

3. The use of static converter for electrical drives leads to iron losses increase.

4. Magnetic saturation can be viewed as a reduction in the value of mutual induction and thus
will increase magnetizing current.

5. The addition of 𝑅𝑐 will make the model more accurate than a simplified model.

2.3.1How to decrease the 𝐑 𝐜 effect:

Eddy current losses are minimized by using lamination. By using lamination the core
area decreases and resistance increases, which results in decrease in eddy currents. Hysteresis
losses are minimized by using high grad silicon steel.

13
 2.3.2 The 𝑹𝒄 effect on 7.5-hp induction motor model:

𝑅𝑐 value calculated was 63.5 Ω. when we applied the 𝑅𝑐 to simulink model the output
was almost the same as without 𝑅𝑐 and that can be see below in figure 2.2 and 2.3.

Figure 2.2 effect with and without 𝑅𝑐

Figure 2.3 𝑅𝑐 Effect

2.4 System Model of Induction Motor:

In this research the study will be done on a balanced three phase induction motor. The voltage and
current are at synchronous rotating reference frame. The rotor parameters are referred to the stator.
From stator current and rotor current, flux can be measured [45]. Table 1, 2 summarizes the name of
each parameter that will to be used in our research.

14
Table 1: Induction Machine Model Parameters

𝑅1 =𝑅𝑠 stator resistance

𝑋1=𝑋𝑙𝑠 stator leakage-reactance

𝑋𝑚 magnetizing reactance

𝑋′2=𝑋′𝑙𝑟 rotor leakage -reactance

𝑅′2 = 𝑅′𝑟 rotor resistance

Table 2: Induction Machine Model Parameters

15
 Figure 2.4 dqo equivalent circuit for induction motor

The stator currents for induction machine are given by;

1
𝑖𝑞𝑠 = (𝜓 − 𝜓𝑚𝑞 )
𝑋𝑙𝑠 𝑞𝑠
1
𝑖𝑑𝑠 = (𝜓 − 𝜓𝑚𝑑 )
𝑋𝑙𝑠 𝑑𝑠
The rotor currents are given by:

16
 ′
1
𝑖𝑞𝑟 = ′ (𝜓′𝑞𝑠 − 𝜓𝑚𝑞 )
𝑋𝑙𝑟
′
1
𝑖𝑑𝑟 = ′ (𝜓′𝑑𝑠 − 𝜓𝑚𝑑 )
𝑋𝑙𝑟
The mutual flux is;
𝜓𝑚𝑞 = 𝑋𝑚 (𝑖𝑞𝑠 + 𝑖′𝑞𝑟 )

𝜓𝑚𝑑 = 𝑋𝑚 (𝑖𝑑𝑠 + 𝑖′𝑑𝑟 )

The stator and rotor flux are given by;

𝑑𝜓𝑞𝑠 𝜔 𝑟𝑟
= 𝜔𝑏 [𝑣𝑞𝑠 − 𝜓𝑑𝑠 + ( 𝜓𝑚𝑞 − 𝜓𝑞𝑠 ) ]
𝑑𝑡 𝜔𝑏 𝑋𝑙𝑠

𝑑𝜓𝑑𝑠 𝜔 𝑟𝑟
= 𝜔𝑏 [𝑣𝑞𝑠 + 𝜓𝑞𝑠 + ( 𝜓𝑚𝑑 − 𝜓𝑑𝑠 ) ]
𝑑𝑡 𝜔𝑏 𝑋𝑙𝑠

𝑑𝜓′𝑞𝑟 𝜔 − 𝜔𝑟 𝑟𝑟′
= 𝜔𝑏 [𝑣′𝑞𝑠 − 𝜓′𝑑𝑠 + ′ ( 𝜓𝑚𝑞 − 𝜓′𝑞𝑟 ) ]
𝑑𝑡 𝜔𝑏 𝑋𝑙𝑟

𝑑𝜓′𝑑𝑟 𝜔 − 𝜔𝑟 𝑟𝑟′
= 𝜔𝑏 [𝑣′𝑑𝑠 + 𝜓′𝑞𝑠 + ′ ( 𝜓𝑚𝑑 − 𝜓′𝑑𝑟 ) ]
𝑑𝑡 𝜔𝑏 𝑋𝑙𝑟

𝜓𝑞𝑠 𝜓′𝑞𝑟
𝜓𝑚𝑞 = 𝑋𝑎𝑞 ( + ′ )
𝑋𝑙𝑠 𝑋𝑙𝑟
𝜓𝑑𝑠 𝜓′𝑑𝑟
𝜓𝑚𝑑 = 𝑋𝑎𝑑 ( + ′ )
𝑋𝑙𝑠 𝑋𝑙𝑟

1
𝜓𝑎𝑞 = 𝑋𝑎𝑑 = 1 1 1
+ 𝑋 + 𝑋′
𝑋𝑚 𝑙𝑠 𝑙𝑟

The Torque, Te is given by

𝑑 𝜔𝑟
𝑇𝑒 = 2𝐻 + 𝑇𝐿
𝑑𝑡 𝜔𝑏

17
 𝑇𝑒 = 𝜓𝑑𝑠 𝑖𝑞𝑠 − 𝜓𝑞𝑠 𝑖𝑑𝑟

The input voltage applied to stator are balanced and rotor voltage is zero, which lead to bring up
speed and stator current

2.4.1 Objective

Identification of induction motor parameters are the main objective of our experiment .The basic
idea is to compare time dependent response of the model and model estimated by using error
criterion. The objective function can be obtained from the output error and estimated model. It tells
how good the estimated model response is and detects the match with system model response. The
structure of the estimated model is same as the system model except that the estimated model has
unknown parameter

18
Chapter 3: Particle Swarm optimization

Particle Swarm Optimization theory was created in 1995 by Dr. Eberhat and Dr. Kenny .The idea
was inspired from nature of the social behavior and dynamic movement in the communication of
birds flocking or fish schooling [37].

3.1 Theory
Particle swarm optimization was used to solve several problems .The idea of PSO algorithm came
from the animal society that didn’t have leader in their group, like fish schooling and bird flocking.
For example, each bird in the flock will search randomly for food, when a bird finds food at a certain
position, it leads the whole flock to the closet food place. The success of the team depends on the
best direction and shortest distance in finding food, which in turn depends on good communication
between them. The process of searching food happens continuously until food is discovered. The
technique of partial swarm optimization is used to find the optimal value and best solution.
Particle swarm optimization consists of a number of particles. Each particle has a certain position
in space [38]. The particle status act:
1) To keep its inertia
2) To change condition according to its most optimization position
3) To change condition according to the swarm’s most optimist position.
Each particle in space adjusts its movement according to its own moving experience with another
particle.
There are two experiences in particle swarm optimization, the ‘individual experience’ and the
‘near experience’. The ‘individual experience’ is the position of each particle in swarm and it is
affected by the most optimist position during its movement. The ‘near experience’ is the position of
the most optimist particle in its surrounding.
Each particle keeps track of where it was at the best fitness. This is called ‘Pbest’. Another "best"
value is tracked by the particle swarm optimizer. It is obtained by any particle in the population.
This best value is the global best and is called gbest. When a particle takes part of the population as
its topological neighbor, the best value is then the local best and is called lbest. Each particle has a
position and velocity. The information can be updated by changing the velocity and position.
In the process, we initiate the population including the value and velocity, we look at the fitness
then we modify the velocity based on the pervious pbest, global best and neighborhood best and we
loop around terminating a good round solution.

𝑣 = 𝐾[𝑣 + 𝑐1 ∗ 𝑟𝑎𝑛𝑑(𝑝𝑏𝑒𝑠𝑡 − 𝑃𝑎𝑟𝑡𝑖𝑐𝑎𝑙) + 𝑐2 ∗ 𝑟𝑎𝑛𝑑(𝑔𝑏𝑒𝑠𝑡 − 𝑃𝑎𝑟𝑡𝑖𝑐𝑙𝑒)]

𝜃 = 𝜃 + 𝑣𝜙 = 𝑐1 + 𝑐2

19
 2
𝐾=
|2 − 𝜙 − √𝜙 2 − 4𝜙|

The states for measuring for induction machine are:

3.2 Parameters identification

Particle update rule: 𝑝 = 𝑝 + 𝑣
Where
p: particle’s position
v: velocity
c1: weight of local information
c2: weight of global information
pbest: best position of the particle
gbest: best position of the swarm
rand: random variable

3.3 PSO steps of operation:

Algorithm of PSO:
- Step 1: Define fitness function.
- Step 2: Generate initial population of N particles with random position and velocities.
- Step 3: Calculate fitness: Evaluate the fitness values of current particle using the objective
function
- Step 4: Update pbest- Compare the fitness values of each particle with its pbests. If current
value is better than pbest then set it to current value.
- Step 5: Update gbest- Compare the fitness values of each particle with its pbests. If current
value is better than gbest then set it to current value.
- Step 6: If the number of iterations reaches the maximum, then go to step 8 otherwise go to
step 4.
- Step 7: The particle that generates the latest gbest is the solution of the problem.
- Step 8: Stop

20
3.4 Advantage and disadvantage of PSO:
Advantages:
1) Scaling insensitive of design
2) Its implementation is very simple
3) Few algorithms are used
4) Very Efficient

Disadvantages:
1) Slow convergence in refined search stage (weak local search ability)

21
Chapter 4: Genetic Algorithm

Dr. John Holland introduced Genetic Algorithm in1970's with the help of his colleagues and some
students at university of Michigan. The idea of GA came from genetic and reproduction behavior
observed in biological population. The ideology of GA came from 'survival of the fitness'. It searches
a process to select and create individual that can be able to survive on the environment. After a
number of iterations, design characteristic evolves and remains in the genome composition of the
population over traits with weaker undesirable characteristics. GA has been known for its success in
solving sophisticated optimization as it can work with nonlinear objective and can work with discrete
and continues variables.
In IT application, GA are used widely in most applications and especially parallel genetic
algorithm. The prospective that parallels Genetic Algorithm and deals with existing information
technology is huge, which may control very large and variable data in real world. GA is idealistic to
do this.
GA is the best method for optimization especially when little data is available and large size
problems are present. GA uses the method of direct analogy of a natural evolution. It is assumed
that any problem can be solved by having a number of parameters. These parameters are like genes
of a chromosome which has a string of value, which correlates with the objective function of the
problem.

In 1970, Dr. Fraser came up with idea of replicating a genetic system by using genetic operators.
Dr. John Holland first introduced a simple Genetic Algorithm and discovered the usage of genetic
operators schemes which lead to production of a new generation and the old generation died. Dr.
Lawrence Davis highlighted the significance of evaluation function as the connection between
Genetic Algorithm and problem to be solved. Dr. Lawrence recommended the following:

1. Initialization of a population of chromosomes.

2. Creation of new chromosomes by mating current chromosomes.
3. Mutation and reproduction as the parent chromosomes mate.
4. Evaluation of the new chromosomes and insertion into the population.
5. If time is up, stop and return the best chromosome. If not, go to step three.

Genetic Algorithm is divided into two operators which makes it unique and successful. These
operators are 'mutation' and 'crossover' .It is predicted from these procedures that a better
chromosome will create a large number of offspring which at the end is guaranteed a higher chance
of surviving in subsequent generation. The sequence of evolution is repeated until desired execution
measured is achieved.

22
4.1 Mutation and Crossover
It was mentioned before that Genetic Algorithm has 2 operators, the first is mutation and second is
crossover.
Mutation: Is to modify the value of one or more gene in a chromosome.
Crossover: Is to reproduce propagation where two parents give birth to a new Offspring

4.1.1 Crossover
Crossover is a genetic operator used to do some changes in chromosomes to make it better. The
methodology of Crossover is to take genes from different parents to create a child with better
functionality and with no defects. Parents must be healthy There are different ways to approach this.
Some types are popular and used a lot in real like one-point crossover, two-point cross over, and
uniform Crossover. Other least used crossovers are cut and splice and half uniform crossover. More
over Crossover is also expressed in binary encoder method.

Most popular:
 One-point crossover:
It is a single crossover point from both parents. All points after certain point are identically
swapped and the result is a child of both parents

Figure 5: One point Crossover

Figure 4.6: One point crossover for offspring

Or

23
 Parent 1: X X | X XXXX
Parent 2: Y Y | Y YYYY

Offspring 1: X X Y YYYY
Offspring 2: Y Y X XXXX

o Encoding Part
 Parent A: 1 0 0 1 0| 1 1 1 0 1
 Parent B: 0 1 0 1 1 |1 0 1 1 0

 Child AB: 1 0 0 1 0 1 0 1 1 0
 Child BA: 0 1 0 1 1 1 1 1 0 1

 Two Point Crossover:

It is the choice of a certain partition from parents and then identically swapped it.

Figure 4.7: Two Point Crossover

Figure 4. 8: Continuous two point crossover for offspring

Or

- Parent 1: X X | X XX | X X

24
 - Parent 2: Y Y | Y YY | Y Y

- Offspring 1: X X Y YY X X
- Offspring 2: Y Y X XX Y Y

o Encoding Part
 Parent A: 1 0 0|1 0 1 1| 1 0 1
 Parent B: 0 1 0* 1 1 1 0* 1 1 0

 Child AB: 1 0 0 1 1 1 0 1 0 1
 Child BA: 0 1 0 1 0 1 1 1 1 0

 Uniform crossover:
Different partitions will be taken from both parents with a probability of 0.5.

Figure 4.9: Uniform crossover

Continuous Uniform crossover:

Figure 4.10: Continuous Uniform Crossover

- Parent 1: XXXXXXX

25
 - Parent 2: Y YYYYYY

- Offspring 1: X Y X Y Y X Y
- Offspring 2: Y X Y X X Y X
o Encoding Part

 Least popular Crossover:

Cut and slice:
This is another type of crossover which changes in Length of children. This happen due to
taking different length of data from each parent.

Figure 4.11: Least popular crossover

 Arithmetic Crossover:
It is a combination of different data from different parent which lead to totally different
Offspring

26
 11001011 + 11011111 = 11001011 (AND)
Figure 4.12: Arithmetic Crossover

4.2.2 Mutation:
Genetic algorithm is a binary code .Each bit in the chromosome is mutated with probability Pbm
known as the mutation rate

Parent1 1 0 0 0 0 1 0
Parent2 1 1 1 0 0 0 1

Child1 1 0 0 1 0 0 1
Child2 0 1 1 0 1 1 0

An Example of Sing le-point Crossover Between the

Third and Fourth Genes with a Mutation Rate of
0.01 Applied to Binary Coded Ch ro mosomes

Figure 4.13: Single Point Crossover and Mutation

27
 Figure 4.14: Multi-point Crossover and Mutation

The mutation operator ensure that gene pool does not become too restricted. GA is carried out by
randomly changing one or more of bits in an individual. This may lead to creation of a totally new
solution from the previous one, this means that GA can be better by using mutation

There are different types of mutation:

- Bit string mutation

- One of bits is chosen randomly and be flipped
.
- Flip Bit
- One genome will be chosen and change all its bits, which mean if a bit is 1 it will be flipped
to 0 and vice versa.

- Boundary
The genome will be replaced with a random upper or lower bounds

28
4.3 Fitness Function:
The fitness Function measures the quality of represented solution over Genetic Representation. Also
it is problem dependent. The illustration of the result might be an array of bits, where each bit
represent a different object. The bit value donates an objective but not every illustration is valid.

Fitness function is used in each iteration of algorithm to value the superiority of all projected solution
to the problem in current population. The fitness function rule is to evaluate the goodness of each
solution in a population.

Figure 4.15: Example for fitness function

29
Chapter 5: Data Acquisition System

In this chapter a programmed method to find the input and output variables of an induction motor
is labeled. The process starts by finding the hardware devices that are used for the data acquisition
system (DAQ). The technique, created as a virtual instrument (VI), is done in the Lab VIEW setting
to measure and save the signals of the rotor speed, three - phase currents and three - phase voltages,
of the induction motor. The technical features of the Lab VIEW and VI observing system supply
huge help in parameter identification of an induction machine since it gives material on developed
signals needed for the identification problem to be cracked.

5.1 Data Acquisition System (DAQ)

The data acquisition (DAQ) VI library attains waveforms and produces signals with all National
Instruments (NI) plug-in and remote data acquisition products. The data acquisition system used here
has a NI DAQ Card (6040E Families) [35]). The Input/output Connector of the DAQ card has 68
pins for the analog and digital signals. Six analog inputs in the range of 10 volts To Measure three-
phase voltages and three-phase current. So, certain type of transducers must be employed to convert
high amplitude signals within the range. For voltage VT7 and two different types of transducers are
used for currents, CTL – 51 and CTA201R. Analog output is provided by a digital tachometer (speed
transducer) to measure the speed, which converts speed in the range of 50 – 10000 rpm to 0 to 5
volts. The way in which analog inputs are connected to the DAQ card is given in Appendix A.31..

Figure 5.16: Data acquisition block diagram for a three-phase system

30
 DAQ systems are used in a different ranges of applications in the research workroom, in the field,
and in manufacturing plants. A general DAQ system includes:
Transducers - connected to induction motor being measured. For three phase voltages of induction
motor – transducers convert input voltage in the range of 600 V to analog output voltage in the range
of 10 Volts.

5.2 Lab VIEW Software

Lab VIEW is a visual programming language for display of measured data and analysis. The
software was firstly produced by NI. These programs are called Virtual Instruments (VI) since they
actually imitate physical instruments. VIs look like the functions in textual programming language,
where the functions are used instead of icons. Each VI in Lab VIEW consists of:
- Front Panel
- Block Diagram

5.2.1 Front Panel User Interface

Front panel is the GUI for creating commands for showing instruments and measurements results.
This interface gathers user input and shows program output. Control objects are used for data input,
while indicator objects are used for data output. Front panel is very suitable to create and to work
with. One example of Front Panel and VI designed in the power laboratory for the goal of obtaining
stator currents, stator voltages and speed of the rotor shaft of induction motor is shown in the Figure
5.18. Figure 5.19 shows front panel configurations for reading and displaying saved data from the
file.

5.2.2 Block Diagram

The block diagram has the graphical source code of VIs. All the functions shown in the front
panel must be controlled and performed with in block diagram. The block diagrams designed for
obtaining signals from the induction motor are shown in the Figures (5.17
5.2.3 Lab VIEW Features
Lab VIEW VIs are modular in design, so any VI can run on its own or be used as part of another
VI (subVI). With this modularity, it gives the operator flexibility to design a whole hierarchy of VIs
and subVIs that serve as building blocks in any number of applications. Also, it provides easy
change, interchange, and combination of VIs in order to provide accommodations for other
applications. Furthermore enhance performance [38].

31
Figure 5.17: GUI in LabVIEW

32
Figure 5.18: Three phase voltage and current Output from LabVIEW for
7.5-hp machine

33
Figure 5.19: Rotor speed data from LabVIEW for 7.5-hp machine

34
Chapter 6: Case Study and Results

In this chapter the results for parameter estimation of induction machine using GA and PSO are
presented. Also the results from the hardware test of machine are presented.

6.1 Case Study

This chapter discusses the experimental testing to the problem identification of electric parameters
of induction motor. The standard test procedure of DC resistance, no load and blocked rotor is
applied to the 7.5-hp induction motor to obtain parameters. Then, a free acceleration test is
performed, and with the help of DAQ and Lab VIEW Software three - phase voltages, three - phase
currents and speed are recorded. Next, the GA method and the PSO method is used to estimate
parameters of induction motors. Thus, the method is based on the following:
- Experimental Test Procedure.
- Implementation of GA to the problem identification.
- Implementation of PSO to the problem identification.

6.2 Experimental Test Procedure

The first test applied for determination of parameters is based on IEEE standard procedure. The
following table gives information on parameters obtained after performing DC resistance, no – load
and blocked rotor tests of a three – phase, 7.5-hp, 60 Hz, 220 V, induction motor.

Table 2: No load, blocked rotor and DC resistance test

results

No Load Test Blocked Rotor Test DC Resistance

Vnl = 126.5V Vbl = 18.63 V Vdc = 11.74 V

Inl = 8.8 A Ibl = 18.62 A Idc = 20 A

Pnl = 274.4 W Pbl = 221.33 KW

Both no load and blocked rotor test are performed at 60 Hz frequency. According to the IEEE
Standard Test Procedure given in [36], this case applies for the situation when the slip is near unity

35
thus, right after the machine start – up, when the starting currents have the largest values. No attempt
was made in this work to compose the test at reduced frequency since there was no adjustable
frequency drive in equipment setup.
The detailed methodology for calculating the parameters of the induction machine is discussed
below:
6.2.1 Calculations:

No load test:
𝑃𝑝ℎ 274.4
cos(𝜙𝑜 ) = = = 0.24
𝑉𝑝ℎ 𝐼𝑜 126.3 ∗ 8.8
𝜙𝑜 = 76.17

𝐼𝑚 = 𝐼𝑜 sin(𝜙𝑜 ) = 8.8 sin(76.17) = 8.74 𝐴

𝐼𝑐 = 𝐼𝑜 cos(𝜙𝑜 ) = 8.8 cos(76.17) = 2.16 𝐴
𝑉𝑝ℎ 126.3
𝐿𝑚 = = = 0.039 𝐻
2 𝜋 𝑓𝑠 𝐼𝑚 2 ∗ 𝜋 ∗ 60 ∗ 8.74
𝑋𝑚 = 14.35𝛺
Rc calculation:
Pfwc = Pnl − PIo 2 Rs
PIo 2rs = Io 2 R s
2
3 ∗ |Vph |
Pfwc =
Rc
Pnl = 823.2 W
R s = 0.3 Ω
Io = 8.8A
PIo 2 rs = 3 ∗ Io 2 R s = 3 ∗ (8.8)2 ∗ 0.3 = 69.696W
Pfwc = Pnl − PIo 2 Rs = 823.2 − 69.696 = 753.504W
R c = 63.5 Ω

36
Locked rotor test
𝑃𝑝ℎ 221.33
cos(𝜙𝑠𝑐 ) = = = 0.64
𝑉𝑝ℎ 𝐼𝑠𝑐 18.63 ∗ 18.62
𝜙𝑠𝑐 = 50.35
𝑉𝑝ℎ 18.63
𝑍𝑠𝑐 = = ≈ 1𝛺
𝐼𝑠𝑐 18.62
𝑅𝑟= 𝑍𝑠𝑐 cos(𝜙𝑠𝑐 ) − 𝑅𝑠 = (1 ∗ 0.64) − 0.3 = 0.34 𝛺
𝑋𝑒𝑞 = 𝑍𝑠𝑐 sin(𝜙𝑠𝑐 ) = 1 sin(50.13) = 0.77 𝛺
𝑋𝑒𝑞 = 𝑋𝑙𝑠 + 𝑋𝑙𝑟 = 0.77𝛺

𝑋𝑙𝑠 = 𝑋𝑙𝑟= 0.385𝛺

Using the said methodology, the parameters of the induction machine were calculated and are
tabulated in following table.

Table 3: Standard test parameters for 7.5-hp induction

motor.

Rs 𝑋1 Xm 𝑋2 Rr

0.3 0.385 14.35 0.385 0.34

The next step is then applied to measure three - phase voltages, three – phase currents and rotor
speed. Machine variables are recorded during the start- up conditions. Three phase voltages are
generated from the Motor – Generator Set (MG-Set 1) available in Power Lab. A three-phase wound
rotor induction motor rated at 7.5-hp is tested.
The configuration test for the experiment can be found in Chapter 5, where Data Acquisition
System and Lab VIEW Software are employed for recording the waveforms. Voltage and current
signals in the range of [-10 V, +10 V] are acquired by the means of voltage and current transducers.
A digital transducer – tachometer is used to generate analog voltage input at the range of 0 to 5 volts.
After employing Lab VIEW Software, acquired signals are saved. Data collected are transformed
to the files suitable for the identification method by the GAs. Source code of the GAs is listed in
Appendix.

37
 The boiler – plate data from the test of induction motor is given in Appendix. No mechanical load,
except for the rotor inertia and windage, was attached to the motor.

6.4 Results
Before applying the genetic algorithm to the real experimental test data, a simulation for induction
machine was implemented to prove the validity of the approach. The theoretical examples are chosen
with widely different horsepower ratings in order to confirm applicability of the method to motors
with various sizes. The following tables summarize the results obtained from running the GA and
PSO for each machine. The estimated parameters agree very closely with the actual value.

Table 4: Test parameters for 3-hp induction motor (J = 0.089kgm 2 )

Rs 𝑋1 Xm 𝑋2 Rr

From GA 0.40 0.5615 25.3539 0.8161 0.7353

-0.09 -0.25 -0.03 0.08 -0.098

From PSO 0.4445 0.7754 26.9727 0.7997 0.7990

0.02 0.03 0.03 0.06 0.02

Actual 0.435 0.754 26.13 0.754 0.816

Table 5: Test parameters for 50-hp induction motor (J = 1.662kgm 2)

Data Set Rs 𝑋1 Xm 𝑋2 Rr

From GA 0.019 0.3558 12.2320 0.3374 0.4507

-0.78 0.2 -0.06 0.1 1

From PSO 0.0113 0.4992 12.7521 0.4967 0.2742

-0.89 0.67 -0.015 0.67 0.18

Actual 0.087 0.302 13.08 0.302 0.228

38
 Table 6: Test parameters for 2250-hp induction motor (J =63.87kgm 2 )
Data Set Rs 𝑋1 Xm 𝑋2 Rr

From GA 0.0200 0.1085 13.5626 0.2096 0.0317

Percentage -0.31 -0.52 0.04 -0.07 0.44

error for
GA

From PSO 0.0209 0.1025 14.9895 0.2364 0.0372

Percentage -0.28 -0.5 0.15 0.09 0.5

error for
PSO

Actual 0.029 0.226 13.04 0.226 0.022

From the tables given above it is obvious that the GA was able to identify parameters with a good
accuracy. The errors between the responses of the induction motor with the actual parameters and
the genetic algorithm identified parameters are small.

Figures (6.19 – 6.21) compare the actual parameter response with that predicted by using the GA
and PSO identified parameters for 3-hp machine. Rotor speed, three-phase stator current and torque
during transients are shown. The Stator voltages are the only input to the induction motor for these
results.

Figures (6.22 -6. 25) compare the actual parameter response with that predicted by using the GA
and PSO identified parameters for 2250-hp machine. Rotor speed, three-phase stator current and
torque during transients are shown. The Stator voltages are the only input to the induction motor for
these results.

39
 Figure 6.20: Rotor Speed for 3-hp machine

Figure 6.21: Speed vs Torque characteristics for 3-hp machine

40
 Figure 6.22: Stator current Ias, for 3-hp machine

Figure 6.23: Stator current Ias, for 2250-hp machine

41
 Figure 6.24: Torque, Te for 2250-hp machine

Figure 6.25: Rotor Speed vs Torque, for 2250-hp machine

42
 Figure 6. 26: Rotor Speed, for 2250-hp machine

6.4 Results of 7.5-hp Motor

Free acceleration of 7.5-hp induction machine was performed and the data for 3-phase voltage and
currents was recorded. Also the data for speed was measured for the said machine. The data was
stored in Excel file and then transferred to Matlab. Using the data, the parameters of the machine
were calculated and verified against the standard parameters of 7.5-hp induction machine. The results
came very close to the values calculated using the blocked rotor test and no load test.

43
 Table 7 Results for 7.5-hp machine (J = 0.5kgm 2)

Rs 𝑋1 Xm 𝑋2 Rr
From No-load and 0.3 0.385 14.35 0.385 0.34
blocked-rotor test

From GA 0.2506 0.3616 14.1444 0.3811 0.2440

Percentage error -0.1647 -0.0608 -0.0143 -0.0102 -0.2824
for GA
From PSO 0.2770 0.3607 14.7466 0.3682 0.2115
Percentage error -0.0767 -0.0631 0.0276 -0.0436 -0.377
for PSO

Figure 6.27 Simulated and measured stator currents of 7.5-hp induction

motor

44
 Figure 6.28. GAs estimated and no-load and blocked-rotor test data
estimated stator currents of 7.5-hp induction motor

Figure 6. 29. Estimated and measured rotor speeds of 7.5-hp induction

motor.

45
6.5 Comparison between GA and PSO Algorithms

GA PSO

Advantages: Advantages:
1) Very large number of parameters 1) Scaling insensitive of design
2) Solves problems with multiple solutions 2) Its implementation is very simple
3) GA is very easy to understand. 3) Few algorithms are used
4) The Percentage error for GA is better in case 4) Very Efficient
of 50-hp,2250-hp and 7.5-hp 5) The Percentage error for PSO is better in case
of 3-hp

Disadvantages: Disadvantages:
GA may not find the global optimum Slow convergence in refined search stage
(weak local search ability)

46
 Chapter 7 Conclusion and future work

This thesis has presented intelligent approaches to estimate the Induction Machine electrical
parameters based on Particle Swarm optimization and Genetic Algorithm heuristic optimization
approach, a simulation model and experimental transient measurements. Based on both the
simulation and experimental measured results, it is concluded that the use of the proposed strategy
is an effective and reliable method for induction motor parameter identification. The reliability of
this technique in estimating the machine parameters of different rating of induction motor has been
presented successfully.

Future work:
1. 𝑅𝑐 should be included in dqo induction motor model and its effect should be investigated
through simulation.
2. Induction motor parameters should be estimated using Nonlinear least square approach and
its results should be compared with Genetic algorithm (GA) , Particle swarm optimization
(PSO) and real experiment.

47
Appendix A: Hardware & Software

Information on hardware tools and equipment’s, which are used for acquiring the signals in the
experiment is given in this section. The second part defines an overall purpose of Lab VIEW
Software intended to simply obtain outside signals. It is included here because of the ability to further
work on this subject.

Setup
The identification technique labeled in the previous chapters performed a 7.5-hp Induction Motor.
The next figure demonstrations the image of the motor along with the plate information from the
motor. When the machine is at halt without mechanical load, the three – phase AC power is supplied
to the stator stations. The experimental setup along with current and voltage transducers is presented
in the following Figure A.31. The three - phase stator voltages, three – phase stator currents and the
rotor speed are recorded by the means of voltage and current transducers respectively and
tachometer, which is used to measure the speed of the rotor shaft.

Figure A. 30: Name plate for 7.5-hp Induction machine

48
 DAQ card used is 6040E .this DAQ has 68 pins as shown in the figure 28

Figure A.31: Physical Connection for Transducers

DAQ card in this experiment is used to measure three-phase voltage, three-phase current and
speed. The maximum voltage that can be applied to the DAQ card is 10 volt. If this input voltage
exceed 10 Volts it may damage DAQ card. Transducers used in our experiment step down voltage
from 220 to 5v maximum. This output of transducer will be measured through the DAQ card.
As it was said before, there are three phase voltage and currents are measured. So 6 pins are going
to be used.
Pins for 3-phase voltage
Ai0 pins (+68 , -34) voltage 1
Ai1 pins (+33, -66) voltage 2
Ai2 Pins(+65,-31) voltage 3
Pins for 3phase current
Ai3 pins (+30,-63) Current 1
Ai4 pins(+28,-61) Current 2
Ai5 pins(+60,-26) Current 3
Pin For speed
Ai6 pin(+25,-58)

49
Transducer:
Two different types of transducers are used
- Voltage transducer
- Current transducer

Voltage transducer:
Voltage transducer are used to step down high voltage (0-220) volts to (0 to 5) volts. Connections
will be in figure A.32 are shown.

Figure A.32: Voltage Transducer

Current transducer:
Current transducer are used to measure high currents in the range of -35/+35 A and produce an output
signal proportional to the input. The output range is a -10/+10 volts. The connections are shown in
figure A.33

Figure A.33: Current Transducer

50
Appendix B: Code

GA Code
%Objective is to solve nonlinear differential equation (3.10) and (3.14). The
file "Flux" calculated the derivative of the state
%variables, which are defined as fluxes.
globalfo Vs w wbWrCoeffTeCoeff
globalrsXrrp D XmXssXlrpXlsrrp% Machine parameters
choice = menu('Select Induction Machine','3-hp Induction Motor','2250-hp
Induction Motor','50-hp Induction Motor','500 hp Induction Motor');

if choice == 1
% Parameters for 3-hp machine
P = 4; fo = 60;
Vs = 220/sqrt(3);
rs = 0.435; Xls = 0.754; Xm = 26.13; Xlrp = 0.754; rrp = 0.816; J = 0.089;
Pn=3*746.7;
Tf = 0.5;
Mach = '3-hp machine';

elseif choice == 2
% Parameters for 2250-hp machine
P = 4; fo = 60;
Vs = 2300/sqrt(3);
%rs = 0.029; Xls = 0.226; Xm = 13.04; Xlrp = 0.226; rrp = 0.022;
rs=0.0359; Xls= 0.2003; Xm=13.0694; Xlrp=0.2020; rrp=0.0215; J = 63.87; Tf =
3.5;
Mach = '2250-hp machine';
elseif choice ==3
% Parameters for 50-hp machine
P = 4; fo = 60;
Vs = 460/sqrt(3);
rs = 0.087; Xls = 0.302; Xm = 13.08; Xlrp = 0.302; rrp = 0.228; J = 1.662; Tf
= 1.8;
Mach = '50-hp machine';
elseif choice ==4
% Parameters for 500-hp machine

51
 P = 4; fo = 60;
Vs = 2300/sqrt(3);
rs = 0.262; Xls = 1.206; Xm = 54.02; Xlrp = 1.206; rrp = 0.187; J = 11.06; Tf
= 1.8;
Mach = '500-hp machine';
end

we=2*pi*fo;
w = we; % Work in the synchronous /stationary reference frame
Tstep = Tf/2000;
tv = 0:Tstep:Tf; % Time vector for Vqdos

%Define voltages of three phases a, b, c as input to the model

Vas = sqrt(2)*Vs*cos(we*tv);
Vbs = sqrt(2)*Vs*cos(we*tv - (2*pi/3));
Vcs = sqrt(2)*Vs*cos(we*tv + (2*pi/3));
Vabcs = [Vas; Vbs; Vcs];

% Convert Vabcs to the QDO reference frame.

Vqdos = zeros(3, length(tv));
forpt = 1:length(tv);
th = w*tv(pt);
Ks = 2/3*[cos(th) cos(th-(2*pi/3)) cos(th+(2*pi/3)); ...
sin(th) sin(th-(2*pi/3)) sin(th+(2*pi/3)); .5 .5 .5];
Vqdos(:,pt) = Ks*Vabcs(:,pt);
end

%Solution for the Equations (3.10) and (3.14) will follow directly after
calling ode solver. Results of the state space equation are
%flux variables.
Xss = Xls + Xm;
Xrrp = Xlrp + Xm;
D = Xss*Xrrp - Xm^2;
wb = we;
WrCoeff = P/(2*J);
TeCoeff = .75*P*Xm/(D*wb);
% matrix to find currents
C = 1/D*[ Xrrp 0 0 -Xm 0 0;

52
 0 Xrrp 0 0 -Xm 0;
0 0 D/Xls 0 0 0;
-Xm 0 0 Xss 0 0;
0 -Xm 0 0 Xss 0;
0 0 0 0 0 D/Xlrp];

FL0 = zeros(8,1);
[t, FL] = ode15s('Flux',tv,FL0);

Iqdos = C(1:3,:)*FL(:,1:6).'; % Equation (3.14)

Iqdor = C(4:6,:)*FL(:,1:6).'; % Equation (3.14)

wr = FL(:,7);
theta = FL(:,8); % Theta-r
rpm = wr*(2/P)*60/(2*pi); % Mechanical rpm
Te = TeCoeff*(FL(:,1).*FL(:,5) - FL(:,4).*FL(:,2)); % equation

% Convert Iqdo back to Iabc

Iabcs = zeros(3, length(t));
Iabcr = zeros(3, length(t));

th1=w*t;
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
forpt = 1:length(t);
th = th1(pt);
Ks_inv = [cos(th) sin(th) 1; cos(th-(2*pi/3)) sin(th-(2*pi/3)) 1;
cos(th+(2*pi/3)) sin(th+(2*pi/3)) 1];
Iabcs(:,pt) = Ks_inv * Iqdos(:,pt);
thr = theta(pt);
b = th-thr;
Kr_inv = [cos(b) sin(b) 1; cos(b-(2*pi/3)) sin(b-(2*pi/3)) 1; ...
cos(b+(2*pi/3)) sin(b+(2*pi/3)) 1];
Iabcr(:,pt) = Kr_inv * Iqdor(:,pt);
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

bits = [10 10 10 10 10]; %Define the number of bits for GA

vlb=[0.1 0.2 20 0.5 0.5];

53
vub=[1 1 30 1 1];
options = foptions([1 1e-4]); % First:printingsecond:Terminate
options(11)=30; % Number of population - before it was 30
options(13) = 0.033; %Mutation probability
options(12)=0.70; %Crossover probability
options(14)=300; % Max number of gen

% load numeric3-hp.mat % 3-hp machine data

x = [rs, Xls, Xm, Xlrp, rrp];

%load numeric2250-hp.mat %2250-hp machine data

%load numeric50-hp.mat %50-hp machine data
%load numeric7.5-hp.mat %7.5-hp machine data
% we = 2*pi*60; % Omega-e, the angular frequency of the synchronous frame
% % w=we; % w
% Vqdos = zeros(3, length(tv)); % Initialize to 0.
% for pt = 1:length(tv);
% th = w*tv(pt);
% % If omega is not constant, theta = integral of omega
% Ks = 2/3*[cos(th) cos(th-(2*pi/3)) cos(th+(2*pi/3)); ...
% sin(th) sin(th-(2*pi/3)) sin(th+(2*pi/3)); .5 .5 .5];
% Vqdos(:,pt) = Ks*Vabcs(:,pt);
% end

FitnessFunction = @IMfitness;
nvars = 5;
[x,stats,options,bf,fgen,lgen]=ga(FitnessFunction,nvars,[],[],[],[],vlb,vub)
% x = ga(FitnessFunction,nvars);
% x % best fitness

54
FLUX

----------------------------------------------------------------------------------------------------------------------

function Flux = Flux(t,FL);

P = 4; fo = 60;
we = 2*pi*fo; %Omega-e, the angular frequency of the synchronous frame
w=we; % Work in the synchronous reference frame
wb = we;

Vs = 220/sqrt(3);
rs = 0.435; Xls = 0.754; Xm = 26.13; Xlrp = 0.754;
rrp = 0.816; J = 0.089; Pn=3*746.7;
Tf = 0.5;
Xss = Xls + Xm;
Xrrp = Xlrp + Xm;
D = Xss*Xrrp - Xm^2;
WrCoeff = P/(2*J);
TeCoeff = .75*P*Xm/(D*wb);

% FL = a 8-vector = [The 6 flux linkages in (3.10); Omega-r; Theta-r];

% This function calls ode15s (one of the differential solvers) to
%obtain solution for the fluxes Equation (3.10).
fo = 60;
Vs = 1;
we = 2*pi*fo; %Omega-e, the angular frequency of the synchronous frame
w=we; % Work in the synchronous reference frame
wb = we;
% Conversion of three phase voltages to synchronous QDO frame
Vas = sqrt(2)*Vs*cos(we*t);
Vbs = sqrt(2)*Vs*cos(we*t - (2*pi/3));
Vcs = sqrt(2)*Vs*cos(we*t + (2*pi/3));
Vabcs = [Vas; Vbs; Vcs];
th = w*t;
Ks = 2/3*[cos(th) cos(th-(2*pi/3)) cos(th+(2*pi/3)); ...
sin(th) sin(th-(2*pi/3)) sin(th+(2*pi/3)); .5 .5 .5];

55
Vqdos = Ks*Vabcs;
Vqdor = [0;0;0];
% Equation (3.10) in the state space form:
wr=FL(7);
A = -wb*[rs*Xrrp/D w/wb 0 -rs*Xm/D 0 0;
-w/wbrs*Xrrp/D 0 0 -rs*Xm/D 0;
0 0 rs/Xls 0 0 0;
-rrp*Xm/D 0 0 rrp*Xss/D (w-wr)/wb 0;
0 -rrp*Xm/D 0 -(w-wr)/wbrrp*Xss/D 0;
0 0 0 0 0 rrp/Xlrp];
Flux = zeros(8,1);
Flux(1:6) = A*FL(1:6) + wb*[Vqdos; Vqdor];
Te = TeCoeff*(FL(1)*FL(5) - FL(4)*FL(2)); % (equation 3.16)
Flux(7) = WrCoeff*Te; % The equation for d(wr)/dt is (3.17)
Flux(8) = FL(7); % Theta-r = integral(wr)
----------------------------------------------------------------------------------------------------------------------

function [rs,Xls,Xm,Xlrp,rrp] = IMfitness(x);

%IM fitness simulates the induction machine and as an output gives
%stator currents and rotor speed. It calls the main genetic function
%which further calls additional sub functions (reproduc, mate, xover
%etc.) It calculates the fitness at every step after calling genetic
%main function.
% global fo Vs w wbWrCoeffTeCoeff we Iabcs_actual
% global rsXrrp D XmXssXlrpXlsrrpindexxVabcstvwr_actual J P Vqdostvwr
% indexx =0
% Declare parameters as variables to be optimized:
rs= x(1) ;
Xls=x(2);
Xm=x(3);
Xlrp=x(4);
rrp=x(5) ;
% J=x(6);
% fitness = [rs, Xls, Xm, Xlrp, rrp,J];

56
PSO Code
% This file descibes the setup parameters for PSO.
% Partical Swarm Optimization for Matlab
% Copyright (C) 2003 AmerAlhinai
%
% This program is free software; you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation; either version 1, or (at your option)
% any later version.
%
% This program is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details. A copy of the GNU
% General Public License can be obtained from the
%%%%%%%%%%%%%%%%%%%%%%%%%
% This configuration depends in the problem, the given example is to design
% Ppower System Stabilizer parameters: K1 T11 T12 K2 T21 T22
global K1 T11 T12 K2 T21 T22
%%
c1=2.0;c2=2.01;
n=50;M=5;tmax=50;
bounds =[ 0 60;
0.1 0.5;
0 0.1;
0 60;
0.1 0.5;
0 0.1]; % Parameters max and min intervels
evalFN = 'objectives';
[Jss,xss,Jsss,t] = psomax(bounds,evalFN,tmax,c1,c2,n,M);
function [Jss,xss,Jsss,t] = psomax(bounds,evalFN,tmax,c1,c2,n,M)
% PSOmax run a Partical Swarm Optimization to maximize the objective
function.
% function [x,endPop,bPop,traceInfo]=ga(bounds,evalFN,tmax,c1,c2,n,M)
%
% Output Arguments:
% Jss - the best solution found during the course of the run.

57
% xss - the final population.
% Jsss - a trace of the best population.
% t - number of iteration.
%
% Input Arguments:
% bounds - a matrix of upper and lower bounds on the variables.
% evalFN - the name of the evaluation .m function.
% tmax - maximum number of iteration.
% c1 - weighting factor.
% c2 - weighting factor.
% n - number of popualtion.
% M - maximum number of iteration of unchanged soltion.
% Partical Swarm Optimization for Matlab
% Copyright (C) 2003 AmerAlhinai
%
% This program is free software; you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation; either version 1, or (at your option)
% any later version.
%
% This program is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details. A copy of the GNU
% General Public License can be obtained from the
%%%%%%%%%%%%%%%%%%%%%%%%%
x_int= bounds;
t=1; % 1st iteration.
%%%%%%%%%%%%%%%%%%%%%%%%%
phi=c1+c2;
K=2/(abs(2-phi-sqrt(phi^2-4*phi)));
fori=1:length(x_int) % initial population position
x(i,:)=x_int(i,1)+(x_int(i,2)-x_int(i,1))*rand(1,n);
end
fori=1:length(x_int) % initial velocities
Vmax(i)=max(x(i,:));Vmin(i)=min(x(i,:));
vk(i)=(Vmax(i)-Vmin(i))/n;

58
V(i,:)=-vk(i)+(vk(i)+vk(i))*rand(1,n);
end
vmax=vk;vmin=-vk;
size_V=size(V);
xs=x; %vs=V;
fori=1:size_V(2) % evaluate the initial population
xc=x(:,i);
e1str=['jc='evalFN'(xc);'];
eval(e1str);
J(i)=jc;
end
Js=J; % initial best individual
Jss=max(Js);xss=x(:,find(Jss==Js));
Jsss(t)=Jss; % initial best global
m=0;
while t<tmax% check maximum # of iteration
if m>M % check the number of iterations since the last change
break% of the best solution "m" is greater than "M"
end
t=t+1
m
r1=rand(1);r2=rand(1);
fori=1:size_V(1) % update the velocity
for k=1:size_V(2)
v_cal=K*(V(i,k)+c1*r1*(xs(i,k)-x(i,k))+c2*r2*(xss(i)-x(i,k)));
if (v_cal>=vmin(i) &v_cal<=vmax(i))
V(i,k)=v_cal;
else
end
end
end
fori=1:size_V(1) % update position
for k=1:size_V(2)
x_cal=V(i,k)+x(i,k);
if (x_cal>=x_int(i,1) &x_cal<=x_int(i,2))
x(i,k)=x_cal;
else

59
end
end
end
fori=1:size_V(2) % evaluation
xc=x(:,i);
e1str=['jc='evalFN'(xc);'];eval(e1str);
J(i)=jc;
end
fori=1:size_V(2) % update best individual
if J(i)>Js(i)
Js(i)=J(i);
xs(:,i)=x(:,i);
else
end
end
Jmax=max(Js);
ifJmax>Jss% update best global
m=0;
Jss=Jmax
xss=xs(:,find(Jss==Js));xss=xss(:,1);
else
m=m+1;
fprintf('%2.0f ',m)
end
Jsss(t)=Jss;
end
return
function [val] = objectives(sol)
% This function compute the Partical Swarm Optimization objective function.
% function [val] = objectives(sol)
%
% Output Arguments:
% val - PSO performance index value
%
% Input Arguments:
% sol - PSO candidate solution
% Partical Swarm Optimization for Matlab

60
% Copyright (C) 2003 AmerAlhinai
%
% This program is free software; you can redistribute it and/or modify
% it under the terms of the GNU General Public License as published by
% the Free Software Foundation; either version 1, or (at your option)
% any later version.
%
% This program is distributed in the hope that it will be useful,
% but WITHOUT ANY WARRANTY; without even the implied warranty of
% MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
% GNU General Public License for more details. A copy of the GNU
% General Public License can be obtained from the
%%%%%%%%%%%%%%%%%%%%%%%%%
% This configuration depends in the problem, the given example is to design
% Ppower System Stabilizer parameters:
% K1 Tw1 T11 T12 T13 T14 K2 Tw2 T21 T22 T23 T24 K3 Tw3 T31 T32 T33 T34
global K1 Tw1 T11 T12 T13 T14 K2 Tw2 T21 T22 T23 T24 K3 Tw3 T31 T32 T33 T34
K1=sol(1); Tw1=sol(2);
T11=sol(3);T12=sol(4);
T13=sol(5); T14=sol(6);

K2=sol(7); Tw2=sol(8);
T21=sol(9);T22=sol(10);
T23=sol(11);T24=sol(12);

K3=sol(13);Tw3=sol(14);
T31=sol(15);T32=sol(16);
T33=sol(17);T34=sol(18);
%%%%%%%%%%%%%%%%%%%%%%
sim('m_IEEE37_2MT') % Nonlinear Simulations
if tout <7 % If the guided constrains interepted the simulation ignor this
solution
val=0
else% Good candidate solution
j=min(sum(J)); % J is the simulation performance index computed within PAT
val=1e6/j
end

61
References

[1] ZbigniewMichalewicz. Genetic Algorithms + Data Structures = Evolution Programs. Springer –

Verglar 3rd edition, 1996.

[2] David E. Goldberg. Genetic Algorithms in Search, Optimization and Machine Learning. Addison –
Wesley Publishing Company, Inc. 1989.
[3] Steven R. Shaw. Numerical Methods for Identification of Induction Motor Parameters. Master’s
thesis, MIT, January 1997.
[4] Dennis J. Evangelista. Parameter Estimation of Induction Motors using PWM Inverters.Master’s
thesis, MIT, May 1999.
[5] A. E. Fitzgerald, C. Kingsley, S. D. Umans. Electric Machinery. McGraw-Hill, fourth edition, 1983.
[6] Paul C. Krause, Oleg Wasynczuk, Scott D. Sudhoff. Analysis of Electric Machinery. IEEE Press,
1995.
[7] Lawrence Davis. Handbook of Genetic Algorithms. International Thomson Computer Press, 1996.
[8] P. C. Krause, C. H. Thomas. Simulation of Symmetrical Induction Machinery. IEEE Transactions on
Power Apparatus and Systems, November 1965.
[9] Richard R. Bishop, Gill G. Richards. Identifying Induction Machine Parameters Using a Genetic
Optimization Algorithm. IEEE Proceedings – 1990 Southeastcon.
[10] P. Ju, E. Handschin, Z. N. Wei. Application of genetic algorithm to nonlinear dynamic modeling.
1995 Elsevier Science B. V.
[11] Seung-Ill Moon, Ali Keyhani, Srinivas Pillutla. Nonlinear Neural – Network Modeling of an
Induction Machine. IEEE Transactions on Control Systems Technology, March 1999.
[12] H.B. Karayaka, A. Keyhani. Induction Machine Parameter Tracking from Test Data via PWM
Inverters. IEEE Industry Applications Society, October 1997.
[13] JA de Kock, FS van der Merwe, HJ Vermeulen. Induction Motor Parameter Estimation through an
Output Error Technique. IEEE Transactions on Energy Conversion, March 1994.
[14] Simon Hart, AbolfazlRanjbar, B.E. Mulhall. Practical Combined Parameter Identification and State
Estimation of Machines. Proceedings volume from the 8th IFAC Symposium, UK, September 2000.
[15] Robin Biesbroek. GA Tutorial. http://www.estec.esa.nl/outreach/gatutor/Default.htm.
[16] MATLAB, High Performance Computation and Visualization Software, The Mathworks Inc., 2001.
[17] Genetic Algorithms. Implementation based on GAs from David E. Goldberg. Andrew F Potvin. The
MathWorks, Inc. January 1994.
[18] Basics on GAs. http://www.d.umn.edu/ece/lis/GA/ga_1_basics.html.
[19] LabVIEW Data Acquisition Basics Manual. National Instruments. January 1998.
[20] LabVIEW User Manual. National Instruments. January 1998.
[21] DAQ PCI E Series User Manual. National Instruments. July 1997.
[22] KouroshSeghisigarchi. DAQ User Manual. WVU 1999.

62
[23] IEEE Standard Test Procedure for Polyphase Induction Motors and Generators. Electric Machines
Committee of the IEEE Power Engineering Society, USA. IEEE Std 112-1996. INSPEC Accession
Number: 5621352.
[24] Yu-hua-Wang, Birdwell-JD. Dynamic identification of the model parameters for an induction motor.
Conference Proceedings of IEEE SOUTHEASTCON, 1982.
[25] National Instrument and LAB View Software. http://www.ni.com/.

[26] P. C. Krause, O. Wasynczuk, S. D. Sudhoff, Analysis of Electric Machinery, Institute of Electrical &
Electronics Engineer; , January 1995.

[27] Y. Koubaa, “Recursive identification of induction motor parameters,” Journal of Simulation

Modeling Practice and Theory, Vol. 12, Issue 5, pp. 363-381, August 2004.
[28] J. Stephan, M. Bodson, and J. Chiasson, “Real Time Estimation of Induction Motor Parameters,”
IEEE Trans. on Industrial Applications, Vol. 30, No. 3, pp. 746-759, May-June 1994.
[29] K. Wang, J. Chiasson, M. Bodson, L. M. Tolbert, “A Nonlinear Least-Squares Approach for
Identification of the Induction Motor Parameters,” IEEE Trans. on Automatic Control, Vol. 50, No. 10,
pp. 1622-1628, October 2005.
[30] H. A. Toliyat, E. Levi, and M. Raina, “A Review of RFO Induction Motor Parameter Estimation
Techniques,” IEEE Trans. on Energy conversion, Vol.18, No. 3, pp. 271-283, June 2003.
[31] R. R. Bishop, G. G. Richards,“Identifying Induction Machine Parameters Using A Genetic
Optimization Algorithm,” Proceedings. Southeastcon IEEE, Vol.2, pp. 476-479, New Orleans, LA, USA,
April 1990.
[32] F. Alonge, F. D´ lppolito, G. Ferrante, F. M. Raimondi,“ParameterIdentification of Induction Motor
Model Using Genetic Algorithms,” IEE Proc. Control Theory Applications, Vol. 145, No. 6, November
1998.
[33] K. S. Huange, Q. H. Wu, D. R. Turner,“Effective Identification of Induction Motor Parameters
Based on Fewer Measurements,” IEEE Trans. on Energy Conversion, Vol. 17, No. 1, March 2002.
[34] E. Bajrektarevi´c, M. A. Choudhry, “Parameter Identification of Induction Motor using a Genetic
Algorithm and Lab View Software,” Proceedings of 34th North American Power Symposium, October
14-15, 2002.
[35] R. K. Ursem, P. Vadstrup, “Parameter Identification of Induction Motors Using Differential
Evolution,” The 2003 Congress on Evolutionary Computation CEC ’03, Vol. 2, pp. 790- 796, 8-12
December 2003.
[36] IEEE Standard Test Procedure for Polyphase Induction Motors and Generators.
Electric Machines Committee of the IEEE Power Engineering Society, USA. IEEE Std112-1996.
INSPEC Accession Number: 5621352.
[37] D.P.Rini,S,M.shamsuddin,S.S.Yuhaniz,”Particle Swarm Optimization: Technique, System and
Challenges” Volume14-No.1,January2011
[38].Qinghai Bai”Analysis of Particle Swarm Optimization Algorithm”Vol.3 No.1 February 2010
[39] J. Kennedy, R. C. Eberhart, Y. Y. Shi, Swarm Intelligence, The Morgan Kaufmann Series in
Artificial Intelligence, March 2001.
[40] S. R. Shaw, S. B. Leeb “Identification of Induction Motor Parameters from Transient Stator Current
Measurements,” IEEE Trans. on Industrial Electronics, Vol. 46, No. 1, February 1999.

63
[41]NadeekaYapa “Genetic Algorithm in Induction Motor Efficiency Detemination” May 2004
[42]Rania Hassan,Babakcohanim,Olivier De Weck “A comparison of Particle Swarm Optimization And
and the Genetic Algorithm”
[43] http://www.talkorigins.org/faqs/genalg/genalg.html#strengths

[45]Ali Karimi, M.A,Choudhry , Ali Feliachi "PSO-based Evolution Optimization for Parameter
Idenetification of an induction Motor" IEEE,2007

[46] Edina Bajrektarevic "Parameter identification of induction motor using a Genetic Algorithm" Masters
thesis ,2002

[47 ]Phaderm Nangsue, Pragasen Pillay, Susan E. Conry. Evolutionary Algorithms for Induction Motor
Parameter Determination.

[48 K.S. Huang, W. Kent, Q.H. Wu, D. R. Turner, Parameter Identification of an

Induction Machine Using a Genetic Algorithms. Proceedings of the 1999 IEEE,
August 1999.]

64