GROUP ACTIVITY FOR AC MACHINES

Genetic Algorithm
Determining the parameters of Induction Motor

Date : 22.03.2018 Submitted By

BATCH 1 : Abishek K R - 107116002
Aishwarya R - 107116004
Aman Barnwal - 107116006
Amritha S B - 107116008
 ABSTRACT

An approach is presented for determining the equivalent circuit parameters of induction

motors by genetic algorithms. An equivalent circuit without considering the steel losses is
analyzed. The sensitivity of the approach is discussed by using one, two and three sets of
data. The accuracy of the proposed approach is analyzed by determining the relative error
in the parameters, obtained by genetic algorithms, with regard to analytical values. Genetic
Algorithm based optimal design of three phase induction motor is proposed. Optimal
design of induction motor is taken as an optimization problem with efficiency and active
material cost as objective function. Different constraints and different variables are
imposed to achieve the best design within a specific range of variables. A genetic algorithm
is used to achieve optimal design. This paper covers the application of genetic algorithms to
use motor nameplate data along with rated values such as voltages, currents and torque
data to determine unknown operating motor parameters for an AC induction motor. These
parameters can be used with textbook equations to offer solutions to load requirements
and capabilities.

1
INTRODUCTION
Electric drives using induction motors are one of the main fields of interest to the control
systems and electrical engineering specialists. The quality and effective control of induction
motors (IM) is based on their equivalent circuits. The conventional method for estimation
of IM equivalent circuit parameters is based on the no-load and blocked rotor tests which
are a time-consuming task, especially if the motor is already coupled to driving equipment.
This report presents a more sophisticated approach for determining the equivalent circuit
parameters of induction motors. The approach is based on only a set of measured or
analytically obtained data such as stator voltage, current, power factor and slip. A genetic
algorithm model for determining the parameters is synthesized that enables the
simultaneous satisfaction of steady-state stator current and power factor by defining an
objective function. The results obtained by the synthesized model are compared with
analytical data. An analysis is performed that proves the validity and adequacy of such
models in IM control systems.

PROCEDURE
It determines the equivalent circuit parameters based on the recorded time variations of
voltage, current, power and speed from start-up till no-load. It is applicable only to large
high voltage induction machines rated 1MW and above. Some of the are based on the motor
nameplate and catalog data. Determining the equivalent circuit parameters when taking
into account the steel losses are described in [6]. The methods are based on several
assumptions such as:

● zero value of the referred leakage inductance of rotor winding when determining
copper losses
● zero value of the referred leakage inductance of stator winding when determining
steel losses

All three methods require knowing the rated supply voltage, stator current, rated power,
power factor, rated speed and rated efficiency. The errors of the different parameters when
using these methods vary from 4% to 60%.

2
GENETIC ALGORITHM:
Genetic algorithms (GA) belong to optimization methods for solving sets of non - linear
equations. They use objective functions based on some criterion which is most often the
calculated error. Sometimes the reciprocal value of this criterion, called fitness function, is
used. GA are based on natural selection and natural genetics. Using random numbers, they
do not need a good initial guess for unknowns. The mechanisms of the most elementary GA
consist of the following steps:

1. create an initial population

2. evaluate the fitness of each population member
3. invoke natural selection
4. select population members for mating;
5. generate offspring
6. mutate selected members of the population
7. terminate or go to step 2.

The substantial advantages of genetic algorithms over other optimization methods are:

● GA are able to find the fitness function's global minimum instead of a local
minimum. This is due to the fact that GA simultaneously explores many points in the
search space. Covering the whole search space, they are less likely to stop at a local
minimum
● GA do not require problem-specific auxiliary knowledge such as derivatives of the
function
● initial guesses for estimated parameters do not need to be close to the actual values
● GA perform very well for large-scale optimization problems.

3
DEVELOPMENT OF THE GENETIC ALGORITHM MODEL:

The set of equations of the genetic algorithm model is based on the T-shaped induction
motor equivalent circuit without considering the steel losses (Fig. 1).

There are five unknowns in this circuit, namely: stator resistance and stator leakage
reactance , rotor resistance and rotor leakage reactance (both referred to stator), and
magnetizing reactance . The known quantities from measurement are the input voltage that
equals the rated voltage, the input current , power factor and the slip .

Based on the circuit in Fig. 1, the stator current and power factor can be computed:

where Zeq is the equivalent circuit impedance. The equivalent circuit resistance and
reactance in (1) are equal to:

4
Next the power factor can be computed

The criterion for selecting the best individuals in the genetic algorithm is the objective
function. An adequately chosen objective function guarantees that the next generation is
usually closer.

The objective function , used in the present study, is the average error in the power factor
and input current for various load points.

Here and are the values computed by (1) and (4) and are measured or analytical values.
The variable varies from 1 to 3 in our case.

The aim of GA is to minimize the error of the objective function.

5
MATLAB CODE:

●// CODE FOR MAKING THE FUNCTION FOR GENETIC ALGORITHM

function z= twofunc(x)

V1=415;

Irat=6.9;

PFrat=0.746;

Req= x(1)+ (15*x(2)*(x(5)^2))/((15*x(2))^2

+(x(5)+x(4))^2);

Xeq= x(3) + (((15*x(2)*x(2)*x(5))+(x(4)*x(5)*(x(5)+x(4))))/((15*x(2))^2 +

(x(5)+x(4))^2));

Zeq=((Req^2 +Xeq^2)^(0.5));

Ic= V1/Zeq; m=Req/(Zeq);

PFc= cos(angle(m)); z= ((Ic-

Irat)/Irat)^2 + ((PFc-

PFrat)/PFrat)^2; end

6
 ● // CODE FOR PLOTTING THE GRAPH BETWEEN PENALTY VALUE
AND GENERATION IN ORDER TO ANALYZE ACCURACY

lb = [2.0 0.0 3.0

3.0 40]; ub = [9 4

20 20 190]; opts

= gaoptimset(...

'PopulationSize', 500, ...

'Generations', 750, ...

'EliteCount', 1, ...

'TolFun', 1e-13, ...

'SelectionFcn',{@selectiontournament,2},...

'PlotFcns', @gaplotbestf);

rng(0, 'twister');

[xbest, fbest, exitflag] = ga(@twofunc, 5, [], [], [], [], lb,

ub,[],[1:1], opts); display(xbest);

%%

% We can also ask |ga| to return the optimal volume of the beam.
fprintf('\nCost function returned by ga = %g\n', fbest);

7
// CODE FOR GRAPHS IN ORDER TO COMPARE THE TWO VALUES, ONE

FROM USING GENETIC ALGORITHMS and OTHER THE EXPERIMENTAL

VALUES
clc;

R1=7.0000;

R2=3.8187;

X1=17.2616;

X2=15.3638;

Xm=175.4813;

Rm=249.72;

V=415;

Ns=1500;

Ws=157.14; s=-1:0.001:2;

Z1=(7+17.2616i);

Z2=((3.8187)./s)+(15.3638i);

Zm=((249.72*(175.4813i))/(249.72+175.4813i));

Zt=Z1+((Z2*Zm)./(Z2+Zm));

Nr=(1-s)*Ns;

Wr=(1-s)*Ws;

I1=abs(230./Zt);

I2=abs((I1*Zm)./(Z2+Zm));

Im=I1-I2;

pf=cos((angle(Zt)));

Pg=((3*(3.158).*(I2.*I2))./s);

Po=(1-s).*Pg;

T=Po./Wr;

Pin=3*415*I1.*pf; n=(Po*100)./Pin;
8
figure(1);

plot(Po,pf,Output_Power,Power_factor);

hold on; title('Power factor vs Output

power') grid on; xlabel("Output Power");

ylabel("Power Factor");

figure(2);

plot(Po,n,Output_Power,EFficiency);

title('Efficiency vs Output Power');

grid on; xlabel("Output Power");

ylabel("Efficiency");

figure(3);

plot(Po,Nr,Output_Power,Speed);

title('Speed vs Output Power');

grid on; xlabel("Output Power");

ylabel("Speed");

figure(4);

plot(Po,T,Output_Power,Torque);

title('Torque vs Output Power');

grid on; xlabel("Output Power");

ylabel("Torque");

9
figure(5);

plot(Po,I1,Output_Power,Line_current);

title('Line Current vs Output Power');

grid on; xlabel("Output Power");

ylabel("Line Current");

figure(6);

plot(Po,s,Output_Power,Slip);

title('Slip vs Output Power');

grid on; xlabel("Output

Power"); ylabel("Slip");

GRAPHS :
10
RESULTS :
 The proposed approach is more simple, faster, less intrusive and cheaper than the conventional
experimental or computational methods for estimating the equivalent circuit parameters of induction
motors.
 The values of the parameters that we got from this genetic algorithms approach by using the required
machine details :
R1 = 7.0000 Ω
R2 = 3.3817 Ω
X1 = 17.2616 Ω
X2 = 15.3638 Ω
Xm = 175.4813 Ω
Parameters Values from Genetic Values from Error
Algorithm Approach experimentation

R1 7.0000 6.72 0.1143

R2 3.8187 3.1 0.18805

X1 17.2616 17.6 0.01923

X2 15.3638 17.6 0.12706

Xm 175.4813 166.67 0.05287

CONCLUSION:

An efficient approach for determining the equivalent circuit parameters of induction motors is presented. Based on genetic algorithms, it
needs only a few sets of electrical input data (voltage, current, power factor) and slip of the motor. The sensitivity and accuracy of the
approach are analyzed. The results show that the approach is sensitive to the number
of input data sets. In order to achieve acceptable accuracy, at least two sets of input data are required. When three sets of data are used,
the approach yields excellent accuracy, the maximum relative error of the estimated parameters with regard to analytical values being
less than 1%. The proposed approach is more simple, faster, less intrusive and cheaper than the conventional experimental or
computational methods for estimating the equivalent circuit parameters of induction motors.

REFERENCES
1. http://www.uni-miskolc.hu/~microcad/publikaciok/2016/C4_8_Gyorgy_Tamas.pdf
2. https://www.hindawi.com/journals/je/2014/148204/
3. http://ieeexplore.ieee.org/document/7853678/
4. http://ieeexplore.ieee.org/document/1216896/