You are on page 1of 8

Temperature Control In Dry Type

Transformer Using Fuzzy Logic

AUTHORS:
N.GURURAJAN
N.HARIHARASUDHAN
III YEAR E&I
VELAMMAL ENGINEERING COLLEGE

E-MAIL: ngururajan_86@yahoo.com
Abstract

The transformer being a very versatile device finds great applications. The oil filled
transformers offer high efficiency and better cooling and are mostly used. However, their
use becomes hazardous in mines and high-rise buildings. This is where the dry type
transformers being safer find great application.

The temperature is a very important parameter in any transformer due to its influence on
the transformer insulation. In a dry type transformer its significance further increases, as
here air is the sole cooling medium. Thus for the most efficient utilization of a DTT, the
control of the temperature within the prescribed limits becomes paramount. Thus by
using the air blast cooling through a fan the cooling can be accomplished. The important
parameter to be controlled becomes the fan speed that needs to be different for different
loads. In the conventional methods of control, the control is exercised within the strict
prescribed limits.

Hence, for any intermediate values within these limits the required real time value is
never obtained. This paper has been framed with the aim to achieve predictive control.
This predictive control is accomplished by the use of fuzzy logic. Thus the fan speed can
be controlled such that the transformer is never under loaded even with temperature rise.
Most importantly, in conventional systems, the accuracy is affected due to the
approximations made during Linearization. Fuzzy Logic enables us to work with the non-
linear parameters without approximating them. Further Fuzzy control provides us with
smaller and more efficient programs, which reduces the response time of the system.
Control process

Prerequisites for Implementation

The first thing to realize is the temperature rise can be controlled in two ways: by
reducing the load and by increasing the fan speed. However in the former the operations
of the circuit breakers on the load side have to be controlled, which is a tedious process.
Instead it is relatively easier to control the fan speed.
The use of Fuzzy Logic enables us to consider the non-linearities of the system and hence
its ambiguities. Therefore, the all the fuzzy rules are formulated based on this condition.
While framing the rules, the various possible intermediate states have to be considered in
addition to the extreme values of each and every parameter. The formation of the rule is
the basic step, for which "If X AND/OR if Y Then Z" relation has to be specified,
wherein X, Y are the inputs and Z is the controlled output. In this case, the input variables
are the Temperature and Load, while Fan speed is the output.
Few sample rules used are:

To comprehend these sets of rules we have to assign the degree to each


membership function. These degrees are assigned to each membership function based on
the Linguistic Rules. This step enables us to decipher the Fuzzy Triangle for each
membership function with the intermediate conditions well specified. These Fuzzy
triangles are a pictorial representation of the membership function, giving its limits and
also the trueness of each linguistic rule. For each membership function the Fuzzy
Triangles are indicated below:
Fuzzy Matrix Formation
Fuzzy matrix formation involves the use of fuzzy arithmetic. Our problem
presents us with two input variables that in combination determine the output. Therefore,
we use the MAX operation. This operation is performed on the corresponding degree of
inputs to get the output. Thus, the 5 X 5 matrix that is sufficient for our problem is
generated.

Temperature
Cool amb Warm Hot vhot
L vlow Zero slilow Med qhigh vvhigh
O Small Low slilow Med qhigh vvhigh
A Rated modlow modlow Med qhigh vvhigh
D High modhigh modhigh modhigh qhigh vvhigh
vhigh vvhigh vvhigh vvhigh vvhigh vvhigh

This is the heart of the problem; this specifies the operations that need to be performed by
the software when given the corresponding inputs.

Implementation
This includes the Defuzzification process, which can be achieved in two ways: manually
and by using software. The former process is tedious and time consuming, whereas the
software gives spontaneous and accurate results. The C program that can be used to
predict the output for the various intermediate states is formulated to give the results in
terms of the membership functions.

#include <stdio.h>
# include <conio.h>
/* sample main program */

void main()
{
float fuzzy(float in1, float in2);
float in1, in2;

scanf("%f",&in1);
scanf("%f",&in2);
printf("Output = %g\n",fuzzy(in1,in2));
getch();
}

float min(float x,float y);


float fuzzy(float in1, float in2) /* fuzzy function */
{
float trapezium(float val, float p1, float p2, float p3, float p4);
float output, m1, m2, nsum=0.0, dsum=0.0;
int i, j, k;
float memin1[] = {
0.000000, 0.100000, 0.100000, 0.200000,
0.100000, 0.200000, 0.200000, 0.400000,
0.200000, 0.400000, 0.400000, 0.800000,
0.400000, 0.800000, 0.800000, 1.000000,
0.800000, 1.000000, 1.000000, 1.000000,
};
float memin2[] = {
0.000000, 0.100000, 0.100000, 0.200000,
0.100000, 0.200000, 0.200000, 0.400000,
0.200000, 0.400000, 0.400000, 0.800000,
0.400000, 0.800000, 0.800000, 1.000000,
0.800000, 1.000000, 1.000000, 1.000000,
};

float memout[] = {
0.000000, 0.100000, 0.100000, 0.200000,
0.100000, 0.200000, 0.200000, 0.400000,
0.200000, 0.400000, 0.400000, 0.800000,
0.400000, 0.800000, 0.800000, 1.000000,
0.800000, 1.000000, 1.000000, 1.000000,
};
int mat[5][5] = {
1, 1, 1, 1, 1,
1, 2, 2, 2, 2,
1, 2, 3, 3, 3,
1, 2, 3, 4, 4,
1, 2, 3, 4, 5,
};

for(i=0; i<5; ++i)


for(j=0; j<5; ++j)
if(mat[i][j]>0)
{
float
m1=trapezium(in1,memin1[i*4],memin1[i*4+1],memin1[i*4+2],memin1[i*4+3]);
float
m2=trapezium(in2,memin2[j*4],memin2[j*4+1],memin2[j*4+2],memin2[j*4+3]);
m1=min(m1,m2);
for(k=0; k<4; ++k)
{
nsum=nsum+m1*memout[(mat[i][j]-1)*4+k];
dsum=dsum+m1;
}
}
if(dsum==0.0)
output=0.0;
else
output=nsum/dsum;
return output;
}
Output is obtained both graphically and in the text format, which is shown below.

Interfacing with the Hardware


The process described so far, will result only in determining the fan speed for the various
intermediate conditions. However, in the real time application, the fan has to attain the
calculated speeds in order to effectively control the temperature rise. This is achieved by,
interfacing the fuzzy system with hardware.
Limitation
There should be sufficient time lag in the temperature rise in the core for the fuzzy
system to predict and help maintain the temperature within the prescribed limits.
We have also neglected the effect of Pressure, Moisture content, change in Dielectric
Constant of the insulation, etc.

Conclusion
This system demonstrates the temperature control in a dry type transformer. It
offers a system which results in the optimum usage of a DTT.It being a predictive
methodology produces near real time values. This is only a two input system, but for
complete practical implementation, it would be essential to increase the no. of inputs to
suit the corresponding real time situation.

Bibliography

1.”Introduction To Fuzzy Logic For Engineering Applications”, Zadeh Lofti,2nd Edition.


2.”Fuzzy Set Theory And Its Applications”Zimmermann H.J.,2nd Edition,Kluwer
Academic Publishers.
3.”Fuzzy Sets,Uncertainity And Information”George Klir & Tina A.Folger,Prentice Hall
Of India.

You might also like