SINERGI Vol. 23, No. 2, June 2019: 99-106
Accredited SINTA 2 by KEMENRISTEKDIKTI, Decree No: 10/E/KPT/2019
http://publikasi.mercubuana.ac.id/index.php/sinergi
http://doi.org/10.22441/sinergi.2019.2.002
.

ASSESS THE RISK LEVEL OF POWER TRANSFORMER DUE
SHORT-CIRCUIT FAULTS BASED ON ANFIS
Azriyenni Azhari Zakri1, Mohd Wazir Mustafa2, Hari Firdaus1, Ibim Sofimieari2
1
Department of Electrical Engineering, Faculty of Engineering, Universitas Riau
Jl. Subrantas km. 12.5, Simpang Panam, Pekanbaru 28293, Indonesia
2
School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia
Sultan Ibrahim Chancellery Building, Jalan Iman, 81310 Skudai, Johor, Malaysia
Email: azriyenni@eng.unri.ac.id wazir@utm.my hari.firdaus@student.unri.ac.id
Abstract -- A power transformer is an electrical machine that converts electrical power at different
voltage levels. Faults, occur in power transformers, inhibit electrical power distribution to the consumer.
Protection, therefore, of the power transformers is essential in power systems reliability. The power
system can be reliable if the protection devices work well when there is a fault. A hybrid intelligent
technique, which is a combination of Artificial Neural Network (ANN) and Fuzzy known as Adaptive
Neuro-Fuzzy Inference Systems (ANFIS), was used in this research. The objective of this paper is the
simulation of differential relays as a protection device on power transformers using Matlab/Simulink.
Performance of differential relays for power transformers protection is carried out with internal and
external fault scenarios. The input data were classified into three different input for ANFIS such as
internal and external 1, internal and external 2, internal, external 1, and external 2, respectively. The
error results of ANFIS training for the type of fault internal and external 1 is 9.46*10-7, and types of fault
internal and external 2 is 1.09*10-6 internal, external 1 and external 2 are 8.59*10-7. The results obtained
from the simulation were accurate and shows that the ANFIS technique is an efficient method that gives
less error and a great value. Finally, the technique can minimize faults with power transformers. Finally,
to prove this method can reduce faults in the power transformer, the assess of this model has been
carried out through the RMSE that has been generated which is zero.
Keywords: ANFIS; Differential relay; Power transformer; Resetting; Faults
Copyright © 2019 Universitas Mercu Buana. All right reserved.
Received: May 2, 2019

INTRODUCTION
Fault diagnosis aims to estimate the
occurrence faults in power systems using a
hybrid intelligent technique, namely ANFIS. The
problem is that the power transformer system
does not detect short circuit fault caused by not
working the relay properly as a protection
device. Therefore, there is a need for an
assessment of the protection relays on power
transformers, wherein this study, the focus is
differential relays. To reduce the fault occurred
and the continuous supply of electricity, it is vital
to recognize the type of fault before starting the
recovery operation. Therefore, a fault diagnosis
system intelligence is an alternative way to
restore fault in protecting the system (Kim et al.,
2019; Li, Wang, & Song, 2019; Kaur, Brar, &
Leena, 2019).
ANFIS is a neuro-fuzzy adaptive inference
system
based
on
artificial
intelligence
techniques based on the Takagi-Sugeno model.
ANFIS technique is a combination of two
methods - the Fuzzy and Neural Networks,

Revised: May 29, 2019

Accepted: June 2, 2019

designed to produce a new hybrid intelligent
technology. By applying the ANFIS technique, it
is expected to minimize short circuit fault in
power transformers, both internal and external
(Salama, Abdel-Latif, Ismail & Mousa, 2018; Ge et
al., 2018; Lei et al., 2007).
There are several categories of ANFIS
methods which are based on input learning,
neural
networks,
fuzzification,
and
defuzzification. The Fuzzy method uses If-Then
rules by having input and output exercises for the
same data as the Back Propagation (BP) method
of neural networks. If-Then rules on actual
values can be trained with minimum operator
requirements to calculate their suitability. This
study will discuss several studies with the hybrid
intelligent techniques related to fault diagnosis in
power systems.
Fault diagnosis detects a fault in the power
system using relay information and Circuit
Breaker (CB). If the relay and CB failed to
operate, then the backup protection will work,
which requires a robust system of fault diagnosis

A.A. Zakri et al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

99

SINERGI Vol. 23, No. 2, June 2019: 99-106

that might eliminate faults and helps protect the
power system. Fault diagnosis becomes a
complex decision process that is important for
awakening several ways that can provide fault
analysis. The diagnosis can be achieved with a
fault diagnosis system for power systems using
the Energy Management System (EMS).
In the electrical power system, it is a crucial
element to achieve the continuity of electrical
service from the power plant to consumers as end
users. Protection systems need to be used to
detect fault from power systems and isolate fault
from components or protective devices to the
system. The system can also function sufficiently
to separate the mistake. The proposed method
operates only using data obtained from
substations consisting of voltage and three phase
currents. An active fault diagnosis algorithm for
modifying the operating mode has a complicated
fault with the electric power system. Automatic
retrieval of incomplete information will be
uploaded to the control center in the electrical
power system. This mechanism can lead to loss of
information services in the control center. If one or
more of the relays and CB do not function
properly, the algorithm model is needed correctly
to assess the fault (Zhangjun et at., 2012).
Liping Qu and Haohan Zhou have
introduced a Support Vector Machine (SVM)
method to analyze and diagnose transformer
errors. Estimates of transformer errors are playing
an essential role in the operation of power systems
to be safe and stable. To detect an initial error on
the transformer as early as possible, accumulation
of error data, the SVM forecast model takes RBF
using kernel functions and utilizes patterns to
overcome data and to reduce imbalances. The
final experimental results show that SVM makes a
better diagnosis accuracy (Qu & Zhou, 2015).
Intelligent hybrid techniques have also been
introduced in research into power protection in
transmission. This hybrid technique is an ANFIS
technique that is applied for distance relay
protection, especially for transmission lines. When
a fault occurs during identification on a
transmission line caused by an unwanted fault, the
power delivery to the consumer does not work
correctly. Therefore, they have provided
alternative solutions to fix this problem. The
transmission line impedance has determined how
long distance relays will protect the channel
distance. Simulation results have shown that
excellent testing results can contribute to
alternative algorithms that have good performance
to protect the system in the transmission line
(Azriyenni & Mustafa, 2015).
Gang Peng et al. has presented a
diagnostic method for electrical equipment

100

compatible with using multilayer fuzzy vectors.
This method mining information about equipment
features using the fuzzy logic method, many
vector support tools use multilayer features as
input and determining the class of devices that are
problematic by contributing to the voting method.
Through the application of algorithms in fault
diagnosis using circuit breakers and transformers,
explain this method can overcome abnormal data.
Therefore, overcoming the sensor difficulties for
unusual conditions and increasing the accuracy of
equipment diagnosis (Peng et al., 2017).
Azriyenni et al. in his research have
examined the Backpropagation of Neural
Networks to detect the location of a fault in the 150
kV transmission line. Distance relay is one of the
protective equipment that becomes the object of
research for protection devices that often used in
power system transmission. The Backpropagation
of neural networks is a computational model that
uses the training process can be used to solve the
problem of the limitations of distance relay
protection work. The Backpropagation also does
not have limited impedance range settings. When
the output gives the wrong result, the weight can
be minimized and the response from the error. The
Backpropagation technique has proven that the
output is closer to the correct value. Therefore,
The Backpropagation structure successfully
detects the fault location and identifies the output
current and CB status that trip (Azriyenni & Dame,
2017; Azriyenni et al., 2014).
Soloot et al. have presented internal faults
in transformers in many of the failure cases.
Developing a method to detect the location of
internal short circuits inside the transformer
winding can be helpful for a more cost-effective
repair and enhancing the transformer design.
Frequency response analysis can estimate the
location of an internal fault inside the winding. The
results showed that there are systematic changes
in the frequency response of High Voltages (HV)
to Low Voltage (LV) or vice-versa when the
position of the internal fault moves from input
layers of the winding to the output layers. The
input impedance from HV and LV terminals are not
sensitive to the location of short circuits in the
frequency range of 10 kHz-10 MHz (Soloot et al.,
2015).
In the last few years, intelligent techniques
are a method that is often used in providing
alternative solutions to solving problems in
forecasting fault in the power system. This study
will propose a fault modeling using a differential
relay to protect the power transformer via
Matlab/Simulink. Differential relay performance is
also used to simulate short circuit fault internal and
external faults. The application of the ANFIS

A.A. Zakri at al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

technique for forecasting the intended power
transformer system minimizes internal and
external fault in the power transformer. ANFIS
differential relay can assess fault with input data,
that is; the fault of Internal, External 1, and
External 2 at RMSE values that are very accurate,
which is zero. The ANFIS model is run using the
Membership Function (MF) Gbell and triangle.
Comparisons will be made between the two MF to
reach the target of correct errors, and to reach the
precise error target, and several iterations are
carried out as much as possible.
METHOD
The Power Transformer
A short-circuit occurs at the power
transformer; the short-circuit current can flow
through the windings. This current is several times
the rated current of the transformer. Due to this
increased short-circuit current, the transformer
winding is prone to undergo mechanical fatigue
due to the electromagnetic forces, which are
proportional to the square of the short-circuit
current. Several methods have been described for
short circuit interference in power transformers.
The method of using the probability of a Monte
Carlo simulation for external fault has also been
introduced by Flores et al. (2008).
The electric power system built has one
generator unit, two measuring transformer units,
and a 25 MVA power transformer using differential
relays. Fig. 1 shows a single line diagram model
of a power transformer using differential relays,
which will be simulated when a fault occurs.
Simulations of fault occur at three points of fault
location, which consists of; internal fault, external
fault one, and external fault two. A few faults will
be simulated with a type of short circuit faults:
there are; one phase to ground fault (AG), two
phases to the ground (BCG), phase to phase (BC),
and three phases (ABC).
11 kV / 20 kV
Bus 11 kV

Bus 20 kV

25 MVA

G
52 G

N2XSY

52 L

CB

Figure 1. Single line diagram of the power
transformer
Fig. 2 explains that the differential relay
configuration needs to be done to provide
protection to the power transformer and improve
relay performance. CT limits the input of CT1 and
CT2 for the protection area. On the 11 kV side, the
transformer is connected in delta, and the voltage
side of 20 kV is connected in a star.

1000 /
5A

25 MVA
11 kV / 20 kV
Ynd5

Y
87 T

2000 /
5A

Figure 2. Differential Relay Configuration
This transformer has a Ynd5 connection
group vector which will experience a phase shift of
150 degrees. To ensure the current phase angle
is shifted, the relationship of the current
transformer is designed differently.
Mismatch Error
The calculation of ratio the CT1 on the 11 kV
side is 2000 to 5, and the CT2 ratio on the 20 kV
side is 1000 to 5. Because of differences, there will
be an error in reading the difference in current,
voltage, and phase shift on the primary and
secondary side of the power transformer. To
calculate the previous mismatch error, calculate
the ideal CT value of one side of the power
transformer. For example for the voltage side of 20
kV (CT2), the ideal value is 2000/5 * (11 kV) / (20
kV). Then, the CT2 ratio on the voltage side is 20
kV when the maximum load is 1100 to 5. The ratio
of CT used on the side of 20 kV is 1000 to 5, and
ideally, CT for the side of 20 kV is 1100 to 5. The
mismatch error for differential relays is 1100 /
1000n = 1.1%.
Thus, the differential relay error in securing
the power transformer from the fault is 1.1%. This
mismatch error in the differential relay can be
corrected by increasing or decreasing the tap on
CT. Mismatch errors are expected to be as small
as possible, so that differential relay protection
works optimally to protect power transformers.
The condition of differential relay sensitivity in
operation mismatch error cannot be more than
5%. This requirement is determined for protection
to maintain the electric power system from fault
optimally.
The internal fault is the location of the fault
that is in the protection zone of the differential
relay, which is limited between CT1 and CT2. An
external fault is the location of the fault that is
outside the differential protection zone that is
outside CT1 and CT2. External fault one is
simulated to occur before CT1. External fault two
is simulated after CT2. This design aims to

A.A. Zakri et al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

101

SINERGI Vol. 23, No. 2, June 2019: 99-106

produce two external faults to improve differential
relay performance. The duration of the fault
simulation carried out in this simulation is 0.2
seconds. So the timing of the fault starts at (2/50)
* 0.2 = 0.04 seconds, then the fault occurs at 0.04
seconds. Fig. 3 shows the modeling of internal and
external fault that using differential relay simulated
via Matlab/Simulink
The setting membership functions using
Gbell by entering MF [3 3 3 3 3 3]. Fig. 4 shows

the ANFIS structure having six entries. These six
inputs are composed of current fault on CT 1 and
CT2, namely; IA, IB, IC. The input of the
membership function is each of the three
membership functions, so the total number of
membership functions is 18. For fuzzy rules that
run as many as 729 for Input Membership
Function (IMF) and Output Membership Function
(OMF) and output targets, that is 1.

Figure 3. Modeling of internal and external fault using differential relay simulated via Matlab/Simulink

Figure 4. Structure of power transformer system
RESULTS AND DISCUSSION
Results and Discussion should be a The
simulation of short circuit fault for the location of

102

an internal fault in the protection zone using
differential relays that is the area between CT1 and
CT2. The differential relay will provide information

A.A. Zakri at al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

on the trip signal to CB at 0.0437 seconds. By the
differential relay working principle that sends a trip
to CB command because the differential current is
greater than the current setting. It can be observed
that the secondary currents of CT1 and CT2 values
decrease to a time of 0.0566 seconds. In the
analysis of the value of the differential current for
each fault, the value of the differential current is
compared with the current value of the setting. The
internal fault of power transformers, differential
relays will operate if:
Iset  Id = differential relay operated
The external 1 and external 2 faults including fault
outside the differential relay protection area, the
differential relay does not operate if:
Iset  Id = differential relay not operated
Table 1 has shown that differential current
(Id) is the difference between secondary current
CT1 and CT2 in differential relays. Iset for an
internal fault is obtained 2.805 A. Iset for an

external fault is obtained 11.22 A. The differential
relay operates when internal fault and does not
operate on the external fault.
In this section, we will explain the simulation
of the results of fault outside the protected area for
a differential relay. The fault simulation is carried
out via software; this data will be used as an input
for training data to the ANFIS, as seen in Table 1.
Table 2 shows the simulation results of fault in 25
MVA power transformers that run with various
types of fault and fault locations. Then, the
simulation produces short circuit data, which is
used as an input data of ANFIS. The data of short
circuit fault is carried out for faults - internal, an
external one, and external two. The data is
generated as input for ANFIS and simulation of a
fault in the power transformer. The data has been
trained and tested through ANFIS; the more
iterations are carried out the smaller error rate.
Each group input analyzed in the ANFIS and, it
has been run with two membership functions
(membership of Gbell and triangle). These results
can improve the accuracy of the ANFIS technique
and provide excellent recommendations on power
transformers.

Table 1. Differential Current Setting
Type of
Fault
AG
BCG
BC
ABC
AG
BCG
BC
ABC
AG
BCG
BC
ABC

Fault
Location

Internal

External 1

External 2

Differential Current (Ampere)
A
14.29
2.23
2.211
12.18
0.29
1.22
0.47
1.53
0.33
0.62
0.62
1.075

B
3.92
9.61
5.26
13.56
1.135
1.38
1.54
1.08
0.87
0.945
0.97
0.67

C
2.37
3.93
6.08
3.8
1.47
0.98
0.94
1.62
1.16
0.484
0.685
0.98

Current
Setting
(Ampere)
2.805

11.22

CB
1
1
1
1
0
0
0
0
0
0
0
0

Table 2. Input Data of ANFIS in Short Circuit Simulation
Type of
Fault
AG
BCG
BC
ABC
AG
BCG
BC
ABC
AG
BCG
BC
ABC

Fault
Location

Internal

External 1

External 2

Current of CT1 (Ampere)
A
B
C
1753,788
1258,317
595,200
998,600
1377,215
1789,889
432,497
1398,640
1789,890
2636,700
2460,660
1789,889
2072,550
1524,915
668,710
1042,780
2361,103
2464,134
595,185
2107,809
2683,976
4072,786
3614,873
2939,647
1753,786
1258,317
595,197
998,600
1377,215
1789,889
998,497
1398,639
1789,889
2636,698
2460,660
1789,889

Current of CT2 (Ampere)
A
B
C
4143,470
395,265
734,939
555,582
2984,624
3526,392
327,332
2258,289
2575,818
4144,438
2985,005
3527,691
1138,776
327,340
843,791
780,413
1520,751
990,807
492,057
1520,751
1037,604
2243,474
1520,751
2003,175
946,626
359,280
734,954
555,598
906,463
831,645
327,350
1011,142
696,765
1490,458
1059,754
1260,205

A.A. Zakri et al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

CB
1
1
1
1
0
0
0
0
0
0
0
0

103

SINERGI Vol. 23, No. 2, June 2019: 99-106

ANFIS training data for internal, external
one and external two faults that are run using
Matlab is a combination of internal & external one
fault, internal & external two, and internal, external
one & internal two as shown in Table 2.
Initialization of initial values for techniques ANFIS
is run for data training. The results of the training
begin training data on iterations 30, 50, 80, 100,
130, 150, 180, 200, and 230. Observation of the
error values in each training iteration of the data is
carried out seven times. The error analysis carried
out at each stage with the lowest error value is

produced, at 200 iterations with an error value of
8.59*10-7.
Table 3 shows the accuracy of ANFIS can
be calculated using the Root Mean Square Error
(RMSE). The RMSE is the least square root error
value of a simulation, where, yi’ is the actual target
value, yi’ is the predicted output value, and n is the
amount of data. The RMSE equation is as follows:
∑𝑛 (𝑦 −𝑦𝑖 ′ )2

𝑅𝑀𝑆𝐸 = √ 𝑖=1 𝑛𝑖

(1)

Table 3. The result of ANFIS for various fault
Input
Data

1
2
3
4
5
6
7
8
9
10
11
12
RMSE

Type of
Fault

AG
BCG
BC
ABC
AG
BCG
BC
ABC
AG
BCG
BC
ABC

Fault
Location

Internal

External 1

External 2

Target

1
1
1
1
0
0
0
0
0
0
0
0

1
1
1
1
0
0
0
0
0

Internal &
External 1
Faults

Internal &
External 2
Faults

Output of
ANFIS

Output of
ANFIS
1
1
1
1
0
0
0
0
0

Internal,
External 1, dan
External 2
Faults
Output of ANFIS
1
1
1
1
0
0
0
0
0
0
0
0
0

Table 4 Comparison error of membership function between Gbell and triangle for Internal, External 1,
and External 2 faults
Iteration
30
50
80
100
130
150
180
200
230

Error (pu)
GBell
Triangle
0,0000018972
0,0000020678
0,0000016721
0,0000020653
0,0000014223
0,0000020614
0,0000012236
0,0000020576
0,0000010641
0,0000020576
0,00000094961 0,0000020576
0,00000086146 0,0000020576
0,00000085893 0,0000020576
0,00000085893 0,0000020576

The error percent is valued by calculating
the difference between the actual target and the
ANFIS output value and divided by the real value.
The ANFIS technique for securing power
transformers using differential relays results in
good forecasting of fault, cause the value of RMSE
is very small. This analysis can be suggested that
the lower the RMSE, the higher the success rate
of applying the ANFIS technique.
Table 4 shows the error value from the
ANFIS training via Matlab. The membership
function Gbell had the smallest error value is

104

8.6*10-7. The Gbell membership function training
obtained the smallest error value in training with
200 iterations.
When, the membership function triangle,
the smallest error value was obtained at 2.06*106
. The triangle membership function training got
the lowest error value at 100 iterations. Based on
training data that has been run for internal,
external 1 and external 2 faults, it can be observed
that the smallest error value generated is the
membership function Gbell.

A.A. Zakri at al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

p-ISSN: 1410-2331 e-ISSN: 2460-1217

CONCLUSION
The ANFIS technique has been applied to
predict fault with 25 MVA power transformers with
a short circuit current data. The calculated
resetting current of the differential relay is 2.805
Amperes to secure the internal fault of 11.22
Ampere and can work properly. The results of the
error concluded that the more the number of
iterations the smaller the errors obtained. For
internal and external fault 1 the target error has
been reached in the 180 iterations (Gbell). For
internal and external fault 2, the target error has
been achieved at 230 iterations (Gbell). The
internal, external 1, and external 2 faults the target
error has been reached in the 130 iterations
(Gbell). Therefore, from the simulation results via
ANFIS, it was concluded that Gbell membership
function is better than the triangle membership for
three short circuit fault scenarios. Finally, ANFIS
differential relays can predict for all types of faults
at excellent and accurate RMSE, which is zero.
This ANFIS method has also been run by using
other relays such as; distance relays that have
been explained in the literature review above, and
then this method is very accurate and efficient to
be applied in terms of the assessment of errors
tested.
ACKNOWLEDGMENT
The authors would like to thank the Institute
for Research and Community Services (LPPM),
Riau University.
REFERENCES
Azriyenni, A. & Dame, M.E. (2017). Pemodelan
Struktur Teknik Cerdas Untuk Sistem Proteksi
Daya Listrik. SINERGI, 21(1), 31-38.
http://doi.org/10.22441/sinergi.2017.1.005
Azriyenni, A. &, Mustafa, M.W. (2015). Application
of ANFIS for Distance Relay Protection in
Transmission Line. International Journal of
Electrical and Computer Engineering (IJECE),
5(6), 1311-1318.
Azriyenni, A., Mustafa, M.W., Sukma, D.Y., &
Dame, M.E. (2014). Backpropagation Neural
Network Modeling for Fault Location in
Transmission Line 150 kV. Indonesian Journal
of Electrical Engineering and Informatics
(IJEEI),
2(1),
1-12.
http://doi.org/10.11591/ijeei.v2i1.92
Flores, W.C., Mombello, E., Jardini, J.A., & Ratta,
G. (2008). Fuzzy risk index for power
transformer failures due to external shortcircuits. Electric Power Systems Research,
79(4),
539-549.
http://doi.org/10.1016/j.epsr.2008.06.021
Ge, C.H., Cui, H.Y., Huo, S.J., & Guo, W.C.
(2018). Improved DGA Methods of Power

Transformer Fault Diagnosis: A Review. In 7th
International Conference on Energy and
Environmental Protection (ICEEP 2018),
Advances in Engineering Research, 170. (pp.
1765-1768).
http://doi.org/10.2991/iceep18.2018.321
Kari, T. et al. (2018). An integrated method of
ANFIS and Dempster-Shafer theory for fault
diagnosis of power transformer. IEEE
Transaction on Dielectrics and Electrical
Insulation,
25(1),
360-371.
http://doi.org/10.1109/2018.006746
Kaur, A., Brar, Y.S., & Leena, G. (2019). Fault
detection in power transformers using random
neural networks. International Journal of
Electrical and Computer Engineering, 9(1), 7884. http://doi.org/10.11591/ijece.v9i1.pp78-84
Kim, Y., Park, T., Kim, S., Kwak, N., & Kweon, D.
(2019). Artificial Intelligent Fault Diagnostic
Method for Power Transformers using a New
Classification System of Faults. Journal of
Electrical Engineering & Technology, 14(2),
825-831. http://doi.org/10.1007/s42835-01900105-0
Lei, Y., He, Z., Zi, Y., & Hu, Q. (2007). Fault
diagnosis of rotating machinery based on
multiple ANFIS combination with GAs.
Mechanical Systems and Signal Processing,
21(5),
2280-2294.
http://doi.org/10.1016/j.ymssp.2006.11.003
Li, E., Wang, L. & Song, B. (2019). Fault Diagnosis
of Power Transformers with Membership
Degree. IEEE Access, 7, 28791-28798.
http://doi.org/10.1109/ACCESS.2019.2902299
Peng, G., Tang, S., Lin, Z., & Zhang, Y. (2017).
Applications of fuzzy multilayer support vector
machines in fault diagnosis and forecast of
electric power equipment. In the 2nd Advanced
Information Technology, Electronic and
Automation Control Conference (IAEAC),
Chongqing,
China.
(pp.457-461).
http://doi.org/10.1109/IAEAC.2017.8054056
Qu, L. & Zhou, H. (2015). The Multi-class SVM is
Applied in Transformer Fault Diagnosis. In 14th
International Symposium on Distributed
Computing and Applications for Business
Engineering and Science (DCABES), Guiyang,
China.
(pp.
477-480).
http://doi.org/10.1109/DCABES.2015.125
Salama, A.M., Abdel-Latif, K.M., Ismail, M. &
Mousa, S.M. (2018). A New Hybrid Protection
Algorithm for Protection of Power Transformer
Based on Discrete Wavelet Transform and
ANFIS Inference System. International Journal
of Emerging Electric Power System, 19(3), 18. http://doi.org/10.1515/ijeeps-2017-0248
Soloot, A.H., Khanali, M., Hoidalen, H.K., &
Jayaram, S. (2015). A method to detect the

A.A. Zakri et al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …

105

SINERGI Vol. 23, No. 2, June 2019: 99-106

position of internal short circuit in power
transformers. In International Conference on
Power Systems Transients (IPST2015),
Cavtat, Croatia. (pp.1-6).
Zhangjun, G. Nuo, G., Lei, W., & Zhaofei, L.
(2012). Power System Fault Diagnosis Based
on Power Grid. In 11th IET International

106

Conference on Developments in
Systems
Protection
(DPSP
Birmingham,
UK.
(pp.
http://doi.org/10.1049/cp.2012.0140

Power
2012),
1-4)

A.A. Zakri at al., Assess the Risk Level of Power Transformer Due Short-Circuit Faults …