Electrical Power and Energy Systems 64 (2015) 35–40

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

A new wavelet based fault detection, classification and location in

transmission lines
Abdul Gafoor Shaik a,⇑, Ramana Rao V. Pulipaka b
a
Centre for Information and Communication Technology, Indian Institute of Technology, Jodhpur, India
b
Department of Electrical and Electronics Engineering, ANU College of Engineering and Technology, Guntur, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper deals with the application of wavelet transforms for the detection, classification and location
Received 15 November 2007 of faults on transmission lines. A Global Positioning System clock is used to synchronize sampling of
Received in revised form 15 June 2014 voltage and current signals at both the ends of the transmission line. The detail coefficients of current
Accepted 20 June 2014
signals of both the ends are utilized to calculate fault indices. These fault indices are compared with
Available online 31 July 2014
threshold values to detect and classify the faults. Artificial Neural Networks are employed to locate the
fault, which make use of approximate decompositions of the voltages and currents of local end. The
Keywords:
proposed algorithm is tested successfully for different locations and types of faults.
Fault Index
Multi resolution analysis
Ó 2014 Elsevier Ltd. All rights reserved.
Transmission line protection

Introduction possible to develop communication-aided high-speed digital

The performance of a power system is affected by faults on
transmission lines, which results in interruption of power flow.
Quick detection of faults and accurate estimation of fault location,
help in faster maintenance and restoration of supply resulting in
improved economy and reliability of power supply. Wavelet
Transform (WT) is an effective tool in analyzing transient voltage
and current signals associated with faults both in frequency and
time domain.
Kim, et al. [1] have used Wavelet Transforms to detect the high
impedance arcing faults. Jiang, et al. [2] have used Haar Wavelet to
detect dc component for identifying the faulty phases. Distance
protection schemes using WT based phasor estimation are
reported in [3,4]. Wavelet multi resolution analysis for location
of faults on transmission lines is reported in [5]. A powerful high
speed traveling wave-based protection technique using wavelet/
PCA analysis is proposed in [6]. Artificial Neural Networks (ANN)
with their excellent pattern recognition capabilities, are used to
estimate the fault location [7,8]. Silva et al. [9] have Wavelet
Transform and ANN for detection and classification of faults in
power transmission lines. Computer based relaying system for
power system has been reported in [10]. Roy et al. [11] presented
software reliability allocation of digital relay for transmission line
protection using a combined system hierarchy and fault tree
approach. Due to fast developing communication techniques, it is
protection scheme, which suits the EHV transmission. Better
performance can be achieved using two terminal synchronized
sampling of signals. Global Position System (GPS) based algorithms
with better performance and accuracy have been proposed in
[12,13].
Wavelet analysis
Wavelet Transform (WT) is an efficient means of analyzing
transient currents and voltages. Unlike DFT, WT not only analyzes
the signal in frequency bands but also provides non-uniform
division of frequency domain, i.e. WT uses short window at high
frequencies and long window at low frequencies. This helps to
analyze the signal in both frequency and time domains effectively.
A set of basis functions called Wavelets, are used to decompose the
signal in various frequency bands, which are obtained from a
mother wavelet by dilation and translation. Hence the amplitude
and incidence of each frequency can be found precisely.
Wavelet Transform is defined as a sequence of a function
{h(n)}(low pass filter) and {g(n)}(high pass filter). The scaling func-
tion u(t) and wavelet w(t) are defined by the following equations
pffiffiffi X
uðtÞ ¼ 2 hðnÞuð2t  nÞ
pffiffiffi X
wðtÞ ¼ 2 gðnÞuð2t  nÞ

⇑ Corresponding author. Tel.: +91 291 244 9039; mobile: +91 7742725795. Where gðnÞ ¼ ð1Þn hð1  nÞ
E-mail address: saadgafoor@iitj.ac.in (A.G. Shaik).

http://dx.doi.org/10.1016/j.ijepes.2014.06.065
0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.
36 A.G. Shaik, Ramana Rao V. Pulipaka / Electrical Power and Energy Systems 64 (2015) 35–40

GPS Satellite

BUS 2
BUS 1
Transmission Line

V&I V&I

WT based Detector WT based Detector

and Classifier and Classifier

Communication Channel
ANN based Locator ANN based Locator

Fig. 1. System considered along with the proposed scheme.

4000
4000 Ia2
Ia1 Ib2
Ib1 Ic2

Amperes
Amperes

Ic1
0 0

-4000 -4000
0 10 20 30 40 50 60 0.00 10.00 20.00 30.00 40.00 50.00 60.00

Time in ms Time in ms
(a) Three Phase currents at Bus-1 of AG fault at 40% (b) Three Phase currents at Bus-2 for AG fault at 40%
of the line of the line

3000 3000
D1_Coeff. of Ia1

D1_Coeff. of Ia2

0 0

-3000 -3000
0.0 10.0 20.0 30.0 40.0 50.0 60.0 0.0 10.0 20.0 30.0 40.0 50.0 60.0
Time in ms Time in ms
(c) Local D1-coefficients at Bus-1of Phase-A (d) Local D1-coefficients at Bus-2 of Phase-A

Fig. 2. Fault current transient and their detail coefficients.

(a) Variation of Effective Coefficients of Phase-A (b) Variation of Fault Index for Phase-A

4000
Ia
Ib
Fault Index If1

Ic
Threshold(Th1)

0
0 10 20 30 40 50 60
Time in ms
(c) Variation of fault indices for AG fault
Fig. 3. Computation of Fault Index and detection of fault phases.
 A.G. Shaik, Ramana Rao V. Pulipaka / Electrical Power and Energy Systems 64 (2015) 35–40 37

8000 of the transmission line are 500 kV, 500 MW and 60 Hz [3]. The
Ia three phase currents and voltages of both the terminals are sam-
Ib
pled at frequency of 1920 Hz. Synchronized sampling is achieved
Fault Index (If1)

Ic
with the help of a GPS clock. The detail D1 coefficients used for
detection and classification of the type of fault are transmitted
through the fiber-optic communication channel to the remote end.
The three phase currents of the local terminal are analyzed with
Bior.2.2 mother wavelet to obtain the detail coefficients (D1L) over
0 a moving window of half cycle length. The selection of mother
0 10 20 30 40 50 60
wavelet was based on hit and trial method. These D1L coefficients
Time in ms are then transmitted to the remote end. The detail coefficients
(a) Variation of fault indices for ABG fault received from the remote bus (D1R) are added to the local detail
coefficients (D1L) to obtain effective D1 coefficients (D1E). The
4000
Fault Index (If1) of each phase is then calculated as If1 = R |D1E|.
ABG Fault Fig. 2(a) and (b) shows the waveforms of three phase currents
Fault Index (If2)

AB Fault for AG fault at Local and Remote terminals. Fig. 2(c) and (d) illus-
Threshold(Th2)
2000 trates the variation of D1 coefficients of Phase-A at Local and
Remote terminals for the same fault. The variation of effective D1
coefficients of Phase-A at both the terminals is shown in Fig. 3(a)
0 and the variation of Fault Index of Phase-A is shown in Fig. 3(b).
0 10 20 30 40 50 60 The Fault Index of three phase currents for an AG fault at 40% of
Time in ms the transmission line is shown in Fig. 3(c). It is observed that the
(b) Variation of Ground Fault Index for AB and Fault Index of faulty phase is large compared to those of healthy
ABG faults phases. Fig. 4(a) shows the variation of Fault Index If1 incase of
ABG fault. Thus the number of faulty phases is determined by com-
Fig. 4. Discrimination of LL and LLG faults. paring the Fault Index (If1) with a Fault Threshold (Th1). It is evident
that by using Fault Index, If1 alone cannot discriminate between LL
A sequence of {h(n)} defines a Wavelet Transform. There are and LLG faults. Hence a Ground Fault Index (If2) given by If2 = 1/3 R
many types of wavelets such as Haar, Daubachies, and Symlet If1 is further computed and compared with a ground fault thresh-
etc. The selection of mother wavelet is based on the type of old (Th2) to discriminate between LL and LLG faults. This is illus-
application. In the following section a novel method of detection trated in Fig. 4.
and classification of faults using multi resolution analysis of the The proposed algorithm has been tested for all types of faults,
transient currents associated with the fault is discussed. considering variations in fault locations and fault incidence angles
(h) in the range 0–180°. Fig. 5 illustrates the variation of Fault Index
Detection and classification of faults for solid faults with respect to location for an incidence angle of 0°.
Arcing high impedance faults which are difficult to detect has also
Fig. 1 shows the single line diagram of the system considered been simulated. The proposed scheme is proved to be successful in
along with the various blocks of the proposed scheme. The ratings detecting these faults also as depicted in Fig. 6. The proposed

8000 Ia Ib Ic Threshold
Ia Ib Ic Threshold 10000

6000 8000
Fault Index

Fault Index

6000
4000
4000

2000 2000

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance in KM Distance in KM
(a) For AG Fault (b) For AB Fault

12000 Ia Ib Ic Threshold
12000 Ia Ib Ic Threshold
Fault Index

8000
Fault Index

8000

4000 4000

0 0
0 50 100 15 0 20 0 25 0 30 0 0 50 100 150 200 250 300
Distance in KM Distance in KM
(c) For ABG Fault (d) For ABCG Fault
Fig. 5. Variations in Fault Indexes of three phase currents for solid faults at an incidence angle of 60o.
38 A.G. Shaik, Ramana Rao V. Pulipaka / Electrical Power and Energy Systems 64 (2015) 35–40

8000 8000
Ia Ib Ic Threshold Ia Ib Ic Threshold

6000 6000

Fault Index

Fault Index
4000 4000

2000 2000

0 0
0 100 200 300 0 50 100 150 200 250 300
Distance in KM Distance in KM
(a) For AG Arcing Fault (b) For AB Arcing Fault

Ia Ib Ic Threshold
8000
Ia Ib Ic Threshold

8000
Fault Index

Fault Index
4000
4000

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance in KM Distance in KM
(c) For ABG Arcing Fault (d) For ABCG Arcing Fault
Fig. 6. Variations in Fault Indexes of three phase currents for arcing faults at an incidence angle of 90o.

6000

Ia Ib Ic Threshold 6000
Ia Ib Ic Threshold

4000
Fault Index

Fault Index

4000

2000
2000

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance in KM Distance in KM
(a) For AG Fault (b) For AB Fault

Ia Ib Ia Ib Ic Threshold
8000 Ic Threshold
8000
Fault Index

Fault Index

4000 4000

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Distance in KM Distance in KM
(c) For ABG Fault (d) For ABCG Fault
Fig. 7. Variations in Fault Indexes of three phase currents for different faults with Zf = 5 X.

Table 1
ANN structure for Ground Fault Location.

Layer Number of neurons Transfer function Input

Input 24 Log-Sigmoid A3 decomposition of faulty phase voltage and current & zero sequence current
Hidden-1 12 Log-Sigmoid
Hidden-2 4 Log-Sigmoid
Output 1 Log-Sigmoid
 A.G. Shaik, Ramana Rao V. Pulipaka / Electrical Power and Energy Systems 64 (2015) 35–40 39

Table 2
ANN structure for Phase Fault Location.

Layer Number of neurons Transfer function Input

Input 16 Log-Sigmoid A3 decompositions of difference of faulty phase voltages and currents
Hidden-1 8 Log-Sigmoid
Hidden-2 4 Log-Sigmoid
Output 1 Log-Sigmoid

Fig. 8. Flow Chart for Proposed Scheme.

Table 3
scheme has also been tested successfully for finite impedance Maximum Value of Error.
faults. The same is demonstrated in Fig. 7.
Type of faults Average error (%) Max error (%)
LG faults 0.1 1.84
Estimation of fault location LLG faults 0.16 1.89
LL faults 0.01 1.88
Subsequent to detection and classification of fault, estimation of 3-Phase faults 0.004 2.3
fault location is carried out using ANNs. For this purpose the three
phase voltages and currents of the local terminal are decomposed
with Bior4.4 mother wavelet and the third level approximate ANNs. The input vector for ANN used to locate ground faults, com-
decompositions A3 (120–0 Hz), over a half cycle window are prises of the A3 decompositions of faulty phase voltage and current
obtained. This is done concurrently with the detail decomposition and zero sequence current. For an LLG fault, A3 decomposition of
used for fault detection and classification to speed up estimation of any one faulty phase can be used to estimate fault location. The
fault location. In case the fault is detected, the A3 decompositions details of the ANN architecture used are depicted in Table 1.
are obtained for another half cycle. The input vector for ANN used to locate Phase faults includes A3
Two separate ANNs, one for ground faults and another for phase decompositions of difference of faulty phase voltages and currents.
faults, are employed for estimation of fault location. The A3 decom- In case of 3-phase fault the input vector is obtained from any two
positions obtained over a complete cycle are used as inputs to these phase quantities. The details of the ANN architecture used are
40 A.G. Shaik, Ramana Rao V. Pulipaka / Electrical Power and Energy Systems 64 (2015) 35–40
depicted in Table 2. The Flow Chart of the proposed scheme is
illustrated in Fig. 8.
The two ANNs detailed above, are trained with the data
obtained from all type of fault simulations at various locations
and with variation in incidence angle. The performance of both
the ANNs are tested for various cases. Table 3 represents the
performance of trained ANNs in terms of maximum percentage
error for different types of faults.
Conclusions
WT based multi resolution analysis approach can be success-
fully applied for effective detection and classification and location
of faults in transmission lines. Fault detection and classification can
be accomplished within half cycle using detail coefficients of cur-
rents at both the ends. Fault location can be estimated with a fair
degree of accuracy from the approximate decomposition of phase
voltages and currents of local end using ANNs. The proposed pro-
tection scheme is found to be fast, reliable and accurate for various
types and nature of faults on transmission lines, at different loca-
tions. It has been proved that the variation in fault incidence angle
has no effect on proposed scheme.
References
[1] Kim Chul-Hwan, Kim Hyun, Ko Young-Hun, Byun Sung-Hyun, Aggarwal Raj K,
Johns Allan T. A novel fault-detection technique of high-impedance arcing
faults in transmission lines using the wavelet transform. IEEE Trans Power
Delivery 2002;17(4):921–9.
[2] Joe-Air Jiang, Ping-Lin Fan, Ching-Shan Chen, Chi-Shan Yu, Jin-Yi Sheu. A Fault
Detection and Faulted Phase Selection Approach for Transmission Lines with
Haar Wavelet Transform. In: Transmission and distribution conference and
exposition 2003, vol. 3, IEEE PES, 7-12, September 2003. p. 285–9.
[3] Liang Feng, Jeyasura B. Transmission line distance protection using wavelet
transform algorithm. IEEE Trans Power Delivery 2004;19(2):545–53.
[4] Osman AH, Malik OP. Transmission line distance protection based on wavelet
transform. IEEE Trans Power Delivery 2004;19(2):515–23.
[5] Chanda D, Kishore NK, Sinha AK. A Wavelet multiresolution analysis for
location of faults on transmission lines. Electrical Power Syst Res
2003;25:59–69.
[6] Jafarian Peyman, Sanaye-Pasand Majid. A traveling – wave-based protection
technique using Wavelet/PCA analysis. IEEE Trans Power Delivery 2010;25(2):
588–99.
[7] Martin Fransisco, Aguado Jose A. Wavelet-based ANN approach for
transmission line protection. IEEE Trans Power Delivery 2003;18(4):1572–4.
[8] Tawfik MM, Marcos MM. ANN-based techniques for estimating fault location
on transmission lines using prony method. IEEE Trans Power Delivery
2001;16(2):219–24.
[9] Silva KM, Souza BA, Brito NSD. Fault detection and classification in
transmission lines based on wavelet transform and ANN. IEEE Trans Power
Delivery 2006;21(4):2058–63.
[10] Phadke Arun G, Throp James S. Computer relaying for power systems. Research
Study Press Ltd.; 1994.
[11] Roy DS, Mohanta DK, Panda AK. Software reliability allocation of digital relay
for transmission line protection using a combined system hierarchy and fault
tree approach. IET Software 2008;2(5):437–45.
[12] Li HY, Potts S. A new type of differential feeder protection relay using the
global positioning system for data synchronization. IEEE Trans Power Delivery
1997;12(3):1090–9.
[13] Jiang Joe-Air, Fan Ping-Lin, Chen Ching-Shan, Liu Chin-Wen. A new protection
scheme for fault detection, direction discrimination, classification, and
location in transmission lines. IEEE Trans Power Delivery 2003;18(1):34–42.