Chapters

CHAPTER 1
=================================================================
INTRODUCTION
=================================================================
1.1 Multiterminal transmission line
The power system consists of thousands of high voltage transmission lines transmitting
electrical power between generators and load centers which represent the foundation of the
power system. The majority of transmission line construction is of overhead type and therefore,
is easily susceptible to various transient and permanent faults. These faults can lead to damage
of the line itself and can cause power system instability. It is of the utmost importance that
protective relay systems are capable of clearing all faults within the designed operating time,
and have a high degree of dependability and security.
Typically, there are three types of line configurations used within the industry. These
line configurations include radial configuration that are (a) one-terminal, (b) two-terminal, and
(c) multi-terminal of which three-terminal is possibly the most prominent multi-terminal type. It
should be noted that "terminals" in this context, refers to source terminals and not-tapped
transformer terminals or stations. The two-terminal line configuration is the most dominant type
followed by radial, and the three-terminal lines are the exceptions.
Three-terminal and other multi-terminal line construction projects are generally a trade-
off of planning economics and protection complexities, and can lead to compromises in
reliability. Two-terminal lines with long tap(s) supplying remote load from the main line may
display many of the same protection and loadability issues as three-terminal lines. These types
of configurations and those with multiple tapped transformer stations (low voltage tie breaker
closed) are beyond the scope of this discussion. However, it should be noted that some of the
same types of complexities may be experienced with these types of configurations as three-
terminal lines.
Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 1

The complexity of protecting these line configurations increases from the relatively
simple radial, to the more difficult two-terminal, and to the still more difficult three-terminal.
Relaying three-terminal lines has been and continues to be a challenge for protection engineers.
There are a number of factors that influence the decision to configure a transmission line
with three terminals, such as economics, constrained lead time, regulatory approvals, right-of-
way availability, line overloads, and system performance requirements.
• There is an economic benefit in the construction of three terminals because it avoids the
expense of all or a portion of a substation and typically reduces the transmission line miles.
• Use of three-terminal lines may be more expeditious in addressing system needs.
• Right-of-way may be limited or not obtainable for new lines and stations.
• Regulatory approvals may be problematic. There may be opposition to the construction of new
facilities and the construction of a three-terminal line may reduce the overall project impact.
• Three-terminal line configuration may mitigate the possibility of transmission line overloads
due to single contingency events. However, this is very dependent on system topology.
1.2 Faults in Overhead Transmission Lines

To spread power from generating stations to remote load centers, transmission lines are
used. Due to lightening, miss-operation, overload, short circuits, human errors, faulty
equipments and ageing, faults may occur on these lines. When fault occurs, the faulted phase
voltage decreases and huge currents will flow which can burn out the components if not
interrupted quickly.
1.2.1 Nature and Causes of Faults

Either insulation failure or failures of conducting path are the major causes for the
occurrence of faults. In addition to this, faults are also caused due to over voltages which are
occurring due to switching surges and lightening. Falling of conducting objects on overhead
lines, encounter of flying birds, tree branches, direct lightening strokes, ice loading, creepers,
storms etc. are the other reasons which can cause faults in overhead lines. Moisture in the soil,

heat of earth, ageing of cables may lead to the solid insulation failure in cables, transformers and
generators [1].
Types of faults:
1. Symmetrical faults
2. Unsymmetrical faults
Table 1.1. Types of faults [1]
Types of faults Symbol % Occurrences Severity
Line to Ground L-G 75 – 80% Very less severe
Line to Line L–L 10 – 15% Less Severe
Double Line to Ground L-L-G 5 – 10% Severe
Three Phase 3-ø 2 -5% Very Severe
1.2.2 Effects of Faults

Following are the ill effects caused by a fault in a power system.
Severe short circuit current may occur in the system due to fault which may prove fatal to
the several equipments of the power system and lead to the overheating of the system. Heavy
current is also the reason behind the setting up of very high mechanical stresses.
Failure of industrial loads, due to drop in the voltage of healthy feeders.
Heating of rotating machines may occur due to unbalancing of currents and supply voltages
arising due to short circuit.
Loss in system stability.
Continuity of power supply is adversely affected.

1.3 Protection of conventional transmission line.
Just as transmission lines vary widely in their characteristics and configurations, so too
do their protection schemes. Several fundamental factors influence the choice of protection
schemes applied to a given line.
• Type of line: overhead, cable, line length, single line, parallel line, radial, two-ended, three-
ended, etc.
• Line function and importance effect on service continuity and timing requirement for isolation
from the system.
• Coordination and compatibility with associated lines and systems.
Most high-voltage transmission lines are protected by distance relays. Compared to

over-current relays, distance relays are inherently directional, less susceptible to source
impedance variations, and have higher loadability limits.
The three most commonly used communication assisted distance protection schemes in
the industry are Direct Under-reaching Transfer Trip (DUTT), Permissive Overreaching
Transfer Trip (POTT), and Directional Comparison Blocking (DCB).
DUTT has the advantages of minimal susceptibility to power system swings. DUTT has
the disadvantage of dependency on communication channels for faults external to overlapping
coverage regions.
The POTT scheme has the advantage of being more secure, as it requires permission
from the remote relays to trip, and it can provide higher-speed tripping. It has the disadvantage
of being dependent upon the communication channel time for all line faults.
The DCB is the most trip dependable, because its operation is not dependent on the
communication channel or operation of the remote relays. It is the least secure in that a loss of
communication can result in line trips for faults not on the line.

Generally, a complimentary use of high-speed schemes is used for the protection of most 200
kV and above transmission lines.
1.4 Difficulty in protection of multiterminal transmission line

The differential relaying scheme using a communication link between the TL ends could
provide a secure protection for multiterminal TLs [2]–[4]. However, the reliability of such a
protection scheme depends upon the reliability of the communication link. Moreover, the
measurement infrastructure of the tapping lines could be poor or there could be no
communication channel for sending measurements from the far end of the tapping lines [5].
Therefore, a protection algorithm, which is based only on the local information obtained at the
relay bus, is greatly useful for the protection of multiterminal TLs. Even when a quite reliable
communication-based technique is employed for the protection of a TL, a second relay, which
makes the decision only based on the local information, would be helpful to increase the
reliability of the protection scheme.
The conventional TL protection schemes (e.g., distance relays) are based on the
fundamental frequency components of the fault signals. It is well known that the fundamental
frequency components of the local signals do not provide the sufficient information required for
discriminating between the internal and external faults for a multiterminal TL. This is due to the
fact that the infeed or outfeed currents contributed from the tap points could affect the fault-loop
impedance estimated by a distance relay. As a result, the distance relay may under-reach or
overreach depending on the transmission system configuration and parameters [6].
The fault-induced high-frequency (HF) transients contain extensive information about

the fault by which the fault direction and location could be identified in a few milliseconds [7]–
[10]. The traveling-wave positional protection technique uses polarities and time intervals
between arriving waves at the relay bus to estimate the fault location [9], [10]. Nevertheless, this
technique cannot be easily employed for protection of multiterminal lines due to the reflections
of traveling waves at the tap points.

Recently, some non-communication-based techniques for the protection of two-terminal lines
have been introduced which use the HF components of the fault signals measured at the relay
bus to discriminate between the internal and external faults [11]–[15]. These algorithms rely on
the fact that the substation equipment affects the characteristics of the HF transients generated
by the faults. The bus stray capacitances have a low-pass filtering effect on the fault-generated
transients. As a result, in the case of external faults beyond the TL ends, the HF components
involved in the fault signals are attenuated when passing through the remote substation bus,
whereas in the case of internal faults, the fault signals measured at the TL ends contain a higher
amount of HF components. When the bus equivalent capacitance is sufficiently large, these
algorithms could provide a satisfactory response. Nevertheless, for small values of the
capacitance, discrimination between the internal and external faults might become difficult due
to the small difference between the characteristic features of these two cases. In addition,
variation of the power system parameters (e.g., due to the disconnection of a transformer or a
transmission line) could affect the performance of these algorithms.
1.5 Outline of dissertation

Six chapters are included in this dissertation. Overview and all information included into
these chapters are specified in the following:
Chapter-1: It includes the introduction, faults in overhead transmission lines, aim of fault
classification and it also includes the contribution of this dissertation and organization of
dissertation.
Chapter-2: It explains the literature review pertaining to protection of multiterminal

transmission line using wavelet transform and Neural Network. The literature on classification
of faults by multi-resolution wavelet analysis and neural network is discussed in this chapter.
Chapter-3: It presents the concept of protection of multiterminal transmission line and fault
classification technique for multiterminal transmission line. Also presents the introduction of
wavelet transform. Discrete Wavelet Transform, wavelet multi-resolution analysis, and working
of wavelet transform.

Chapter-4: It presents the MATLAB/SIMULINK based simulation models and simulation
results are discussed.
Chapter-5: It presents the conclusions of the work presented in dissertation and future scope of
this work.

CHAPTER 2
=================================================================
LITERATURE REVIEW
=================================================================
2.1 Use of wavelet transforms analysis for fault classification
In order to detach only faulted line, it is crucial to differentiate the faults zone precisely
and indicate exact fault type with the aid of one end data only [16]. Transient current waves
generated by faults contains distinct frequency bands and to capture two bands of frequencies
from the transient current signal discrete wavelet transform db1 as a mother wavelet is used.
Fault zone is determined by using the frequencies of these two bands. The mother wavelet Haar
is used to select faulted phase. Faulted phase was classified by computing the average value of
the coefficients of each current wave. A model signal is obtained using db6 as mother wavelet.
The decision regarding fault to be external or internal was taken by determining the ratio of two
energies for the modal signal. A new scheme for the solution of the parallel transmission line
protection problems which depends on the six phase line currents and three phase line voltages
of the two parallel circuit lines at both ends is proposed [17]. Fault analysis is done by wavelet
transform. And internal faults on double circuit line are recognized by comparing current
phasors magnitudes of corresponding phases on each line. It is shown that at different loading
conditions each type of fault can be properly recognized.
For the power to be efficiently distributed to different locations, it is necessary to

accurately detect and classify the different faults [18]. Active tripping of circuit breaker ensures
the accurate protection of transmission line and circuit breakers tripping action depends on the
current and voltages waveforms during the fault. For analysis of waveforms of current during
fault, Discrete Wavelet Transform (DWT) is used. The evaluation of discrete wavelet analysis
for identification and classification of faults on a transmission line network is done. According
to energy level percentage, classification of faults has been done.
The use of wavelet transform for protecting the series compensated line by Current
Differential pilot Relay (CDPR) is discussed [19]. Simulation results are obtained using

MATLAB and analysis is done using db4 as mother wavelet. Fault classification is done by
detecting different types of faults using wavelet based approach.
Probability based technique of Bayesian linear discrimination can also be used to

differentiate between the different types of faults [20]. An adaptive wavelet algorithm (AWA) is
used to generate the wavelets using probability based method of Bayesian linear discrimination.
It is shown that adaptive wavelets can be used in the transmission lines of high speed protection
system as analysis filters.
Power need to be transmitted from the power station to the load centers located far away
[21]. So, the possibility of fault in the transmission lines is considerable. Here, comes the use of
signal processing in the digital distance protection. Fourier transform and wavelet transforms are
used for locating faults. Simulation is done with MATLAB/SIMULINK. Simulation result
shows that wavelet method is more robust tool to locate the faults in the transmission lines.
Further it is showed that both wavelet transform and Fourier transform methods can be used to
find the characteristics of disrupt signals irrespective of the noise levels present.
The discrete wavelet analysis has been used for the protection of high speed EHV
transmission line [22]. An algorithm for fault detection and classification based on discrete
wavelet analysis has been presented. By comparing different wavelet coefficients of all three
phase signals, type of fault is identified. And simulation is done using ATP-EMTP and
MATLAB Wavelet toolbox. Such an algorithm is presented that is not dependent not only on
fault location but also on fault inception angle and fault impedance. The algorithm is suitable,
strong and quick and this is very prolific for EHV transmission line protection.
For the fault classification and boundary protection of series-compensated transmission

lines, a new technique is proposed [23]. Different frequency bands of the wave of transient fault
current are detected in order to have the suitable boundary protection. In order to amass the two
frequency bands of transient fault current signal, db4 as a mother wavelet is used. Whether the
fault is internal or external, it is determined by calculating the spectral energies of two bands of
frequencies. Faulted phases are classified by calculating the average value of the wavelet

coefficients of every current wave. A simple modal signal is obtained using the fault current
values of three phases for all types of faults. Analysis of modal signal is done by using db4 as a
mother wavelet, then detail 1 and detail 6 coefficients are calculated of the modal signal. To
distinguish whether the fault is internal or external, the ratio of spectral energy is obtained and
average values of d6 coefficients of three phase currents and ground current are obtained which
is further applied to classify the type of fault.
A new approach for protection of TEED transmission lines and use wavelet transforms
for accurate detection, classification and to locate faults in TEED transmission lines is presented
[24]. The three phase currents at each terminal are decomposed at single level by using Bior2.2
as mother wavelet to obtain D1 detail coefficients. Then these detail coefficients at the three
ends of TEED transmission line are added to obtain the resultant detail coefficients which are
further compared with threshold values to detect and classify distinct types of faults in TEED
transmission lines. The use of Fuzzy logic and neural network for protection of double circuit
series compensated transmission line is described [25]. Fuzzy logic and Neural Network are
used for accurate decision making and to estimate the actual power system condition
respectively, which increases the selectivity of protection system that further improves the
reliability of power system. Effect of mutual zero-sequence coupling, series compensation and
fault resistance is studied. The effect of the mutual coupling of parallel circuit and of series
capacitor impedance on the relays accuracy depends on actual condition of power system.
Ultimately, it is shown that with change in power system condition, the relay sensitivity is
reduced to almost zero with decision making system of Fuzzy logic.
An impedance based calculation method to locate fault on transmission line is of

immense importance [26]. Results get changed by changing line parameters. This is showed by
carrying out the analysis of the two widely used methods in real faults. Most commonly used
fault location methods are compared and their relative disadvantages and advantages are
described. In this manner it becomes simple for the users to go for the most accurate method.
Ultimately, it is shown that the two end methods are stronger than the one end methods as
sensitivity to errors in two end methods is less as compared to one end methods.

Discrete Wavelet Transform (DWT) is used to extract the concealed factors from the fault
signals by decomposition at distinctive levels [27]. Daubechies db6 wavelet is used for
decomposition at single level. For ground faults a threshold is calculated to classify and detect
the faulted phase. The fault location is determined by getting the local fault information, remote
fault information and the length of transmission line. The system is considered with negligible
fault resistance.
The protection scheme of double circuit transmission line based on artificial neural
network (ANN) has been proposed [28]. Three stages are involved in this scheme to detect and
classify different types of faults. Data from one end of the double circuit transmission line has
been utilized to calculate the wavelet coefficients. The primary protection is provided to entire
transmission line by using one end data only. For forward and backward adjacent transmission
line, back up protection is provided. This technique improves the first zone reach setting up to
99% of the length of line for protection of transmission line.
A scheme based on wavelet transform for fault classification is proposed [29]. Currents
samples from the three lines are used to calculate dW Iabc. For different fault inception angles,
different fault locations, and different fault distances and for different fault parameters
simulation is done by EMTP software. It is shown that magnitude of wavelet transform is
valuable to set threshold to discriminate between different types of faults hence to classify the
faults.
An algorithm based on discrete wavelet transform is developed with C programming

[30]. A 500kv, 200km single line is simulated by using MATLAB. It is shown that as the fault
resistance increases, the percentage error increases rapidly. And when, the reactance of the
circuit is considered to calculate the distance to fault, then the percentage error in the
measurement of distance increases with increase in fault resistance.
2.2 Use of Neural Network for fault classification

Various applications of Neural Networks have been used in the past to improve the
protection scheme in the transmission lines [31]. They have been used in fault classification,

fault section estimation, adaptive relaying and fault diagnosis. Many of these methods are based
on back propagation, Radial basis function and Finite Impulse response neural networks. A few
of these approaches are discussed here.
A typical Back Propagation Neural Network is a non-linear regression technique which

attempts to minimize the global error. Its training includes both forward and backward
propagation, with the desired output used to generate the error values for back propagation to
iteratively improve the output. The back propagation neural network must have at least one
input layer and one output layer. The hidden layers are optional.
The Back propagation neural network consists of three layers: an input layer with three
neurons, a hidden layer with two neurons and an output layer with two neurons. In the figure,
we see that the output of a neuron in a layer goes to all neurons in the following layer and each
neuron has its own weights. Initially the weights of the input layer are assumed to be 1 for each
input. The output of the back propagation neural network is reached by applying input values to
the input layer, passing the output of each neuron to the following layer as input.
The number of neurons in the input layer depends on the number of possible inputs we
have, while the number of neurons in the output layer depends on the number of desired outputs.
The number of hidden layers and how many neurons in each hidden layer cannot be well
defined in advance, and could change per network configuration and type of data. In general the
addition of a hidden layer could allow the network to learn more complex patterns, but at the
same time decreases its performance [32]. Ideally, we could start a network configuration using
a single hidden layer, and add more hidden layers if we notice that the network is not learning as
well as we like.
The Back propagation training algorithm [33] could be summarized as follows: The
Input data sample is first presented to the network and then the network’s output taken from the
output layer is compared with the desired output and the error is calculated in each output
neuron. And now for each neuron, a scaling factor called the local error is calculated which
indicates how much higher or lower the output must be adjusted to match the desired output.

The weights are modified to lower this local error. This process gets repeated until the
error falls within the acceptable value (pre-defined threshold) which would indicate that the
neural network has been trained successfully. On the other side, if the maximum number of
iterations is reached, then it indicates that the training was not successful.
2.3 Multiterminal transmission line protection techniques
A multiterminal TL is a line which ends at more than one terminal with sources behind
them and taps at the other locations of the line. Such a configuration poses some difficulties
from the protection point of view due to the possibility of multiple sources feeding short-circuit
faults [34].
The differential relaying scheme using a communication link between the TL ends could
provide a secure protection for multiterminal transmission lines (TLs) [2]–[4]. However, the
reliability of such a protection scheme depends upon the reliability of the communication link.
Moreover, the measurement infrastructure of the tapping lines could be poor or there could be
no communication channel for sending measurements from the far end of the tapping lines [4].
Therefore, a protection algorithm, which is based only on the local information obtained at the
relay bus, is greatly useful for the protection of multiterminal TLs. Even when a quite reliable
communication-based technique is employed for the protection of a TL, a second relay, which
makes the decision only based on the local information, would be helpful to increase the
reliability of the protection scheme.
The conventional TL protection schemes (e.g., distance relays) are based on the
fundamental frequency components of the fault signals. It is well known that the fundamental
frequency components of the local signals do not provide the sufficient information required for
discriminating between the internal and external faults for a multiterminal TL. This is due to the
fact that the infeed or outfeed currents contributed from the tap points could affect the fault-loop
impedance estimated by a distance relay. As a result, the distance relay may under-reach or
overreach depending on the transmission system configuration and parameters [6].

The fault-induced high-frequency (HF) transients contain extensive information about
the fault by which the fault direction and location could be identified in a few milliseconds [7]–
[10].
The traveling-wave positional protection technique uses polarities and time intervals
between arriving waves at the relay bus to estimate the fault location [9], [10]. Nevertheless, this
technique cannot be easily employed for protection of multiterminal lines due to the reflections
of traveling waves at the tap points.
Recently, some non-communication-based techniques for the protection of two-terminal

lines have been introduced which use the HF components of the fault signals measured at the
relay bus to discriminate between the internal and external faults [11]–[15]. These algorithms
rely on the fact that the substation equipment affects the characteristics of the HF transients
generated by the faults. The bus stray capacitances have a low-pass filtering effect on the fault-
generated transients. As a result, in the case of external faults beyond the TL ends, the HF
components involved in the fault signals are attenuated when passing through the remote
substation bus, whereas in the case of internal faults, the fault signals measured at the TL ends
contain a higher amount of HF components. When the bus equivalent capacitance is sufficiently
large, these algorithms could provide a satisfactory response. Nevertheless, for small values of
the capacitance, discrimination between the internal and external faults might become difficult
due to the small difference between the characteristic features of these two cases. In addition,
variation of the power system parameters (e.g., due to the disconnection of a transformer or a
transmission line) could affect the performance of these algorithms.
2.4 Objective of present work
This dissertation presents a new non-communication protection technique for

multiterminal transmission lines using the fault-generated high-frequency transients. To
minimize the influence of the variable power system parameters and conditions on the relay
performance, appropriately designed line traps are installed at the TL terminals. The dyadic

wavelet transform (DWT) is used to decompose the fault signals into different frequency bands.
Then, the signal spectral energy within each frequency band is computed. The Artificial Neural
Network (ANN) is used to classify the faults into two categories of internal and external faults
according to their frequency spectrum.
Extensive simulation studies are carried out using the power system simulation toolbox
software MATLAB to evaluate the performance of the proposed algorithm under different
conditions. The current transformers (CTs) and capacitor voltage transformers (CVTs) are also
modeled precisely. The obtained results confirm that the proposed algorithm provides a
satisfactory and reliable scheme for the protection of the multiterminal TLs even for a small
amount of bus equivalent capacitance and variation of the power system parameters. The
proposed technique considerably improves the first zone coverage of the protective scheme and
covers almost the total length of the multiterminal TL under protection.

CHAPTER 3
=================================================================
PROPOSED APPROACH
=================================================================
3.1 Fault classification scheme for Multiterminal transmission line
3.1.1 Objective of system
The dyadic wavelet transform (DWT) is used to decompose the fault signals into
different frequency bands. Then, the signal spectral energy within each frequency band is
computed. The Artificial Neural Network (ANN) is used to classify the faults into two
categories of internal and external faults according to their frequency spectrum.
3.1.2 Overview of system

3.1.2.1 Discrete wavelet transform for fault classification
Information which is not available directly from the original signal can be obtained by
applying mathematical transformation. Most of the signals in their raw format are the time-
domain signals. It means, whatever the signal is measuring will be a function of time or we can
say that while plotting the signal, one axes will represent time which is an independent variable
and the other axes will represent the other variable which is dependent on time. Usually, this
dependent variable is amplitude.
So, a signal in time-domain is plotted, representation of signal in time-amplitude form is

obtained. Further for many applications related to signal processing, this representation is not
useful. In several cases, the most important information is concealed in the frequency contents
of the signal. The frequency spectrum of any signal is nothing but frequency components of the
signal. The frequencies existing in the signal are shown by frequency spectrum of the signal.
Basically, frequency is rate of change of a given variable and is measured in cycles per second.
If this variable is changing rapidly, then we can say that the frequency of variable is very high
and if the value of this variable is changing slowly, then we can say that the frequency of the
variable is low.

Further, if there is no change in the value of variable then the frequency is zero or no
frequency. So, to measure the frequency of any variable or in other words to find frequency
content of any signal we can use Fourier transform.
For analyzing the frequency contents of any signal, FT can be applied efficiently.
Though, if the Fourier transform of the entire time axis is taken, then we cannot notify that at
what instant a specific frequency rises. Short Time Fourier Transform (STFT) which uses a
sliding window can also used to find information of both frequency and time. However, window
length limits the resolution of frequency. So, the Wavelet transform is the ultimate solution of
these problems. Small wavelets having limited duration are considered in wavelet transform.
Figure 3.1. Block diagram of fault classification scheme for multiterminal transmission line
A wavelet-based signal-processing technique is an effective tool for power system

transient analysis and feature extraction [38]. 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 [39]. This helps to
analyze the signal in both frequency and time domains effectively. And, WT is able of providing
the frequency and time information concurrently, thus provides the time-frequency
representation of the given signal. The time-domain signal passes through numerous low pass
and high pass filters. And these filters, filters either the low or high frequency part of the signal
[40]. This process is repeated, and each time a certain part of the signal equivalent to some
frequencies is removed from the signal. 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.
The WT can detect the low frequency and high frequency components precisely. One of
the main properties of the WT that it has the great ability to locate the signals short-time high
frequency features and determine the low frequency performance.
The convolution of the signal f (t), which is under analysis, is taken with a wavelet 𝜓, to
obtain wavelet transform [41].
∞
C (Scale, position) = ∫ f(t)ψ (Scale, position, t) dt
−∞
Where,
𝜓 belongs to a special wavelets family to compare with f(t) is known as “mother wavelet”.
There are various types of wavelets i.e. Haar, Symlets, Meyer, Daubechies, and Discrete
Meyer. To analyze the portion of the signal which is not known using convolution, 𝜓 is selected
i.e. WT can detect if under a determined position and scale, the analyzed signal intimately
correlated with 𝜓. Basically, wavelet transform is used to evaluate the non-stationary signals,
those frequency response change in time.
3.1.2.1.1 Discrete Wavelet Transform

Signal is filtered and then sampling is done in DWT. This process reduces the data but
the necessary information is kept as such [42]. Signal can be analyzed at different frequency

bands in DWT process. Two data sequences cA1 and cD1 are obtained after the process as
shown in Figure 3.2. Approximation (cA1) part of the signal consists of low frequency and high
scale components of the signal while detail (cD1) consists of high frequency and low scale
components of the signal [43].
Let, V(t) is the original signal at 20 kHz and we want to obtain its discrete wavelet
transform coefficients. So, this signal is then passed through a low pass and high pass filters.
The output of these filters is sub-sampled by a factor of 2. And the output of these filters gives
the first level DWT coefficients i.e. approximation and detail coefficients. They represent the
signal at 10 kHz having 1004 simples. Approximation (cA1) contains the components of the
signal which are at less than 10 kHz frequency (low frequency, high scale) while detail (cD1)
contains greater than 10 kHz (high frequency, low scale) components of the signal. The
approximation and detail coefficients of level 1 are further decomposed by passing them
through low pass and high pass filters to give the level 2 coefficients of DWT which represents
the signal at 5 kHz frequency having 502 samples. Approximation (cA2) coefficient of level 2
contains the components of the signal which are at below 5 kHz frequency and detail (cD2)
coefficient of level 2 contains the components of the signal which are at greater than 5 kHz
frequency. This process is recurring until the required level for a particular application is
achieved. In a nutshell, in DWT process number of data is reduced at each step but it contains
the entire useful information.
Figure 3.2. Discrete Wavelet Transform

3.1.2.1.2 Wavelet Multi-Resolution Analysis
Wavelets can present multiple resolutions in both time and frequency. The signal is
decomposed at distinctive resolution levels by the use of wavelet and scaling functions in
multiple resolution analysis [44]. The detail form of the decomposed signal will be generated by
the wavelet function and the approximated form of the decomposed signal will be generated by
the scaling function. It means that the wavelet function consists of the high pass filter and low
pass filter is contained in the scaling function [45].
Let V(t) is the original signal obtained from a measuring device. V(t) is decomposed into
detail and approximation. cA1 and cD1 are the decomposed signals of level 1 in the multi
resolution technique. Where, cA1 is the approximation of the original signal and the cD1 is the
detail version of the original signal. cA1 and cD1 are defined as given in equations (4.2) and
(4.3) respectively [46].
cA1 (t) = ∑ V(t) . Ld (k − 2t)

k
cD1 (t) = ∑ V (t) . Hd(k − 2t)

k
Where,
Ld is the low-pass filter and,
Hd is the high-pass filter.
These filters are related to mother wavelet 𝜓. High frequency components of the signal
are contained in cD1 whereas cA1 contains the low frequency components of the signal. When
the original signal V(t) is passed through the low pass and high pass filters, it gets decomposed
into cA1 and cD1 coefficients of the signal. Here the cA1 and cD1 are level1 coefficients.
Further approximation cA1 is decomposed into cA2 and cD2 coefficients of the signal which
are the level 2 coefficients. And this procedure is repeated again and again until the required
level is obtained for a particular application as shown in Figure 3.3.

Figure 3.3 Wavelet decomposition tree
3.1.2.1.3 Working of Wavelet

Let the signal having sampling frequency 20 kHz is passed through the low pass and
high pass filters where it is decomposed into two different frequency bands i.e. cA1 which
contains less than 10kHz frequency portion of the signal and cD1 which contains greater than
10kHz frequency portion of the signal. Hence, cA1 contains the low frequency portion and cD1
contains high frequency portion of the signal. This decomposition is known as level1
decomposition. Further, cA1 is decomposed into two frequency bands i.e. low frequency portion
(less than 5 kHz) and high frequency portion (more than 5 kHz). Similarly, cD1 is decomposed
into two frequency bands i.e. low frequency portion (less than 5 kHz) and high frequency
portion (more than 5 kHz). This decomposition is known as second level decomposition. In this
way, next level decompositions are also obtained by dividing the parent frequency with a factor
of 2 as shown in Figure 3.4 This process is continual until the required level of application is
achieved.

Figure 3.4. Wavelet decomposition of signals sampled at 20 kHz shows association of levels to components
3.1.2.2 Artificial Neural Network for fault classification

3.1.2.2.1 Introduction to Neural Network
It is well known that biological systems can perform complex tasks without recourse to
explicit quantitative operations. In particular, biological organisms are capable of learning
gradually over time. This learning capability reflects the ability of biological neurons to learn
through exposure to external stimuli and to generalize. Such properties of nervous systems make
them attractive as computation models that can be designed to process complex data. For
example, the learning capability of biological organisms from examples suggests possibilities
for machine learning.
Neural networks or more specifically, artificial neural networks are mathematical models
inspired from our understanding of biological nervous systems. They are attractive as
computation devices that can accept a large number of inputs and learn solely from training
samples. As mathematical models for biological nervous systems, artificial neural networks are
useful in establishing relationships between inputs and outputs of any kind of system.
Roughly speaking, a neural network is a collection of artificial neurons. An artificial

neuron is a mathematical model of a biological neuron in its simplest form. From our
understanding, biological neurons are viewed as elementary units for information processing in

any nervous system. Without claiming its neurobiological validity, the mathematical model of
an artificial neuron is based on the following theses:
1. Neurons are the elementary units in a nervous system at which information processing occurs.
2. Incoming information is in the form of signals that are passed between neurons through
connection links.
3. Each connection link has a proper weight that multiplies the signal transmitted.
4. Each neuron has an internal action, depending on a bias or fring threshold, resulting in an
activation function being applied to the weighted sum of the input signals to produce an output
signal.
Thus, when input signals x1, x2, . . . , xn reach the neuron through connection links with
associated weights w1,w2, . . .,wn, respectively, the resulting input to the neuron, called the net
input, is the weighted sum ∑𝑛𝑖=1 𝑤𝑖. 𝑥𝑖 . If the firing threshold is b and the activation function is f,
then the output of that neuron is
𝑦 = 𝑓(∑𝑛𝑖=1 𝑤𝑖. 𝑥𝑖 )
In the first computational model for artificial neurons, proposed by McCulloch and Pitts
[43], outputs are binary, and the function f is the step function.
Figure 3.5. First model for artificial neuron

Defined by
1 𝑖𝑓 𝑥 ≥ 0
𝑓(𝑥) = {
0 𝑖𝑓 𝑥 < 0
so that the activation of that neuron is

𝑛
𝑛 1 𝑖𝑓 ∑ 𝑤𝑖. 𝑥𝑖 ≥ 𝑏
𝑖=1
𝑓(∑(𝑤𝑖. 𝑥𝑖 − 𝑏) = 𝑛
𝑖=1
0 𝑖𝑓 ∑ 𝑤𝑖. 𝑥𝑖 < 𝑏
{ 𝑖=1
This is depicted in Figure 3.5.
An artificial neuron is characterized by the parameters θ = (w1,w2, . . . ,wn, b, f)

The bias b can be treated as another weight by adding an input node x0 that always takes
the input value x0 = +1 and setting w0 = −b (see Figure 3.6). With this representation, adjusting
bias and adjusting weights can be done in the same manner.
We will consider here only feed-forward neural networks, that is, information propagates
only forward as indicated by the direction of the arrows. Mathematically speaking, a
feedforward neural network is an acyclic weighted, directed graph. Viewing artificial consists of
an input layer of input nodes and one output layer consisting of
Figure 3.6. Artificial neuron with bias as weight
neurons. This is referred to as a single-layer neural network because the input layer is not a layer
of neurons, that is, no computations occur at the input nodes. This single-layer neural network is
called a perceptron.

Figure 3.7. Perceptron
A multi-layer neural network is a neural network with more than one layer of neurons.
Note that the activation functions of the different neurons can be different. The neurons from
one layer have weighted connections with neurons in the next layer, but no connections between
neurons of the same layer. A two-layer neural network is depicted in Figure 3.8. Note that
activation functions of different neurons can be different. The input layer (or layer 0) has n+ 1
node, the middle layer, called the hidden layer, has p nodes, and the output layer has m nodes.
This is called an n-p-m Neural network. Neurons (nodes) in each layer are somewhat similar.
Neurons in the hidden layer are hidden in the sense that we cannot directly observe their output.
From input patterns, we can only observe the output patterns from the output layer.
Off course, a multi-layer neural network can have more than one hidden layer. The two-
layer neural network depicted in Figure 3.8 is a typical multi-layer perceptron (MLP), a
multilayer neural network whose neurons perform the same function on inputs, usually a
composite of the weighted sum and a differentiable nonlinear activation function, or transfer
function, such as a hyperbolic tangent function. Multi-layer perceptrons are the most commonly
used neural network structures for a broad range of applications.

Figure 3.8. Two layer Neural Network
3.1.2.2.2 The Learning Process

All learning methods used for adaptive neural networks can be classified into two major
categories:
Supervised learning which incorporates an external teacher, so that each output unit is told
what its desired response to input signals ought to be. During the learning process global
information may be required. Paradigms of supervised learning include error-correction
learning, reinforcement learning and stochastic learning. An important issue concerning
supervised learning is the problem of error convergence, i.e. the minimization of error between
the desired and computed unit values. The aim is to determine a set of weights which minimizes
the error. One well-known method, which is common to many learning paradigms, is the least
mean square (LMS) convergence.
Unsupervised learning uses no external teacher and is based upon only local
information. It is also referred to as self-organization, in the sense that it self-organizes data
presented to the network and detects their emergent collective properties. Paradigms of
unsupervised learning are Hebbian learning and competitive learning.
From Human Neurones to Artificial Neuron Esther aspect of learning concerns the
distinction or not of a separate phase, during which the network is trained, and a subsequent
operation phase. We say that a neural network learns off-line if the learning phase and the
operation phase are distinct. A neural network learns on-line if it learns and operates at the same

time. Usually, supervised learning is performed off-line, whereas unsupervised learning is
performed on-line.
3.1.2.2.3 Transfer Function

The behavior of an ANN (Artificial Neural Network) depends on both the weights and
the input-output function (transfer function) that is specified for the units. This function
typically falls into one of three categories:
-linear (or ramp)
-threshold
-sigmoid f  a  1 1  exp(a)
For linear units, the output activity is proportional to the total weighted output.
For threshold units, the output are set at one of two levels, depending on whether the total input
is greater than or less than some threshold value.
For sigmoid units, the output varies continuously but not linearly as the input changes. Sigmoid
units bear a greater resemblance to real neurons than do linear or threshold units, but all three
must be considered rough approximations.
We can teach a three-layer network to perform a particular task by using the following
procedure:
1. We present the network with training examples, which consist of a pattern of activities for the
input units together with the desired pattern of activities for the output units.
2. We determine how closely the actual output of the network matches the desired output.
3. We change the weight of each connection so that the network produces a better
approximation of the desired output.
In order to train a neural network to perform some task, we must adjust the weights of
each unit in such a way that the error between the desired output and the actual output is
reduced. This process requires that the neural network compute the error derivative of the
weights (EW). In other words, it must calculate how the error changes as each weight is
increased or decreased slightly. The back propagation algorithm is the most widely used method
for determining the EW.

The back-propagation algorithm is easiest to understand if all the units in the network are
linear. The algorithm computes each EW by first computing the EA, the rate at which the error
changes as the activity level of a unit is changed. For output units, the EA is simply the
difference between the actual and the desired output. To compute the EA for a hidden unit in the
layer just before the output layer, we first identify all the weights between that hidden unit and
the output units to which it is connected. We then multiply those weights by the EAs of those
output units and add the products. This sum equals the EA for the chosen hidden unit. After
calculating all the EAs in the hidden layer just before the output layer, we can compute in like
fashion the EAs for other layers, moving from layer to layer in a direction opposite to the way
activities propagate through the network. This is what gives back propagation its name. Once
the EA has been computed for a unit, it is straight forward to compute the EW for each
incoming connection of the unit. The EW is the product of the EA and the activity through the
incoming connection.
3.1.2.2.4 The backpropagation algorithm

The following popular learning algorithm, referred to as the backpropagation algorithm,
is a generalization of the delta rule. If we look closely at the delta rule for single-layer neural
networks, we realize that to update a weight wik when the learning input pattern xq is presented,
we need ∆wqik = δiq.xkq
Since backpropagation uses the gradient descent method, one needs to calculate the
derivative of the squared error function with respect to the weights of the network. Assuming
one output neuron, the squared error function is:
,
where
is the squared error,
is the target output for a training sample, and
y is the actual output of the output neuron.

The factor of is included to cancel the exponent when differentiating. Later, the expression
will be multiplied with an arbitrary learning rate, so that it doesn't matter if a constant
coefficient is introduced now.
For each neuron , its output is defined as
.
The input to a neuron is the weighted sum of outputs of previous neurons. If the
neuron is in the first layer after the input layer, the of the input layer are simply the
inputs to the network. The number of input units to the neuron is . The variable
denotes the weight between neurons and .
The activation function is in general non-linear and differentiable. A commonly used

activation function is the logistic function:
which has a nice derivative of:
1) Finding the derivative of the error

Calculating the partial derivative of the error with respect to a weight is done using
the chain rule twice:
In the last term of the right-hand side, only one term in the sum depends on , so that
.
The derivative of the output of neuron with respect to its input is simply the partial derivative
of the activation function (assuming here that the logistic function is used):
This is the reason why backpropagation requires the activation function to be differentiable.

The first term is straightforward to evaluate if the neuron is in the output layer, because
then and
However, if is in an arbitrary inner layer of the network, finding the derivative with
respect to is less obvious.
Considering as a function of the inputs of all neurons receiving input
from neuron ,
and taking the total derivative with respect to , a recursive expression for the derivative is
obtained:
Therefore, the derivative with respect to can be calculated if all the derivatives with respect
to the outputs of the next layer – the one closer to the output neuron – are known. Putting it
all together:
with
To update the weight using gradient descent, one must choose a learning rate. The change
in weight, which is added to the old weight, is equal to the product of the learning rate and the
gradient, multiplied by -1:
The -1 is required in order to update in the direction of a minimum, not a maximum, of the error
function.

3.1.3 Working of fault classification scheme
This algorithm uses the phase current signals measured at the relay bus. The current
signals are first passed through antialiasing filters, and then are sampled at the sampling
frequency of 160 kHz. The measured signals are decomposed into different frequency bands
using DWT. Meanwhile, using the wavelet transform, a denoising process is performed on the
signals to increase the noise immunity of the proposed protection algorithm. Considering the
specified sampling frequency, the signals are decomposed up to five stages. The obtained detail
coefficients d1, d2 , d3 , d4, and d5 correspond to the frequency bands of 40–80 kHz, 20–40
kHz, 10–20 kHz, 5–10 kHz and 2.5–5 kHz, respectively. The dyadic wavelet transform is
implemented using the filter bank approach which does not impose considerable computation
burden. The signal energy at each frequency band is calculated for a 1/4-cycle sliding window.
The calculated signal energy at the defined frequency bands is normalized using the
signal energy at the fourth frequency band, and is used as the ANN inputs to discriminate
between internal and external faults. The ANN is well capable of constructing appropriate
hyperplanes for clustering the input data in a high dimension space.
The processing burden of the proposed algorithm is higher than that of some
conventional techniques employed in the impedance-based relays. Nevertheless, considering the
advancements in the digital technology and processing speed, this is not a serious problem.
Since the proposed algorithm uses a 1/4-cycle sliding data window to calculate the signal
spectrum energy, if an appropriate processor is used, the operation time of the proposed
algorithm would be about 1/4 cycle of the power system frequency.
3.2 Protection scheme of Multiterminal transmission line

Figure 3.9. Block diagram of protection scheme of multiterminal transmission line
3.2.1 Objective of system

This dissertation presents a new non-communication protection technique for
multiterminal transmission lines using fault-generated high-frequency transients. To minimize
influence of the variable power system parameters and conduction on relay performance,
appropriately design line traps are installed at the transmission line terminals.
This dissertation improves the performance of distance relay by introducing line traps for
providing discrimination between internal fault and external fault. Due to line trap impedance
relay can trip during internal fault only. It means that impedance distance relay protect the main
transmission line when fault occurs on that line, and not trip the line when external fault occur
in outside protected transmission line.
3.2.2 Overview of system

3.2.2.1 Protective relay
Fault current is the expression given to the current that flow in the circuit when load is
shorted i.e. flow in a path other than the load. This current is usually very high and may exceed
ten times the rated current of a piece of plant. Faults on power system are inevitable due to
external or internal causes, lightning may struck the overhead lines causes insulation damage.
Internal overvoltage due switching or other power system phenomenon may also cause an over
voltage leads to deterioration of the insulation and faults. Power networks are usually protected
by means of two main components, relays that sense the abnormal current or voltage and a
circuit breaker that put a piece of plant out of tension power system.
Protection is the art and science of the application of devices that monitor the power line
currents and voltages (relays) and generate signals to de-energize faulted sections of the power
network by circuit breakers. Goal is to minimize damage to equipment and property that would
be caused by system faults, if residues, and maintain the delivery of electrical energy to the
consumers. Many types of protective relays are used to protect power system equipments, they
are classified according to their operating principles; over current relay senses the extra (more
than set) current considered dangerous to a given equipment, differential relays compare in and
out currents of a protected equipment, while impedance relays measure the impedance of the
protected piece of planet. For a good performance of a relay in a power system it must have the
following characteristics; dependability, security, selectivity, sensitivity and speed.
Distance Relays: Distance relays, as the name sounds, should measure distance. In fact this is
true, as in case of transmission line, distance relay measures the impedance between the relay
point and the fault location. This impedance is proportional to the length of the conductor, and
hence to the distance between the relaying point and the fault.
Principle of operation: The basic principle as illustrated in figure 3.10, involves the division of
the voltage at the relaying point by the measured current. The apparent impedance is compared
with the reach point impedance. If the measured impedance is less than the reach point
impedance, it is assumed that a fault exists on the line between the relay and the reach point.
The reach point of the relay is the point along the line impedance locus that is intersected by the

boundary characteristics of the relay. Distance relay is the broader name of the different types of
impedance relay.
Figure 3.10. Principle of operation of distance relay
The relay is connected at position, R and receives a secondary current, iF, equivalent to a
primary fault current, IF. The secondary voltage, VF, is equivalent to the product of the fault
current “IF” and impedance of the line up to the point of fault, ZF. The operating torque of this
relay is proportional to the fault current “IF”, and its restraining torque is proportional to the
voltage “VF”. Taking into account the number of turns of each coil, there will be a definite ratio
of V/I at which the torque will be equal. This is the reach point setting of the relay. The relay
will operate when the operating torque is greater than the restraining torque. Thus any increase
in current coil ampere-turns, without a corresponding increase in the voltage coil ampere-turns,
will unbalance the relay. This means the V/I ratio has fallen below the reach point. Alternatively
if the restrain torque is greater than the operating torque, the relay will restrain and its contacts
will remain open. In this case the V/I ratio is above the reach point. The reach of a relay is the
distance from the relaying point to the point of fault. Voltage on the primary of voltage
transformer, VT, is

𝐸 𝑍𝑓
𝑉𝑇 =
(𝑍𝑠 + 𝑍𝑓)
Fault current If is
𝐸
𝐼𝑓 =
(𝑍𝑠 + 𝑍𝑓)
The relay compare the secondary values of V and I, as to measure their ratio which is an
impedance Zm ,
𝑉⁄
𝑍𝑚 = 𝑉. 𝑇. 𝑅𝑎𝑡𝑖𝑜
𝐼⁄
𝐶. 𝑇. 𝑅𝑎𝑡𝑖𝑜
Zm is the measured impedance called secondary impedance.
Zones of protection: Basic distance protection will comprise instantaneous directional Zone 1
protection and one or more time delayed zones. Numerical distance relays may have up to five
zones, some set to measure in the reverse direction. Numerical relays usually have a reach
setting of up to 85% of the protected line impedance for instantaneous Zone 1 protection. The
resulting 15% safety margin ensures that there is no risk of the Zone 1 protection over-reaching
the protected line due to errors in the current and voltage transformers, inaccuracies in line
impedance data provided for setting purposes and errors of relay setting and measurement of the
distance protection must cover the remaining 15% of the line.
The reach setting of the Zone 2 protection should be at least 120% of the protected line
impedance. In many applications it is common practice to set the Zone 2 reach to be equal to the
protected line section +50% of the shortest adjacent line. Zone 3 reach should be set to at least
1.2 times the impedance presented to the relay for a fault at the remote end of the second line
section. Typical reach for a 3-zone distance protection are shown in Figure 3.11.

Figure 3.11. Typical 3 zone distance protection
Residual factor: The measuring element of the distance relay is principally laid out such that
for each fault type the line impedance of the fault loop is determined. In three phase system the
zone-1 of the relay will have six elements responsible for detecting both phase and earth faults.
For phase faults elements, the difference between the two relevant phase signals are used, e.g. a-
b elements is supplied with samples of Va – Vb voltage and Ia - Ib current. For earth elements,
the relevant phase voltage is supplied e.g. Va, but the corresponding current is residually
compensated. The earth faults compensation factor may be calculated considering the sequence
networks connection for the phase A-to-ground fault on a transmission line. Table indicates
calculation formula for phase and line to line faults. In order for the relay to be correctly
operated, residual factor shall be introduced as shown in the following equations
𝑉𝑘
𝑅𝑘 =
𝐼𝑘 + 3 𝑅𝑒. (𝐾0). 𝐼𝑜𝑘
𝑉𝑘
𝑋𝑘 =
𝐼𝑘 + 3. 𝐼𝑚(𝐾𝑜). 𝐼0𝑘
Where; K0, is the compensation factor
I0, is zero sequence current
Vk, Ik are the sampled voltage and current respectively

Table 3.1: Calculation formula for phase and line to line faults
Distance Element Voltage signal Current signal
Phase A Va Ia + 3K0I0
Phase A-PhaseB Va – Vb Ia - Ib
Effect of fault resistance on relay coverage: The earth fault resistance reduces the effective
earth-fault reach of a mho Zone 1 element to such an extent that the majority of faults are
detected in Zone 2 time. Figure 3.12 illustrates the effect of arc resistance on the relay reach.
The effect of fault resistance on the reach of distance relays is better discussed with the
simulation results.
3.2.2.2 Circuit Breaker

A circuit breaker is an automatically operated electrical switch designed to protect
an electrical circuit from damage caused by overload or short circuit. Its basic function is to
detect a fault condition and interrupt current flow. Unlike a fuse, which operates once and then
must be replaced, a circuit breaker can be reset (either manually or automatically) to resume
normal operation. Circuit breakers are made in varying sizes, from small devices that protect an
individual household appliance up to large switchgear designed to protect high voltage circuits
feeding an entire city.
All circuit breaker systems have common features in their operation, although details
vary substantially depending on the voltage class, current rating and type of the circuit breaker.

Figure 3.12. Effect of arc resistance on relay coverage
The circuit breaker must detect a fault condition; in low voltage circuit breakers this is
usually done within the breaker enclosure. Circuit breakers for large currents or high voltages
are usually arranged with protective relay pilot devices to sense a fault condition and to operate
the trip opening mechanism. The trip solenoid that releases the latch is usually energized by a
separate battery, although some high-voltage circuit breakers are self-contained with current
transformers, protective relays and an internal control power source.
Once a fault is detected, contacts within the circuit breaker must open to interrupt the
circuit; some mechanically-stored energy (using something such as springs or compressed air)
contained within the breaker is used to separate the contacts, although some of the energy
required may be obtained from the fault current itself. Small circuit breakers may be manually
operated; larger units have solenoids to trip the mechanism, and electric motors to restore
energy to the springs.
The circuit breaker contacts must carry the load current without excessive heating, and
must also withstand the heat of the arc produced when interrupting (opening) the circuit.
Contacts are made of copper or copper alloys, silver alloys and other highly conductive
materials. Service life of the contacts is limited by the erosion of contact material due to arcing
while interrupting the current. Miniature and molded-case circuit breakers are usually discarded

when the contacts have worn, but power circuit breakers and high-voltage circuit breakers have
replaceable contacts.
3.2.2.3 Line Trap

Line traps are connected in series with HV transmission lines to prevent dissipation of
the power-line carrier (PLC) signal in the substation or in other transmission lines. The main
function of the line trap is to present high impedance at the carrier frequency band while
presenting negligible impedance at the power system frequency. The high impedance is required
to reduce the carrier signal attenuation due to the division among the several transmission lines
terminated at the same bus. A line trap consists of three major components in parallel (i.e., mail
coil, tuning device, and surge arrester). The main coil carries the rated current of the
transmission line and is designed to withstand the maximum short-circuit current. The tuning
device is connected across the main coil and provides high impedance over a specified PLC
frequency band. The surge arrester limits the lightning and switching over voltages applied to
the line trap.
Depending on the type of tuning (i.e., single-frequency, double frequency, and

wideband), the tuning device consists of capacitors, inductors, and resistors, all having relatively
low-power ratings compared to the main coil [35].
When line traps are installed at the TL ends, most of the fault-induced HF components
are confined to the faulty transmission line regardless of configuration of the primary system
switchgear. As a result, a protective algorithm based on the spectral energy of the fault-induced
transients can well discriminate between internal and external faults. In other words, the line
trap provides the required high impedance at the high frequencies so that the reflection and
refraction coefficients of the TL termination points are not affected considerably by variation of
the substation parameters.
It should be considered that, in practice, there are some limitations in selection of the
values of the line trap parameters. Since the main coil carries the TL current, a high inductance
main coil could develop into a physically very large dimension.

Moreover, the undesired voltage drop across the line trap at the fundamental frequency
increases with an increase of the main coil inductance. According to the IEC 60353 standard,
the value of the main coil inductance should not be more than 2 mH [36]. For this value, the
voltage drop across the line trap is negligible. For high-frequency studies, the stray capacitance
of the main coil should also be modeled. The open, air-insulated design of the main coil results
in small inherent capacitances of about 30 to 100 pF, depending on the size of the line trap [37].
Fig. 3.13 depicts the designed line trap installed at the protected TL ends. In this figure L1,R1
and C1, and denote the inductance, resistance, and the equivalent stray capacitance of the main
coil, R2 and C2 denote the resistor and tuning capacitor, respectively.
Figure 3.13. Circuit diagram and parameters of the designed line trap
It should be noted that, in practice, CVTs are also installed at the TL ends to measure the phase
voltages. CVTs are installed after the line trap at the TL side. Thus, in this Proposed approach,
CVTs are appropriately modeled to consider their influence on the frequency characteristics of
the line trap.
3.2.2.4 Current transformer (C.T.)

The transformer used for measurement of current is called as current transformer. The
current transformer is used with its primary winding connected in series with line carrying the
current to be measured and, therefore, the primary current is dependent upon load connected to

system and is not determine by the load (burden) connected on the secondary winding of the
current transformer. The primary winding consists of very few turn and, therefore, there is no
appreciable voltage drop across it. The secondary of current transformer has larger number of
turns, the exact number being determined by the turns ratio. The ammeter or wattmeter current
coil, are connected directly across secondary winding terminals. Thus a current transformer
operates its secondary winding nearly under short circuit condition. One of the terminals of the
secondary winding is earthed so as to protect equipment and personnel in the vicinity in the
event of insulation breakdown in current transformer.
3.2.2.5 Capacitance voltage transformer (C.V.T.) or Potential transformer (P.T.)

The transformer used for measurement of voltage is called as current transformer.
Potential transformers are used to operate voltmeters, the potential coils of wattmeter and relays
from high voltage lines. The primary winding of transformer is connected across the line
carrying the voltage to be measured and the voltage circuit is connected across the secondary
winding.
The design of potential transformer is quite similar to that of a power transformer but the
loading of potential transformer is always small, sometimes only a few volt-ampere. The
secondary winding is design so that a voltage of 100 to 120V is delivered to the instrument load.
The normal secondary voltage rating is 110V.
3.2.3 Working of protection scheme

When the internal fault occurs impedance relay measured the impedance of line, if
impedance of line is less than actual set point impedance then relay send the trip signal to the
circuit breakers which are installed at each end of multiterminal transmission line. This circuit
breakers isolates the multiterminal line from fault. At same time high frequency generated fault
signals are reflected towards each terminal of multiterminal transmission line. That high
frequency transient surge are not reflected outside i.e. external portion of main protected
transmission line with the help of line trap. Hence external line relay not trip and other outside
lines are work normally.

When external faults occurs impedance relay measured the impedance of line, if impedance of
line is more than actual set point impedance then relay not send trip signals and main protected
line works without any disturbance. But impedance measured by outside line relay is less than
actual set point impedance then that line get isolate with the help of circuit breakers. At same
time fault signal generated outside the main protected line are not refracted toward the main
protected multiterminal transmission line bus due to line traps. Hence relay of main protected
line not sense any less impedance value and line normally operates.

CHAPTER 4
=================================================================
TEST SYSTEM AND EXPERIMENTAL RESULTS

=================================================================
4.1 Simulated Model
Figure 4.1 Simulated model of complete system

Figure 4.2 Simulation model of zone 1

Figure 4.3. Simulation model of zone 2

Figure 4.4. Simulation model of zone 3

Figure 4.5.Simulation model of zone 4
4.2 Specification of simulated components

(1) Generators 1,2,3; Phase to phase rms voltage = 153e3 V; Frequency = 50 Hz; 3 phase short
circuit level = 100e6 V; Base voltage = 153e3 V; X/R=7

(2) Transformer 1; Power= 100e6 VA; Primary voltage = 153e3 V; Secondary voltage = 34.5e3
V ; Frequency = 50Hz
(3) Transformer 2; Power = 80e6 VA; Primary voltage = 34.5e3 V; Secondary voltage = 400 V;
Frequency = 50Hz
Frequency = 50 Hz
(5) Transformer 4; Power = 80e3 VA; Primary voltage = 400 V; Secondary voltage= 34.5e3 V;
Frequency = 50Hz
Frequency = 50Hz
(7) Transformer 6; Power = 100e6 VA; Primary voltage = 153e3 V; Secondary voltage = 34.5e3
V; Frequency = 50Hz
(8) Transformer 7; Power = 100e6 VA; Primary voltage = 153e3 V; Secondary voltage = 34.5e3
V; Frequency = 50Hz
(9) Source impedance1; R1 = 0.381 Ω; L1 = 1e-3 H
(10) Source impedance2; R2 = 3.466 Ω; L2 = 1e-3 H
(11) Source impedance 3; R3 = 0.821 Ω; L3 = 1e-3 H
(12) Load 1 ; Phase to phase voltage = 400 V; Frequency = 50 Hz; Active power = 0.1 W;
Inductive reactive power = 3.5 Positive VAR; Capacitive reactive power = 2.4 Negative VAR

(13) Load 2 Phase to phase voltage= 440; Frequency = 50 Hz; Active power = 0.1 W; Inductive
reactive power = 25 +VAR; Capacitive reactive power = 2.4 -VAR
(14) Line Trap 1,2,3; L1 = 2 mH; R1 = 0.5 Ω; C1 = 50 pF; R2 = 50 Ω; C2 = 5 nf
(15) Transmission line model Phase to phase voltage 34.5 KV; Frequency = 50 Hz; Zone1 (AB)
= 50Km; Zone2 (BC) = 100Km; Zone3 (CD) = 60Km; Zone4 (EF) = 70 Km
4.3 Wavelet transforms subsystem
Figure 4.6. Wavelet transform subsystem blocks in simulated model
The specification of each blocks in Discrete Wavelet Transform subsystem are following way:
(1)DWT block: Mother wavelet = Daubechies; Wavelet order = 2; Number of levels = 5; tree
structure = symmetric; output = multiple ports
(2)Buffer: output buffer size (per channel) = 128; buffer overlap = 0; initial condition = 0

(3)Unbuffer: Initial condition = 0
Figure
4.7. DWT detail blocks in simulation model
4.4 Neural Network Subsystem
Figure 4.8. Neural network subsystem
The specification of neural network subsystem are as follows:

Name of neural network = nn1
Neural network type = Feed-forward backpropagation
Input data = inputdata1
Target data = targetupdated1

Training function = TRAINLM (Levenberg-Marquard)
Performance function = Mean Square Error
Number of layers = 2;
Numbers of neurons in first layer = 10;
Transfer function = TANSIG
Figure 4.9. Neural network configuration window
4.4.1 Training of Neural Network

To provide the required training and testing data sets, various internal and external faults
were simulated on the transmission system under study by changing the fault type, the fault
resistance, the fault location, and the earth resistance. In total, 500 fault cases were simulated,
taking into account all eight fault types, several fault resistances in the range of 0.01–50, 15
different fault locations, and several earth resistances 0.001-5.

The fault signals were processed to extract their HF spectral energy at different frequency
bands. The Daubechies db3 mother wavelet was used for the signal decomposition. The data
sets were formed so that each row consists of five inputs (i.e., the signal HF energies at the four
specified frequency bands as well as the corresponding class to which the input vector belongs).
The corresponding classes were defined as “1” for the case of internal faults and “-1” for the
case of external faults. Out of the 500 simulated fault cases, 94 faults were randomly selected to
train the ANN, and the remaining cases were used to test the trained ANN.
Steps of training the neural network are following way:

Step (1): Select training information for Neural Network.
In this we create perceptron network to classify the internal and external fault. It has input
vector named as input data1 which contain 5×94 matrix dataset. 5 represent five wavelet
decomposition signal coefficients and 94 represents 94 fault condition based on various
parameters. Those parameters are types of fault, fault resistance, fault location and kilometer
length from reference bus, earth resistance. And target vector named as targetdata1 which
contain 1×94 matrix dataset. For external fault target is -1 and for internal fault target is 1.
Figure 4.10. Training information selection window for Neural Network training
Step (2): Set the training parameter for Neural Network

Figure 4.11. Training parameter window of neural network
Step (3): Select Train Network
Figure 4.12. Training performance window of neural network

Step (4): Analyze the performance of Neural Network during training

Figure 4.13. Neural Network training performance window
Step (5): Analyze Neural Network training States
Figure 4.14. Neural Network training state window

Step (6): Analyze the Neural Network training regression

Figure 4.15. Neural Network training regression
4.4.2 Testing of Neural Network
Steps for testing the Neural Network is following ways:
Step (1): First select the sample data1 input for testing neural network output. Then simulate the
network. Observe the nn3_output.
Figure 4.16. Testing window of Neural Network

Step (2): Observe nn3_output for sampledata1 input
Figure 4.17.Output of neural network for sampledata input

Step (3): Then select another fault condition for testing Neural Network output for example
faultcase07.
Figure 4.18.Testing window of Neural Network for fault case 07
Step (4): Observe the output of Neural Network for faultcase07 input.
Figure 4.19.Output of Neural Network for fault case 07
Step (5): Again select another input data for testing Neural Network output for example
faultcase52.
Figure 4.20. Testing window of Neural Network for fault case 52
Step (6): Observe the output of Neural Network for faultcase52.

Figure 4.21.Output of Neural Network for fault case 52
4.5 Simulation Results
4.5.1 Results from line trap

The frequency response of the designed line trap is shown in figure 4.22 considering the
frequency band of the fault-induced HF transients, which is mostly in the range of 10–100 kHz,
the resonance frequency of the line trap is tuned at 50 kHz.
Figure 4.22. Impedance of line trap at high frequency transients
The main function of the line trap is to present high impedance at the carrier frequency
band while presenting negligible impedance at the power system frequency. The high
impedance is required to reduce the carrier signal attenuation due to the division among the
several transmission lines terminated at the same bus.

Figure 4.22 shows that line trap is present high impedance at the carrier frequency of 100 kHz
while presenting negligible impedance at the power supply frequency 50 Hz.
4.5.2 Results from discrete wavelet transform decomposition

Case (1): External LLL (ABC) fault at 30 km from bus B on transmission line AB.
Figure 4.23 depict the frequency spectrum of combine 3 phase current signal measured
at the relay location decomposed by DWT for case of external fault. In that case, an external
fault occurs at 30 km from bus B on transmission line AB. As can be seen, for external fault,
most of high frequency (HF) components involved in the current signal at frequencies above 20
kHz are not permitted to enter the transmission line under the protection.
Figure 4.23. Frequency spectrum of 3 phase current signal for an external fault at 30km from bus B on transmission
line AB.

Similarly figure 4.24, 4.25 and 4.26 depicts the frequency spectrums of phase A, phase B and
phase C current signal respectively measured at relay location decomposed by DWT for external
fault at 30 km from bus B on transmission line AB.
Figure 4.24. Frequency spectrum of phase A current signal for an external fault at 30km from bus B on
transmission line AB.

Figure 4.25.Frequency spectrum of phase B current signal for an external fault at 30km from bus B on transmission
line AB
Figure 4.26. Frequency spectrum of phase C current signal for an external fault at 30km from bus B on
transmission line AB.
Case (2): Internal LLG (ABG) fault at 75km from bus B on transmission line BE.

Figure 4.27.Frequency spectrum of 3phase current signal for an internal fault at 75km from bus B on line
transmission line B
Figure 4.28. Frequency spectrum of phase A current signal for an internal fault at 75km from bus B on transmission
line BE.

Figure 4.29. Frequency spectrum of phase B current signal for an internal fault at 75km from bus B on transmission
line BE.
Figure 4.30. Frequency spectrum of phase C current signal for an internal fault at 75km from bus B on transmission
line BE

Similarly figure 4.28, 4.29 and 4.30 depicts the frequency spectrums of phase A, phase B and
phase C current signal respectively measured at relay location decomposed by DWT for internal
fault at 75 km from bus B on transmission line BE.
4.5.3 Results from Neural Network

To find out bottlenecks of the HF-based protection technique, initially, the ability of the
extracted feature in discrimination between internal and external faults was evaluated through
training Artificial Neural Network. The operation time of the proposed technique for all the fault
cases is less than 4 msec.
As can be seen, the maximum fault classification accuracy is 93.63% for all the faults
cases. Table 4.1 present some of the fault cases for which the design ANN is not able to
correctly classify the fault. In this table, the fault location is given from bus B. Most of the
misclassified faults are those which occur very close to the remote buses C or E. Indeed, it is
difficult to distinguish between faults occurring close to the remote buses on the TL under
protection or on the next TLs.
Table 4.1. Results from Neural Network for different faults condition
Fault Ground
Fault Location resistance Resistance ANN ANN Target
Sr.No. type (km) (Ω) (Ω) Output error output
1 AG AB(30) 0.01 0.01 -1 -2.92E-07 -1
2 BG AB(30) 0.2 0.01 -0.9998 -0.0002 -1
3 CG AB(30) 0.25 0.01 -1 0 -1
4 ABG AB(30) 0.3 0.01 -1 -5.74E-10 -1
5 BCG AB(30) 0.2 0.01 -1 -1.36E-11 -1
6 ACG AB(30) 2 0.01 -1 0.00E+00 -1
7 ABCG AB(30) 0.3 0.01 -1 0.00E+00 -1
8 ABC AB(30) 5 0.01 -1 0.00E+00 -1
9 AG AB(45) 0.2 0.01 -1 -4.19E-13 -1

10 BG AB(45) 0.7 0.01 -1 -3.09E-06 -1
11 CG AB(45) 2.5 0.01 -1 -3.60E-13 -1
12 ABG AB(45) 8 0.01 -1 -2.37E-07 -1
13 BCG AB(45) 10 0.01 -1 -2.92E-09 -1
14 ACG AB(45) 50 0.01 -1 -7.11E-15 -1
15 ABCG AB(45) 30 0.01 -1 -6.58E-11 -1
16 ABC AB(45) 3.5 0.01 -0.9994 -6.45E-05 -1
17 AG AB(5) 0.22 0.01 -0.9993 -6.95E-05 -1
18 BG AB(5) 0.05 0.01 -0.9996 -3.68E-05 -1
19 CG AB(5) 0.36 0.01 -1 -3.06E-13 -1
20 ABG AB(5) 26 0.01 -0.9991 -1.42E-05 -1
21 BCG AB(5) 3 0.03 -1 -2.55E-09 -1
22 ACG AB(5) 19 0.07 -1 -6.55E-15 -1
23 ABC AB(5) 0.08 0.002 -1 -1.22E-10 -1
24 ABCG AB(5) 0.26 0.8 -1 -5.75E-11 -1
25 AG CD(122) 10 0.9 -1 0 -1
26 BG CD(122) 22 0.01 -1 0 -1
27 CG CD(122) 25 0.02 -1 0 -1
28 ABG CD(122) 37 0.03 -1 -1.11E-16 -1
29 BCG CD(122) 40 0.04 -1 -1.11E-16 -1
30 ACG CD(122) 0.2 0.05 -1 0 -1
31 ABC CD(122) 0.08 0 -0.9999 -1.42E-05 -1
32 ABCG CD(122) 0.88 0.08 -1 0 -1
33 AG CD(143) 29 0.09 -1 -1.91E-12 -1
34 BG CD(143) 0.03 0.1 -1 -2.14E-12 -1
35 CG CD(143) 0.07 0.12 -1 -2.14E-12 -1
36 ABG CD(143) 0.83 0.13 -1 -2.16E-12 -1
37 BCG CD(143) 0.62 0.14 -1 -2.17E-12 -1
38 ACG CD(143) 44 0.15 -1 -1.78E-12 -1
39 ABC CD(143) 33 0 -0.9994 -6.46E-05 -1

40 ABCG CD(143) 42 0.17 -1 -4.41E-12 -1
41 AG CD(154) 21 0.18 -0.9998 -0.000122 -1
42 BG CD(154) 0.21 0.19 -0.9996 -4.18E-05 -1
43 CG CD(154) 0.44 0.2 -0.9996 -4.26E-05 -1
44 ABG CD(154) 0.88 1 -0.9996 -4.28E-05 -1
45 BCG CD(154) 24 2 -0.9993 -0.000166 -1
46 ACG CD(154) 0.98 3 -0.9995 -4.67E-05 -1
47 ABC CD(154) 4 0 -1 -4.82E-10 -1
48 ABCG CD(154) 8 4.5 0.9993 -6.77E-05 -1
49 AG BE(75) 16 5 1 0 1
50 BG BE(75) 32 4.2 1 0 1
51 CG BE(75) 36 0.02 0.9995 5.39E-05 1
52 ABG BE(75) 40 0.03 1 0 1
53 BCG BE(75) 45 0.01 1 0 1
54 ACG BE(75) 48 0.03 0.9992 0.0004838 1
55 ABC BE(75) 49 0 1 9.78E-09 1
56 ABCG BE(75) 0.86 3.5 0.9994 0.0002624 1
57 AG BE(25) 0.77 4.2 1 0 1
58 BG BE(25) 0.89 4.8 1 0 1
59 CG BE(25) 0.46 2.1 1 3.35E-12 1
60 ABG BE(25) 2 2.5 1 0 1
61 BCG BE(25) 3 3.2 1 1.00E-09 1
62 ACG BE(25) 18 1.2 1 3.33E-09 1
63 ABCG BE(25) 7 1.4 1 8.00E-09 1
64 ABC BE(25) 21 0 0.9996 3.51E-05 1
65 AG BE(112) 0.08 0.03 1 0 1
66 BG BE(112) 0.18 0.02 1 0 1
67 CG BE(112) 0.22 0.1 1 0 1
68 ABG BE(112) 31 0.28 1 0 1
69 ACG BE(112) 41 0.29 1 0 1

70 ABCG BE(112) 0.41 0.95 1 0 1
71 ABC BE(112) 0.01 0.08 1 0 1
72 AG BC(63) 0.02 1.4 0.9999 0.0001007 1
73 BG BC(63) 0.03 2.8 0.9998 2.29E-05 1
74 CG BC(63) 0.04 4.5 0.9991 0.0001862 1
75 ABG BC(63) 0.04 4.5 1 6.34E-10 1
76 BCG BC(63) 0.06 4.9 1 3.24E-09 1
77 ACG BC(63) 0.07 0.9 1 8.75E-08 1
78 ABCG BC(63) 0.08 0.8 1 3.30E-10 1
79 ABC BC(63) 0.09 0.01 1 0 1
80 AG BC(54) 0.99 0.02 1 7.64E-10 1
81 BG BC(54) 0.1 0.03 1 0 1
82 CG BC(54) 0.67 0.04 1 0 1
83 BCG BC(54) 0.92 0.05 1 1.78E-15 1
84 ACG BC(54) 0.99 1.2 0.9997 3.39E-05 1
85 ABCG BC(54) 1 1.4 1 0 1
86 ABC BC(54) 7 2.2 1 0 1
87 AG BC(98) 12 2.4 0.9993 0.0003662 1
88 BG BC(98) 16 3.2 0.9992 0.0001787 1
89 CG BC(98) 21 3.4 0.9992 8.19E-05 1
90 ABG BC(98) 24 4.2 0.9996 4.31E-05 1
91 BCG BC(98) 26 4.4 0.9997 3.15E-05 1
92 ACG BC(98) 28 4.8 0.9998 2.09E-05 1
93 ABCG BC(98) 30 0.2 0.9997 3.11E-05 1
94 ABC BC(98) 0.26 0.8 -1 2 1
To increase the separating margin of the trained SVMs in discrimination between the
two classes of internal and external faults, in this paper, the training data set is modified so that
it does not contain fault cases occurring near the remote buses on the three-terminal TL under

protection. In other words, to increase the reliability of the proposed fault classification scheme,
the protected zone was selected to protect 95% of the TL length.
An alternative approach which helps to improve the protective technique performance in

the classification of remote faults is to install line traps on the other TLs terminated at the
remote buses. This way, the HF transients generated by external faults would be attenuated
twice and better results would be obtained for discrimination between faults occurring before or
after the TL remote buses.
In other words, by reducing the protected zone reach, the separating margin between the
signatures of internal and external faults increases which prevents the relay mal-operation for
external faults at the beginning of the next TLs.
4.6 Discussion
The performance of the proposed technique was further investigated using many other
fault cases. It was found that desirably the proposed relay never operates for external faults at
the beginning of the next TLs. However, as designed, some internal faults at the end parts of the
TL and some single-phase-to ground faults occurring at fault resistance may not be detected by
the trained ANN.
For faults occurring at the end part of the TL, the relay installed at the remote bus
operates quickly because the fault is seen as a close-up fault by this relay. Consequently, the
remote circuit breaker (CB) opens to isolate the fault. A supplementary algorithm based on
detecting the open switching of the remote CB can be used to cover the total length of the
protected TL [47]–[49]. It must be considered that if the conventional distance relay is used to
protect the three-terminal TL under study, its first protection zone could not be extended more
than 5–10 km over the tap point (i.e., only 50–54% of the CD transmission line is covered in the
first zone). This is due to the fact that the length of the tapping line is usually small and,
therefore, the reach of the first protection zone of the distance relay should be selected so that it
does not react for faults beyond the bus at the end of tapping line. Mean while, the proposed

algorithm increases the first zone reach to 95% even without using the supplementary algorithm.
This zone coverage increase is of great importance for the protection of TL.
In practice, short-circuit faults often occur when the instantaneous voltage is high.
Therefore, in most conditions, the designed classifier is well capable of identifying internal
faults occurring on the protected TL. For some rare conditions in which an internal fault occurs
at a fault resistance, similar to the other HF-based protection techniques, the designed classifier
may not be able to identify the fault because the generated HF transients are too small.

CHAPTER 5
=================================================================
CONCLUSION AND FUTURE SCOPE

=================================================================
5.1 Conclusion
This dissertation presented a non-communication protection technique for multiterminal

transmission lines. The proposed technique uses the fault-generated high–frequency transients to
discriminate between the internal and external faults. To minimize the effect of variable power
system parameters on the proposed algorithm performance, appropriately designed line traps
were installed at the TL ends. The ANN technique was employed to classify the faults according
to the fault currents HF components decomposed by the wavelet transform.
The performed simulation studies show that the ANN classifier provides the most
accurate result for discrimination between the internal and external faults. Using this algorithm,
the reach of the first protected zone can be extended so that only a very small portion at the
remote-end side is not covered by the proposed relay. For this small portion, a supplementary
algorithm based on the open switching of the remote CB can be used. This is done to maximize
the separating margin between the two classes of internal and external faults and provide the
required security for the protection scheme. The ANN classifier is able to reliably identify the
internal faults up to 96% of the TL length. The proposed algorithm provides high noise
immunity and works well even under changes in the power system parameters and conditions
due to the installation of appropriately designed line traps at the ends of the protected TL.
5.2 Future scope

In this work, types of external and internal faults are classified for multiterminal
transmission line. Further, protection of multiterminal transmission line improve by using ANN
and line traps.
Based on this dissertation, the following area of work is suggested for further exploration:

 This work can be extended for classification of three phase faults on double circuit series
compensated line and FACT based transmission line.
 It is proposed to find the location of series faults i.e. open conductor faults on double
circuit series compensated line.
It is proposed to find the fault zone, fault classification and fault identification on
multiterminal transmission line and double circuit transmission line.

REFERENCES
[1] Mohammad, A.B., “Travelling Waves for Finding the Fault Location in Transmission
Lines,” Journal Electrical and Electronic Engineering, vol. 1, no. 1, pp.1-19, 2013.
[2] R. K. Aggarwal and A. T. Johns, “Digital differential relaying scheme for teed circuits
based on voltage and current signal comparison,” Proc. Inst. Elect. Eng., Gen., Transm.
Distrib., vol. 137, no. 6, pp. 414–423, Nov. 1990.
[3] R. K. Aggarwal and A. T. Johns, “The development of a new high speed 3-terminal line
protection scheme,” IEEE Trans. PowerDel., vol. PWRD-1, no. 1, pp. 125–134, Jan.
1986.
[4] H. Gao and P. A. Crossley, “A new current differential protection scheme for teed
transmission lines,” presented at the Power Eng. Soc. Gen. Meeting, Montreal, QC,
Canada, Oct. 2006.
[5] J. Izykowski,E.Rosolowski,M.M. Saha,M.Fulczyk, and P. Balcerek, “A fault-

locationmethod for application with current differential relays of three-terminal lines,”
IEEE Trans. Power Del., vol. 22, no. 4, pp. 2099–2106, Oct. 2007.
[6] J. Daniel, R. K. Aggarwal, and A. T. Johns, “Three terminal line protection based on a
superimposed component impedance relay,” Proc. Inst. Elect. Eng., Gen., Transm.
Distrib., vol. 140, no. 6, pp. 447–454, Nov. 1993.
[7] V. Pathirana and P. G. McLaren, “A hybrid algorithm for high speed transmission line
protection,” IEEE Trans. Power Del., vol. 20, no. 4, pp. 2422–2428, Oct. 2005.
[8] W. Chen, O. P. Malik, X. Yin, D. Chen, and Z. Zhang, “Study of wavelet-based ultra
high speed directional transmission line protection,” IEEE Trans. Power Del., vol. 18,
no. 4, pp. 1134–1139, Oct. 2003.
[9] D. J. Zhang, Q. Henry Wu, Z. Q. Bo, and B. Caunce, “Transient positional protection of
transmission lines using complex wavelets analysis,” IEEE Trans. Power Del., vol. 18,
no. 3, pp. 705–710, Jul. 2003.
[10] P. Jafarian andM. Sanaye-Pasand, “A traveling-wave-based protection technique using

wavelet/PCA analysis,” IEEE Trans. Power Del., vol. 25, no. 2, pp. 588–599, Apr. 2010.
[11] N. Zhang and M. Kezunovic, “Transmission line boundary protection using wavelet
transformand neural network,” IEEE Trans. Power Del., vol. 22, no. 2, pp. 859–869,
Apr. 2007.

[12] T. Johns, R. K. Aggarwal, and Z. Q. Bo, “Non-unit protection technique for EHV
transmission systems based on fault generated noise part 1: Signal measurement,” Proc.
Inst. Elect. Eng., Gen., Transm. Distrib., vol. 141, no. 2, pp. 133–140, Mar. 1994.
[13] R. K. Aggarwal, A. T. Johns, and Z. Q. Bo, “Non-unit protection technique for EHV
transmission systems based on fault generated noise part 2: Signal processing,” Proc.
Inst. Elect. Eng., Gen., Transm. Distrib., vol. 141, no. 2, pp. 141–147, Mar. 1994.
[14] Z. Q. Bo, “A new non-communication protection technique for transmission lines,”

IEEE Trans. Power Del., vol. 13, no. 4, pp. 1073–1078, Oct. 1998.
[15] D. Xingli, G. Yaozhong, and D. Xinzhou, “A wavelet and traveling waves based non-
communication high speed transmission line protection,” Autom. Elect. Power Syst.,
vol. 10, 2001.
[16] Sharma, R., Ahmad, A. and Shailendra, K. S., “Protection of Transmission Lines
using Discrete Wavelet Transform,” International Journal of Innovative Technology and
Exploring Engineering (IJITEE), vol. 3, Issue-1, June, 2013.
[17] Osman, A.H. and Malik, O.P., “Protection of Parallel Transmission Lines Using Wavelet
Transform” IEEE Transactions on power delivery, vol. 19, no. 1, 2004.
[18] Ashok, V., Bangarraju, K. G. V. S. and Murthy, V.V.N., “Identification and

Classification of Transmission Line Faults Using Wavelet Analysis,” ITSI Transactions
on Electrical and Electronics Engineering, vol. 1, no. 1, pp. 117-122,
2013.
[19] Pandya, V. J. and Kanitkar, S.A., "A novel unit protection scheme for protection of
series compensated transmission line using wavelet transform," Power Engineering
Conference, 2007.
[20] Perez, F.E., Orduna, E. and Guidi, G., “Adaptive wavelets applied to fault classification
on transmission lines,” IET Generation, Transmission and Distribution, vol. 5, no. 7, pp.
694–702, 2011.
[21] Roshni, U., Niranjan, V., Prakash, C.D. and Srinivas, R., “Location of Faults In
Transmission Line Using Fast Fourier Transform And Discrete Wavelet Transform In
Power Systems,” Undergraduate Academic Research Journal (UARJ), vol. 1, Issue-1,
pp. 84-86, 2012.
[22] Solanki, M. and Song, Y.H., “Transient Protection of EHV Transmission Line Using
Discrete Wavelet Analysis,” Power Engineering Society General Meeting, IEEE, vol. 3,
pp. 1868-1873.

[23] Megahed, A.I., Monem Moussa, A. and Bayoumy, A.E., "Usage of wavelet transform in
the protection of series-compensated transmission lines," Power Delivery, IEEE
Transactions on, vol. 21, no. 3, pp. 1213-1221, July, 2006.
[24] Samapth, B. and Vijayabhaskar, K., “Wavelet Based Protection Schemes for TEED
Tramission Circuits,” International Journal of Engineering Research and Applications
(IJERA), vol. 2, no. 6, pp.931-940, Nov. - Dec., 2012.
[25] Nidhi, B., Kumar, R. and Sinha, A., “Neural Network and Fuzzy Logic Based Protection
of Series Compensated Double Circuit Transmission line” International Journal of
Engineering Research and Applications (IJERA), vol. 3, no. 1, pp.1545- 1551, 2013.
[26] Andrade, L.D. and de Leao, T.P., "Impedance-based fault location analysis for
transmission lines," Transmission and Distribution Conference and Exposition (T&D),
2012 IEEE PES, pp. 1-6, May, 2012.
[27] Saravanababu, K., Balakrishnan, P. and Sathiyasekar, K., “Transmission Line Faults
Detection, Classification, and Location Using Discrete Wavelet Transform,”
International Conference on Power, Energy and Control (ICPEC), pp. 233-238, 2013.
[28] Yadav, A. and Swetapadma, A., “Improved first zone reach setting of artificial neural
network-based directional relay for protection of double circuit transmission lines” IET
Generation, Transmission and Distribution., vol. 8, Issue-3, pp. 373-388, 2014.
[29] Youssef, O.A.S., “Fault classification based on wavelet transform,” Transmission and
Distribution Conference and Exposition IEEE/PES, vol. 1, pp. 531-536, 2001.
[30] Reddy, R.B., Kumar, M.V., Suryakalavathi, M. and Prasanth Babu, C., "Fault detection,
classification and location on transmission lines using wavelet transform," Electrical
Insulation and Dielectric Phenomena, pp.409-411, Oct., 2009.
[31] Robi Polikar, “The Engineer’s ultimate guide to Wavelet analysis: The Wavelet
Tutorial”, http://users.rowan.edu/~polikar/WAVELETS/WTtutorial.html
[32] Philippe Crochat and Daniel Franklin, “Back-Propagation Neural Network

Tutorial”,Internet“http://ieee.uow.edu.au/~daniel/software/libneural/BPN_tutorial/BPN_
English/BPN_Eng lish/BPN_English.html”, April 2003.
[33] R.Aggarwal, Y.Song, “Artificial neural network in power systems, I. General

introduction to neural computing”, Power Engineering Journal, Vol 11, No.23 pp. 129-
134, June 1997.
[34] S. M. Brahma, “New fault-location method for a single multiterminal transmission line
using synchronized phasor measurement,” IEEE Trans. Power Del., vol. 21, no. 3, pp.
1148–1153, Jul. 2006.

[35] IEEE Guide for Power Line Carrier Applications, An American National Standard,IEEE,
ANSI/IEEE Standard 643–1980, 1980.
[36] Line Traps for Ac power Systems, IEC Standard 60353 Ed. 2.0 b, , 1989.
[37] “Line Traps, Product Catalog” Areva T&D Ltd. Co., India. [Online]. Available:
http://arevausitr.com
[38] Youssef, O.A.S., “Fault classification based on wavelet transform,” Transmission and
Distribution Conference and Exposition IEEE/PES, vol. 1, pp. 531-536, 2001.
[39] Shaaban, S.A. and Hiyama, T., “Transmission Line Faults Classification Using Wavelet
Transform,” 14th International Middle East Power Systems Conference (MEPCON’10),
Cairo University, Egypt, pp. 532-537, December, 2010.
[40] Saravanababu, K., Balakrishnan, P. and Sathiyasekar, K., “Transmission Line Faults
Detection, Classification, and Location Using Discrete Wavelet Transform,”
International Conference on Power, Energy and Control (ICPEC), pp. 233-238, 2013.
[42] Martín, F. and Aguado, J.A., “Wavelet-Based ANN Approach for Transmission Line
Protection,” IEEE Transactions on power delivery, vol. 18, no. 4, pp. 1572-1574, Oct.,
2003.
[43] Megahed, A.I., Monem Moussa, A. and Bayoumy, A.E., "Usage of wavelet transform in
the protection of series-compensated transmission lines," Power Delivery, IEEE
Transactions on, vol. 21, no. 3, pp. 1213-1221, July, 2006.
[44] Chanda, D., Kishore, N.K. and Sinha, A.K., “Application of Wavelet Multi-resolution
Analysis for Classification of Faults on Transmission lines” Conference on convergent
Technologies for the Asia-Pacific Region, vol. 4, pp.1464 – 1469, 2003.
[45] Roshni, U., Niranjan, V., Prakash, C.D. and Srinivas, R., “Location of Faults In
Transmission Line Using Fast Fourier Transform And Discrete Wavelet Transform In
Power Systems,” Undergraduate Academic Research Journal (UARJ), vol. 1, Issue-1,pp.
84-86, 2012.

[47] A.Sharafi, M. Sanaye-Pasand, and P. Jafarian, “Non-communication protection of
parallel transmission lines using breakers open-switching traveling-waves,” IET Gen.,
Transm. Distrib., 2011.
[48] Z. Q. Bo, “Adaptive non-communication protection for power lines BO scheme 1- the
delayed operation approach,” IEEE Trans. Power Del., vol. 17, no. 1, pp. 85–91, Jan.
2002.
[49] Z. Q. Bo, “Adaptive non-communication protection for power lines BO scheme 2- the
instant operation approach,” IEEE Trans. Power Del., vol. 17, no. 1, pp. 92–96, Jan.
2002.
WEBSITES VISITED
[1] www.ieeesystem journals.com

[2] www.IJCSI.org
[3] www.analog.com/library/Buck-Boost
[4] www.wpi.edu/education
[5] www.researchgate.net
[6] www.iajes/journals.com
[7] www.ijser.com
[8] www.youtube.com
[9] www.electrical4u.com

APPENDIX A: Database for Neural Network
Figure A.1. Model for generating the dataset of Neural Network
Table A.1 Dataset for training the Neural Network
Sr No. d1 d2 d3 d4 d5 Target
1 14.3168 8.5336 4.5292 2.8019 -0.557 -1
2 16.5785 11.869 5.0882 3.6668 -1.411 -1
3 25.8944 15.4392 8.1167 4.9237 -1.5395 -1
4 14.0217 8.6217 4.2892 2.6613 -0.5009 -1
5 23.5487 14.0459 7.2315 4.3085 -1.1615 -1
6 22.2599 11.8045 6.9637 3.7533 -0.5973 -1
7 21.4893 11.2049 6.3991 3.4425 -0.2273 -1
8 21.0086 11.0176 6.4235 3.3416 -0.3216 -1
9 21.3301 13.7054 6.7616 4.4767 -1.4705 -1
10 21.0248 13.6851 6.5606 4.4529 -1.5055 -1

11 21.997 14.1605 6.9659 4.6104 -1.5635 -1
12 20.5368 13.1966 6.5052 4.3081 -1.3903 -1
13 20.8427 13.5367 6.588 4.3991 -1.4881 -1
14 21.6217 13.8657 6.8476 4.5162 -1.5196 -1
15 20.5021 13.1912 6.4887 4.2969 -1.4203 -1
16 20.3113 13.1044 6.4273 4.2655 -1.414 -1
17 19.3948 12.2751 6.1663 4.0345 -1.2409 -1
18 19.2392 12.8775 6.0865 4.1866 -1.5323 -1
19 18.7945 12.9711 5.9305 4.1952 -1.6461 -1
20 13.5009 8.3008 4.2858 2.7533 -0.538 -1
21 9.0099 7.9626 2.7382 2.4706 -1.2989 -1
22 20.961 12.5182 6.6243 4.0317 -1.3066 -1
23 5.618 3.6334 1.7925 1.2021 -0.373 -1
24 4.0441 3.0741 1.2849 1.0078 -0.4285 -1
25 23.4751 14.6533 7.4172 4.7493 -1.5078 -1
26 23.473 14.6527 7.4168 4.7493 -1.5077 -1
27 23.473 14.6526 7.4166 4.7493 -1.5077 -1
28 23.4625 14.6495 7.4138 4.7487 -1.5078 -1
29 23.4623 14.6494 7.4138 4.7487 -1.5078 -1
30 23.4728 14.6522 7.4164 4.7488 -1.5079 -1
31 21.375 13.9314 6.7754 4.5464 -1.5575 -1
32 23.4726 14.6521 7.4163 4.7488 -1.5079 -1
33 22.3228 14.2592 7.0678 4.6452 -1.5059 -1
34 22.3305 14.2616 7.0701 4.646 -1.5039 -1
35 22.3312 14.2619 7.0703 4.6461 -1.5039 -1
36 22.3304 14.2615 7.0701 4.646 -1.5039 -1
37 22.3309 14.2618 7.0702 4.6461 -1.5039 -1
38 22.3156 14.2569 7.0657 4.6444 -1.5071 -1
39 21.3459 13.9188 6.767 4.5432 -1.5578 -1
40 22.3165 14.2371 7.066 4.6445 -1.5069 -1

41 19.4843 13.1326 6.1687 4.2747 -1.5753 -1
42 19.3717 13.0941 6.1345 4.2646 -1.5768 -1
43 19.373 13.0947 6.1349 4.2648 -1.5769 -1
44 19.3791 13.0966 6.1368 4.2652 -1.5768 -1
45 19.5116 13.1419 6.177 4.2771 -1.5749 -1
46 19.3899 13.1004 6.14 4.2662 -1.5767 -1
47 21.2884 13.7924 6.7478 4.5029 -1.5035 -1
48 19.4345 13.1155 6.1535 4.2701 -1.5761 -1
49 15.7211 11.6549 5.0153 3.8508 -1.6108 1
50 13.424 10.2736 4.319 3.453 -1.4104 1
51 20.2273 13.3952 6.4012 4.3622 -1.5499 1
52 12.856 9.5946 4.162 3.2539 -1.2531 1
53 17.6368 11.9714 5.6084 3.939 -1.4074 1
54 20.1171 12.8863 6.3818 4.2128 -1.395 1
55 17.3704 11.2119 5.5714 3.7066 -1.1839 1
56 14.4579 9.7452 4.6319 3.2463 -1.0948 1
57 16.7474 11.3131 5.3536 3.7633 -1.2448 1
58 17.4382 12.393 5.5736 4.1364 -1.3708 1
59 19.1055 12.9706 6.0292 4.2131 -1.1564 1
60 18.53 12.7274 5.8783 4.1877 -1.4057 1
61 20.0396 13.2735 6.2777 4.2631 -1.3668 1
62 17.5684 11.7623 5.5685 3.8405 -1.3776 1
63 19.6264 12.9172 6.1461 4.1349 -1.4187 1
64 19.2573 12.609 6.0319 4.0199 -1.4802 1
65 16.3192 12.051 5.1843 3.9418 -1.6106 1
66 16.3188 12.0509 5.1842 3.9419 -1.6101 1
67 16.319 12.051 5.1843 3.9422 -1.6094 1
68 16.6494 12.1846 5.3035 4.0272 -1.4485 1
69 16.8955 12.2746 5.384 4.0651 -1.4101 1
70 16.3152 12.0504 5.1837 3.9437 -1.6033 1

71 16.3199 12.0512 5.1845 3.9419 -1.6107 1
72 25.1474 16.6126 7.9665 5.4234 -1.9029 1
73 21.7545 13.7212 7.0426 4.655 -1.4426 1
74 24.0553 14.8741 7.6853 4.9509 -1.3369 1
75 24.984 16.6492 8.03 5.5531 -2.0813 1
76 22.344 14.0625 7.303 4.8726 -1.3648 1
77 25.0332 16.652 7.9598 5.4968 -1.7684 1
78 25.1802 16.8161 8.1582 5.7125 -1.9443 1
79 23.0021 15.9298 7.3933 5.3112 -2.0909 1
80 18.858 13.395 6.0041 4.4121 -1.7857 1
81 23.1384 15.5778 7.3986 5.1343 -1.9238 1
82 13.0484 10.8878 4.1773 3.6631 -1.5689 1
83 17.475 13.3255 5.6598 4.5217 -1.8589 1
84 13.2247 11.8427 4.2461 3.9892 -2.0431 1
85 19.1527 15.3371 6.2142 5.1863 -2.5769 1
86 20.1999 15.6181 6.5533 5.3292 -2.6634 1
87 19.7428 13.2472 6.2629 4.3315 -1.5665 1
88 19.7524 13.2512 6.2664 4.3334 -1.5655 1
89 19.7644 13.2562 6.2706 4.3356 -1.5645 1
90 19.7741 13.2601 6.274 4.3374 -1.5637 1
91 19.7795 13.2624 6.2759 4.3383 -1.5632 1
92 19.7864 13.2652 6.2783 4.3395 -1.5627 1
93 19.779 13.2622 6.2757 4.3383 -1.5632 1
94 21.3547 13.8177 6.7665 4.5088 -1.4995 1

APPENDIX B: Three phase current and voltage waveform for external and internal fault
at different transmission line buses.
Case (1): External LLL (ABC) fault at 30 km from bus B on transmission line AB.
Figure B.1. 3phase voltage and current at bus A for External fault LLL(ABC) at 30km from bus B on line AB
Figure B.2. 3phase voltage and current at bus B for External fault LLL(ABC) at 30km from bus B on line AB

Figure B.3. 3phase voltage and current at bus C for External fault LLL(ABC) at 30km from bus B on line AB
Figure B.4. 3phase voltage and current at bus D for External fault LLL(ABC) at 30km from bus B on line AB

Figure B.5. 3phase voltage and current at bus E for External fault LLL(ABC) at 30km from bus B on line AB
Case (2) : Internal LLG (ABG) fault at 75 km from bus B on transmission line BE.
Figure B.6. 3phase voltage and current at bus A for internal fault LLG(ABG) at 75km from bus B on line BE

Figure B.7. 3phase voltage and current at bus B for internal fault LLG(ABG) at 75km from bus B on line BE
Figure B.8. 3phase voltage and current at bus C for internal fault LLG(ABG) at 75km from bus B on line BE

Figure B.9. 3phase voltage and current at bus D for internal fault LLG(ABG) at 75km from bus B on line BE
Figure B.10. 3phase voltage and current at bus E for internal fault LLG(ABG) at 75km from bus B on
line BE

Chapters

Uploaded by

Copyright:

Available Formats

Chapters

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapters

Uploaded by

Copyright:

Available Formats

CHAPTER 1

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 1

1.2 Faults in Overhead Transmission Lines

1.2.1 Nature and Causes of Faults

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 2

Table 1.1. Types of faults [1]

Types of faults Symbol % Occurrences Severity

Line to Ground L-G 75 – 80% Very less severe

Line to Line L–L 10 – 15% Less Severe

Double Line to Ground L-L-G 5 – 10% Severe

Three Phase 3-ø 2 -5% Very Severe

1.2.2 Effects of Faults

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 3

Most high-voltage transmission lines are protected by distance relays. Compared to

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 4

1.4 Difficulty in protection of multiterminal transmission line

The fault-induced high-frequency (HF) transients contain extensive information about

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 5

1.5 Outline of dissertation

Chapter-2: It explains the literature review pertaining to protection of multiterminal

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 6

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 7

For the power to be efficiently distributed to different locations, it is necessary to

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 8

Probability based technique of Bayesian linear discrimination can also be used to

For the fault classification and boundary protection of series-compensated transmission

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 9

An impedance based calculation method to locate fault on transmission line is of

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 10

An algorithm based on discrete wavelet transform is developed with C programming

2.2 Use of Neural Network for fault classification

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 11

A typical Back Propagation Neural Network is a non-linear regression technique which

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 12

2.3 Multiterminal transmission line protection techniques

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 13

Recently, some non-communication-based techniques for the protection of two-terminal

2.4 Objective of present work

This dissertation presents a new non-communication protection technique for

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 14

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 15

3.1.2 Overview of system

So, a signal in time-domain is plotted, representation of signal in time-amplitude form is

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 16

A wavelet-based signal-processing technique is an effective tool for power system

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 17

3.1.2.1.1 Discrete Wavelet Transform

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 18

Figure 3.2. Discrete Wavelet Transform

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 19

cA1 (t) = ∑ V(t) . Ld (k − 2t)

cD1 (t) = ∑ V (t) . Hd(k − 2t)

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 20

3.1.2.1.3 Working of Wavelet

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 21

3.1.2.2 Artificial Neural Network for fault classification

Roughly speaking, a neural network is a collection of artificial neurons. An artificial

Department of Electrical Engineering, S.S.C.E.T, Bhadrawati. 22

Figure 3.5. First model for artificial neuron