김지수
(Ji-Soo Kim)
1iD
송진솔
(Jin-Sol Song)
1iD
신광수
(Gwang-Su Shin)
1iD
김호영
(Ho-Young Kim)
1iD
김철환
(Chul-Hwan Kim)
†iD
-
(Dept. of Electrical and Computer Engineering, Sungkyunkwan University, Korea.)
Copyright © The Korean Institute of Electrical Engineers(KIEE)
Key words
Neural network, Fault contribution, Intelligent Protective Method, Microgrid, Signal processing
1. Introduction
The conventional power system has a radial structure, owing to which, the impedance
increases with the distance between the point at which the fault occurs and the main
power source; consequently, the fault current decreases. Accordingly, depending on
the difference in the magnitude of the fault current, the operating time of the protective
relay can be set; as a result, the main protective device, which is the closest protective
device where the fault has occurred, can be set to operate first. If the operation
of the main protective devices fails, then the backup protective device, which is
the second closest protective device to the fault point, is set for operation. This
series of protective processes is called protective coordination, which can be performed
smoothly because the fault current decreases as the fault point moves away from the
main power source.
However, this approach is not suitable in microgrids (MGs) owing to the presence of
distributed generation (DG) systems such as RESs and energy storage systems (ESSs).
MGs, which can be self-sufficient regarding generation-load balance in small areas,
are useful as they are capable of integrating the RESs into the power system. In other
words, an MG is independent and a next-generation power grid that combines RESs such
as photovoltaics (PVs), wind power and ESSs. Furthermore, owing to the inclusion of
various power sources in the MGs, the fault current is not reduced even if the distance
between fault point and main power source is farther away. On the contrary, various
problems are caused by contribution of individual power sources to the fault current.
Owing to these structural problems in MGs, the conventional protective system becomes
ineffective (1).
Therefore, numerous studies are aimed at investigating solutions to these problems
of MG protection. In particular, several studies have been conducted to prevent problems
caused by fault contribution and islanding operation of RESs. Recently, methods to
protect MGs through pattern analysis using machine learning have been proposed (3). However, there are few review papers regarding each of these topics and how the
studies are being conducted. This paper describes the challenges related to protection
in MGs, which are different from those in the existing power system, and discusses
the recent research trends and different approaches available for solving the related
problems. The following are the main contributions of this paper:
(1) Comprehensive review of various challenging issues and methods of conventional
MG protection.
(2) Comprehensive review of intelligent protective method (IPM) using machine learning
to protect MG.
(3) Comparison of various IPMs based on their principles of operation, advantages,
and disadvantages.
2. Challenging issues of Microgrid Protection
The challenging issues concerning protection of the MG can be classified. Problems
arise due to the fault contribution of RESs and the reverse current when RESs are
installed in the middle or at the end of the feeder. Moreover, the change in the structure
of the MG and islanding operation of RESs also cause some issues. This section identifies
those issues that are related to the MG protection and studies the recent research
trends and possible solutions (2).
2.1 Fault contribution of DG
When some of the RESs are connected to the power system in parallel, the total system
impedance decreases, causing an increase in the total fault current. In addition,
fault current is generated from various power sources because RESs are distributed
throughout power system. Therefore, the fault characteristics of MGs are different
from those of the existing power systems, and as a result, several problems may occur;
these are discussed as follows.
2.1.1 Blinding protection
When a fault occurs in a feeder in which an RES is installed, the fault contribution
of the main source is reduced owing to the fault contribution of the RESs. As a result,
the overcurrent relay that was previously functional may not operate. There are two
ways to solve the problem of blinding protection. First approach is to reset the methods
of protective relay using optimization algorithms considering various scenarios (3). The second is to reduce the fault contribution of RESs by using fault current limiter
so that the operation of the protective relay is untroubled (4).
2.1.2 Sympathetic tripping
The blinding protection is caused by fault contribution of RESs in the fault phase,
whereas sympathetic tripping a result of the fault contribution of RESs in the healthy
feeder. When a fault occurs, a fault contribution from RESs flows through the healthy
feeder. As a result, an overcurrent relay in the healthy feeder, which should not
operate, can operate (5). Various solutions have been proposed to prevent the occurrence of this phenomenon;
correspondingly, there exist relay resetting methods such as counter measurement of
blinding protection (6), and methods for analyzing fault characteristics of healthy and faulty signals through
signal processing techniques (7). In addition, there is also an approach that does not block the reverse fault current
by including a direction detection method in an existing overcurrent relay (8).
2.1.3 Problems of auto-reclosing
80% of faults in a power system are temporary faults with a lifetime of several milliseconds.
In case of such a temporary fault, effective protection can be performed through a
recloser. In the conventional power system, after the recloser is opened, the part
of power system blocked by the recloser is in a no-voltage state. However, in the
case where the RESs are injecting power into blocked part of the power system, the
blocked part is not in a no-voltage state even when the recloser opens. Therefore,
during reclosing, an asynchronous situation may occur at both ends of the recloser,
which may cause transients, such as overcurrent and overvoltage surges (9). This situation should be prevented, and an algorithm for determining a temporary
fault and a permanent fault is currently being developed in several studies (10). In addition, studies on performing synchronization naturally are also being conducted
(11).
2.1.4 Fault characteristics of inverter-based DG
Inverter-based RESs and synchronous-based RESs have different fault characteristics.
In general, inverter-based RESs have different fault response times compared to those
of a synchronous machine; therefore, it is necessary to consider these characteristics
when analyzing the fault before establishing the protective system. Therefore, in
the virtual inertia analysis (12), the fault characteristics of the inverter based RESs are analyzed by dividing them
into a voltage source or current source (13).
2.2 Reverse power flow
In the conventional power system, the load current flows only in one direction. However,
when the RESs are connected to the MG, the load current can exhibit bidirectionality
(14). Therefore, the current flowing opposite to the direction of the existing load current
is called reverse current; this may cause the following problems.
2.2.1 Misoperation of non-directional devices
Algorithms for determining the directionality are not basically included in existing
protective devices, and these existing protective devices that determine only the
current magnitude to perform protection may cause various malfunctions when reverse
current flows through them. In particular, it may cause malfunction of the sectionalizer,
which separates power systems. This is because a sectionalizer can only separate power
systems in the event of no-voltage state (15). This problem is solved by introducing an algorithm for determining current directionality
of existing protective relay (16). In addition, by controlling the output of the RESs, it is possible to prevent reverse
current flow (17); moreover, the output of the RESs may not exceed the load owing to the maintenance
of supply and demand balance by controlling the demand response (18).
2.2.2 Overvoltage and over-ampacity in feeder
In the conventional power system, power is supplied from the substation, and the voltage
drop due to line impedance increases the direction towards the end the feeder. However,
as RESs are installed in the middle or at the end of the feeder line, overvoltage
may occur near the installation location of RESs (19). In MGs, there are various voltage regulation devices to adjust the voltage within
the allowable range. Furthermore, the connection of RESs through smart inverter has
its own voltage regulating function such as Volt-Var function. A typical method for
voltage regulation, uses the line drop compensation (LDC) of on load tap changer (OLTC)
installed in a substation. OLTC predicts the voltage drop based on the amount of load
current from the substation and decides whether to compensate for the voltage. However,
as the RESs are connected to the power system, the value of the load current from
the main source decreases; consequently, the OLTC may compensate the voltage incorrectly.
Therefore, coordination algorithms for voltage regulation among OLTC, shunt capacitor
and smart inverters are being studied (20). In addition, studies on hosting capacity, i.e., the maximum number of RESs that
will not cause an overvoltage, are being actively conducted (21).
2.3 Changing of microgrid configuration
The advantage of an MG is that it can flexibly change the system structure according
to various situations. This advantage helps to maintain reliability of power system
and minimizes damage in case a fault occurs. However, changes in the structure of
the MG from the perspective of the protective system will lead to new protective coordination
concerns. The following sections describe the various problems in the implementation
of protective systems caused by changes in grid structure.
2.3.1 Frequent change in configuration
Depending on the structure of the specified MG, the maximum load and fault currents
at each protective device are calculated by the line impedance (22). However, if the structure of the MG is changed, the magnitude of these currents
will vary as well; moreover, frequent structural changes of the MG due to the automatic
power system would make it impossible to perform protection using a uniform protective
system. In order to solve these problems, research is underway to set relay setting
values that can be commonly used even when the system configuration is changed (23). Moreover, research on an automatic resetting method of protective relays when the
system configuration is changed, is actively progressing (24). Finally, a method of constructing a protective system regardless of the structure
of an MG, using a traveling wave is also being studied (25).
2.3.2 Varied fault level in dual mode
MGs can be divided based on modes of operation: islanding mode MG and grid-connected
mode MG. Naturally, magnitude of the fault current varies depending on the presence
or absence of the main power (26). Therefore, in each case, it is necessary to set the protective relays’ setting values.
In this case, there is a method of setting the relay using an optimization algorithm
in a method similar to that discussed earlier in Section 2.3.1 (27). Moreover, it is also possible to determine the current mode of the MG through signal
processing techniques to apply the corresponding protective relay setting values (28).
2.4 Islanding
An islanding operation refers to a phenomenon in which RESs supply power to a power
system grid while being separated from the main power source. Unintentional islanding
operation may cause system instability and lead to accidents to humans; therefore,
its detection and resolution is paramount (29). Methods for islanding detection can be classified into three types: passive methods,
active methods, and methods using communication.
3. Comprehensive Review of Intelligent Protective Method
Most recently, MG protection using machine learning became a trending research topic.
When a fault occurs, the magnitude of the voltage, current, and other power quality
fluctuates from the normal state, and the pattern thus obtained is different depending
on the type of the fault. The method of protecting MG through each machine learning
is similar. Each fault characteristic (voltage, current, energy, ntropy, etc.) shows
different characteristics according to the fault location and fault type. Each machine
learning method learns the corresponding characteristics and when an actual fault
occurs, analyzes the learned characteristics to classify the location and type of
the fault. However, there are advantages and disadvantages to each machine learning
method, and the comparison has proceeded in Section 4.
3.1 ANN-SVM method
An artificial neural network (ANN) is a statistical learning algorithm inspired by
neural networks of biology that are applied in machine learning. It refers to the
entire model with problem solving ability to perform machine learning. In general,
it receives the voltage and current data as input for each fault situation and learns
which voltage and current occur when any fault situation occurs. Moreover, support
vector machine (SVM) is one of the areas of machine learning for pattern recognition
and data analysis. Given a set of data belonging to either category, the SVM algorithm
creates a non-stochastic binary linear classification model that determines which
category the new data belong to; this classification is based on the given data set.
The created classification model is expressed as a boundary in the space where the
data are mapped. The SVM algorithm finds the boundary with the largest width. Moreover,
SVM can be used in both linear and nonlinear classification. In order to perform non-linear
classification, it is necessary to map the given data into a high-dimensional feature
space; kernel tricks are used to do this efficiently. Using this method, the input
data for the fault are learned, and when a fault occurs, the situation determined
with the highest probability is recognized as the current situation (30).
3.2 Sparse autoencoder and deep neural network
Neural networks have two types learning methods. One is called supervised learning,
where learning is performed is a state where both the input value and target value
of the data are given. In contrast, learning that finds the characteristics of data
in a state where only the input value of data is given is called non-supervised learning.
The SVM method belongs to supervised learning, whereas the sparse autoencoder (SAE)
belongs to unsupervised learning. A proposed SAE based deep neural network scheme
has the ability to automatically learn features from the unlabeled dataset consisting
of instantaneous values of voltage and current signals without specifically extracting
attributes for different fault cases. SAE is a neural network that simply copies inputs
to outputs. It appears to be a simple neural network, but it is made into a complicated
neural network by constraining the network in various ways. For example, the number
of neurons in the hidden layer is smaller than the number of neurons in the input
layer to compress the data or add noise to the input data to restore the original
input. Owing to the effective performance of SAE in discovering the system structure
information from input dataset with reduced computation effort, it has been successfully
implemented in various classification applications (31).
3.3 Hilbert transform and machine learning techniques
Hilbert transform (HT) is utilized to calculate various functions for the MG fault
classification process. Through HT, it is possible to derive various functions (energy,
entropy, etc.) as required in various conditions in the power system, which can be
learned to determine the fault situations. Input data are required to use HT. In general,
voltage or current is used to obtain the input data through decomposition into a mono
component signal called intrinsic mode function (IMF) through empirical mode decomposition
(EMD); finally, this IMF value is used as input data of HT. Various features that
are obtained through HT are learned through various machine learning techniques such
as SVM, which further become indicators for determining a fault situation. The features
that can be obtained through HT, include maximum and minimum values related to the
size of HT, root-mean square, energy, standard deviation, skewness, kurtosis, and
entropy (32).
3.4 Convolutional neural Network
The input data of a general ANN is limited to a one-dimensional (array) form. However,
if multiple items of input data are required, multiple dimensions must be compressed
into single dimension; information could be lost during this compression process.
As a result, ANN has limitations in extracting and learning features and increasing
accuracy owing to lack of information. Therefore, convolutional neural network (CNN)
is proposed as a model that can be trained to wolve this problem while maintaining
information. CNN can be divided into the parts that extract the features and the parts
that classify. The feature extraction area is composed of multiple layers of the convolution
layers and pooling layers. A convolution layer is an essential element that reflects
the activation function after applying a filter to the input data. In contrast, the
pooling layer is applied to the feature produced by the convolutional layer, and is
an optional layer. Using the CNN as described above, it is possible to perform more
effective fault diagnosis by receiving individual data on three-phase current or voltage
(33).
Table 1 Comparison of various IPMs (Merits)
|
Type
|
Merits
|
Demerits
|
|
a)
|
SVM
|
1) Data analysis is considerable fast.
2) It is also applicable when it is difficult to classify data through a linear model.
|
1) The more the number of samples, the slower the speed and the larger the memory
allocation; this ultimately decreases the performance.
2) It is difficult to understand how predictions were decided and how the models were
analyzed.
|
|
SAE
|
1) Enables efficient data representation.
2) Excellent effect on data compression and noise rejection.
|
1) Increased the number of parameters in proportion to the size of the data.
2) Taking advantage of data-specific attributes is difficult.
|
|
CNN
|
1) Owing to the convolution characteristics, it is easier to input and learn more
than two dimensions of data compared to a normal neural network.
2) Multi-dimensional analysis can be performed better than other algorithms.
|
1) Requires innumerable computations.
2) Continuous re-learning is required as the environment varies.
|
|
b)
|
HT
|
1) It works well with noisy signals.
2) It has an ability to process non-stationary and non-linear data.
|
1) The performance of composite signals is low.
2) It is limited to interpreting a narrowband signal.
|
|
WT
|
1) It is simple for frequency analysis.
2) It is effective for analysis of discontinuous signals.
|
1) In case of detailed analysis, it becomes computationally intensive.
2) It is less efficient.
|
3.5 Wavelet-based deep neural network
Numerous wavelet transform (WT) techniques are already being applied to detect the
fault in the power system. The method of determining the type of fault situation using
the wavelet transform is not significantly different from the HT method discussed
earlier in Section 3.3. This method also requires the process of extracting features
such as maximum and minimum values related to the size of HT, root-mean square, energy,
standard deviation, skewness, kurtosis, and entropy using signal processing and learns
them through the neural network (34).
4. Comparison of Various IPMs
After a comprehensive review and in-depth analysis, a comparison of various IPMs considering
different capability parameters is presented in Table 1 and 2. These IPMs are primarily used to detect faults and determine the location of faults.
Each protective method can be classified according to the a) structure of neural network
and type of b) signal processing technique the method applies (30-34).
Table 2 Comparison of various IPMs (Characteristic)
|
|
Operation
Time
|
Accuracy
|
Memory
Allocation
|
|
SVM
|
Slow
|
Low
|
Small
|
|
SAE
|
Fast
|
Middle
|
Middle
|
|
CNN
|
Middle
|
High
|
Large
|
5. Conclusion
This paper presents a comprehensive review of challenging issues of MGs, and various
IPMs. Most protection-related issues that can occur in MGs are caused by the presence
of RESs, especially in the event of a fault, owing to the fault contribution of the
RESs. Furthermore, the fault characteristics of the power system are changed owing
to the fault contribution of the RESs; consequently, the reliability of the existing
protective system decreases. Therefore, it is necessary to establish a protective
system that takes into account the fault contribution of the RESs. In addition, IPMs
use machine learning to learn the fault characteristics, detect the type of fault,
and determine the location of occurrence of the fault. Currently, in the topic of
MGs, IPMs will become the future trend in MG protection. In the case of studying IPMs,
research into new neural network structures or research into the appropriate type
of signal processing method to extract the features, which are to be used as input
data for machine learning, are being conducted extensively.
This review paper is believed to be useful for development of MG protection systems
in the future. In addition, this paper presents the classification of IPMs for MGs
that were not previously classified, thus helping in the construction of a more systematic
MG protection system.
Acknowledgements
This work was supported by the National Research Foundation of Korea(NRF) grant funded
by the Korea government(MSIP) (No. 2018R1A2A1A05078680).
References
International Renewable Energy Agency, April 2018, Renewable energy polices in a
time of transition
I. Almutairy, Dec 2016, A review of coordination strategies and techniques for overcoming
challenges to microgrid protection, 2016 Saudi Arabia Smart Grid
Niraj Kumar Choudhary, Soumya Ranjan Mohanty, Ravindra Kumar Singh, May 2015, Protection
coordination of over current relays in distribution system with DG and superconducting
fault current limiter, 2014 18th National Power Systems Conference
Kyle Jennett, Campbell Booth, Martin Lee, Oct 2011, Analysis of the sympathetic tripping
problem for networks with high penetrations of Distributed Generation, 2011 International
Conference on Advanced Power System Automation and Protection
Hossam Sabra, Doaa Khalil Ibrahim, Mahmoud Gilany, Oct 2018, Field experience with
sympathetic tripping in distribution networks: problems and solutions, The Journal
of Engineering, Vol. 2018, No. 15, pp. 1181-1185
Ankita Sharma, Bijaya Ketan Panigrahi, Feb 2019, Interphase fault relaying scheme
to mitigate sympathetic tripping in meshed distribution system, IEEE Transactions
on Industry Applications, Vol. 55, No. 1, pp. 850-857
S. F. Tan, S. K. Salman, Sep 2008, Investigation into the implementation of auto reclosing
scheme in distribution networks with high penetration of DGs, 43rd International Universities
Power Engineering Conference
Jonathan D. Glidewell, Manish Y. Patel, Nov 2012, Effect of high speed reclosing on
fault induced delayed voltage recovery, 2012 IEEE Power and Energy Society General
Meeting
F. T. Dai, Apr 2010, Impacts of distributed generation on protection and autoreclosing
of distribution networks, 10th IET International Conference on Developments in Power
Systems Protection
Olga V. Gazizova, Alexander P. Sokolov, Nikolay T. Patshin, Aleksey V. Malafeev, Alexander
L. Karyakin, Jan 2020, The use of non-synchronous automatic reclosing in power plants
of large industrial enterprises with a complex network configuration, 2019 International
Conference on Electro-technical Complexes and Systems
N. Rajaei, M. H. Ahmed, M. M. A. Salama, R. K. Varma, July 2014, analysis of fault
current contribution from inverter based distributed generation, 2014 IEEE PES General
Meeting
Mpeli Rampokanyo, Pamela Kamera, June 2018, Impact of increased penetration levels
of distributed inverter-based generation on transient stability, 2018 IEEE PES/IAS
Power Africa
Dimitrios I. Doukas, Paschalis A. Gkaidatzis, Aggelos S. Bouhouras, Kallisthenis I.
Sgouras, Dimitris P. Labridis, Nov 2016, On reverse power flow modelling in distribution
grids, Mediterranean Conference on Power Generation, Transmission, Distribution and
Energy Conversion
Ali Hooshyar, Reza Iravani, Nov 2018, A New Directional Element for Microgrid Protection,
IEEE Transactions on Smart Grid, Vol. 9, No. 6, pp. 6862-6876
Gustavo Ramos, David Celeita, Tatiana Quintero, Sep 2019, Reverse power flow analyzer:
a tool to assess the impact of PVs in distribution systems, 2019 IEEE Industry Applications
Society Annual Meeting
Ingmar Leisse, Olof Samuelsson, Jörgen Svensson, Sep 2012, Coordinated voltage control
in distribution systems with DG — Control algorithm and case study, CIRED 2012 Workshop:
Integration of Renewables into the Distribution Grid
Hiroyuki Hatta, Masahiro Asari, Hiromu Kobayashi, Aug 2010, Study of energy management
for decreasing reverse power flow from photovoltaic power systems, 2009 IEEE PES/IAS
Conference on Sustainable Alternative Energy (SAE)
Giovanni Brusco, Alessandro Burgio, Daniele Menniti, Anna Pinnarelli, Nicola Sorrentino,
Sep 2014, Energy management system for an energy district with demand response availability,
IEEE Transactions on Smart Grid, Vol. 5, No. 5, pp. 2385-2393
Liciane Otremba, Jonas R. Pesente, Rodrigo B. Otto, Rodrigo A. Ramos, Oct 2015, A
procedure to analyze the impact of three-phase unbalanced conditions on switching
overvoltages in systems with Distributed Generation, 2015 IEEE Power & Energy Society
General Meeting
Laura Wieserman, T. E. McDermott, Sep 2014, Fault current and overvoltage calculations
for inverter-based generation using symmetrical components, 2014 IEEE Energy Conversion
Congress and Exposition (ECCE)
H. S. V. S. Kumar Nunna, Balzhan Azibek, Anvar Khamitov, Prashant K. Jamwal, Akshay
K Rathore, Apr 2020, Increasing hosting capacity of distribution networks by microgrid
reactive power management, 2020 IEEE International Conference on Power Electronics,
Smart Grid and Renewable Energy (PESGRE2020)
Carlos Augusto M. Gomes, Hugo P. Ferreira, May 2018, Hosting capacity evaluation of
distributed generation systems with genetic algorithm, 2018 Simposio Brasileiro de
Sistemas Eletricos (SBSE)
O. V. Gnana Swathika, S. Hemamalini, Dec 2016, Prims-aided dijkstra algorithm for
adaptive protection in microgrids, IEEE Journal of Emerging and Selected Topics in
Power Electronics, Vol. 4, No. 4, pp. 1279-1286
Emilio C. Piesciorovsky, Noel N. Schulz, Jan 2017, Fuse relay adaptive overcurrent
protection scheme for microgrid with distributed generators, IET Generation, Transmission
& Distribution, Vol. 11, No. 2, pp. 540-549
Ankita Sharma, Bijaya Ketan Panigrahi, May 2018, Phase fault protection scheme for
reliable operation of microgrids, IEEE Transactions on Industry Applications, Vol.
54, No. 3, pp. 2646-2655
Waleed K. A. Najy, H. H. Zeineldin, Wei Lee Woon, Apr 2013, Optimal protection coordination
for microgrids with grid-connected and islanded capability, IEEE Transactions on Industrial
Electronics, Vol. 60, No. 4, pp. 1668-1677
Kai Yu, Qian Ai, Shiyi Wang, Jianmo Ni, Tianguang Lv, Mar 2016, Analysis and optimization
of droop controller for micro-grid system based on small-signal dynamic model, IEEE
Transactions on Smart Grid, Vol. 7, No. 2, pp. 695-705
Tatiana Chakravorti, Rajesh Kumar Patnaik, Praditpta Kishor Dash, July 2017, Advanced
signal processing techniques for multi-class disturbance detection and classification
in microgrids, IET Science, Measurement & Technology, Vol. 11, No. 4, pp. 504-515
J. S. Kim, C. H. Kim, Y. S. Oh, G. J. Cho, J. S. Song, Oct 2020, An islanding detection
method for multi-RES systems using the graph search method, IEEE Transactions on Sustainable
Energy, Vol. 11, No. 4, pp. 2722-2731
Hengwei Lin, Kai Sun, Zheng-Hua Tan, Chengxi Liu, Josep M. Guerrero, Juan C. Vasquez,
Oct 2018, Adaptive protection combined with machine learning for microgrids, IET Generation,
Transmission & Distribution, Vol. 13, No. 6, pp. 770-779
Murli Manohar, Ebha Koley, Subhojit Ghosh, Oct 2018, Enhancing the reliability of
protection scheme for PV integrated microgrid by discriminating between array faults
and symmetrical line faults using sparse auto encoder, IET Renewable Power Generation,
Vol. 13, No. 2, pp. 308-317
Manohar Mishra, Pravat Kumar Rout, Sep 2017, Detection and classification of Microgrid
faults based on HHT and machine learning techniques, IET Generation, Transmission
& Distribution, Vol. 12, No. 2, pp. 388-397
Syed Basit Ali Bukhari, Chul-Hwan Kim, Khawaja Khalid Mehmood, Raza Haider and Muhammad
Saeed Uz Zaman, Jan 2020, Convolutional neural network-based intelligent protection
strategy for microgrids, IET Generation, Transmission & Distribution, Vol. 14, No.
7, pp. 1177-1185
N. Narasimhulu, D. V. Ashok Kumar, M. Vijay Kumar, May 2020, LWT based ANN with ant
lion optimizer for detection and classification of high impedance faults in distribution
system, Journal of Electrical Engineering and Technology
저자소개
He received a B.S degree from the College of Information and Communication Engineering,
Sungkyunkwan University, Korea, in 2016. At present, he is enrolled in the combined
master’s and doctorate program. His research interests include power system transients,
wind power generation and distributed energy resource.
He received a B.S degree from the College of Information and Communication Engineering,
Sungkyunkwan University, Korea, in 2017.
At present, he is enrolled in the combined master’s and doctorate program.
His research interests include distributed generation and power system protection.
He received a B.S, degrees in electrical engineering from Kangwon national University,
in 2019.
At present, he is enrolled in the combined master’s and doctorate program of the College
of Information and Communication, Sungkyunkwan University.
His research interests include power system protection and power system transients.
He received a B.S degree from College of Information and Communication Engineering,
Sungkyunkwan University, Korea, in 2020.
At present, he is enrolled in the master program.
His research interests include power system transients, distributed energy resource.
He received the B.S., M.S., and Ph.D. degrees in electrical engineering from Sungkyunkwan
University, Suwon, Korea, in 1982, 1984, and 1990, respectively.
In 1990, he joined Jeju National University, Jeju, Korea, as a Full-Time Lecturer.
He was a Visiting Academic with the University of Bath, Bath, U.K., in 1996, 1998,
and 1999.
He has been a Professor with the College of Information and Communication Engineering,
Sungkyunkwan University, since 1992, where he is currently the Director of the Center
for Power Information Technology.
His current research interests include power system protection, artificial intelligence
applications for protection and control, modeling/protection of underground cable,
and electromagnetic transients program software.