1898 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO.

4, AUGUST 2015

Half-Interval Method Applied in Feeder Terminal

Unit Overcurrent Detecting Curve Setting
Chao-Rong Chen, Senior Member, IEEE, and Chi-Juin Chang

Abstract—Currently, self-healing is one of the most important

functions in the smart grid. Meanwhile, fault detection, isola-
tion, and restoration of feeder automation systems dominate the
self-healing function in power distribution systems. The steps of
the aforementioned function depend on the fault flag status of
the feeder terminal unit. The conditions for setting this flag are
judged by the feeder terminal unit overcurrent detecting curve.
This paper found an efficient approach to calculate this curve via
a half-interval method. Versatile application software with the
curve plotting capability was also developed and deployed on the
web server of the information-management department of Taiwan
Power Company and is running successfully.

Index Terms—Computer aided analysis, fault currents, IEEE

standards, piecewise linear approximation, power distribution
protection.

I. INTRODUCTION

R ECENTLY, smart grid causes a vast discussion and was

diffusely implemented in many power utilities though
the comprehensive functions of smart grid are still underde- Fig. 1. Result of the manual setting of the FTU overcurrent detecting curve.
veloped or several standards remain to be updated. The self-
healing, however, is the most acknowledged core function used
in the smart grid [1]–[6]. So far, the smart-grid plan of Taiwan addition, many of the existing manual setting overcurrent de-
Power Company (Taipower) includes a smart power distribu- tecting curves in Taipower FA system were found where sev-
tion project which is mainly constructed by the feeder automa- eral time-current points of the curve are located under the total
tion scheme. For the sake of rapid power restoration as well clearing curve of the fuse, for example, Fig. 1. Thus, the FTU
as high-power quality, fault detection, isolation, and restora- flags will be possibly kept unchanged even when a real feeder
tion (FDIR) is included in the most implementation of auto- incident has occurred and, consequently, causes the malfunction
mated feeders. The parameters set in intelligent electronic de- of FDIR.
vices (IEDs) connected to the feeder circuit breaker (CB) are Many research results have been studied for finding the total
the important references for adjusting the feeder terminal unit solution of the aforementioned problem. Lin et al. [7] developed
(FTU) overcurrent detecting curve which affects the correct- a multiagent-based approach for service restoration of distribu-
ness of the fault flag setting. The feeder automation (FA) plan- tion systems after a fault contingency. Ren et al. [8] presented
ning employees are hard to set each time-current point of the two kinds of algorithms for failure judgment. The first is the
curve due to the lacking of an adequate computer-aided tool. In matrix which is fit only for the master station. The second is
the nonterminal-locating algorithm which is applied to the dis-
tribution station. Gill [9] analyzed either the post isolation al-
Manuscript received July 01, 2014; revised October 09, 2014; accepted tered system model or “zone” model for alternate sources with
November 21, 2014. Date of publication December 02, 2014; date of current
version July 21, 2015. This work was supported in part by the National Science available capacity, and checked for voltage violations to restore
Council in Taiwan, under the Project Title: Caltech-Taiwan collaboration downstream customers. Fan et al. [10] addressed the potential
on energy research-uncertainty mitigation for renewable energy integration, problems associated with uncoordinated optimization applica-
Project No: NSC 101-3113-P-008-001. Paper no. TPWRD-00798-2014.
The authors are with the Electrical Engineering Department, National tions that include FDIR. However, all of the aforementioned
Taipei University of Technology, Taipei, Taiwan (e-mail: crchen@ntut.edu.tw; research seems not to offer the way for setting the FTU over-
u145387@taipower.com.tw). current detecting curve via a computer-aided analysis approach.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. This paper therefore tried to find out an efficient method for this
Digital Object Identifier 10.1109/TPWRD.2014.2375878 purpose.

0885-8977 © 2014 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/
redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
CHEN AND CHANG: HALF-INTERVAL METHOD APPLIED IN FEEDER TERMINAL UNIT OVERCURRENT DETECTING CURVE SETTING 1899

Fig. 2. FTU overcurrent detecting curve.

II. PROBLEMS FORMULATION

The FTU overcurrent detecting curve is constructed by sev-
eral time–current points in a stepping style, see Fig. 2. These
points should be located under the time–current curve (TCC)
of the IED of feeder circuit breaker (FCB) in order to detect the Fig. 3. Typical phase FTU overcurrent detecting curve.
fault current as well as setting the fault flag before IED tripping.
For the sake of avoiding the influence from the feeder transient
state of large load switching, all of these points should also be in the Taipower distribution system. The service range of FDIR
set as close to the TCC of FCBs IED as possible. An adequate is designed only for the whole trunk of the feeder in order to
planning procedure is therefore needed to find an optimal result reduce FA deployment cost. As far as the phase FTU overcurrent
subject to the upper limit, that is, the TCC of FCBs IED, and the detecting curve is concerned, its fault flag should not be set if
number of the time–current points. the power fuse is not totally clear totally in case of a branch
phase fault as well as avoiding unnecessary FDIR operation.
III. FTU OVERCURRENT DETECTING CURVE From a different point of view, if the power fuse is melted and
Currently, all of the FTUs used in Taipower comply with the totally cleared, it means that the fault is located on the branch
following overcurrent detecting algorithm. of the feeder and FDIR is no longer necessary to start due to the
If the current-time points to form an overcur- fault being isolated by the fuse being burned. Thus, the phase
rent detecting curve in a current increment and trip time decre- FTU overcurrent detecting curve should be located above the
ment sequence, an overcurrent case , that is, the fault cur- fuse total clearing curve. The fuse total clearing curve therefore
rent occurs and lasts seconds, and will cause the fault flag forms the lower bound of the phase FTU overcurrent detecting
set once and , where . curve.
Suppose a four, 4, set-point FTU overcurrent detecting curve From the ground FTU overcurrent detecting curves point of
is shown in Fig. 2. These points form a stepping path. The fault view, the TCC curve of the ground IED mounted on the ground
will not be set as soon as overcurrent case 1 (OC1) occurs due line of the feeder is therefore considered as the upper bound.
to and . Accordingly, overcurrent cases Similarly, the actual upper bound is moved down by several per-
2 and 3 will cause the FTU fault flag set since and centages (e.g., 20%), of IED TCC of the FCB for FCB trip time
. tolerance purposes. The lower bound is evaluated by a higher
Once the feeder fault occurs, the FCB will be opened by the level for avoiding the influence of the possible transient un-
connected IED trip signal based on the power distribution pro- balanced load. Consequently, the lower bound curve is chosen
tection mechanism. The fault current will vanish after the FCB by empirical results that are formed by moving the TCC of the
opens and the FTU cannot detect the fault anymore and it is im- ground IED down by a specific percentage, say 50%.
possible for the fault flag to be set. In summary, the FTU fault Based on the studying result of this paper, the typical phase
flag will not be set correctly if any overcurrent detecting curve and ground FTU overcurrent detecting curves and their cur-
set point, that is, in a solid block, or its derivative point, that is, rent-time values are shown in Figs. 3 and 4 and Tables I and
in a hollow block, is located on the upper side of the IED TCC II, respectively.
of FCB in Fig. 2. Thus IED TCC of FCB forms an upper bound If an instantaneous overcurrent relay (device 50) is used (see
to the FTU overcurrent detecting curve. For the sake of the tol- Fig. 3), the start point of the overcurrent detecting curve should
erance of FCB operation time, the actual upper bound moves be set at the point , where is the fault current with a
down by several percentages of IED TCC of FCB (e.g., 20%). current skew, say 5%, to the instantaneous startup overcurrent
Recently, 125E standard speed power fuse is usually mounted and is the trip time of the instantaneous overcurrent relay. The
on the head end of the branch or on the terminal of the feeder current skew is to avoid the situation where the fake fault flag
1900 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 4, AUGUST 2015

Fig. 4. Typical ground FTU overcurrent detecting curve. Fig. 5. Typical eight-points FTU overcurrent detecting curve.

TABLE I
TYPICAL POINT VALUES OF PHASE FTU OVERCURRENT DETECTING CURVES
Notes: Careful algorithm checks for 0, limits i, and
other possible stopping conditions (e.g., ).
The purpose of adopting the half-interval method is to find
the time and current value of all points of the FTU overcurrent
detecting curve with a given point number and making the first
and the last point cover the entire fault current range. It can be
easily found that the width occupied by the FTU overcurrent
detecting curve is inversely proportional to the point number of
the curve. An eight (8)-points FTU overcurrent detecting curve
is shown in Fig. 5 in order to compare to Fig. 3 for verifying the
aforementioned statement.
Based on this proportional relationship, the half-interval
is set due to the time–current characteristic of the instantaneous method can be applied to compute the time and current values
overcurrent relay. of all points constrained by a given point number. The detailed
algorithm for finding an eight-point FTU overcurrent detecting
curve is shown as follows.
IV. HALF-INTERVAL METHOD 1) Set the upper bound curve as a specific percent, say 20%,
of the down displacement of TCC [14] of the FCB IED.
A typical half-interval, also named bisection, algorithm can
Thus
be found in [11] and [12]. This is for finding a nonlinear and con-
tinuous equation root. The discrete problem can also be solved
via an adequate modification of this algorithm for finding the
key from an ordered table [13].
The basic idea of the half-interval method comes from
solving the root finding problem. Given a computable function (1)
, the problem is to find the root and
. We first find a subinterval , with where
, set . Then, the procedure of the
half-interval method goes as follows: IED trip time;
Step 1) Set ; IED time dial setting;
Step 2) If , set ;
otherwise, set ; current flow through the circuit breaker;
Step 3) If , set and go to Step IED pickup current;
1);
IED parameters.
Step 4) Stop with .
CHEN AND CHANG: HALF-INTERVAL METHOD APPLIED IN FEEDER TERMINAL UNIT OVERCURRENT DETECTING CURVE SETTING 1901

2) Set the lower bound curve as the fuse total clearing curve. where is chosen to be 0.5 at the first step of the half-interval
Thus iteration.
As a matter of fact, (5) is easy to find the inverse function as

(2)
(7)
where and stand for two terminals of a
line segment of the fuse total clearing curve, where On the contrary, (6) is quite difficult to find its inverse func-
. is the iteration factor of the half-interval method tion. It is then necessary that the Newton Raphson [16] iteration
and is for the initial step of the half- interval method should be applied to this situation.
iteration. Suppose is the root of the equation can be
3) Let 1. obtained by the following iteration:
4) Compute and
and (8)
to 1, where (i.e., 4500 A in Fig. 5, and
is the maximum fault current and is the where if where is a user-defined
iteration step number tolerance value and is usually set to 0.1, depending on the scale
where of fault current, in the Taipower system.
If the Newton-Raphson iteration is applied in (6), the iteration
model can be derived as follows:

(3)

5) If and , where is the

minimum fault current, then terminate the iteration and let
(take the lower-half interval).
6) If and 1, then cease the iteration and
let (take the upper half interval).
7) Repeat 4) until and . (9)
If GE IAC standard [15] is used for the FCB IED, (1) is then
revised as (4). The iteration will be ceased once or 100.
In case of 100, the iteration is diverge and no solution can
be found. Otherwise, , described in (3), can be obtained by
setting its value to the result of the Newton–Raphson iteration,
that is, , in Step 4) of the half-interval algorithm. Usually,
the initial value is chosen as the minimum fault current. The
(4) IEDs used in Taipower have been totally tested with its TCC
It is quite difficult to use a previous known trip time to find and no diverge situation occurred.
the fault current via (4), that is, finding out the inverse function
of (4). For the sake of avoiding this situation, the iteration step V. TCC AND FUSE TOTAL CLEARING CURVE
number is arranged from 8 to 1. INTERSECTION CHECK
As far as the ground FTU overcurrent detecting curve is con-
cerned, no fuse curve can be adopted as the lower bound curve. In case of the phase FCB analysis situation, its IED TCC
A dummy lower bound curve is chosen by moving down the should not intersect with the fuse total clearing curve. A revised
original FCB IED TCC with 50% off to the trip time. Thus, (2) partial differential approach (PDA) [17] was developed for this
is then revised as follows: checking.
The original PDA approach is used for checking two IED
TCCs' intersections. The revised approach is changed by re-
(5) placing the downstream protection device from IED to fuse.
Since the original fuse curve data are composed by many line
segments, this revised PDA approach, shown as follows, must
be carried out for each line segment of the fuse total clearing
curve in order to obtain the result of full fault current range in-
tersection investigation.
Based on the IEEE C37or IEC60255-4standard, the equation
of the time–current curve of IED or solid-state relay can be
(6) shown similar to (1). In addition, a piecewise linear approxi-
mation is usually used for fuse total clearing curve modeling.
1902 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 4, AUGUST 2015

The time interval between FCB IED TCC and one of the line Newton–Raphson iteration is necessary again for solving (12)
segments of fuse total clearing curve is shown as due to its difficulty of being solved by conventional fractional
arithmetic or transposition operation.
As the result of solving the value of , the intersection
situation of FCB IED TCC and fuse total clearing curve can
be exactly evaluated. The existence of the intersection is true
if . The parameters of FCB IED TCC should be
adjusted at this time for eliminating the intersection situation.
(10) Consequently, the FTU overcurrent detecting curve setting can
be planned correctly once the intersection of FCB IED TCC and
where and are the terminal points of this line fuse total clearing curve does not exist.
segment.
Take the first-order partial differentiate of (10), then VI. FEEDER RESTORATION APPLICATION
For a two main transformers (MTR) distribution substation
example, a load bank of one MTR, named MTR1, with FCB1
and IED1, is always switched to another MTR, called MTR2,
with FCB2 and IED2 via bus tie breaker operation during the
(11)
MTR1 maintenance period. The power of feeder loads of MTR1
is then transferred to MTR2. The lever value of IED2 should
be recalculated due to the change of the maximum load cur-
Based on the PDA approach, three intersection indicators and rent of MTR2. This lever recalculation can be easily performed
their first-order partial differentiate are defined as follows: by the optimal approach described in [17]. Consequently, the
overcurrent detecting curve of the total FTUs mounted on the
feeders must be changed accordingly. The half-interval method
developed by this paper can then be applied to the calculation of
this change with more precise and rapid result. Thus, the feeder
restoration based on the FTU fault flag setting can also be done
precisely during feeder fault after MTR load transformation.

VII. WEB-BASED APPLICATION SYSTEM

An actual web-based application system for FTU overcurrent
detecting curve planning had been developed and deployed on
the web server of the information-management department of
where and are the minimum and maximum Taipower. The user interface of this system is shown as Fig. 6.
fault current, respectively. The radio buttons “FCB_CO” and “FCB_LCO” stand for the
For the sake of intersection evaluation, the midpoint min- phase or ground IED being planned. “SW_RY” stands for the
imum time interval existing principle (MMEP) is defined as solid-state relay of the feeder switch. The dropdown list names
follows: are described as ollows:

“CO-IED MNF” phase IED manufacture or brand of IED;

“TYP” phase IED type;
“CO-TCC STD” phase IED TCC standard;
If MMEP is false, that means the midpoint minimum time in-
terval exists and can be calculated by the simultaneous “VER” phase IED TCC version;
equation shown in (12), at the bottom of the page. “CO-TCC TYP” phase IED TCC type;
If MMEP is true, can be simply assigned by
“No of PT” FTU point number.
or .

(12)
CHEN AND CHANG: HALF-INTERVAL METHOD APPLIED IN FEEDER TERMINAL UNIT OVERCURRENT DETECTING CURVE SETTING 1903

Fig. 6. User interface of the web-based application system.

Fig. 7. Web-based application system architecture.

All of the ground IEDs and feeder switch solid-state relays

parameters are similarly shown in Fig. 6. TABLE II
The text box “FCB TCC DOWN” stands for the upper bound TYPICAL POINT VALUES OF THE GROUND FTU
curve formed by moving the FCB TCC curve down by the filled OVERCURRENT DETECTING CURVES

percentage. In order to avoid the FTU fault flag from incorrect

operation in using the indicating instantaneous trip (IIT) situa-
tion, a current skew is introduced in FTU overcurrent detecting
curve planning. The “SHIFT” field stands for this current skew
which is counted by the percentage of the maximum fault cur-
rent. “IIT_A” stands for the fault current setting of IIT. “IIT_T”
stands for the trip time setting of IIT. The current skew is used
for the last point of the FTU overcurrent detecting curve setting
in case of the IIT mechanism being used.
The user can fill in any message in the text box “Note” and
this message will be shown on the left down corner of the over-
current detecting curve figure as shown in Fig. 3 (e.g., “Tsaotun system. The users' information is then recorded in a database
AP21”). table for auditing purposes.
The core FTU overcurrent detecting curve calculating algo- The web browser used by the client must support HTML ver-
rithms are coded by the Microsoft Visual Basic .NET program- sion 5 (e.g., Firefox or Google chrome browser). The several
ming language. All of the protection devices TCC parameters external plug-in programs are necessary to install on the web
are built in a Microsoft SQL server database. After all of the server for the incompatibility of Microsoft Internet Explorer
input fields are filled and the button “OK” is clicked, the core pa- with HTML version 5.
rameters calculation program coded in Microsoft Visual Basic
is then started to make the related calculation and shows the A. Case Study
calculated result on the screen. All of the input fields can be A Siemens 7SJ62 IED with ANSI/IEEE Extremely Inverse
cleared during the input session once the “CLEAR” button is TCC is installed on the Taipower Tsaotun secondary substation
pressed. As soon as the button “OK” is clicked, all of the cal- AP21 feeder for overcurrent protection purposes. The CT ratio
culated results are converted to the curve data embedded in a and tap position are 500/5 and 5, respectively. The time dial
JavaScript program with CANVAS code format and transferred setting of this IED is 1.9. The maximum/minimum fault current
to the client to be shown on the screen in a graphical format. is 4500 A/550 A. An indicating instantaneous trip mechanism
The entire web-based application system architecture is shown is set with a 0.07-s trip time when the fault current is greater
in Fig. 7. than 4300 A. The original manual 8 points FTU overcurrent
As far as user management is concerned, the user must log in detecting curve planning result for the phase FTU is shown in
via a single sign-on interface in order to enter this application Table III and its curve figure is shown to be the same as Fig. 1.
1904 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 4, AUGUST 2015

TABLE III standard, the rapid and minimum user intervention feeder load
TSAOTUN AP21 FTU OVERCURRENT DETECTING CURVE
switching in the substation can be fulfilled by transmitting
MANUAL PLANNING RESULT
the related calculated parameters from the control center to
corresponding IED and FTU in order to precisely set the pro-
tection and coordination parameters and overcurrent detecting
curve after the feeder load has been switched. The half-interval
method proposed in this paper has been verified in that it is
superior than the FTU overcurrent detecting curve calculation.

REFERENCES
TABLE IV
TSAOTUN AP21 FTU OVERCURRENT DETECTING CURVE [1] S. A. Arefifar, Y. A.-R. I. Mohamed, and T. H. M. El-Fouly, “Com-
COMPUTER AID PLANNING RESULT prehensive operational planning framework for self-healing control ac-
tions in smart distribution grids,” IEEE Trans. Power Syst., vol. 28, no.
4, pp. 4192–4200, Nov. 2013.
[2] S. Liu, Y. Hou, C. C. Liu, and R. Podmore, “The healing touch: Tools
and challenges for smart grid restoration,” IEEE Power Energy Mag.,
vol. 12, no. 1, pp. 54–63, Jan./Feb. 2014.
[3] Y. Yan, Y. Qian, H. Sharif, and D. Tipper, “A survey on cyber se-
curity for smart grid communications,” IEEE Commun. Surveys Tuto-
rials, vol. 14, no. 4, pp. 998–1010, 4th qtr. 2012.
[4] H. Gharavi and B. Hu, “Multigate communication network for smart
grid,” Proc. IEEE, vol. 99, no. 6, pp. 1028–1045, Jun. 2011.
[5] R. Podmore and M. R. Robinson, “The role of simulators for smart grid
development,” IEEE Trans. Smart Grid, vol. 1, no. 2, pp. 205–212, Sep.
2010.
In Fig. 1, it is clear that the 800-A transient inrush current [6] Z. Jiao, X. Wang, and H. Gong, “Wide area measurement/wide area
information-based control strategy to fast relieve overloads in a self-
that lasts 0.5 s will cause a fake fault flag set. However, the
healing power grid,” IET Gen., Transm. Distrib., vol. 8, no. 6, pp.
CAP tool developed in this paper will provide an adequate or 1168–1176, Jun. 2014.
even optimal planning result for avoiding the fake fault flag set [7] C. H. Lin, H. J. Chuang, C. S. Chen, C. S. Li, and C. Y. Ho, “Fault
situation. Table IV and Fig. 3 show the computer-aided analysis detection, isolation and restoration using a multiagent-based distribu-
tion automation system,” in Proc. 4th IEEE Conf. Power, Energy, Ind.
result and curve figure, respectively. Appl., 2009, pp. 2528–2533.
[8] R. Ren, Z. Wang, B. Song, and Q. Fang, “The realization of the feeder
automation function in distribution automation systems,” in Tech. Ses-
VIII. CONCLUSION sion 3, Protect., Control, Commun. Autom. Distrib. Netw., 2008, pp.
1–4.
[9] H. M. Gill, “Smart grid distribution automation for public power,” in
With an optimal FTU overcurrent detecting curve being set by Proc. IEEE Power Energy Soc. Transm. Distrib. Conf. Expo., 2010, pp.
using the web application software developed in this paper, the 1–4.
FDIR function is obtaining more correct and feeder power and [10] H. Fan, P. Backscheider, J. Fan, and J. D. McDonald, “Collaborative
optimization of distribution grid operation,” in Proc. IEEE Power En-
is therefore restored more precisely and rapidly. Meanwhile, the ergy Soc. Innovat. Smart Grid Technol., 2012, pp. 1–6.
smart grid is wildly constructed in many power companies with [11] R. Butt, Introduction to Numerical Analysis Using MATLAB, ser. Math-
IEC61850 communication protocol standards [18]. Based on the ematics. Canada: Jones and Bartlett, 2010, pp. 23–28.
[12] S. Aboreshaid, R. Billinton, and M. Fotuhi-Firuzabad, “Probabilistic
infrastructure of this standard, all of the time-current points can transient stability studies using the method of bisection [power sys-
be easily calculated and automatically downloaded to the cor- tems],” IEEE Trans. Power Syst., vol. 11, no. 4, pp. 1990–1996, Nov.
responding FTU once the feeder peak load is changed or a load 1996.
[13] S. Nishihara and H. Nishino, “Binary search revisited: Another advan-
transformation occurs. tage of fibonacci search,” IEEE Trans. Comput., vol. C-36, no. 9, pp.
One of the main purposes of DA in Taipower is the fault 1132–1135, Sep. 1987.
process for a remote-area electric power household in order to [14] IEEE Standard Inverse-Time Characteristic Equations for Overcurrent
Relays, IEEE Standard C37.112-1996, IEEE Standards Board, .
avoid the rugged journey. With the sophisticated RF planning [15] “750/760 Feeder Management Relay Instruction Manual,” General
[19], optimal signal coverage that ensures stable two-way com- Electric, 2010.
munication can be achieved. Based on this stability and two-way [16] G. Labelle, “On extensions of the Newton-Raphson iterative scheme
to arbitrary orders,” in Proc. DMTCS AN, 2010, pp. 713–724.
communication as well as the research result of this paper, an ef- [17] C. R. Chen, C. H. Lee, and C. J. Chang, “Overcurrent relay coordi-
fective self-healing feeder load transformation is feasible for the nation optimization with partial differentiation approach for the val-
remote area. idation of coordination violation,” Elect. Power Compon. Syst., pp.
933–947, 2011.
For the sake of smart-grid deployment purposes, numeral [18] IEC, “Communication networks and systems in substations—Part 1:
standards have been declared (e.g., IEC 61850). In addition, introduction and overview, Geneva, Switzerland, IEC TR 61850-1,
CIM is also used in the data exchange of an energy-management 2003.
[19] C. M. Chang, “Telecommunications network design: single node opti-
system (EMS) among the dispatching control centers. With the mization and the location selection problem,” Ph.D. dissertation, Elect.
help of the aforementioned communication and data-exchange Eng. Dept., Univ. Texas Arlington, Arlington, TX, USA, 1985.
CHEN AND CHANG: HALF-INTERVAL METHOD APPLIED IN FEEDER TERMINAL UNIT OVERCURRENT DETECTING CURVE SETTING 1905

Chao-Rong Chen (SM’11) received the B.S., M.S., Chi-Juin Chang received the M.S. degree in
and Ph.D. degrees in electrical engineering from Na- business administration from National Chiao Tung
tional Taiwan University, Taiwan, in 1983, 1988, and University, Hsinchu, Taiwan, in 1994 and is currently
1991, respectively. pursuing the Ph.D. degree in electrical engineering
In 1991, he joined the National Taipei University from National Taipei University of Technology,
of Technology as a faculty member, and is currently Taipei.
an Associate Professor in the Electrical Engineering He has been an Electric Engineer with Taiwan
Department. From 1995 to 1996, he was a Visiting Power Company since 1978 for planning super-
Scholar at the University of Washington, Seattle, visory control and data-acquisition systems. His
WA, USA. His research interests include power research interests include power system protection
system stability analysis, artificial neural networks, and coordination, substation and feeder automations,
and pattern recognition. and smart grid.