1-s2.0-s0142061599000356-main
TRANSCRIPT
-
8/12/2019 1-s2.0-S0142061599000356-main
1/8
An adaptive load shedding method with time-based design for isolatedpower systems
S.-J. Huang*, C.-C. Huang
Department of Electrical Engineering, National Cheng Kung University, Tainan, Taiwan, 70101
Abstract
When an isolated power system encounters a serious disturbance or a large generator unit trip, the system frequency may drop if total
generating power is not able to supply the load demand sufficiently. Since an isolated power system possesses a lower inertia with limited
reserves, load shedding becomes a critical solution to restore system frequency. In this paper, the initial rate of change of frequency is firstdetected. A designated time interval and a modification algorithm that considers the maximum frequency change at the last step are
integrated to complete a load shedding task. Features of the method include the easy determination of shedding loads and fast trip of the
load in anticipation of avoiding the occurrence of over-shedding or under-shedding. Numerical simulations have solidified the effectiveness
of the proposed method for the application. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Load shedding; Isolated power system
1. Introduction
When an isolated power system encounters a serious
disturbance or a large generator trip, the system frequency
may drop once the total generating power could not supply
the load demand sufficiently. It is known that an isolatedpower system should be self-sufficient to provide security,
reliability and economy of electricity. In such a system, the
speed of frequency decay due to loss of generation is parti-
cularly significant. Therefore, more attention should be paid
to the load shedding scheme design [13].
Several studies have been proposed to restore the system
operation frequency after serious disturbances. One of these
strategies is throughthe under-frequencyloadshedding design
[4,5]. This approach is simple and easy to implement;
however, its poor coordination capability may deteriorate the
overall performance. Efforts to improve under-frequency load
shedding using the rate of change of frequency as additional
control variables have been suggested [610]. These methods
primarily shed the load when therate of change offrequency is
higher than a designated value. Among these methods, a low-
order linear system frequency response model based on the
idea of average frequency was proposed, where the essential
characteristics of the system frequency response was seen
accurately estimated in their test cases [6]. This is followed
by an adaptive methodology that was applied for the setting of
under-frequency relays based on the initial rate of change of
the frequency at the relay [7]. Because conventional under-
frequency loadshedding schemes canonlyhandle a maximum
of 50% load, a hybrid load-shedding scheme was also
suggested. This hybrid system includes at least one time-coor-
dinated step in addition to several frequency-coordinatedsteps. By this hybrid, it is able to shed 70% and more of the
loadfrom the system[11]. On accountof tediouscomputations
required for determining the amount of overload and the load
shed per step, a neural network approach has also been
employed to calculate the minimum frequency during a forced
outage of a generating unit [12].
This paper proposes the detection of the initial rate of
change of frequency with predetermined time intervals to
formulate an appropriate load shedding strategy. Mean-
while, to avoid load-shedding failures, a modification
method considering the maximum change of frequency at
the last step is embedded in order to adjust the required total
amount of shed load. By this modification, the load-shed-
ding scheme can be more effective in restoring the system
frequency, which also overcomes the coordination difficulty
of the problems. Numerical simulations support the
proposed method in good agreement.
2. Problem formulation
2.1. Analysis model
Fig. 1 shows the linear frequency response model that
Electrical Power and Energy Systems 22 (2000) 5158
0142-0615/00/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved.
PII: S0142-0615(99) 00035-6
www.elsevier.com/locate/ijepes
* Corresponding author. Tel.: 886-6-276-0711; fax: 886-6-234-
5482.
E-mail address:[email protected] (S.-J. Huang)
-
8/12/2019 1-s2.0-S0142061599000356-main
2/8
represents the frequency characteristics of an isolated power
system, where all parameters are in per unit on an MVA
base equal to the total rating of the generating units in the
island. This frequency response model is formulated based
on neglecting non-linearity and all but the largest time
constants in the generating unit inertia and reheat time
constant of generating units in the isolated system. Note
that in modeling, the generation is assumed dominated by
reheat steam turbine generators. In Fig. 1, the change in thespeed changer setting Prefand the change in load demand
PLare two inputs. Other parameters in this model are FHP,
GM and s, where FHP is the high-pressure turbine power
fraction, GM the mechanical power gain factor and s the
differential operator. Most steam turbine generators in
service today are of the reheating type, in which about
one-third of the power is developed in the high-pressure
section. When the steam leaves the high pressure turbine,
it returns to the reheater and then flows to the intermediate
and low pressure sections of the turbine, where the remain-
ing two-thirds of the power is developed.
In the above model, the inertia constant of H and thereheater time constant ofTRH are important. For the inertia
constant of H, it is defined as the ratio of the moment of
inertia of the generator rotating components to the generat-
ing capacity of the unit. This constant is available from the
manufacturer of generating units, with the typical value of
28 s. Calculation of this constant in a system is also
expressed as follows:
HH1MVA1 H2MVA2 HnMVAn
MVA1 MVA2 MVAn1
The other time constant that is critical to the system is the
reheater time constant, TRH. This constant is usually of therange between 6 and 12 s, and tends to dominate the
response of the largest fraction of turbine power output
[6]. The remaining parameters of this model shown in Fig.
1 are the incremental frequency off, the damping factor of
D, and the inverse of the governor regulation of 1R
2.2. Formulation
With the above analysis model, let us now consider a
simple situation where Pref 0 and a step change happens
in load demand of PL. The frequency variation in the
frequency domain is then calculated as follows:
fs bR1 TRHs
ss2 2ns 2n
PL 2
where
b 2HRtRH1
n bDR GM05
b2n
2HR DR GMFHPTRH3
By applying the inverse Laplace transform, the change of
frequency as function of time when a step of change in the
load can be written:
ft bR
2n1 rent cosht PL 4
where
r1 2nT
2RH 2nTRH
1 2
05
h n1 2
05
tan1 nTRH
1 2
Since Eq. (4) is expressed in the closed form, we can readily
compute the initial slope of the response at t 0Therefore,
m0 dft
dt t0
PL
2H 5
By differentiating Eq. (4) with respect to the time variable of
tand let the post-differentiated result be equal to zero, the
result oftmreveals the time when the variation of frequencyis maximum. This result is also expressed as below:
tm 1
htan1
hTRH
nTRH 1
6
Substitutetminto Eq. (4), the outcome can be simplified as:
fmax kPL 7
where k is a constant. This equation illuminates that the
maximum frequency variation can be deemed proportional
to the system disturbances.
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 515852
frefP
-
+
sT
F
RH
HP
+
1
1
R
1
MG
+
+
-
+
LP
DHs+2
1
HPF
Fig. 1. Analysis model.
-
8/12/2019 1-s2.0-S0142061599000356-main
3/8
3. Design of load-shedding scheme
The load-shedding scheme is designed to determine a
workable balance between providing maximum system
protection and minimal service interruption. The design
procedure for the proposed load-shedding scheme is listed
as follows:
Step1 (obtain system parameters): As an example of the
computation of the proposed method, consider an isolated
area whose system parameters are D 10R 006H
45 GM 090 FHP 033 and TRH 75 We can then
calculate the following values, including b 02469 n
04869 08158 r 50046 h 02816 and
13696By substituting these values into Eq. (4), we can
obtain the frequency variations in the function of time:
ft 00625PL1 50046e03972t cos02816t
13696 8
Moreover, by Eq. (6), the value oftm can be calculated as2.9044 s. By substituting thistminto Eq. (8) and comparing
the results with Eq. (7), the value of k was found to be
0.1196.
Step 2 (select the first shedding frequency(f1) and mini-
mum allowable operating frequency (fmin)): Many utilities
have set the first step of load shedding at 59.5 Hz. This
choice was made based on several reasons [7,14]. First, all
of the larger turbine-generators on the system are not rated
for continuous operation below 59.5 Hz. Thus, setting the
initial load shedding frequency at a relatively high value,
such as 59.5 Hz, tends to limit the maximum frequency
deviation. Second, a load-shedding program starting at59.5 Hz would be more effective in minimizing the depth
of the under-frequency response for a heavy overload than
would a similar program which had a lower first shedding
frequency. Third, the first shedding frequency should not be
too close to normal frequency. In this way, the tripping on
severe but non-emergency frequency swings can be
avoided. In this paper, the 59.5 Hz was also selected as
shedding frequency at the first step shedding frequency.
As for the selection of minimum allowable frequency,
because the TPC limits system frequency deviation within
^ 4% (i.e. ^ 2.4 Hz for 60 Hz base), the minimum allow-
able operating frequency of fmin is equal to 57.6 Hz. Thus,
the value of the maximum change of frequency fmaxbecomes576 595595 00319 pu
Step 3 (determine the number of steps (N) and the time
interval between consecutive steps (t)): Steps of the shed
load should be carefully chosen. Over- or under-shedding
might happen under a small disturbance when a large step
size is selected, while coordination problems might happen
when many steps are decided. Experience has shown that
three to five steps are adequate selections [13,14]. Hence, in
this paper, the three, four and five steps were all adopted for
the evaluation of the proposed method.
An under-frequency relay is often incorporated with a
time delay in order to decrease the impact caused by the
system surges. This time delay may result in a longer trip
reaction such that the frequency may decline, and hence
result in unnecessary generator protection relay trips. To
deal with this problem, a time-based procedure was
proposed to shed the load in each predetermined time inter-
val. By this method, because each length of time step was
known, the time delay will not be required and the coordi-
nation problem can also be solved. This time interval can be
determined based on the system characteristics and types of
under-frequency relays. The length of each time step is not
necessarily equal to each other; while in this paper, time
intervals were all set at t 01 s for easy manipulation.
Step 4 (calculate the total amount of load shed PLS):
The worst possible disturbance is subject to the speculation,
while the load-shedding plan is inherently limit based on
this assumption. Although the larger system upsets are seen
less probable than small upsets, there is a certain probability
that a large disturbance happens [15,16]. In the load-shed-
ding study, once the frequency deviation curve is known, wecan obtain the initial rate of change of frequency ofm0based
on this curve. The overload calculated by using Eq. (5)
becomes PL 2Hm0
With the time tm obtained in Step 1 and the maximum
change of frequency fmaxin Step 2, we can solve the PL,minwith the aid of Eq. (8); and then, the total amount of load
shed can be obtained by the following equation:
PLS 2Hm0 PLmin 9
Step 5 (determine the amount of load shed per step
(PLS,i)): Once the total amount of load to be shed is
known, the load shed per step can be obtained throughdividing by the number of shedding steps. In the method,
each successive step was designed to shed a larger amount
of load than the preceding step [17,18]. This design feature
allows the protection scheme to drop smaller amounts of
load when the degree of overload is less serious. Note that
for an isolated system, fewer steps and larger load shed per
step are required when compared with the interconnected
systems.
Step6 (plot the system frequency change curve): By the
information on the amount of load shed per step, the over-
load PL, and the time interval t, the system frequency
change curve off(t) and f(t) can be plotted through the aid
of Eqs. (4) and (9), respectively. Therefore, we can evaluatethe minimal operation frequency value. If the minimum
operation frequency is seen higher than the minimum allow-
able value offmin, the load shedding is claimed accomplished
and the system has reached a new steady state. Otherwise,
an additional modification step should be activated. As for
the modification task, it will be discussed in the next step.
Step7 (modification) A modification task is required for
the case when the minimum operation frequency is lower
than the minimum allowable operation frequency fmin.
Under this scenario, it reveals that the amount of load to
be shed is still insufficient to restore the system frequency.
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 5158 53
-
8/12/2019 1-s2.0-S0142061599000356-main
4/8
To justify the total amount of shed load, a necessary task is
to get the shedding frequency fi at each step. When the
frequency at the last step is termed as fL, this value can be
computed by the following Eqs. (10)(12) iteratively:
PL1 PL PLS1 10
PLi1 PLi PLSi1 i 1 2 3 N 1 11
fi1 fi1
f
ti i 1 2 3 N
1 12where PLis the initial system overload, PL,ithe overload
after thei-th step, PLS,ithe amount of shed load at the i-th
step, fti frequency variation after the i-th step during
time interval t, andfi the frequency at the i-th step.
Now, the frequency at the last step is obtained; therefore,
we can immediately calculate the maximum allowable
change of frequency at the last step, namely, fmax
fmin fLfL in p.u./s. Similar to Eq. (7), the modified
expression can be rewritten as below:
fmin fL
fLkPmax 13
Eq. (13) can be also rearranged as:
Pmax
1
k
fmin
fL 1
14
At this stage, the modified amount of shed of load is calcu-
lated as follows:
PLS PL Pmax 15
By Eqs. (5) and (14), Eq. (15) can be further formulated as
follows:
P
LS
2Hm0
1
k
fmin fL
fL
16
whereP LSrepresents the total amount of shed load after the
modification, and Pmax the maximum allowable overload
after the last step. This equation informs the new amount of
load to be shed in anticipation of restoring the system
frequency.
4. Numerical simulations
To verify the proposed algorithm, numerical simulations
have been performed. For an isolated system, the maximum
load subject to a shedding schedule may include all loads.
Therefore, in the simulation, the maximum disturbance at
100% of the generation (i.e. PL 10 pu) in an isolated
power system was investigated. Under this situation, the
initial rate of change of frequency m0 can be obtained
through Eq. (5),m0 01111 pusThe value ofPL,minwas thus computed as 0.267 p.u. By Eq. (9), the total
amount of shed load PLS was concluded to be 0.733 p.u.
As mentioned above, the 3, 4 and 5 steps were selectedas the number of shedding steps. In each case, two
strategies (Strategy A, B) for shedding the load are
considered:
Strategy A: The loads are shed equally in each step.
Strategy B: The loads are shed in the geometry mean
form. Namely, the load shed at the second step is of the
half amount of load shed at the first step.
Tables 1 and 2 tabulate the load shed by strategy A and
strategy B, respectively. In Table 1, the load can be seen
shed equally at each step. For example, if the 3-step shed-
ding-load is selected, the amount of the shed load at eachstep is equal to 07333 0244 pu While in Table 2, by
strategy B, each step of shed load will be 0733 47
0419 pu 0733 27 0209 pu and 0733 17
0105 pu In these tables, we use the scheme 1A-3 to
symbolize the 3-step load-shedding by strategy A, and the
scheme 1B-3 to symbolize the 3-step load-shedding by strat-
egy B. Figs. 2 and 3 plot the frequency change curve by
these strategies, respectively. From these figures, the mini-
mum frequency is seen lower than fmin 576 Hz by both
strategies. The load shedding process was not successfully
accomplished.
To improve the above load-shedding process, the Step 7
mentioned in Section 3 has been adopted. In this modifica-tion process, the shedding frequency at the last step, fL, was
first calculated. Table 3 shows the shedding frequency at
each step. In the table, the fL of 58.694, 58.294 and
57.899 Hz that is seen for the scheme 1A-3, 1A-4 and 1A-
5 at their individual final step. Similarly, the fL values of
58.895, 58.648 and 58.431 Hz is seen for the scheme 1B-3,
1B-4 and 1B-5, respectively. Now, by substituting these
values of fL into Eq. (16), the amount of total shed load
can be determined. Table 4 lists the required shed-load
before and after modification. For example, after the modi-
fication process, the new amount of load required to shed for
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 515854
Table 1
Amount of load shed per step by Strategy A
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
1A-3 0.245 0.244 0.244
1A-4 0.184 0.183 0.183 0.1831A-5 0.147 0.147 0.147 0.146 0.146
Table 2
Amount of load shed per step by Strategy B
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
1B-3 0.419 0.209 0.105
1B-4 0.391 0.195 0.098 0.049
1B-5 0.378 0.189 0.095 0.047 0.024
-
8/12/2019 1-s2.0-S0142061599000356-main
5/8
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 5158 55
Fig. 2. Load frequency curve of Scheme 1A.
Fig. 3. Load frequency curve of Scheme 1B.
Table 3
Shedding frequency of Strategy A
Strategy Shedding frequency at each step (Hz)
1 2 3 4 5
1A-3 59.5 59.018 58.694
1A-4 59.5 58.979 58.578 58.294
1A-5 59.5 58.955 58.508 58.157 57.899
1B-3 59.5 59.129 58.895
1B-4 59.5 59.111 58.848 58.648
1B-5 59.5 59.103 58.828 58.614 58.431
Table 4
Required shed load before and after modification
Strategy Total amount of load shed (p.u.)
P LS PLS Increment
1A-3 0.733 0.8442 0.1112
1A-4 0.733 0.9005 0.1675
1A-5 0.733 0.9568 0.2238
1B-3 0.733 0.8162 0.0832
1B-4 0.733 0.8506 0.1176
1B-5 0.733 0.8811 0.1481
-
8/12/2019 1-s2.0-S0142061599000356-main
6/8
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 515856
Fig. 4. Load frequency curve of Scheme 2A.
0 2 4 6 8 10 12 14 1657.5
58
58.5
59
59.5
60
60.5
Time [Sec]
Frequen
cy[Hz]
Scheme 2B-5
Scheme 2B-4Scheme 2B-3
Fig. 5. Load frequency curve of Scheme 2B.
Table 5
Modified amount of load shed per step by Strategy A
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
2A-3 0.2814 0.2814 0.2814
2A-4 0.2252 0.2251 0.2251 0.2251
2A-5 0.1914 0.1914 0.1914 0.1913 0.1913
Table 6
Modified amount of load shed per step by Strategy B
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
2B-3 0.4824 0.2412 0.1206
2B-4 0.4803 0.2401 0.1201 0.0600
2B-5 0.4939 0.2469 0.1235 0.0617 0.0308
-
8/12/2019 1-s2.0-S0142061599000356-main
7/8
the scheme 1A-3, P LS, will be:
1A-3 P LS 10 1
01196
576 58694
58694
08442 pu
Hence, the increment amount of total shed load becomes
08442 0733 01112 puSimilarly, the new amount of
load required to shed for the scheme 1B-3, P LS, can be
computed:
1B-3 P
LS
10 1
01196
576 58895
58895
08162 pu
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 5158 57
Fig. 6. Load frequency curve of Scheme 3A.
Fig. 7. Load frequency curve of Scheme 3B.
Table 7
Modified amount of load shed per step by Strategy A
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
3A-3 0.2721 0.2721 0.2720
3A-4 0.2127 0.2127 0.2126 0.2126
3A-5 0.1763 0.1762 0.1762 0.1762 0.1762
Table 8
Modified amount of load shed per step by Strategy B
Scheme Amount of load shed per step (p.u.)
1 2 3 4 5
3B-3 0.4664 0.2332 0.1166
3B-4 0.4537 0.2268 0.1135 0.0566
3B-5 0.4548 0.2274 0.1137 0.0568 0.0284
-
8/12/2019 1-s2.0-S0142061599000356-main
8/8
Now, because the total amount of load to be shed is
different from the above case, the above scheme 1A is modi-
fied to be schemes 2A and 2B, while the scheme 1B is
modified to be 3A and 3B. Tables 5 and 6 list the newly
determined amount of load shed per step after the modifica-
tion by strategy A and strategy B, respectively. By this new
amount of shed load, Figs. 4 and 5 plot the respective
frequency curve by strategy A and strategy B.
Similarly, for the modified amount of shed load from
Scheme 1B, Tables 7 and 8 list the amount of shed load
per step by strategy A and strategy B, respectively. By the
new amount of shed load, Figs. 6 and 7 individually plot the
frequency curve by strategy A and strategy B.
From the above results, the operation frequency of Figs.
4, 5 and 7 is seen higher than 57.6 Hz, which have demon-
strated the effectiveness of the proposed method. However,
seen from Fig. 6, it does not complete the load-shedding
task. The reason why Fig. 6 failed is that Scheme 3A was
modified from Scheme 1B, and the shed load of Scheme 3A
is less drastic than Scheme 1B. Therefore, the load may notbe shed sufficiently in time, and the frequency drops down
to be lower than the value offmin. The frequency restoration
was thus not successfully accomplished. This outcome also
explains the critical importance of the load-shedding
scheme to the system protection performance.
5. Conclusions
A time-based load shedding protection scheme is
proposed in this paper. In the method, the rate of change
of frequency was detected as the initial value. A predeter-
mined time interval and a modification algorithm thatincludes the maximum frequency change at the last step
are embedded to organize a load shedding strategy. By
case studies with numerical simulations, the proposed
method is verified with simplicity, flexibility, and success
for load shedding problems considering power system
dynamics nature and protection component nature. It can
be a useful integration of load shedding algorithm with a
practical value. Currently, by the assistance of Taiwan
Power Company engineers, a research project is carried
on the extension of the proposed method to cover the control
of pump-storage generating units in Taiwan Power Systems.
The results will be reported in the near future.
References
[1] Jones JR, Kirkland WD. Computer algorithm for selection of
frequency relays for load shedding. IEEE Computer Applications in
Power 1988;1(1):2125.
[2] Concordia C, Fink LH, Poullikkas G. Load shedding on an isolated
system. IEEE Transactions on Power Systems 1995;10(3):1467
1472.
[3] Thompson JG, Fox B. Adaptive load shedding for isolated power
systems. IEE ProceedingsGeneration, Transmission and Distribu-
tion 1994;141(5):491496.
[4] Maliszewski RM, Dunlop RD, Wilson GL. Frequency actuated load
shedding and restoration Part Iphilosophy. IEEE Transactions on
Power Apparatus and Systems 1971;PAS-90(4):14521459.
[5] Horowitz SH, Polities A, Gabrielle AF. Frequency actuated load
shedding and restorationPart IIimplementation. IEEE Transactions
on Power Apparatus and Systems 1971;PAS-90(4):14601468.[6] Anderson PM, Mirheydar M. A low-order system frequency response
model. IEEE Transactions on Power Systems 1990;5(3):720729.
[7] Anderson PM, Mirheydar M. An adaptive method for setting under-
frequency load shedding relays. IEEE Transactions on Power Systems
1992;7(2):647655.
[8] Lokay HE, Burtnyk V. Application of under-frequency relays for
automatic load shedding. IEEE Transactions on Power Apparatus
and Systems 1968;PAS-87(5):13621366.
[9] Chuvychin VN, Gurov NS, Venkata SS, Brown RE. An adaptive
approach to load shedding and spinning reserve control during
under-frequency conditions. IEEE Transactions on Power Systems
1996;11(4):18051810.
[10] Shilling SR. Electrical transient stability and under-frequency load
shedding analysis for a large pump station. IEEE Transactions on
Industry Applications 1997;33(1):194201.[11] Hicks KL. Hybrid load shedding is frequency based. IEEE Spectrum
1983;02:5256.
[12] Kottick D, Or O. Neural-networks for predicting the operation of an
under-frequency load shedding system. IEEE Transactions on Power
Systems 1996;11(3):13501358.
[13] Taylor CW, Nassief FR, Cresaf RL. Northwest power pool transient
stability and load shedding controls for generation load imbalances.
IEEE Transactions on Power Apparatus and Systems 1981;PAS-
100(7):34863495.
[14] Smaha DW, Rowland CR, Pope JW. Coordination of load conserva-
tion with turbine-generator under-frequency protection. IEEE Trans-
actions on Power Apparatus and Systems 1980;PAS-99(3):1137
1150.
[15] Prasetijo D, Lachs WR, Sutanto D. A new load shedding scheme for
limiting under-frequency. IEEE Transactions on Power Systems1994;9(3):13711378.
[16] Halevi Y, Kottick X. Optimization of load shedding system. IEEE
Transactions on Energy Conversion 1993;8(2):207213.
[17] IEEE Committee Report. Dynamic models for steam and hydro
turbines in power system studies. IEEE Transactions on Power Appa-
ratus and Systems 1973;PAS-92:19041915.
[18] Nagrath IJ, Kothari DP. Power system engineering, New Delhi: Tata
McGraw-Hill, 1994. p. 339376.
S.-J. Huang, C.-C. Huang / Electrical Power and Energy Systems 22 (2000) 515858
Shyh-Jier Huang received his PhD degree in Electrical Engineering
from the University of Washington, Seattle, in 1994. Currently, he is
with Department of Electrical Engineering and is the project manager
in Computational Intelligence Applied to Power (CIAP) laboratory at
National Cheng Kung University, Taiwan. He has worked research
projects at the University of California, Berkeley from 1989 to 1991.
He received Research Awards from National Science Council, Taiwan,
from 1996 to 1999. He is a member of IEEE PES, CSS, CS and SMC.
His main areas of interest are power system analysis, neural networks
and fuzzy control.
Chin-Chyr Huang received his MS degree from National Cheng Kung
University in 1992. He is pursuing his PhD degree in the same univer-
sity. His major interests are power system analysis and fuzzy controls.