Application of HSA and GA in Optimal Placement of FACTS Devices Considering Voltage Stability and Losses
Application of HSA and GA in Optimal Placement of FACTS Devices Considering Voltage Stability and Losses
Application of HSA and GA in Optimal Placement of FACTS Devices Considering Voltage Stability and Losses
Wh
H
V
L
V
G
Z
LL
ow equation ha
rticular, they fo
a power flow
oposed an inde
ultiple power f
at a pair of mu
d merges at th
nditions get
esented an ind
6]. The index
mount of reac
stability. Kesse
th the power f
wer system
vantage of the
lculation. Thom
ed the minim
composition te
the method
atrix technique
V curves have
udying voltage
nceptually sho
ltage stability
used due to the
Fig. 2 P-V curve
Voltage Stabil
he voltage stab
rcuit theory. it
ay be divided in
L
:load nodes
G
:generation no
he transmission
[ ]
L L
G G
V I
H
I V
=
(
(
here:
H: hybrid matrix
(IL): voltage (
G
(IG): voltage
L
, F
LG
, K
GL
, Y
G
as been presen
focused on the
w solution is
ex that employ
flow solutions
ultiple power
he saddle bifu
heavy-loaded
dex with the o
shows that a
ctive power
el and Glavitsc
flow calculatio
conditions w
e index is to
mas and Tiran
mum singular
echnique [18].
of Thomas an
[19,20].
been tradition
e stability in e
ows the impact
of a hypotheti
e computationa
es for studying v
lity Index L
ility index L is
is considered
nto
L
o and
G
o
odes
n system may b
LL
G GL
Z
K
=
(
(
x
(current) at loa
(current) at gen
G
: sub-matrix o
nted by Afterw
sign of the dete
stable or not
yed an angle b
[15]. That is b
flow solutions
urcation point
gradually. C
optimal power
a power syste
if it is app
ch developed a
n [17]. The in
with the hybr
require only
nuchit presente
value of the
Afterwards, L
nd Tiranuchit
nally used as g
electric power
t of a synchron
cal node. In th
al efficiency.
voltage stability i
s based on the
d that nodes in
G
.
be written as:
LG L
G GG
F I
V Y
( (
( (
d node
neration node
of matrix H
wards in [14].
erminant to jud
. Tamura, et
between a pair
based on the f
s becomes clo
as power syst
Carpentier, et
flow calculat
em lacks a la
proaches volta
an index called
dex evaluates
rid matrix. T
one power fl
ed a method t
e singular va
Lof, et al. speed
with the spa
graphical tools
r systems. Fig
nous generator
his paper, index
in power system
hybrid matrix
a power netwo
(
In
dge
al.
r of
fact
oser
tem
al.
tion
arge
age
d L
the
The
low
that
alue
ded
arse
for
g. 2
on
x L
ms
x of
ork
14)
The volt
1
j
L =
Where
Therefo
may be
L Ma =
Index L
without
above a
voltage
Over
have
optimiz
existing
metaheu
enginee
meta-he
random
Recentl
(HS) m
techniqu
perform
an aesth
find a g
In musi
the poss
the pitc
each pl
harmon
optimiz
value w
vector [
In re
each de
replaced
The s
Step
paramet
Minim
f(x) :
x : se
Xi :
variable
N : is
The H
These
-
-
-
tage stability in
0 j
j
V
V
+
0
G
j j
i
V F
o e
=
_
ore, the voltag
expressed as:
j
MaxL j o e
L varies from 0
t load and L=
allows the sy
instability.
IV. H
r the last four
been develop
zation problem
g numerical me
uristic algorit
ering optimiza
euristic algorit
mness to imitate
ly, Geem et al.
meta-heuristic
ue inspired
mances seek to
hetic standard,
global solution
ic improvisatio
sible range, tog
ches make a go
layer's memor
ny is increase
zation, each d
within the poss
[23-25].
eal optimizatio
ecision variable
d with each var
steps in the pro
1. Initialize t
ters.
mize F(x) s.t
objective func
et of each desig
set of the po
e (Lxi <Xi < U
s the number o
HS algorithm p
e are:
HMS : harm
solution vect
HMCR : harm
PAR : pitch a
ndex at load nod
ji i
V
ge stability ind
L
o
to 1, where L=
=1 implies vo
stem operator
ARMONY SEAR
decades, a la
ped to solv
ms. The com
ethods have fo
thms based o
ation problems
thms is that
e natural pheno
. [22] develope
algorithm t
by music
find pleasing
just as the opt
n as determined
on, each playe
gether making
ood harmony, t
ry, and the po
d next time.
decision variab
ible range, tog
on, each music
e, and its prefe
riables preferr
ocedure of harm
the optimizatio
xi
e
Xi , i=1,
ction
gn variable (xi)
ossible range
Uxi) .
f design variab
parameters are
mony memory
tors in the harm
mony memory
adjusting rate ;
de j may be writ
dex for the wh
=0 means a po
ltage collapse
r to evaluate a
RCH ALGORITH
arge number o
ve various
mputational dr
orced researche
on simulation
s. The commo
they combine
omena.
ed a new Harm
that is an
phenomenon
harmony as de
timization proc
d by an objecti
er sounds any
one harmony
that experience
ossibility to m
Similarly in
ble initially c
gether making
cian can be re
erred sound pi
red values.
mony are as fol
on problem an
2,3,..N.
)
of values for
bles
also specified
y size or the
mony memory;
y considering ra
;
tten as [21]:
(15)
hole network
(16)
wer network
. The index
a margin to
M
f algorithms
engineering
rawbacks of
ers to rely on
ns to solve
on factor in
e rules and
mony Search
optimization
n. Musical
etermined by
cess seeks to
ive function.
pitch within
vector. If all
e is stored in
make a good
engineering
chooses any
one solution
eplaced with
tches can be
llows;
nd algorithm
each design
in this step.
number of
ate ;
World Academy of Science, Engineering and Technology 56 2009
740
- N : number of decision variables ;
- NI :number of improvisations;
- Stopping criterion.
Step 2. Initialize the harmony memory (HM).
The harmony memory (HM) is a memory location where
all the solution vectors (sets of decision variables) are stored.
This HM is similar to the genetic pool in the GA. HM matrix
is filled with as many randomly generated solution vectors as
the HMS.
1 1 1 1
1 2 1
2 2 2 2
1 2 1
1 1 1 1
1 2 1
1 2 1
...
...
...
...
N N
N N
HMS HMS HMS HMS
N N
HMS HMS HMS HMS
N N
x x x x
x x x x
HM
x x x x
x x x x
(
(
(
(
=
(
(
(
# # # # #
(17)
Step 3. Improvise a new harmony
In this Step, a new harmony vector,
1 2
( , , ... )
n
x x x x
' ' ' '
= , is
generated based on three rules: (1) memory consideration, (2)
pitch adjustment and (3) random selection. Generating a new
harmony is called improvisation.
The value of the first decision variable
1
( ) x' for the new
vector can be chosen from any value in the specified HM
range
1
1 1
( )
HMS
x x . Values of the other design variables
2 3
( , , ... )
n
x x x
' ' ' are chosen in the same manner.
The HMCR, which varies between 0 and 1, is the rate of
choosing one value from the historical values stored in the
HM, while (1-HMCR) is the rate of randomly selecting one
value from the possible range of values.
{ }
1 2
, , ...,
(1 )
HMS
i i i i
i
i i
x x x x with probability HMCR
x
x X with probability HMCR
' e
'
'
e
(18)
For instance, a HMCR of 0.90 indicates that the HS
algorithm will choose the decision variable value from
historically stored values in the HM with the 90% probability
or from the entire possible range with the 10090%
probability. Every component of the new harmony vector,
1 2
( , , ... )
n
x x x x
' ' ' '
= , is examined to determine whether it should
be pitch-adjusted.
This operation uses the PAR parameter, which is the rate of
pitch adjustment as follows:
1 ( )
i
Yes with probability PAR
x
No with probability PAR
'
(19)
The value of (1-PAR) sets the rate of doing nothing. If the
pitch adjustment decision for x
i
' is yes,
i
x
'
is replaced as
follows:
( 1,1)
i
x x bw U ' ' + (20)
bw: arbitrary distance bandwidth for the continuous design
variable
U (-1, 1): uniform distribution between -1 and 1.
In Step 3, HM consideration, pitch adjustment or random
selection is applied to each variable of the new harmony
vector in turn.
Step 4.Update harmony memory
If the new harmony vector,
1 2
( , , ... )
n
x x x x
' ' ' '
= , is better than
the worst harmony in the HM, from the point of view
objective function value, the new harmony is included in the
HM and the existing worst harmony is excluded from the HM.
Step 5.Check stopping criterion
If the stopping criterion (maximum number of
improvisations) is satisfied, computation is terminated.
Otherwise, Steps 3 and 4 are repeated.
V. GENETIC ALGORITHM
GA is an evolutionary computing method in the area of
artificial intelligence. It was pioneered by Holland in the 60's
and 70's and his work is comprehensively presented in [26]. It
is a global search algorithm that is based on concepts from
natural genetics and the Darwinian survival-of-the-fittest code.
Meta-heuristic algorithm-based engineering optimization
methods, including GA, have occasionally overcome several
deficiencies of conventional numerical methods. Genetics is
usually used to reach to a near global optimum solution. In
each iteration of GA (referred as generation), a new set of
string (.i.e. chromosomes) with improved fitness is produced
using genetic operators (i.e. selection, crossover and
mutation). Useful practical details of genetic algorithms are
available in [27] and [28].
A. Binary Encoding
The most common way of encoding is a binary string, which
would be represented as in Fig. 3.
Chromosome 1
Chromosome 2
1
0
1
1
0
0
1
1
0
1
0
1
0
1
0
1
1
1
1
0
0
0
.
.
Chromosome n 1 1 0 1 1 1 0 0 1 0 1
Fig. 3. Typical Chromosome for binary encoding
Each chromosome encodes a binary (bit) string. Each bit in the
string can represent some characteristics of the solution. Every
bit string therefore is a solution but not necessarily the best
solution. Another possibility is that the whole string can
represent a number. The way bit strings can code differs from
problem to problem. Binary encoding gives many possible
chromosomes with a smaller number of alleles. On the other
hand this encoding is not natural for many problems and
sometimes corrections must be made after genetic operation is
completed. Binary coded strings with 1s and 0s are mostly
used. The length of the string depends on the accuracy [29].
B. Selection
Selection is the process of choosing two parents from the
population for crossing. In proposed GA, method of
tournament selection is used for selection.
C. Cross Over
Crossover is the process of taking two parent solutions and
producing from them a child. After the selection process, the
World Academy of Science, Engineering and Technology 56 2009
741
pop
ma
Cr
tha
tec
wh
cor
E.
Af
Mu
mi
gen
inf
los
Ga
ran
F.
Th
in
con
sho
and
ref
In
con
ter
tra
con
PQ
con
ter
G
l
i
V
l
i
Q
A.
In
FA
inv
pro
har
In
Q
F
pulation is en
akes clones of
ossover operat
at it creates a
chniques. In th
here the two
rresponding po
Mutation
fter crossover
utation preven
nimum. Muta
netic materials
formation. It is
ss of genetic m
aussian distribu
ndom number i
Fitness Functi
he goal of optim
order to enhan
nsidering mod
ould be placed
d transformers
ference value.
most of the no
nsidered by d
rms. In this p
ansformers, and
ntrolled variab
Q buses and
nstraints that
rm. The initial
Gi
: Penalty fact
Vi
: Penalty fact
max
lim
min
;
;
i i
i i
V V
V V
=
max
lim
min
;
;
i
i
i
Q Q
Q Q
=
Consideration
this part of the
ACTS devices
vestigated and
oblem has bee
rmony search a
this study belo
- Finding
devices
lim
(
V
Q Vi
i N
f
e
= +
_
nriched with be
f good strings
tor is applied to
better offsprin
his paper single
o mating chr
oints.
, the strings
nts the algorit
ation plays th
s as well as fo
s an insurance
material. In this
ution to the p
is selected.
ion
mization algori
nce power sys
dels of FACT
d to prevent in
s and maintai
onlinear optimi
defining the o
problem the v
d the amount o
bles which are
injected react
are add to the
object function
tor
tor
max
min
i
i
V
V
>
<
max
min
i i
i i
Q Q
Q Q
>
<
VI. SIMULA
ns
e research optim
in a IEEE 30
seven scenari
n solved by th
algorithm (HSA
ow cases have b
the optimal lo
consisting T
lim 2
( )
i i
i N
V V
e
+
_
etter individua
but does not
o the mating p
ng. There are v
e point crossov
romosomes ar
are subjecte
thm to be tra
e role of rec
or randomly d
e policy agains
s paper, a rando
parents. For ea
ithm is to place
tem stability l
TS devices. S
nstability in tr
in bus voltage
ization problem
object functio
voltages of PV
f reactive pow
e self constrai
tive powers o
e object funct
n is expressed a
ATION RESULTS
mal placement
bus power sy
ios have been
he Genetic alg
A).
been considere
ocation of three
TCPAR, UPFC
lim
(
Q
Gi Gi
N
Q Q
_
als. Reproduct
create new on
ool with the ho
various crosso
ver has been u
re cut once
ed to mutati
apped in a lo
covering the l
disturbing gene
st the irreversi
om vector from
ach chromosom
e FACTS devi
evel (Section I
So these devi
ransmission lin
es close to th
ms constraints
on using pena
V buses, tap
wer installation
ned. Voltages
of PV buses
tion as a pena
as:
(
)
(2
(2
S
t of three types
ystem (Fig. 4)
considered. T
gorithm (GA) a
ed.
e types of FAC
C and SVC
lim 2
max
)
Gi
Q L +
tion
nes.
ope
ver
sed
at
ion.
ocal
lost
etic
ible
m a
me,
ces
III)
ices
nes
heir
are
alty
of
are
of
are
alty
21
22)
23)
s of
are
This
and
CTS
by
-
The con
two par
individu
string r
network
FACTS
devices
convert
TCPAR
degrees
tcpar
UPFC:
degrees
upfc =
It shoul
and also
FACTS
B. Netw
The 30
study th
generato
(6,10),
transfor
installat
PV bus
swing b
has assu
source
paramet
of G1-G
genetic algor
different stat
Studding the
the power lo
there optimal
nfiguration of
rameters: the l
ual represents
represents the
k and the secon
S device. Acc
, the rated va
ted in to the rea
R: The working
s to +5 degrees
r =RV 5(degr
The working
s to +180 degre
= RV 180(deg
ld be noticed th
o the power sy
S devices is 17.
work data
0_bus IEEE p
hat is shown
or buses and
(4,12) and
rmer branches
tion buses are
ses as follow:
bus. The variab
umed that the
installation ar
ters limits are
G6 are shown i
Fig. 4
rithm and harm
es.
e impact of FA
oss transmission
l placement.
FACTS devic
location of dev
a string whic
location of t
nd value repres
cording to th
alues (RV) o
al compensatio
g range of the
.
ree)
g range of the
ees.
gree)
hat the RV par
ytem transmiss
.557 MW.
ower system h
in Fig. 4. Con
20 load bus
(27,28) are
s. The possib
3, 10 and 24.
2, 5, 8, 11 an
ble limits have
transformer ta
re continuous
represented in
in Table.III.
IEEE 30 bus po
mony search a
ACTS devices
n and stability
ces have been
vices and rate
ch the first va
the FACTS de
sents the rated
e model of
f each FACT
on as follow:
TCPAR is bet
e UPFC is be
ameter is betw
sion loss in the
has been selec
nsists of 48 b
ses. Four bran
under load
le reactive po
. Six buses are
d 13. The bus
e been given i
aps and the rea
variables. Al
n Table.II. The
ower system
algorithm in
s settings on
y considering
obtained by
value. Each
alue of each
evice in the
value of the
the FACTS
TS device is
tween the 5
(24)
etween -180
(25)
ween -1 and 1
e absence of
cted for this
branches, six
nches, (6,9),
tap setting
ower source
e selected as
s No.1 is the
in Table.I. It
active power
so the other
e generations
World Academy of Science, Engineering and Technology 56 2009
742
TABLE.I
VARIABLE LIMITS (P.U)
Variables V
G
V
PQ
T Q
C
Max 1.1 1.05 1.1 0.3
Min 0.9 0.95 0.9 -0.3
TABLE.I
LIMITATION OF THE PV REACTIVE POWER GENERATION
Bus 1 2 5 8 11 13
Q
C
Max
2.0
1.0 0.8 0.6 0.5 0.6
Q
C
Min
-0.2
-0.2 -0.15 -0.15 -0.1 -.15
TABLE.III
GENERATIONS of G
1
-G
6
P
1
P
2
P
3
P
4
P
5
P
6
99.2MW 80 MW 50 MW 20 MW 20 MW 20 MW
C. Considered scenarios
Three scenarios have been considered and in each scenarios
GA and HSA have been implemented for finding optimal
placement and best rate values for FACTS devices.
Case 1: one-type FACTS allocation using GA and HSA
In this case, allocation of one-type FACTS devices has
been performed using GA and HSA. Three different kinds of
FACTS devices (TCPAR, UPFC, and SVC) have been used to
be placed in optimal location solely to enhance power system
stability index and reduce losses.
Results for one-type FACTS allocation using GA and HSA
have been compared and presented in Tables IV-VI.
TABLE.IV
OPTIMAL LOCATION AND RELATED PARAMETES FOR SVC
Algorithm SVC
Location
(Bus No.)
Size
(MVA)
Losses
(MW)
L (stability
index)
GA 32 21.23 5.28 0.1431
HSA 12 12 5.22 0.1400
TABLE.V
OPTIMAL LOCATION AND RELATED PARAMETES FOR UPFC
Algorithm UPFC
Location
(Branch)
Rate
Losses
(MW)
L (stability
index)
GA 9 -0.6 5.21 0.140
HSA 6 -0.8 5.1 0.138
TABLE.VI
OPTIMAL LOCATION AND RELATED PARAMETES FOR TCPAR
Algorithm TCPAR
Location
(Branch)
Rate
Losses
(MW)
L (stability
index)
GA 10 -0.341 5.36 0.1464
HSA 23 -0.435 5.30 0.1412
Case 2: two-type FACTS allocation using GA and HSA
In this case, allocation of two-type FACTS devices has
been performed using GA and HSA. Two different kinds of
FACTS devices among TCPAR, UPFC, and SVC have been
selected and are located in optimal location two by two to
enhance power system stability index and reduce losses.
Consequences for two-type FACTS allocation by GA and
HSA have been assessed and presented in Tables VII-VII.
TABLE.VII
OPTIMAL LOCATION AND RELATED PARAMETES FOR TWO TYPES OF FACTS
DEVICES (SVC+TCPAR)
SVC TCPAR
GA
Locatio
n (Bus
No.)
Size
(MVA)
Location
(Branch) Rate
Losses
(MW)
L
(Stability
Index)
17 -2 15 0.6453 5.201 0.1382
HSA 8 -22 33 0.7810 5.180 0.1422
TABLE.VIII
OPTIMAL LOCATION AND RELATED PARAMETES FOR TWO TYPES OF FACTS
DEVICES (SVC+UPFC)
SVC UPFC
GA
Location
(Bus No.)
Size
(MVA)
Location
(Branch) Rate
Losses
(MW)
L
(Stability
Index)
17 -2 17 0.4223 5.278 0.1401
HSA 18 -13 21 -0.6681 5.1765 0.1444
Case 3: multi-type FACTS allocation using GA and HSA
In the last scenario, three types controller simultaneously
have been implemented to improve voltage stability index and
losses. Table IX shows optimal location of devices in this
scenario.
This results show that installation of multi-type FACTS
devices can lead to improve in voltage stability index and
reduce in power system losses simultaneously. So multi-type
FACTS devices should be placed in optimal location to both
improve stability margins and reduce losses in the network.
VI. CONCLUSION
In this paper, a method for placement of multi-type
FACTS devices based on Harmony Search (HS) and Genetic
Algorithm (GA) has been presented. Three types of FACTS
devices (SVC, TCPAR, and UPFC) have been modeled. The
criteria for optimization were considered as the voltage
stability index and the minimization of losses. For
investigation of the purposes, Different scenarios have been
considered. In the first scenario TCPAT, UPFS, and SVC are
placed exclusively in transmission lines and indices have been
calculated. Then two types of above controller try to improve
parameters randomly. Next three types controller
simultaneously have been implemented to improve voltage
stability index and losses.
The results of execution of these scenarios on a typical 30-bus
test system were clarified robustness of this method in optimal
and fast placement of FACTS devices. The results showed
efficiency of this method for improvement of voltage profile,
reduction of power losses and also an increase in power
transfer capacity, maximum loading and voltage stability
margin.
World Academy of Science, Engineering and Technology 56 2009
743
TABLE.IX
OPTIMAL LOCATION AND RELATED PARAMETES FOR THREE TYPES OF FACTS DEVICES (SVC+UPFC+TCPAR)
Algorithm
SVC TCPAR UPFC
Location
(Bus No.)
Size(MVA)
Location
(Branch)
Size (MVA)
Location
(Branch)
Size (MVA)
Losses
(MW)
L (stability
index)
GA
16 17.121 9 0.769 7 -1 5.211 0.1333
HSA
25 29.21 6 0.879 12 -0.8453 5.120 0.1301
REFERENCES
[1] S.N. Singh. Location of FACTS Devices for Enhancing Power
Systems Security 2001 IEEE
[2] N G. Hingorani, L. Gyugyi, Understanding FACTS: Concepts and
Technology of Flexible AC Transmission Systems, IEEE Press, New-
York, 2000.
[3] J.G. Singh, S.N. Singh, and S.C. Srivastava. Enhancement of Power
System Security through Optimal Placement of TCSC and UPFC
[4] Sung-Hwan Song, Jung-Uk Lim, and Seung-I1 Moon, FACTS
Operation Scheme for Enhancement of Power System Security Paper
accepted for presentation at 2003 IEEE Bologna PowerTech Conference,
June 23-26, Boloppa, Italy
[5] D. Arabkhaburi, A. Kazemi, M.Yari, J. Aghaei. Optimal Placement of
UPFC in Power Systems Using Genetic Algorithm '2006 IEEE
[6] S. Gerbex, R Cherkaoui and AJ Germond, Optimal location of FACTS
devices to enhance power system security, IEEE Power tech
[7] C. A. Cnizares, Applications of optimization to voltage collapse
analysis, in Proc. IEEE Power Eng. Soc. Summer Meeting, San Diego,
CA, Jul. 1998.
[8] J. V. Milanovic and T. M. David, Stability of distribution networks
with embedded generators and induction motors, in Proc. Winter
Meeting Power Eng. Soc., New York, 2002, vol. 2, pp. 10231028.
[9] K. Sedghisigarchi and A. Feliachi, Control of grid-connected fuel cell
power plant for transient stability enhancement, in Proc. Winter
Meeting Power Eng. Soc., New York, Jan. 2002, vol. 1, pp. 383388.
[10] I. A. Hiskens and Y. V. Makarov, Calculation of power system critical
loading conditions, in Proc. Electrical Engineering Congr., Sydney,
Australia, Nov. 1994, pp. 185189.
[11] Y. Kataoka, A probabilistic nodal loading model and worst case
solutions for electric power system voltage stability assessment, IEEE
Trans. Power Syst., vol. 18, no. 6, pp. 15071514, Nov. 2003.
[12] IEEE Tutorial, FACTS application, IEEE Power Engineering Society
Winter Meeting 1996, Publication No. 96TP116-0
[13] V. A. Venikov and M. N. Rozonov, "The stability of a load," Izv. Akad.
Nauk SSSR. (Energetika i Atromatica), pp. 121-125, 1961.
[14] V. A. Venikov, V. A. Storev, V. I. Idelchick and V. I. Trasov,
"Estimation of Power System Steady-State Stability," IEEE Trans.
Power Syst., Vol. PAS-94, pp. 1034-1041, Mar. Apr. 1975.
[15] Y. Tamura, H. Mori and S. Iwamoto, "Relationship between Voltage
Instability and Multiple Load Flow Solution in Electric Power Systems,"
IEEE Trans. Power App. and Syst., Vol. PAS-102, No. 5, pp. 1115-1123,
May 1983.
[16] J. Carpentier, R. Girard and E. Scano, "Voltage Collapse Proximity
Indicators Computed from Optimal Power Flow," in Proc. 1984 8
th
Power System Computing Conference, Helsinki, pp. 671-678.
[17] P. Kessel and H. Glavitsch, "Estimating the Voltage Stability of Power
System," IEEE Trans. Power Delivery, Vol. PWRD-1, No. 3, pp. 346-
354, July 1986.
[18] A. Tiranuchit and R. J. Thomas, "A Posturing Strategy against Voltage
Instability in Electric Power Systems," IEEE Trans. Power Syst., Vol. 3,
No. 1, pp. 87-93, Feb. 1988.
[19] P. -A. Lof, T. Smed, G. Anderson and D. J. Hill, "Fast Calculation of a
Voltage Stability Index," IEEE Trans. Power Syst., Vol. 7, No. 1, pp.
54- 64, July 1992.
[20] A. A. El-Keib and X. Ma, "Application of Artificial Neural Networks in
Voltage Stability Assessment," IEEE Trans. Power Syst., Vol. 10, No. 4,
pp. 1890-1895, Nov. 1995.
[21] H. Mori, Member, IEEE, and Y. Komatsu '' A Hybrid Method of
Optimal Data Mining and Artificial Neural Network for Voltage
Stability Assessment''
[22] Kang Seok Lee, Zong Woo Geem . A new meta-heuristic algorithm for
continuous engineering optimization: harmony search theory and
practice, 2004 Elsevier B.V. doi:10.1016/j.cma.2004.09.007
[23] Zong Woo Geem, Novel derivative of harmony search algorithm for
discrete design variables Environmental Planning and Management
Program, Johns Hopkins University,729 Fallsgrove Dr. #6133,
Rockville, MD 20850, USA
[24] Geem ZW, Kim JH, Loganathan GV. A new heuristic optimization
algorithm: harmony search. Simulation 2001;76 (2):608.
[25] K. Lee, Z. Geem, A new meta-heuristic algorithm for continuous
engineering optimization: harmony search theory and practice,
Computer Methods in Applied Mechanics and Engineering 194 (2005)
(2005) 39023933.
[26] J. H. Holland, Adaptation in Natural and Artificial Systems. Ann Arbor,
MI: University of Michigan Press, 1975
[27] R. H. Haupt, and S. E. Haupt, Practical Genetic Algorithms, 2nd Ed.
Hoboken, NJ: John Wiley & Sons, 2004
[28] D. L. Chambers, The Practical Handbook of Genetic Algorithms, Vol. I,
2nd Ed. London: Chapman & Hall, 2001
[29] S.N.Sivanandam S.N.Deepa Introduction to Genetic Algorithms
Spinger
World Academy of Science, Engineering and Technology 56 2009
744