Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 1,
April 201
6, pp. 1 ~ 10
DOI: 10.115
9
1
/ijeecs.v2.i1.pp1-10
1
Re
cei
v
ed
Jan
uary 21, 201
6
;
Revi
sed Ma
rch 3, 2
016;
Acce
pted Ma
rch 1
4
, 2016
A Novel Method Based on Teaching-Learning-Based
Optimization for Recloser Placement with Load Model
Consideration in Distribution System
Sina Khajeh
Ahmad
Attar
i
*, Mohamad Bakhs
h
ipou
r, Mahmoudr
eza Sha
kara
m
i,
Farhad Nam
d
ari
Dep
a
rtement o
f
Electrical Eng
i
ne
erin
g, Lores
tan Univ
ersit
y
,
Dan
e
shg
ah Str
eet, 712
34-9
8
6
53, Khorram
a
b
ad, Loresta
n, Iran
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: sinaattari
@g
mail.com
A
b
st
r
a
ct
T
h
is pap
er pro
pose
d
a nove
l
techniq
ue bas
ed on
teac
hin
g
-le
a
rni
ng-b
a
s
ed opti
m
i
z
a
t
io
n (T
LBO)
alg
o
rith
m i
n
or
der to fin
d
o
p
ti
ma
l pl
ace
m
ent
of recl
os
ers i
n
the d
i
stributi
on n
e
tw
orks which is
ap
pli
e
d
to
improve r
e
li
abi
lity. Reclos
ers use to
el
i
m
in
ate transie
nt faul
ts, faults
isolati
on, netw
o
rk
mana
ge
me
nt a
n
d
enh
anc
e re
lia
bi
lity to r
educ
e c
u
stomer
outa
g
e
s. Accord
ing
t
o
recl
oser
rol
e
in
netw
o
rk rel
i
a
b
ility, th
e cost f
o
r
the inst
all
a
tion
and
mai
n
ten
ance
must b
e
sustain
ed
by
distrib
u
tio
n
c
o
mpa
n
ies. T
h
erefore, s
e
lect
in
g
sufficient
nu
mber a
n
d
suita
b
l
e l
o
cati
on for
reclos
ers
ar
e
i
m
porta
nt iss
ue. In this
p
a
per, the
pro
p
o
s
ed
obj
ective fu
nction for
opti
m
al
reclos
er n
u
m
ber a
nd
pl
ace
m
e
n
t has
be
e
n
for
m
ul
ated t
o
i
m
pr
ove thr
e
e
relia
bi
lity ind
i
ce
s w
h
ich consists of three terms; i.e.
System
Averag
e Interruptio
n F
r
eque
n
cy Index (SAIFI),
System Aver
a
ge Interru
ptio
n
Duratio
n
Ind
e
x
(SAIDI
) and
Averag
e Ener
g
y
Not Sup
p
li
ed
(AENS). Besi
des
the loa
d
mode
l
effectiveness
has be
en co
ns
ider
ed to t
he simulati
on. T
o
verify the efficie
n
cy of propos
e
d
m
e
thod, it
has been
co
nducted to IEEE 69-bus radi
al distribution system
.
The ob
tained simula
tion r
e
sults
de
mo
nstrate th
e reli
abi
lity i
m
p
r
ove
m
e
n
t.
Ke
y
w
ords
: opt
imal recl
oser p
l
ace
m
e
n
t, relia
bility i
m
pr
ove
m
ent, Loa
d mod
e
l, TLBO
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
I
n
dist
ribut
io
n
powe
r
sy
st
e
m
s re
clo
s
e
r
s
use t
o
elimin
ate transi
ent faults, faults isolatio
n,
netwo
rk man
ageme
n
t and
enha
nce reli
ability indices.
A recl
oser i
s
a
device
wi
th the ability
to
detect ph
ase
and ph
ase-to
-ea
r
th over
cu
rre
nt con
d
itio
ns, to interru
p
t the circuit i
f
the overcu
rrent
persist
s afte
r
a predete
r
mi
ned time,
and
then to
auto
m
atically
re
cl
ose
to
re-ene
rgized th
e lin
e. If
the fault that origin
ated the
operat
ion
still exists, then
the re
clo
s
e
r
will stay op
e
n
after a p
r
e
s
et
numbe
r of o
peratio
ns, th
us i
s
olatin
g the fault
ed
section from t
he re
st of th
e syste
m
. In an
overhe
ad
dist
ribution
sy
ste
m
bet
ween
8
0
to 9
5
p
e
rce
n
t of the fa
ults a
r
e
of the
tempo
r
ary
nat
ure
and la
st, at the mo
st, for
a few
cycle
s
or sec
ond
s.
Thus, th
e re
close
r
, with its openi
ng/cl
osing
cha
r
a
c
teri
stics, prevent
s a distri
b
u
tion
circuit bein
g
left out of service for temporary faults.
Typically, recl
ose
r
s a
r
e
de
sign
ed to
hav
e up
to th
ree
ope
n/clo
s
e
o
peratio
ns an
d
,
after the
s
e,
a
final open o
p
e
ration to lo
ck out the seq
uen
ce
s. One
further
clo
s
in
g ope
ration b
y
manual me
ans
is usually all
o
we
d. The counti
ng me
chani
sms
regi
ster o
peratio
ns
of the ph
ase o
r
ea
rth-fault
units whi
c
h
can also be initiated
by exte
rnally controlle
d
devices
when app
ro
priate
comm
uni
cati
on mean
s a
r
e
available [1].
Becau
s
e the
high co
st of reclo
s
e
r
the
usage m
u
st be e
c
on
omically ju
stified fo
r
distrib
u
tion compani
es. T
he
n
u
mbe
r
a
nd re
closer
l
o
catio
n
in
clu
de m
any fa
ctors e.g.
cu
st
omer
type, load va
riation, in
stall
a
tion an
d ma
intenan
ce
co
st and
etc.
Many re
se
arche
s
h
a
ve b
een
done for pl
acement of prot
ective device
s
.
Optimum swi
t
ch pla
c
eme
n
t has ca
rri
e
d
out with QEA-ba
sed
algorith
m
to improve
cu
stome
r
se
rvice relia
bility [2]. A new compo
s
ite
obje
c
tive function of inve
stment cost
and
reliability on
optimum placement of line switch is
presented [3]. Si
mulated annealing optimi
z
i
n
g
algorith
m
h
a
s
di
scu
s
sed
[4]. Placem
ent form
ulat
ed by
nonlin
ear
bina
ry p
r
og
rammi
ng
[5].
Sectionali
z
e
r
and
re
clo
s
e
r
pla
c
eme
n
t h
a
s
bee
n d
o
n
e
to in
crea
se
relia
bility by usi
ng
discre
te
event simulat
i
on [6]. Improvements of System
Avera
ge Interru
ptio
n Duration Index (SAIDI) and
System Average Interrupti
on Fre
que
ncy
Index (SAI
FI) have be
en
use
d
as th
e o
b
jective fun
c
tion
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 1– 10
2
for the
optim
al pla
c
em
ent
of DG
an
d
re
clo
s
er
by Ant
Col
ony O
p
timization
(A
CO) m
e
thod.
T
he
IEEE 69 and 394-bus test systems
have been selected as test sy
stem
s and the results of the
prop
osed te
chniqu
e have
bee
n
comp
ared
with
th
e re
sult
s of
GA
in t
h
e
s
a
me
sy
st
em
[
7
]
.
Momenta
r
y Average Inte
rru
ption Event Frequ
en
cy
Index (MAIFI) has put
in the objective
function
of [
7
] whi
c
h i
s
in term
s of
three
co
nditi
ons; i.e. lo
a
d
ing
con
d
itio
ns, g
ene
rati
on
penetration l
e
vel and
po
wer facto
r
[8]
.
Similar
work h
a
s
bee
n
perfo
rmed
by
defining
a n
o
vel
comp
osite
rel
i
ability index [9]. Thre
shol
d
value of
the DG cap
a
city has bee
n
calculate
d
beyo
nd
recl
oser-fuse
coo
r
din
a
tion
is lo
st. The
SAIF
I, SAIDI, Energy No
t Supplied
(ENS) h
a
ve be
en
selected as indices of reli
ability improvement in
the test system
RBTS
bus 2, [10]. A novel
techni
que
ba
sed
on
classi
fy the re
clo
s
er-fu
s
e
coo
r
d
i
nation
statu
s
at fault
co
n
d
ition h
a
s be
en
sug
g
e
s
ted to study the
impact of
distri
b
u
ted-gene
ration penetration on
re
clo
s
e
r
-fuse
coo
r
din
a
tion.
The mai
n
a
d
vantage
of the pro
p
o
s
ed
approa
ch i
s
that usin
g the cl
assification
pro
c
e
s
s for the
coo
r
din
a
tion
status di
scrimi
nate
s
b
e
twee
n the
ca
se
s that
require a
n
a
c
tion
again
s
t
DG
p
enetratio
n
an
d the
case
where
n
o
a
c
tio
n
is requi
red
[11]. A novel
reliability in
d
e
x
has b
een
def
ined an
d the
zon
e
-n
etwo
rk method u
s
e
d
to evaluate
this co
mpo
s
i
t
e index in the
pre
s
en
ce of
DG unit
s
. Th
e simple
GA has b
een imp
r
oved by a m
u
lti-pop
ulatio
n method a
n
d
the
influen
ce of i
m
pro
per
gen
etic pa
ramete
rs
can
b
e
gre
a
tly decrea
s
e
d
and p
r
em
ature
conve
r
ge
nce
can
be
overcome effe
ctively [12]. Re
close
r
a
nd a
u
tose
ctionali
z
e
r
have
allo
ca
ted to imp
r
o
v
e
reliability by
a meth
odolo
g
y ba
sed
on
co
st/ben
efit
analysi
s
. A
h
y
brid m
e
thod
ba
sed
on
IPSO
algorith
m
a
n
d
Monte
Carlo
simulatio
n
i
s
employed. Si
mulation
ha
s
been
do
ne i
n
a p
r
a
c
tical te
st
system i
n
Iran [13]. Optim
a
l num
ber and location
of recl
oser to
improv
e
reliability of feeders
usin
g gen
etic algorithm h
a
s
pe
rforme
d. The propo
se
d method enj
oys the sim
p
l
i
city of config
ure
duratio
n, accura
cy of the result
s an
d re
ductio
n
of
the
time co
nsu
m
ing. The o
b
tai
ned results al
so
sho
w
the a
p
p
licability of the algo
rithm [1
4]. the
effects of demand
resp
on
se p
r
og
ram
s
espe
cia
l
l
y
direct load
control
on sy
stem and nodal reliability of a
deregulat
ed power sy
stem using di
rect
load
cont
rol
and e
c
o
nomi
c
loa
d
mo
del
[15]. A
two-stage Pla
c
em
ent propo
sed
without invol
v
ing
rand
om va
ria
b
le such a
s
f
a
ilure
rate
s a
nd inte
rru
ptio
n du
ration
s o
f
the system.
Ho
weve
r, all
contin
gen
cie
s
are
given
th
e same
wei
g
ht and
no
re
liability index
is
analysed
[16]. A mixed
integer li
nea
r
prog
ram
m
ing
(MILP) a
nd
CPLEX used
to solve the p
r
oble
m
. They
also p
r
e
s
e
n
ted
a sen
s
itivity
analysi
s
o
n
t
he effe
ct of i
n
terrupt
ion
cost (cu
s
tom
e
r da
mage
fu
nction
) n
u
mb
er of
device
s
to be
placed, cust
omer type an
d load den
sit
y
[17].
An Evolutiona
ry Intelligent Meth
od
for
solving
th
e p
r
oble
m
of
locatio
n
a
nd
optimize
d
size of
DG
r
e
s
o
ur
ces w
i
th mult
iple objectives
[18]. The differen
c
e b
e
twe
en the ope
rat
i
ons of
ci
rcuit brea
ke
r and
se
ctionali
z
e
r
has
con
s
id
ered
[19].
Placem
ent so
lutions in th
e above pa
pe
rs typi
cally use
d
the ob
serve
d
sam
p
le me
ans f
o
r
rand
om va
ria
b
les such a
s
failure
rates and
inte
rru
p
t
ion du
ration
s, with
out
co
nsid
erin
g lo
a
d
model
s. In this pa
per
optim
al re
clo
s
er
nu
mber
and pl
a
c
eme
n
t with l
oad mo
del
co
nsid
eratio
n h
a
s
been d
one to
enhan
ce
reli
ability indice
s and tea
c
hin
g
learni
ng ba
sed algo
rithm
prop
osed a
s
an
efficient meth
od to
solve
di
screte
proble
m
. Eventually, simulatio
n
h
a
s
been
carri
ed o
u
t on IE
EE
69-b
u
s
radi
al distrib
u
tion sy
stem.
All the sim
u
l
a
tions
are carri
ed o
u
t in
MA
TLAB software.
The
re
st of the
pape
r i
s
orga
nized a
s
follows: se
ction 2 di
scusse
s fi
ndin
g
optimal recl
o
s
er pla
c
em
e
n
t. It has be
en
formulate
d
wi
th propo
se
d functio
n
whi
c
h
is con
s
i
s
ted
different reli
a
b
ility indices.
Load mo
del a
n
d
forwa
r
d
-
ba
ckward l
oad
flow
have
discu
s
sed i
n
S
e
ction
3
an
d 4,
re
spe
c
t
i
vely. Sectio
n 5
rep
r
e
s
ent
s T
L
BO fo
rmula
t
ion an
d
sol
v
ing meth
od.
The
comp
a
r
iso
n
betwee
n
different lo
ad
model
effecti
v
eness
ha
s
sho
w
n
in Se
ction
6 a
nd
finally co
ncl
u
ding
rem
a
rks are
d
r
a
w
n i
n
se
ction 7.
2.
Optimal Recloser Plac
ement
T
h
e
r
e
as
on
s fo
r
d
i
s
t
r
i
b
u
t
io
n
c
o
mpa
n
i
es
to
in
s
t
a
l
l re
c
l
os
er
ar
e fo
r
incr
e
a
s
i
ng s
y
s
t
em
reliability. Stu
d
ies have
be
en d
one
in
re
clo
s
er
pla
c
e
m
ent
u
s
u
a
lly inclu
de ene
rg
y
not suppli
e
d
a
s
an obje
c
tive
function. Th
e
indice
s of reliability are
categ
o
rie
d
a
s
custom
er i
ndices a
nd l
oad
indices. Sin
c
e taki
ng
on
i
ndex m
a
y ch
ange
othe
r i
n
dice
s, the
r
efo
r
e i
n
thi
s
a
r
ti
cle
and
obj
ective
function
ba
se
d on
the
su
m
of the
weig
hted
reli
ability i
ndices have
been
propo
se
d. The
obj
ecti
ve
function
s in this pa
per, a
c
cord
ing to following relations
h
ip.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Novel M
e
th
od Base
d on
TLBO for Reclose
r
Placem
ent with Loa
d
Model …
(Sina K.A. At
tari)
3
S
A
I
F
I
S
AI
D
I
AENS
TT
T
S
A
I
F
I
S
AI
D
I
AE
NS
OF
W
W
W
S
A
I
F
I
S
AI
D
I
AENS
(1)
w
h
er
e
1
1
.
n
ii
i
n
i
i
P
U
AEN
S
N
(2)
1
1
.
n
i
i
i
n
i
i
N
SA
I
F
I
N
(3)
1
1
.
n
i
i
i
n
i
i
UN
SAI
D
I
N
(4)
N
i
is the
num
ber of custo
m
er at i
th
loa
d
point, U
i
i
s
interruption d
u
ration
at i
th
load poi
nt,
P
i
is average
load at i
th
lo
ad poi
nt, W
x
is weight
coe
fficient and X
T
is ta
rg
et value for
reliabil
i
ty
indices. In th
is arti
cle, W
x
values for
AENS, SAIFI, and SAIDI are 0.3
3
, 0.33, and 0.34.
In
addition for ta
rget value
s
are 350, 10, an
d 100, re
spe
c
tively.
With co
nst
r
ai
nts:
n
Nmax
n: numbe
r of recl
osers
N
m
a
x
: ma
ximu
m n
u
m
be
r
of r
e
c
l
os
er
s
With a simpl
e
example, the method of
calcul
ation o
f
the indicato
rs de
scri
bed.
Figure 1
sho
w
s a
sam
p
le radial
di
stribution
sy
ste
m
of fou
r
lin
e
s
A, B, C an
d
D
with fo
ur l
oad
point
s 1,
2,
3, and 4. Hypothetical recl
ose
r
s
R1, R2,
R3 hav
e bee
n place
d
in a
ppro
p
ri
ate locations. If X
i
=0,
it
mean
s
the ab
sen
c
e and
X
i
=1, me
an
s prese
n
ce of reclose
r
. Unavail
ability and fail
ure
rate of l
o
ad
point ca
n be
cal
c
ulate
d
in the followi
ng a
ppro
a
ch.
D
i
s
t
r
i
but
i
on
s
ubs
t
a
t
i
o
n
1
2
3
A
B
C
D
R1
R2
R3
4
Figure 1. Sample ra
dial di
stributio
n syst
em
First, unavail
ability to each
of lines
A
,
B
,
C,
and
D
can
be cal
c
ulate
d
.
Acco
rdi
ng to
the netwo
rk stru
cture
sh
own
in Fi
gure 1. Load po
int 1, in the followin
g
scena
rio
s
are
without ele
c
tricity:
AA
A
B
B
B
CC
C
D
D
D
Ur
Ur
Ur
U
r
(5)
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 1– 10
4
1)
A fault on the line A.
2)
A fault on the line B in the
absen
ce of R
1
(X
1
'
=1 or X
1
=0).
3)
A fault on the line C in the absen
ce of
R
1
,
R
2
(X
1
'
. X
2
'
=1).
4)
A fault on the line D in the absen
ce of
R
1
,
R
3
(X
1
'
. X
3
'
=1).
Therefore load point 1 unavailability can be formulated in (6).
''
'
'
'
11
1
2
1
3
AB
C
D
uU
U
x
U
x
x
U
x
x
(6)
In this equati
on
X
'
is co
mpl
i
mentary to
X
(X
'
= 1–X
).
Similarly, una
vailability of l
oad poi
nts 2, 3,
and 4 ca
n be formul
ated
as equ
ation
s
(7-9
).
'
23
AB
C
D
uU
U
U
U
x
(7)
''
32
3
AB
C
D
uU
U
U
x
U
x
(8)
'
42
AB
C
D
uU
U
U
x
U
(9)
Failure rat
e
for load point
1 to 4
can
be ac
cessed
by replacing with
line unavailability.
Thus, reliabili
ty indices
can
be cal
c
ulate
d
.
3.
Load Mo
d
e
l
In this pape
r different voltage-dep
end
e
n
t load mod
e
l
s scen
ario
s
are b
e
invest
igated.
Practi
cal volt
age-dep
end
e
n
t load
mo
del
s, i.e., resid
e
n
tial, indu
stri
al an
d
com
m
e
rci
a
l
have
b
een
adopte
d
for in
vestigation
s
. To formulate
the
probl
em e
quation (10
)
and (1
1)
will be used.
io
i
i
PP
V
(10
)
io
i
i
QQ
V
(11
)
P
i
and Q
i
are the
real and reactive
p
o
we
r
at
i
th
load b
u
s, re
sp
ectiv
e
ly.
P
oi
and
Q
oi
are the
real an
d rea
c
tive operating
points at
i
th
load bu
s.
V
i
is
i
th
voltage bus and
α
an
d
β
are real a
n
d
rea
c
tive po
wer exp
one
nts. In the
con
s
tant p
o
wer
model
co
nve
n
tionally u
s
e
d
in
power flow
studie
s
,
α
=
β
= 0 i
s
a
s
sumed. Th
e value
s
of the
real a
nd
rea
c
tive expon
e
n
ts u
s
ed i
n
the
pre
s
ent work
for indu
strial,
resi
dential, a
nd
com
m
erci
al load
s are g
i
ven in Table
1 [20].
Table 1. Re
sults for Pareto Re
config
uration with two
Objective
s
L
o
a
d T
y
pe
α
β
Constant
0
0
Industrial 0.18
6
Residential
0.92
4.0
Commercial 1.51
3.4
In pra
c
tice, l
oad
s are mix
t
ures of different
load type
s, dep
endi
ng
on the n
a
ture of the
area b
e
ing
su
pplied.
Therefore, a
load
cla
ss m
i
x of resid
ent
ial,
indu
strial,
and
comm
e
r
cial l
oad
s i
s
to be
investigate
d
, too [21].
4.
For
w
a
rd-Bak
w
a
e
r
d Loa
d Flo
w
Gau
s
s-seidel
and n
e
wto
n
-raph
so
n a
r
e t
w
o
conve
n
tio
nal metho
d
u
s
ed to
cal
c
ul
ate the
load flo
w
in
power
syste
m
. Lack of p
r
ope
r p
e
rfo
r
mance be
co
mes
clea
r
when e
n
count
er with
load u
nbal
an
ce
s in
distri
b
u
tion net
wo
rk, radial
st
ruct
ure
and
high
pro
portio
n
of
re
sista
n
ce a
nd
rea
c
tan
c
e on
the lines.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Novel M
e
th
od Base
d on
TLBO for Reclose
r
Placem
ent with Loa
d
Model …
(Sina K.A. At
tari)
5
Other m
e
tho
d
is
widely u
s
ed
name
d
f
o
rwar
d-b
a
ckward. Thi
s
m
e
thod i
s
ba
sed on th
e
repe
ated sta
g
e
s that ea
ch i
t
eration con
s
i
s
ts of two ste
p
s:
In the first stage all the b
r
an
che
s
current
are
cal
c
u
l
ated (ba
c
kward
step) a
n
d
in the
se
con
d
stage
, with
cal
c
ulat
ed b
r
an
ch
es
curre
n
t an
d li
ne imp
eda
nce, bu
se
s volt
age
are
o
b
tained
(forward step
). Then bu
se
s voltage and
bran
ch
es
cu
rrent value will
be upd
ated [2
2].
5.
Teac
hing-Learning
-Ba
sed algorith
m
The
ba
sis of
this al
gorith
m
is ba
sed
o
n
the
tea
c
hin
g
of a te
acher i
n
the
cla
s
sro
o
m [23].
The te
acher i
n
the
cla
s
sro
o
m pl
ays
a m
a
jor rol
e
in
st
udent l
earnin
g
an
d b
e
tter
stude
nt lea
r
ni
ng
depe
nd
s on
the tea
c
he
r
spe
e
ch. In a
ddition to th
e
effects of te
ach
e
r,
revie
w
texts bet
we
en
stude
nts the
m
selve
s
are also le
ading t
o
learn thei
r
studie
s
better. This idea fo
rms the b
a
si
s of
TLBO to solv
e optimizatio
n probl
em.
TLBO algo
rithm divided i
n
to two part
s
. The first pa
rt con
s
ist
s
of
teache
r pha
se. The
se
con
d
part consi
s
ts of lea
r
ne
r pha
se.
The tea
c
h
e
r
pha
se m
ean
s le
arni
ng from the te
acher
and
the
learn
e
r
pha
se mea
n
s
learni
ng throu
gh the intera
ction betwe
en
learn
e
rs.
5.1. Teacher
Phase
A good teach
e
r brin
gs hi
s
or her le
arn
e
r
s up to his o
r
her level in terms of kn
o
w
led
ge.
But in pra
c
tice this i
s
not p
o
ssible
and a
teac
h
e
r can only
move
th
e
mean of
a cla
ss up
to some
extent depen
ding on the capability
of the class. Thi
s
follows
a ra
ndom process depe
ndin
g
on
many factors.
,,
,
()
ne
w
D
ol
d
D
t
e
a
c
her
D
F
D
XX
r
X
T
M
(12
)
D
ind
e
x indi
cates th
e nu
m
ber
of course
(vari
able
pro
b
lem),
X
old,D
old member,
still have
to incre
a
se t
heir kno
w
led
ge of
the te
a
c
he
r’s tea
c
h
and i
n
cl
ude
s
1*D vecto
r
which
get
s
re
sults
related
to a
n
y sp
ecifi
c
topi
cs o
r
le
sson
s.
r
i
s
a
ra
ndom
numb
e
r in th
e ra
nge
[0,1].
X
teacher,D
is the
best
memb
er of po
pulatio
n
in ite
r
ation
s
t
hat trie
s fo
r t
he ave
r
ag
e o
f
the
class (p
opulatio
n) to
i
t
s
positio
n. The
value of T
F
can b
e
eithe
r
1 or 2
whi
c
h is
agai
n a heu
risti
c
st
ep and
de
cid
ed
randomly wit
h
equal
probability as T
F
= roun
d[1+ra
n
d
(0,1
){2
-
1}].
M
D
is 1*
D ve
ctor that in
clu
des
the mean values for ea
ch
subj
ect in cla
ss.
X
new
,
D
if compa
r
ed to the old memb
ers a
nd improved
will be accept
ed.
5.2. Student
Phase
Learne
rs in
crease their kn
owle
dge by two diffe
re
nt mean
s. One t
h
rou
gh inp
u
t from the
teach
e
r
and
other th
ro
ugh
intera
ction
b
e
twee
n them
selve
s
. A lea
r
ne
r inte
ra
cts ran
domly
wi
th
other lea
r
ne
rs with the he
lp of group d
i
scus
sio
n
s, p
r
esentation
s
,
formal com
m
unication
s and
etc. A lea
r
ne
r learns
so
me
thing ne
w if t
he othe
r le
arner
ha
s mo
re
kno
w
le
dge
than hi
m o
r
h
e
r.
Learne
r modi
fication is exp
r
esse
d as
For i=1: P
n
Ran
domly sel
e
ct anoth
e
r le
arne
r
X
j
, s
u
ch that
i
j.
,,
,,
–
()
(
)
–
ne
w
i
old
i
i
i
j
ne
w
i
o
ij
ld
i
i
j
i
if
f
X
f
X
El
X
Xr
X
X
X
Xr
X
X
End
i
f
E
se
nd
fo
r
(13
)
A
cce
pt
X
new
if it gives a
better fun
c
tio
n
value. The
flow chart f
o
r TLB
O
met
hod i
s
given
in
Figure 2.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 1– 10
6
6. Simulation Resul
t
s
In order to inves
t
igate th
e performance of the propos
ed approac
h
, the IEEE 69-bus
radial di
strib
u
t
ion test system is utilized
in
this paper. Figure 3 sh
ows t
he singl
e line diagra
m
of
the test
sy
ste
m
. The
total a
m
ounts of th
e
a
ctive a
n
d
re
active lo
ad
s
o
f
the
system
are
3.80
19
M
W
and 2.69
41 M
VAr, resp
ecti
vely.
The forward-backward loa
d
flow is don
e due to obtained voltage
and then po
wer i
n
each bus. In
next step wit
h
lo
ad model
consi
deratio
n and
without that, si
mulation
will be run.
The proc
ess
is
repeated
by
100
iterations
and 20 populations
.
Simulation res
u
lts
for IEEE 69-
bus [24], are indicated in T
able 2.
The system data are given in Ta
ble 3.
R
u
n
t
h
e
P
o
w
e
r
F
l
o
w
&
s
e
t
in
it
ia
l
p
a
ra
m
e
t
ers
o
f
s
y
s
t
em
Set
t
h
e
pa
ra
m
e
t
e
rs
of
T
L
B
O
(P
o
p
u
l
a
t
i
o
n
s
i
ze
,
n
u
m
b
e
r
of
i
t
e
r
at
i
o
n
s
)
C
r
e
a
t
e
t
h
e
in
it
ia
l
p
o
p
u
l
a
t
i
o
n
i
n
a
Ra
n
d
o
m
w
a
y
It
e
r
=
1
Ru
n
t
h
e
P
o
w
e
r
F
l
o
w
C
a
l
c
u
l
a
t
e
t
h
e O
b
j
e
ct
i
v
e F
u
n
c
t
i
o
n
Id
en
t
i
f
y
t
h
e
b
e
st
so
l
u
t
i
o
n
(
t
e
a
c
h
e
r
p
h
a
s
e
)
m
o
di
f
y
s
o
l
u
t
i
o
n
s
i
n
t
h
e
t
h
e
a
c
h
e
r
ph
a
s
e
by
us
i
n
g
E
q
u
a
t
i
o
n
(1
2
)
&
f
i
nd t
h
e
ne
w
be
s
t
s
o
l
u
t
i
o
n
i
n
t
h
i
s
pha
s
e
Se
l
e
c
t
t
w
o
s
o
l
u
t
i
o
n
s
R
a
ndo
m
l
y
(
s
t
u
de
nt
ph
a
s
e
)
m
o
d
i
f
y
s
o
l
u
t
i
ons
b
y
u
s
i
n
g E
q
u
a
t
i
on
(13)
F
i
nd
t
h
e
be
s
t
s
o
l
u
t
i
o
n
Pr
i
n
t
R
e
s
u
l
t
s
Ha
v
e
r
e
a
c
h
ma
x
i
mu
m
it
e
r
a
t
i
o
n
?
It
e
r
=It
e
r
+
1
No
Ye
s
St
a
r
t
Re
a
d
S
y
s
t
e
m
d
a
t
a
Figure 2. Flow ch
art showi
ng the wo
rki
n
g of TLBO algorithm
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Novel M
e
th
od Base
d on
TLBO for Reclose
r
Placem
ent with Loa
d
Model …
(Sina K.A. At
tari)
7
It was perfo
rmed for different asp
e
ct
s of recl
o
s
e
r
pl
acem
ent and
termination
con
d
ition
wa
s sel
e
cted
as the obje
c
ti
ve function in
each
st
age n
o
t much different than the previou
s
sta
g
e
.
Stop conditio
n
is formul
ate
d
as follo
ws:
0
0m
i
n
[]
1
0
0
i
(12
)
is improve
m
ent perce
ntage with i re
cl
ose
r
,
0
is objective functio
n
whe
n
no re
clo
s
er in
netwo
rk.
i
is minimum o
b
jective funct
i
on value wh
en
i
r
e
c
l
os
er in
n
e
t
w
o
rk
,
mi
n
is minimum
obje
c
tive value (wh
en all lin
es have recl
o
s
er).
66
67
1
23
4
5
6
7
8
9
10
1
1
12
13
14
15
1
6
17
18
19
20
21
22
23
24
26
27
28
2
9
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
4
6
47
48
49
50
51
5
2
53
54
55
56
57
58
59
60
61
62
63
64
65
68
69
La
t
r
a
l
1
:
R
e
s
i
d
e
n
t
i
a
l
La
t
r
a
l
2
:
R
e
s
i
d
e
n
t
i
a
l
La
t
r
a
l
3
:
R
e
s
i
d
e
n
t
i
a
l
25
La
t
r
a
l
4
:
R
e
s
i
d
e
n
t
i
a
l
Lat
r
al
5
:
R
e
s
i
d
e
n
t
i
a
l
L
a
t
r
al
6
:
Cm
m
m
e
r
i
c
al
La
t
r
a
l
7:
C
m
m
m
eri
ca
l
L
a
t
r
a
l
8
:
I
n
dus
tr
i
a
l
S
/
S
Figure 3. Single line diag
ram of
69-b
u
s
dist
rib
u
t
i
on t
e
st
sy
st
em
Table 2. Opti
mal placeme
n
t and OF Va
lue for 69
-bu
s
Distrib
u
tion
Test System
Lo
ad Mo
del
Recl
os
er
Nu
mb
er
O
p
t
i
m
a
l P
l
ac
eme
n
t
OF
Value
η
(%
)
0
-
0.561
67
0
1
8-9
0.264
53
67.43
2
26-
27, 61-
62
0.206
64
77.34
3
9-1
0
, 3
-
36
, 4
-
47
0.163
45
86.75
4
34-
35, 42-
43, 9-
5
3
,
12
-6
8
0.140
09
91.84
Consta
nt
5
7-8,
14
-15
,
9
-
53
,
12-
68,
68-
69
0.133
71
93.23
6
7-8,
16
-17
,
21
-2
2
,
28
-29,
9
-
53,
12
-
6
8
0.126
36
94.83
7
9-1
0
, 1
2
-1
3, 2
9
-
3
0, 3
-
36,
36
-37
,
5
3
-5
4, 6
-
65
0.120
85
96.03
8
5-6,
12
-13
,
34
-3
5
,
36
-37,
49
-50
,
8
-
51, 5
6
-
57,
58-
59
0.115
80
97.13
9
2-3,
6-
7,
12-
13,
1
4
-1
5, 1
7
-1
8, 2
5
-
2
6, 42
-4
3, 5
3
-5
4,
58-
59
0.110
98
98.18
10
3-4,
6-
7,
8-9
,
10
-
11, 1
6
-
17,
22-
23,
9-5
3
, 5
6
-
57,
57
-
58, 6
1
-
6
2
0.108
23
98.78
68
All
lines
0.102
63
100
0
-
0.561
62
0
1
10-
11
0.312
53
54.27
2
10-
11,
15-
16
0.261
35
65.42
3
3-4,
9-
10,
40
-41
0.220
36
74.35
4
3-2
8
, 2
9
-3
0, 3
-
3
6
,
64
-65
0.182
26
82.65
Comp
osite
5
25-
26, 36-
37, 40
-
41,
5
6
-
57, 59-
60
0.161
20
87.74
6
3-4,
10
-11
,
40
-4
1
,
42
-43,
59
-60
,
6
2
-6
3
0.141
41
91.55
7
5-6,
9-
10,
17
-18
,
18-
19, 24-
25, 44
-
45,
4
8
-
4
9
0.128
65
94.33
8
9-1
0
, 1
8
-1
9, 1
9
-
2
0, 20
-2
1, 3
-
36
, 3
6
-3
7, 5
8
-5
9, 1
1
-
6
6
0.121
31
95.93
9
20-
21, 22-
23, 34
-
35,
3
8
-
39, 43-
44,
4-4
7
,
5
3
-
54, 58
-
59,
6
2
-
6
3
0.114
89
97.33
10
6-7,
14
-15
,
3
-
28
,
30-
31,
38-
39,
4-
4
7
, 51
-5
2, 9
-
53
, 5
8
-5
9, 6
8
-6
9
0.110
2
98.35
11
4-5,
9-
10,
20
-21
,
23-
24, 30-
31, 37
-
38,
4
0
-
41, 49-
50,
55-
56, 61
-62,
64-
65
0.108
36
98.75
68
All
lines
0.102
62
100
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 1– 10
8
Table 3. 69
-b
us test sy
ste
m
data System
Br.
no
S.
no
R.
no
Le
ng
th
(k
m)
(f/
y
r
)
r
(h
)
Nu
mb
er of
cus
t
o
m
er
Lo
ad
ty
p
e
1
1
2
0.8
0.065
5 200
0
2
2
3
0.7
0.065
5 200
2
3 3
4
0.6
0.065
5
1
2
4 4
5
0.7
0.065
5
1
2
5 5
6
0.6
0.065
5
10
2
6 6
7
0.6
0.065
5
10
2
7
7
8
0.6
0.065
5 210
2
8
8
9
0.8
0.065
5 210
2
9 9
10
0.6
0.065
5
3
2
10
10
11
0.8
0.065
5
3
2
11
11
12
0.8
0.065
5
1
2
12
12
13
0.7
0.065
5
1
2
13
13
14
0.7
0.065
5 200
2
14
14
15
0.6
0.065
5 200
2
15
15
16
0.8
0.065
5
1
2
16
16
17
0.6
0.065
5
1
2
17
17
18
0.6
0.065
5
10
2
18
18
19
0.7
0.065
5
10
2
19
19
20
0.6
0.065
5 210
2
20
20
21
0.7
0.065
5 210
2
21
21
22
0.6
0.065
5
3
2
22
22
23
0.8
0.065
5
3
2
23
23
24
0.6
0.065
5
1
2
24
24
25
0.6
0.065
5
1
2
25
25
26
0.8
0.065
5 200
2
26
26
27
0.7
0.065
5 200
2
27
3
28
.6
0.065
5
1
3
28
28
29
0.6
0.065
5
1
3
29
29
30
0.8
0.065
5
10
3
30
30
31
0.7
0.065
5
10
3
31
31
32
0.7
0.065
5 210
3
32
32
33
0.7
0.065
5 210
3
33
33
34
0.6
0.065
5
3
3
34
34
35
0.8
0.065
5
3
3
35
35
36
0.6
0.065
5
1
3
36
36
37
0.7
0.065
5
1
3
37
37
38
0.6
0.065
5 200
3
38
38
39
0.7
0.065
5 200
3
39
39
40
0.8
0.065
5
1
3
40
40
41
0.8
0.065
5
1
3
41
41
42
0.6
0.065
5
10
3
42
42
43
0.8
0.065
5
10
3
43
43
44
0.7
0.065
5 210
3
44
44
45
0.8
0.065
5 210
3
45
45
46
0.6
0.065
5
3
3
46
4
47
0.6
0.065
5
3
3
47
47
48
0.7
0.065
5
1
2
48
48
49
0.6
0.065
5
1
2
49
49
50
0.8
0.065
5 200
2
50
8
51
0.8
0.065
5 200
2
51
51
52
0.6
0.065
5
1
2
52
9
53
0.8
0.065
5
1
2
53
53
54
0.8
0.065
5
10
1
54
54
55
0.7
0.065
5
10
1
55
55
56
0.6
0.065
5 210
1
56
56
57
0.8
0.065
5 210
1
57
57
58
0.6
0.065
5
3
1
58
58
59
0.8
0.065
5
3
1
59
59
60
0.7
0.065
5
1
1
60
60
61
0.6
0.065
5 200
1
61
61
62
0.8
0.065
5 200
1
62
62
63
0.6
0.065
5
1
1
63
63
64
0.6
0.065
5 210
1
64
64
65
0.7
0.065
5 210
1
65
11
66
0.7
0.065
5 210
1
66
66
67
0.6
0.065
5 210
2
67
12
68
0.6
0.065
5 210
2
68
68
69
0.7
0.065
5 210
2
Note
: N:
num
be
r
of custom
er
, Lo
a
d
t
y
pe:
Co
nstan
t
=0, In
dust
r
ial=1,
Residen
tial=2,
C
o
mme
rcial=3
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Novel M
e
th
od Base
d on
TLBO for Reclose
r
Placem
ent with Loa
d
Model …
(Sina K.A. At
tari)
9
It can be
se
e
n
with
Con
s
t
ant load th
e
differen
c
e b
e
twee
n 9 a
nd
10 reclo
s
e
r
s l
e
ss than
1%. So, the
best nu
mbe
r
for re
clo
s
er place
m
ent
of 69-b
u
s di
stributio
n test system is
9. In
addition
with
load m
odel
consi
deration t
he o
p
ti
mal re
clo
s
er num
be
r an
d pl
acem
ent have
bee
n
cha
nge
d to
10. Re
clo
s
er n
u
mbe
r
a
nd optimal
p
l
acem
ent ha
ve been
sh
o
w
n in
Tabl
e
2.
Acco
rdi
ng to
the results
of Table 2, i
t
c
an be n
o
ted that re
clo
s
er
numb
e
r
and pla
c
e
m
e
n
t
cha
nge
d with
load model
consi
deration.
7.
C
o
nc
lu
s
i
on
A novel method ba
se
d o
n
TLBO alg
o
rithm for o
p
timal numb
e
r an
d pla
c
e
m
ent of
recl
oser
was proposed in this pa
per. The main purpose of this st
udy is reliabilit
y improvement.
Also by u
s
in
g an obj
ectiv
e
functio
n
wi
th impac
t o
n
reliability in
dice
s, the be
st location
s and
numbe
rs of reclo
s
e
r
s a
r
e
sele
cted.
Re
clo
s
ers ha
ve played a fundame
n
tal role in
impro
v
ing the relia
bility of distribution
system. Th
erefore, it is e
s
sential to d
e
term
in
e the o
p
timum num
ber a
nd lo
cat
i
on of re
clo
s
ers
due to
impro
v
e network reliability. Simulation
re
sult
s
sho
w
that
as th
e nu
mb
er of
re
clo
s
e
r
s
increa
sed, n
e
t
work reliabilit
y improved.
Ho
wever,
di
st
ribution co
mp
anie
s
should also pay
high
er
co
st a
s
well.
Several
pa
rameters
ca
n
determi
ne th
e optimu
m
n
u
mbe
r
of
re
cl
ose
r
s.
Here,
the
differen
c
e be
tween two steps in obj
e
c
tive functio
n
value sh
o
u
ld be le
ss
than 1%. If
the
differen
c
e in
obje
c
tive fun
c
tion i
s
le
ss t
han 1%, thi
s
step
will sele
ct as the o
p
timal num
ber
and
locatio
n
of
re
clo
s
ers, b
e
si
des, it
can
rev
eal that lo
ad mo
del
ch
ange th
e o
p
timal num
be
r
and
locat
i
o
n
of
re
clo
s
er
s.
Referen
ces
[1]
Gers, JM, Holmes EJ.
Prote
c
tion of e
l
ectri
c
ity distrib
u
tio
n
netw
o
rks
. T
he Institutio
n
of Electrica
l
Engi
neers, L
o
n
don, Un
ited Ki
ngd
om. 200
8: 133-
141.
[2]
Che
n
G. A No
vel QEA-base
d
Optimum S
w
itch
Placeme
n
t
Method for Improvi
ng Custo
m
er Servic
e
Reli
ab
ilit
y.
IEEE Trans. on Power SysteM.
2
008; 16(
2): 620
-623.
[3]
Che
n
Ch, L
i
n
ChH, Ch
ua
ng
HJ, Li Ch
Sh, H
uan
g MY, Hua
ng CH. Optim
a
l Plac
ement of
Lin
e
S
w
itch
e
s
for Distributi
o
n
Automatio
n
Systems Usi
ng I
mmune Al
gorit
hm.
IEEE Trans. on Power System
. 20
06
;
21(3): 42
5-4
3
0
.
[4]
Billi
nton
R, Jo
nnav
ithul
a S.
Optimal S
w
itc
h
in
g
D
e
vice
P
l
acem
ent i
n
R
adi
al D
i
strib
u
ti
on S
y
stems
.
IEEE Trans. on Power Deliver
y.
1996; 1
1
(3): 164
6-16
51.
[5]
F
e
rreira G, Bre
t
as A. A non
li
n
ear b
i
nar
y
pro
g
r
ammi
ng
mod
e
l
for el
ectric d
i
s
t
ributio
n s
y
ste
m
s reli
abi
lit
y
optimiz
ation.
In
ternatio
nal J
o
u
r
nal of Electric
al Pow
e
r & Energy Syste
m
s
. 199
6; 43(1): 38
4– 39
2.
[6]
Qin Q, Wu N.
Reclos
er a
nd
Section
a
li
z
e
r P
l
ace
m
ent for
R
e
lia
bi
lity Impro
v
ement Us
ing
Discrete Ev
en
t
Simulation
.
PE
S General Me
e
t
ing Co
nfere
n
c
e
& Exp
o
sitio
n
.
1996.
[7]
W
ang
L, Si
ngh
C. R
e
li
abi
lit
y-c
onstrai
ned
o
p
ti
mu
m p
l
acem
e
n
t of rec
l
os
ers
and
distri
bute
d
ge
ner
ators
in
distrib
u
tion
net
w
o
rks
usi
n
g
an
ant c
o
l
o
n
y
s
y
stem
al
gorith
m
.
Systems, M
an, a
nd
Cyb
e
r
netics, Par
t
C: Applic
ations
and Rev
i
ew
s, IEEE Transactions on
. 2
008;
38(6): 75
7-6
4
.
[8]
Popov
i
ć
DH, Greatbanks J
A
, Begovi
ć
M, Pregelj A. Placem
ent of
distribute
d
g
ener
ators an
d
reclos
ers for di
stributio
n net
w
o
rk securit
y
a
n
d
reli
abi
lit
y
.
Int
e
rnati
ona
l Jo
urnal of El
ectrica
l
Po
w
e
r
&
Energ
y
S
y
stem
s. 2005; 27(
5): 398-
408.
[9]
W
ang L, Si
ngh
C.
Reli
abi
lity-c
onstrai
ned opti
m
u
m
r
e
clos
er pl
ac
e
m
ent i
n
di
stributed
gen
er
ation
usin
g
ant colony system
algorithm
. In Po
w
e
r S
y
ste
m
s Confere
n
ce
and E
x
pos
itio
n. 2006:
186
0-1
865.
[10]
Chait
u
san
e
y
S, Yoko
yama A.
Preventi
on of r
e
lia
bi
lit
y
de
gra
datio
n from rec
l
oser
–fuse mis
c
oord
i
nati
o
n
due to distri
but
ed ge
ner
ation.
IEEE Transactions on Power
Deliv
ery
. 200
8; 23(4): 254
5-5
4
.
[11]
Naiem AF
, He
gaz
y
Y, Abdel
aziz AY, Elsh
arka
w
y
MA. A classificati
on
techni
qu
e for reclos
er-fuse
coord
i
nati
on i
n
distri
buti
on s
y
stem
s
w
i
th distrib
u
ted ge
nerati
on.
IEE
E
T
r
ansactio
n
s
on P
o
w
e
r
Deliv
ery
. 201
2; 27(1): 176-
85.
[12]
Li Z
,
Yuqi
n
X, Z
engpi
ng W
.
Researc
h
o
n
opti
m
i
z
a
t
io
n
of recl
os
er p
l
ace
m
ent of D
G
-enha
nce
d
distrib
u
tion n
e
tw
orks
.
T
h
ird Intern
ation
a
l
Co
nferenc
e
on
El
ectric
Utilit
y D
e
re
gul
ation
an
d
Restructuri
ng a
nd Po
w
e
r T
e
chnol
ogi
es. 200
8: 2592-
25
97.
[13]
Abdi S, Afshar
K, Ahmadi S,
Bigd
eli N, Ab
di
M. Optimal recloser a
nd a
u
tosectio
nal
izer
alloc
a
tio
n
i
n
distrib
u
tion
n
e
tw
o
r
ks
usin
g IP
SO–Monte
Car
l
o
appr
oac
h.
In
ternatio
nal
Jo
u
r
nal
of E
l
ectric
al P
o
w
e
r
&
Energy Syste
m
s
. 2014; 55: 60
2-11.
[14]
Deh
gha
ni
N, D
a
shti R. Optimi
zation
of recl
o
s
er
pl
aceme
n
t
to improv
e re
li
abil
i
t
y
b
y
gen
et
ic al
gorithm.
Energy a
nd Po
w
e
r Engine
eri
n
g
. 2011; 3(
4): 508.
[15]
Manso
u
ri A, A
z
ami A, Omi
d
i
an A, M
o
h
a
ma
dia
n
E, Az
ami
R. Eval
uatio
n
of Po
w
e
r S
y
st
em Re
li
abi
li
t
y
Consi
der
ing
Di
rect Loa
d C
ont
rol Effects.
Internatio
nal J
our
n
a
l of El
ec
trica
l
and Co
mp
uter Engi
neer
in
g
(IJECE)
. 2013;
3(2): 254-2
59.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 1, April 2016 : 1– 10
10
[16]
Bezerra JR, Barroso GC, Leao RP.
Sw
itch place
m
ent al
gorith
m
for re
duci
ng custo
m
ers outag
e
impacts o
n
ra
di
al d
i
stributi
on
n
e
tw
orks
. In
T
E
NCON 2
012-
2
012 IEEE
Re
gi
on 1
0
C
onfer
e
n
ce. 20
12:
1-
6.
[17]
Abiri-Ja
hromi
A,
F
o
tuhi-F
iruzaba
d M, Parvani
a M, Mosleh M.
Optimized sectio
nal
i
z
ing s
w
itc
h
plac
ement stra
teg
y
i
n
distrib
u
t
ion s
y
stems.
IEEE Transactions on Power Deliv
ery
. 201
2; 27(1): 362
-
70.
[18]
Kavia
n
i-Ara
n
i
A. Optimal pl
a
c
ement a
nd si
zing
of distrib
u
t
ed ge
ner
ation
units us
ing c
o
-evol
u
tion
a
r
y
particl
e s
w
arm
optimiz
atio
n a
l
gorit
hms.
T
E
LKOMNIKA Indones
ian
Jour
n
a
l of
E
l
ectrica
l
Engi
neer
in
g
.
201
5; 13(2): 24
7-25
6.
[19]
Moradi
A, F
o
tu
hi-F
iruza
b
a
d
M
.
Optimal s
w
itc
h
pl
acem
ent i
n
distrib
u
tio
n
s
ystems usi
ng tri
nar
y
particl
e
s
w
a
rm optimiz
ation a
l
g
o
rithm
.
IEEE
Transac
tions on Power
Delivery
. 2
008
; 23(1): 271-9.
[20]
Singh D, V
e
rm
a KS. Multiobj
ective optimization for
DG planning
w
i
t
h
load
models. IEEE
T
r
ansactions
on Po
w
e
r S
y
st
em. 2009; 2
4
(1
): 427-36.
[21]
Hun
g
DQ, Mit
h
ula
nant
han
N,
Lee
KY. D
e
ter
m
inin
g PV
p
e
n
e
tration
for
dist
ributi
on s
y
st
ems
w
i
th
time-
var
y
in
g loa
d
mode
ls. IEEE
T
r
ansacti
ons o
n
Po
w
e
r S
y
stem
s. 2014; 29(
6): 304
8-57.
[22]
Nan
da J, Sri
n
i
v
as MS, Shar
ma M, De
y SS
, Lai
LL.
New
findings on
r
a
dial distribution system
loa
d
flow
algorith
m
s
. InPo
w
e
r Engi
neer
ing S
o
ciet
y W
i
nt
er Meeti
ng. 200
0; 2:11
57-1
161.
[23]
Rao RV, Sav
s
ani VJ, Vakh
aria DP. T
eachin
g–l
ear
nin
g
-
base
d
optimiz
at
ion: a nov
el
method for
constrai
ned m
e
cha
n
ica
l
des
i
gn optim
izatio
n
probl
em
s. Co
mputer-Ai
ded
Desig
n
. 20
11; 43(3): 30
3-1
5
.
[24]
Aman MM, Jasmon GB, Baka
r AH, Mokhlis
H.
A ne
w
a
ppr
oach for o
p
tim
u
m simulta
n
e
o
u
s multi-DG
distrib
u
ted g
e
n
e
ratio
n
Un
its pl
aceme
n
t an
d s
i
zing
bas
ed
on
maximizati
on
of s
y
stem l
o
a
d
abil
i
t
y
usi
n
g
HPSO (h
ybrid
particl
e s
w
arm
optimiz
ation) a
l
gorithm.
Ener
g
y
. 2014; 66: 20
2-15.
Evaluation Warning : The document was created with Spire.PDF for Python.