Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 1, No. 3,
March 20
16, pp. 480 ~ 4
8
9
DOI: 10.115
9
1
/ijeecs.v1.i3.pp48
0-4
8
9
480
Re
cei
v
ed O
c
t
ober 1
7
, 201
5; Revi
se
d Febru
a
ry 18, 2
016; Accepte
d
February 2
7
, 2016
A Comprehensive Study on Specifying an Intelligent
Approa
ch to Solve Network Reconfiguration Problem
Mahmoud Reza Sha
kara
m
i*, Sina Khajeh Ahmad
Attari, Farha
d
Namdari
Dep
a
rtment of Electrical E
ngi
neer
ing, L
o
rest
an Un
iversit
y
,
Dan
e
shg
ah Str
eet, 712
34-9
8
6
53, Khorram
a
b
ad, Loresta
n, Iran
e-mail: sh
akar
ami.mr@lu.ac.i
r
A
b
st
r
a
ct
T
h
is pa
per pr
e
s
ents an
ap
pro
a
ch b
a
se
d on
bio
geo
gra
phy-
base
d
o
p
ti
mi
z
a
tion (BBO) al
g
o
rith
m t
o
solve th
e distri
butio
n netw
o
rk
reconf
ig
urati
o
n (DNR) pr
obl
em for
mi
ni
mi
zi
ng activ
e
pow
er loss. Also it
is
de
mo
nstrated t
hat w
i
th cons
id
erin
g
the
natur
e of reco
nfig
ur
ation pro
b
le
m, amon
g
the inte
llig
ent
a
l
g
o
rith
ms,
BBO appro
a
ch
could res
u
lt the best
p
e
rfor
ma
nce. One o
f
the remark
a
b
le a
d
va
ntage
s of this study is
compari
ng s
e
ven d
i
fferent
intell
ig
ent a
l
g
o
rith
ms
i
n
so
lving
netw
o
rk
reconfi
gur
ati
on pr
obl
e
m
.
T
h
is
comparis
on,
n
o
t only
inc
l
u
d
e
s
final
fitness
i
n
opti
m
i
z
a
t
i
on process, but
a
l
so
cons
id
ers n
u
mber
of functi
on
eval
uatio
n (NF
E
). T
he effecti
v
eness
of the
BBO me
tho
d
has
be
en tes
t
ed o
n
tw
o dif
f
erent d
i
stributi
o
n
systems a
nd the obta
i
n
ed si
mu
lati
on resu
lts are compar
e
d
w
i
th Genetic algor
ith
m
(GA), Particle sw
ar
m
opti
m
i
z
at
ion (P
SO), Artificial bee c
o
lo
ny (A
BC), Grav
itatio
nal s
earch
alg
o
rith
m (GSA), T
e
chnic
a
l l
ear
nin
g
base
d
opti
m
i
z
a
t
ion (T
LBO)
an
d C
u
ckoo
a
l
go
rithm (CA)
. T
h
e co
mparis
on
r
e
sults s
how
th
at BBO a
ppro
a
c
h
can be a
n
effici
ent and
pro
m
is
ing
meth
od for
solvin
g DN
R pr
obl
e
m
s.
Ke
y
w
ords
: Di
stributio
n n
e
tw
ork reco
nfig
uration, P
o
w
e
r
l
o
ss, Biog
eo
gr
aphy-B
ase
d
Optimi
z
a
ti
on (B
BO),
Nu
mb
er of function eva
l
u
a
tio
n
1. Introduc
tion
Due to op
erating at low volt
age, dist
ributi
on
syste
m
co
ntribute
s
a la
rge amo
unt of power
losse
s
. Also
by increa
sin
g
the total n
e
twork
l
oad,
the requi
red
cu
rre
nt will
increa
se. Su
ch
scena
rio
will cau
s
e
som
e
voltage profile red
u
ctio
n
a
nd po
we
r system losse
s
[1, 2]. Consid
ering
the above m
entione
d pro
b
lems, impl
e
m
enting radi
al
distri
bution
system re
co
nfiguratio
n, is one
of the effective and efficien
t technique
s to dist
rib
u
tion netwo
rk lo
sses re
du
ction,
voltage profile
improvem
ent, load
co
nge
st
ion ma
nag
em
ent an
d
syst
e
m
reli
ability e
nhan
cem
ent.
The el
ect
r
ica
l
energy is deli
v
ered
directly
from th
e int
e
rmedi
at
e tran
sform
e
r sub
s
tations to
con
s
ume
r
s th
rou
gh
the distri
butio
n netwo
rks
whi
c
h a
r
e
al
ways o
p
e
r
ate
with radi
al structu
r
e
s
. Op
erating i
n
ra
dial
config
uratio
n redu
ce
s
th
e sho
r
t-circuit current
si
gnificantly. The re
storatio
n of t
he net
wo
rk f
r
o
m
faults i
s
p
e
rfo
r
med
thro
ugh
the
cutting/cl
osin
g ma
ni
pu
lations of el
e
c
tri
c
al
swit
ch
pairs lo
cate
d
on
the loops, consequ
ently. Therefo
r
e, there are m
any switche
s
on the distribution
syst
em.
Distri
bution
netwo
rk
reconfigur
ation
(DNR) i
s
th
e p
r
o
c
e
s
s
of alteri
ng t
he top
o
logi
cal
arrang
ement
of distributio
n
feeder
s by changi
ng the
status
(op
en/
clo
s
ed
) of se
ctionali
z
ing
a
n
d
tie switche
s
with taking
consi
deration
on syst
em
constraints u
p
on sati
sfying the distributi
o
n
netwo
rk o
perators’
(DNO
s) obje
c
tives. Many studie
s
have been i
n
vestigate
d
to solve network
reconfigu
r
atio
n problem
s
usin
g differe
nt tec
hni
que
s for th
e la
st two
de
ca
des. Exten
s
i
v
e
resea
r
ch
works have
bee
n explo
r
ed
in
the a
r
ea
of
reco
nfiguratio
n of radi
al
di
stributio
n
system
(RDS).
Referenc
e [3] firs
tly reported a method
f
o
r di
stri
butio
n sy
stem
re
config
uratio
n
to
minimize line
losse
s. The
structu
r
e of the me
thod
wa
s based o
n
formulated
the problem
as
integer mixe
d non-li
nea
r
optimizatio
n probl
em an
d
solving it by a discrete b
r
an
ch
-an
d
-b
o
und
techni
que.
Referen
c
e
[4]
pre
s
ente
d
a
new he
uri
s
tic app
ro
ach
of
bran
ch
ex
cha
nge to
redu
ce the
power lo
sse
s
of distrib
u
tion sy
stems
based up
on
the dire
ction
of the bran
ch p
o
we
r flo
w
s.
Referen
c
e
[5
] develop
ed
a b
r
an
ch
exchang
e m
e
tho
d
in
which
lo
ss redu
ction
is a
c
hi
eved
b
y
excha
nge o
p
e
ration th
at correspon
ds t
o
the sele
ctio
n of a pair of
swit
che
s
, one
for openi
ng
and
the other for closi
ng, so
that the resu
lting
network has lower li
ne losses while rem
a
ining
con
n
e
c
ted
a
nd radial.
Re
feren
c
e [6]
i
n
trodu
ce
d g
e
netic
algo
rith
m (GA
)
fo
r reco
nfiguratio
n of
RDS with
co
nsid
erin
g loss minimizatio
n
. A
method based on a shuffled frog l
eapin
g
algorit
hm
(SFLA) h
a
s b
een stu
d
ied t
o
minimize the co
st of
power lo
ss a
nd p
o
we
r of dist
ri
buted ge
ne
ra
tors
[7]. A discrete artificial
be
e col
ony (DA
B
C) h
a
s
pro
p
o
se
d to optim
ize the
distri
b
u
tion net
work [8].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 480 – 489
481
A method b
a
s
ed
on h
a
rm
onic
se
arch
algorith
m
(H
SA) wa
s inv
e
stigate
d
for
DNR p
r
obl
e
m
to
minimize po
wer l
o
ss [9]. A new meth
o
d
based
on
a
dapted a
n
t colony optimization (AACO) is
prop
osed for
minimization
of real p
o
wer loss re
co
nfig
uration [1
0]. The p
r
op
ose
d
algo
rithms
and
method
s for
netwo
rk
re
co
nfiguratio
n problem, ge
nerally, can be
cla
ssifie
d
int
o
two follo
wi
ng
main cla
s
se
s:
Heu
r
isti
c alg
o
rithm
s
[3-5]
,
such a
s
d
i
screte b
r
a
n
ch-and
-b
oun
d tech
niqu
e
and
swit
ch
excha
nge al
g
o
rithm.
Intelligent alg
o
rithm
s
[6–10
], such a
s
GA
, SFLA, DABC, HSA, AACO.
Among the
s
e
algorithm
s,
heuri
s
tic
algo
rithms a
r
e all
gree
dy sea
r
ch alg
o
rithm
s
. These
method
s a
r
e
easy to
be
impleme
n
te
d and
appli
e
d on the
problem
s
with high
sea
r
ch
ing
efficien
cy, but generally cannot
converge to the gl
obal optimu
m
solutio
n
in
the large
-
scale
distrib
u
tion n
e
tworks. Intelligent gro
up
algorith
m
s
ca
n dire
ct sea
r
chin
g pro
c
e
s
s to the glob
al
optimum at the pro
bability
of one hund
red pe
rcent
in theory. But they all inevitably involve a
large n
u
mb
er of computati
on req
u
ireme
n
ts
and
really
have variou
s control p
a
ra
meters.
In
this
pap
er an
a
p
p
r
oa
ch
based on bio
geog
rap
h
y-b
a
se
d
o
p
timization (BBO) algorith
m
is p
r
op
osed
to solve
the
distrib
u
tion
n
e
twork
re
co
n
f
iguration
(DNR) p
r
o
b
lem
for mi
nimizi
ng
active po
we
r losses. Th
e
significant a
d
vantage
of
this study i
s
comp
arin
g seven diffe
re
n
t
heuri
s
tic al
g
o
rithm
s
in solving netwo
rk re
co
nf
igu
r
ation pro
b
le
m. This com
pari
s
on, not only
inclu
d
e
s
final
fitness in
op
timization
pro
c
e
ss,
but al
so con
s
ide
r
s n
u
mbe
r
of fun
c
tion
evaluati
on
(NFE
). The
effectivene
ss
of the BBO app
roa
c
h h
a
s
be
en carried
ou
t on two diffe
rent distri
butio
n
netwo
rks an
d the obtain
ed simul
a
tio
n
results
are
compa
r
ed
with Geneti
c
algorithm (GA),
Particle
swa
r
m optimizatio
n (PSO
), Artificial be
e col
ony
(ABC), Gravitational
sea
r
ch
al
gorit
hm
(GSA), Te
ch
nical l
earnin
g
ba
sed
op
timization
(T
LBO) a
nd
Cu
ckoo al
go
rithm (CA).
By
comp
ari
ng th
e si
mulation
result
s, it is d
e
mon
s
tr
ate
d
that BBO ap
p
r
oa
ch
ca
n
be
an
efficient
a
nd
promi
s
in
g method for sol
v
ing DNR problem
s in
compa
r
ison with the six other mentio
n
ed
intelligent alg
o
rithm
s
.
The re
st of the pape
r is formed as f
o
llow:
Sectio
n 2 depict
s probl
em formulation.
Mean
while,
Section
3
re
pre
s
ent
s th
e
propo
se
d
bi
ogeo
gra
phy-based
optimi
z
ation
(BBO
) for
solving
net
work reconfigu
r
ation
proble
m
. The
sim
u
lation results
are sho
w
n
and discri
be
d
in
se
ction 4 an
d
finally concl
u
si
on are dra
w
n in sectio
n 5.
2.
Formulati
on of the Pr
oblem
2.1.
Po
w
e
r Fl
o
w
M
e
th
od
Duri
ng net
wo
rk reconfigu
r
ation, the po
wer fl
o
w
anal
ysis shoul
d b
e
perfo
rmed.
For ea
ch
prop
osed
co
nfiguratio
n, the po
we
r flo
w
an
alys
i
s
should b
e
imp
l
emented to
evaluate the
nodal
voltage, po
wer lo
ss of
sy
stem a
nd
cu
rrent of
ea
ch
bran
ch. In
thi
s
se
ction, forward/ba
ckwa
rd
sweep te
chni
que ha
s b
e
e
n
sele
cted i
n
this study
d
ue to seve
ral
advantage
s
e.g. Needi
ng
low
memory, high computational performance, simple
st
ructure, high conv
ergence capability [11].
2.2. Po
w
e
r Fl
o
w
F
o
rmulation
In this study, the obje
c
tive function i
s
de
sc
ribe
d for re
al power lo
sses minimi
zati
on:
L
O
b
je
c
t
i
v
e
F
u
n
c
tio
n
m
in
P
Whi
c
h the ex
act real p
o
we
r losse
s
are o
b
tained by th
e followin
g
eq
uation:
bb
NN
Li
j
i
j
i
j
i
j
i
j
j
i
11
P[
a
(
P
P
Q
Q
)
b
(
Q
P
Q
P
)
]
(1)
Whe
r
e
ij
ij
i
j
ij
R
ac
o
s
(
)
VV
and
ij
ij
i
j
ij
X
b
sin(
)
VV
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Com
p
rehe
nsi
v
e Study o
n
Specifyi
ng
an Intelligent
Appro
a
ch to Solve …
(Ma
h
m
oud RS)
482
ij
ij
i
j
Z R
j
X
are the comp
onent
s of impedan
ce mat
r
i
x
and Nb is th
e numbe
r of buses [12].
2.3. Po
w
e
r Fl
o
w
Con
s
tr
aints
The co
nst
r
ain
t
s of objective
function are as follo
ws:
The limitation
of voltage
mi
n
i
ma
x
VV
V
(2)
Whe
r
e V
min
a
nd V
max
indicate the mi
ni
mum a
nd m
a
ximum pe
rmi
ssi
ble volta
g
e (±5%) an
d
V
i
is
the voltage at bus i.
Feede
r capa
bility limits:
im
a
x
b
r
0
I
I
;
i
1
,
2
,
...,
N
(3)
Whe
r
e N
br
is
numbe
r of di
stributio
n sy
stem bra
n
che
s
.
The radial nature of
dist
ribution network
must be mai
n
tained an
d al
so all loa
d
s m
u
st be served
.
2.4. Procedu
r
es
In this se
ction
a four
step
s algorith
m
is p
r
es
ented fo
r checkin
g
the r
adial topol
og
y of trial
solutio
n
s. Th
e method ste
p
s are as foll
ows:
Step 1: Initialize a conn
ect
ed matrix of the loop
di
stri
bution net
work A(a,a
)
with
a is the num
b
e
r
of buse
s
of the distrib
u
tion
netwo
rk. Ea
ch
entry in matrix A is define
d
as bel
ow:
A(i,j)=1, A(j,i)=1, if node i is co
nne
cted t
o
node j.
A(i,j)=0, A(j,i)=0, if node i n
o
t conn
ecte
d to node j.
Initialize a set
of power b
u
ses B=[bu
s
1
, bus
2
,…, bus
k
], with k is the numbe
r of buse
s
in
the distrib
u
tio
n
netwo
rk.
Step 2: Rea
d
the t
r
ial
so
lution
whi
c
h i
s
a
set of tie
-
switch
es tha
t
need
to
ch
eck a
nd m
o
d
i
fy
A(i,j)=0, A(j,i)=0 if the swit
ch on the br
a
n
c
h from n
ode
I to node j is a tie-switch.
Step 3: Evaluate all load
s as bel
ow:
If node n
B
and A(m,n)
= 1, with
m = 1, 2, …, le
n
g
th(B)
and
n = m
+
1, m
+
2, …, b then
the
node n i
s
mo
ved to B, B = B + [node n] and A(m,n
)
=0
, A(n,m)=0.
Step 4: If matrix A is a zero matrix and array
B is equal to the number of bu
se
s, then the trial
solutio
n
ca
n be co
nsi
dere
d
as a
radial netwo
rk confi
guratio
n.
3.
Bioge
ogra
p
h
y
Theor
y
3.1. Based T
h
eor
y
Biogeog
rap
h
y
Based Opti
mization
(BBO) app
ro
ach
is based on
biogeo
gra
p
h
y
theory
[13]. In the scien
c
e
of bio
geog
rap
h
y, a habitat is
an
ecol
ogi
cal a
r
ea that is liv
ed by pa
rticu
l
ar
plant o
r
ani
mal
spe
c
ie
s and
ge
og
ra
phi-cally i
s
ol
ated from ot
her habitat
s
.
Each h
abit
a
t is
orga
nized
by Habitat Suit
ability Index
(HSI).
G
eog
raphi
cal a
r
ea
s, which
are
well
suite
d
as
resi
den
ce
s fo
r biologi
cal
speci
e
s a
r
e sa
id to hav
e a high HSI. Fe
ature
s
that correlate with
HIS
inclu
de rainfa
ll, diversity of vegetation,
divers
ity of t
opog
rap
h
ic f
eature
s
, lan
d
area, tem
p
e
r
a-
ture, etc. If each
of the feature
s
is
assi
gned a val
ue,
HSI is a fun
c
tion of the
s
e
values. Ea
ch
of
these fe
ature
s
that characterize
habita
b
ility is kno
w
n
as Suitability
Index Variab
les (SIV). SI
Vs
are the i
ndep
ende
nt varia
b
les
while
HSI are the de
pend
ent vari
able
s
. Habita
ts with hig
h
HSI
have the l
a
rg
e pop
ulation
and h
a
ve hig
h
emig
ration
rate
μ
,
simply
by virtue of a
larg
e nu
mbe
r
of
spe
c
ie
s that
migrate to
ot
her
habitat
s
. The immi
grati
on rate
λ
is l
o
w for t
hose h
abitats
whi
c
h
are
alrea
d
y satu
rated with
sp
e
c
ie
s. On the
other h
and, h
abitats
with lo
w HSI h
a
ve h
i
gh immig
r
ati
on
rate
λ
, low emigration rat
e
μ
due to sparse p
opula
t
ion. The val
ue of HIS, for low
HSI ha
bitat,
may increa
se with the i
n
flux of spe
c
i
e
s from othe
r ha
bitats a
s
suitability of
a habitat i
s
the
function of its biological diversity. Ho
wev
e
r, if
HSI does not increa
se and re
main
s low, spe
c
ie
s in
that habitat go extinct and
this leads to
additional im
migratio
n. For the sa
ke of simplicity, it
is
safe to a
s
su
me a line
a
r
relation
ship b
e
twee
n habit
a
ts HIS, its i
mmigratio
n a
nd emig
ration
rate.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 480 – 489
483
These rates
are
sam
e
for all the habit
a
ts an
d dep
end up
on th
e numb
e
r of
spe
c
ie
s in t
he
habitats.
Figure
1 d
e
p
i
cts th
e relati
onship
s
bet
ween fitne
s
s of
habitat
s
(sp
e
c
ie
s), e
m
igration rate
μ
and immig
r
ation rate
λ
.
E is the po
ssi
ble maxim
u
m value of
emigration ra
te and S is t
h
e
numbe
r
of
speci
e
s in
the
habitat,
whi
c
h
rel
a
tes to
fitness. S
max
is the
maxi
mum number of
spe
c
ie
s that can be supp
ort
ed by the hab
itat. S
0
is the equilibrium value.
Figure 1. Specie
s model of
a single h
abi
tat
3.2. Propose
d
Method S
t
eps
This
study p
r
opo
sed
an a
ppro
a
ch ba
sed on BBO
algorith
m
whi
c
h i
s
inve
stig
ated to
determi
ne the
netwo
rk
re
co
nfiguratio
n ap
plied to
re
du
ce power lo
sses of
the di
stribution sy
ste
m
.
The propo
se
d algorith
m
st
eps a
r
e pe
rfo
r
med a
s
follo
w:
Step 1: Ente
r the
bran
ch
and l
oad
dat
a an
d al
so
o
pen
switch
es of
the syste
m
(n
etwo
rk
reconfigu
r
atio
n initial data)
and initial dat
a of powe
r
flow.
Step
2:
Initiali
ze po
pulatio
n
size
rand
oml
y
and spe
c
ie
s count p
r
oba
b
ility of each habitat.
Step 3: Evaluate the fitness for ea
ch ind
i
vidual in pop
ulation si
ze.
Step 4: While
The terminati
on crite
r
io
n is not met do.
Step 5: Save
the best ha
bitats in a temp
ora
r
y array.
Step 6: For e
a
ch h
abitat, map the HSI to numbe
r of speci
e
s S,
λ
an
d
μ
.
Step 7: Probabilistically choose the immi
gr
ation i
s
land based on the immigration
rate
μ
.
Step 8: Migra
t
e rando
mly selecte
d
SIVs
based on the
sele
cted i
s
lan
d
in Step 7.
Step 9: Mutate the worst h
a
lf of
the population a
s
permutation algo
rithm.
Step 10: Evaluate the fitness fo
r ea
ch in
dividual in po
pulation
size.
Step 11: Sort the popul
atio
n from be
st to worst.
Step 12: Repl
ace
worst wit
h
best
ha
bitat from temporary array.
Step 13: Go to step 3 for th
e next iteration.
Step 14: end
while
The followi
ng BBO param
eters have been us
ed, population
size
= 15 for IEEE-33 bus
test
sy
stem and populati
on size =
40
for
IEEE-
69 bus test
system, Habit
a
t Modifi
cati
on
Probability = 1, per gene i
mmigrat
ion Probability bounds
= [0, 1],
elitism param
e
ter = 4, step
size rel
a
ted to numerical integrat
ion of probabilitie
s = 1, maximum
λ
and
μ
ra
te
s
fo
r
ea
ch
isla
nd
= 1 and Prob
ability of mutation = 0.05.
To demo
n
strate the performance and e
ffectiven
ess
of the propo
s
ed BBO method, it is
applied to two s
t
andard I
EEE 33,
69
bus tes
t
s
y
s
t
ems
.
The obtained
res
u
lt
s of BBO met
h
od
impleme
n
tation are
com
p
ared
with so
me of well-kn
own intelli
ge
nt algorithm
s includi
ng Ge
netic
algorithm
(GA), PSO, ABC, GSA,
T
L
BO an
d
Cu
cko
o
alg
o
rithm
(CA). All th
e
seven
me
ntio
ned
algorith
m
s
are implem
ente
d
on di
st
rib
u
t
i
on sy
st
ems.
S
i
mulat
i
on
s
were devel
op
ed by MATL
AB
R20
15a in 2
GHz, i3, personal compute
r
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Com
p
rehe
nsi
v
e Study o
n
Specifyi
ng
an Intelligent
Appro
a
ch to Solve …
(Ma
h
m
oud RS)
484
4.
Simulation Resul
t
s an
d Discus
s
io
n
4.1. IEEE-33 Bus Test S
y
stem
The IEEE 33-bus di
stributi
on
system, includes 37
branches, 32 secti
onali
z
ing switches
and
5 tie
switche
s
. T
he li
n
e
an
d lo
ad
d
a
ta of thi
s
sy
stem
are
p
r
e
s
ente
d
in
[14]
. The
diag
ra
m of
this net
work i
s
sho
w
n in Fi
gure
2. The t
o
tal active an
d rea
c
tive po
wer
of load
s
in this net
wo
rk
are 3.7
15M
W and 2.3 MVA
r
, re
spe
c
tivel
y
, also re
al
a
nd re
active p
o
we
r lo
sses f
o
r the initial
case
evaluated f
r
o
m
load
flow
a
r
e 2
10.67
9
kW a
nd 1
43.1
4
kVAr,
re
spe
c
tively. The
si
mulation
re
su
lts
of network
reconfigurat
ion for the 33-bus
system obtaine
d by
seven different
intelligent
algorith
m
s
wi
th popul
ation
size=15 a
n
d
100 ite
r
ati
ons, a
r
e
sho
w
n in T
able
1. It should
be
notice
d
that, this
work e
m
p
hasi
z
e
s
o
n
selectin
g be
st
method fo
r solving net
work reco
nfigu
r
at
ion
probl
em an
d the comp
ari
s
on between
method
s is
b
a
se
d on both
minimum nu
mber of fun
c
tion
evaluation
s
a
nd final
fitness, the
r
efore t
he le
as
t
po
ssible valu
e of
popul
ation
si
ze th
at
coul
d
be
conve
r
ge
d to the final answer i
s
ch
osen. Figure
3, 4 sho
w
the
comp
ari
s
on
of final fitness
conve
r
ge
nce
of BBO ap
p
r
oa
ch
with
si
x other
di
fferent metho
d
s.
The
optimal
config
uratio
n o
f
IEEE-33 bus
system
is
7–9–14–32–37 whi
c
h
i
s
onl
y obtained by
proposed BB
O approach
and
Cu
ckoo al
go
rithm (CA)
with 15 po
pul
ations. Also
the minimum
power lo
ss is obtai
ned
by
perfo
rming B
B
O and cuckoo method
s.
12
34
56
7
8
9
10
1
1
12
1
3
14
1
5
16
1
7
18
26
2
7
28
3
0
29
3
1
32
3
3
23
2
4
25
19
20
21
22
Figure 2. Single line diag
ram of
33-b
u
s
dist
rib
u
t
i
on t
e
st
sy
st
em
Figure 5 illust
rates the
compari
s
on bet
ween
the number
of funct
i
on evaluations
(NFE
s)
of seven diffe
rent alg
o
rith
ms. The
s
e
NFEs are
obtai
ned fro
m
100
iteration
s
for solving n
e
twork
reconfiguration problem f
o
r IEEE-33
Bus
and
150 iterations f
o
r I
EEE-69 Bus
dis
t
ribution
system
s. As
sho
w
n i
n
Fig
u
re
5, the nu
mber
of
funct
i
on evalu
a
tio
n
for
CA is m
u
ch
gre
a
ter t
han
BBO and th
e other i
n
tell
igent metho
d
s
, therefo
r
e
i
t
can b
e
co
nclu
ded th
at for this
spe
c
ific
probl
em i.e. DNR, BBO appro
a
ch is be
tter
than the other six intell
igent method
s.
Table 1. 33
-Bus System Result
s on the
Different Met
hod
s
Metho
d
Ope
n
s
w
itches
Ploss (kW
)
Initial 33,34,35,36,
37
210.84
BBO 7,9,14,32,37
139.55
Cuckoo 7,9,14,32,37
139.55
TLBO 8,9,28,32,33
147.91
PSO 7,11,14,28,3
6
143.15
GA
9,33,34,35,3
6
151.76
GSA
7,11,28,34,3
6
144.28
ABC 7,11,12,28,3
2
150.29
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 480 – 489
485
Figure 3. Co
mpari
s
o
n
of 33-bu
s sy
stem
indice
s for B
B
O, GA, PSO
, ABC and TL
BO in powe
r
loss
Figure 4. Co
mpari
s
o
n
of 33-no
de sy
ste
m
indi
ce
s for
BBO, GSA a
nd Cu
ckoo in
power lo
ss
The po
wer lo
ss of the opti
m
um config
u
r
ation
in co
m
pari
s
on
with power lo
ss o
f
initia
l
config
uratio
n
is red
u
ced
from 210.6
79 kW
to 1
39.55 kW. Another
adva
n
tage of DNR is
improvin
g voltage profile.
Figure 6 re
p
r
esents th
e voltage profil
e of t
he distribution
syste
m
before a
nd af
ter netwo
rk re
config
uratio
n.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Com
p
rehe
nsi
v
e Study o
n
Specifyi
ng
an Intelligent
Appro
a
ch to Solve …
(Ma
h
m
oud RS)
486
Figure 5. Co
mpari
s
o
n
of numbe
r of func
tion evaluatio
n (NFE
) for solving network
reconfigu
r
atio
n probl
em bet
wee
n
PSO, Cuckoo, GA, TLBO, ABC, GSA and BBO method
s for
IEEE-33, 69 s
y
s
t
ems
Figure 6.
Voltage profile for the 33-Bu
s syste
m before and after reconfiguration
It should be n
o
ted that gre
a
ter NFE ca
u
s
e more time consumin
g and slo
w
perfo
rman
ce
during optimiz
ation proc
es
s
.
For example c
pu time c
o
ns
umed during IEEE-33 bus
net
work
reconfigu
r
atio
n, was 6.0
5
s
fo
r BBO approach and 1
7
.12s for
CA.
4.2. IEEE-69 Bus Test S
y
stem
The 69
-Bu
s
distrib
u
tion
system, inclu
d
e
s 6
9
no
de
s and 7
3
bra
n
ch
es. Th
ere
are 68
se
ctionali
z
ing
switch
es a
n
d
5 tie switches
an
d the total loads are 3.802M
W and 3.696 M
VAr
[15]. The
dia
g
ram
of the
system i
s
d
epi
cted i
n
Fi
gure 7. T
he
ope
n switche
s
a
r
e 69,
70, 7
1
,
72,
73. After performin
g the propo
sed reco
nfigurat
io
n b
a
se
d on BBO, switche
s
20, 58, 61, 69, 7
1
are op
ene
d a
nd the netwo
rk lo
sses a
r
e
redu
ce
d from
224.95 kW to 115.88
kW.
These n
e
two
r
k
config
uratio
n
and
po
wer l
o
ss redu
ction
are
only
o
b
tained
by u
s
i
ng cucko
o
a
ppro
a
ch b
u
t with
greate
r
NFE. The simul
a
tion re
sults
of netwo
rk
re
co
nfiguratio
n for the
69-bu
s system o
b
tai
ned
by seven different intellig
e
n
t algorithm
s
with pop
ulatio
n size=40 an
d 150 iteratio
ns, are
sho
w
n in
Table 2.
1500
9537
1715
4515
1489
1500
1515
4040
24532
4440
12040
4097
4000
4040
0
5000
10000
15000
20000
25000
30000
NFE
IEEE-33 Bus
IEEE-69 Bus
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 480 – 489
487
Figure 7. Single line diag
ram of
69-b
u
s
dist
rib
u
t
i
on t
e
st
sy
st
em
Figure 8, 9
show th
e com
pari
s
on
between final
fitne
ss
co
nverg
e
n
c
e
s
of BBO a
ppro
a
ch
with six othe
r different met
hod
s. The op
timal c
onfig
uration of IEEE-69
b
u
s
syste
m
only obtain
ed
by pro
p
o
s
ed
BBO app
roa
c
h an
d Cucko
o
algo
rithm
(CA). But a
s
shown in
Figu
re 5, the
num
ber
of functio
n
ev
aluation fo
r
CA is mu
ch
g
r
eater
th
an B
B
O metho
d
t
herefo
r
e
it ca
n be
co
ncl
u
d
e
d
that for thi
s
spe
c
ific probl
em (DNR),
BBO
app
roa
c
h i
s
better
than the
oth
e
r
six intellig
ent
method
s.
Table 2. 69
-Bus System Result
s on the
Different Met
hod
s
Metho
d
Ope
n
s
w
itches
P
loss
(kW
)
Initial 69,70,71,72,
73
224.95
BBO 20,58,61,69,
71
115.88
Cuckoo 20,58,61,69,
71
115.88
TLBO 20,45,58,64,
69
127.23
PSO 20,42,45,57,
64
128.89
GA
13,26,42,57,
71
133.69
GSA
14,22,57,69,
71
135.01
ABC 20,22,42,45,
57
138.54
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
A Com
p
rehe
nsi
v
e Study o
n
Specifyi
ng
an Intelligent
Appro
a
ch to Solve …
(Ma
h
m
oud RS)
488
Figure 8. Co
mpari
s
o
n
of 69-bu
s sy
stem
indice
s for B
B
O, GA, PSO
, ABC and TL
BO in powe
r
loss
Figure 9. Co
mpari
s
o
n
of 69-Bu
s syste
m
indice
s for B
B
O and Cu
ckoo in po
wer l
o
ss
Figure 10. Voltage profile f
o
r the 69
-Bus
system befo
r
e and after re
config
uratio
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 3, March 20
16 : 480 – 489
489
Figure 10 shows
the voltage pr
ofile
of the IEEE-69
bus
dis
t
ribution s
y
s
t
em before
and
after netwo
rk reconfigu
r
atio
n. As it can b
e
see
n
in Fig
u
re
s 6, 10 net
work reconfig
uration n
o
t on
ly
improve
s
volt
age
profile,
b
u
t also
red
u
ces th
e d
e
viation of volta
g
e
s
. Th
ese affe
cts i
n
cre
a
se t
he
distrib
u
tion sy
stem po
we
r q
uality.
5.
C
o
nc
lu
s
i
on
This pa
pe
r p
r
opo
se
s an a
ppro
a
ch ba
sed on bio
geo
grap
hy-b
ased
optimization
(BBO)
algorith
m
to solve the di
stribution net
work
re
c
onfig
u
r
ation (DNR) proble
m
for redu
cin
g
acti
ve
power l
o
sse
s
.
Voltage
profile imp
r
oveme
n
t after n
e
two
r
k
re
co
nfigura
t
ion is al
so ill
ustrate
d
in
thi
s
study. The m
a
in advanta
g
e
of this
stud
y is com
p
a
r
in
g seve
n different intelligen
ce al
gorith
m
s in
solving network re
co
nfigu
r
ation
p
r
o
b
le
m.
This co
m
pari
s
on i
n
cl
u
des fin
a
l fitne
ss i
n
optimi
z
ation
pro
c
e
ss and numbe
r
of
fu
nction evalua
tion
(NFE).
T
he effectiven
ess of the BB
O app
roa
c
h
has
been
teste
d
on two diffe
rent di
stributi
on n
e
tw
o
r
ks and
the
obt
ained
sim
u
la
tion re
sult
s
are
comp
ared wit
h
Geneti
c
al
gorithm
(GA), PSO,
ABC, GSA, TLBO and Cucko
o
algorithm
(CA).
Comp
ari
ng th
e simulatio
n
result
s verifies that BBO approa
ch
can b
e
an efficient
and promi
s
in
g
method fo
r
solving
DNR pro
b
lem
s
in
com
pari
s
o
n
with the
si
x other m
e
n
t
ioned intelli
gent
algorith
m
s.
Referen
ces
[1]
Dah
a
la
n WM, Mokhlis
H, Ah
mad R, B
a
kar
AA, Mu
sirin
I. Simulta
neo
us
Net
w
ork
Reco
nfigur
ation
a
n
d
DG Sizing Us
ing Evo
l
uti
ona
r
y
Pro
g
rammi
ng an
d Gen
e
tic Algor
ithm to Minimiz
e
P
o
w
e
r L
o
sses
.
Arabi
an Jo
urna
l for Science a
nd Eng
i
n
eeri
n
g
. 2014; 39(
8): 6327-
38.
[2]
Rao GS, Obules
h YP. Voltage Profi
l
e Improv
em
ent of Distributi
o
n
S
y
stem usi
n
g Distrib
ute
d
Generati
n
g
Un
its.
Internatio
n
a
l J
ourn
a
l
of E
l
ectrical
a
nd
C
o
mputer
En
gin
eeri
ng (IJECE)
. 201
3; 3(
3):
337.
[3]
Merlin A, Back
H.
Search for a mi
ni
mal-l
o
ss oper
ating s
pan
ni
n
g
tree confi
gurati
on i
n
an
urba
n pow
e
r
distribution sys
tem
. InProc. 5th Po
w
e
r S
y
stem Computatio
n Conf., Cambridge, UK. 1975:
1-18.
[4]
Martín JA, Gil AJ. A ne
w
h
e
u
ristic a
ppro
a
c
h
for distri
buti
on s
y
stems l
o
ss reducti
on.
Electric Power
System
s Res
e
arch
. 200
8; 78(
11): 195
3-8.
[5]
Civan
lar S, Grain
ger JJ, Yin
H, Lee SS. D
i
s
tr
ibutio
n fee
d
e
r reco
nfigur
ati
on for l
o
ss red
u
ction.
IEEE
Trans. Pow
e
r Deliv
ery
. 198
8; 3(3).
[6]
Nara K, S
h
ios
e
A, Kitaga
w
a
M, Ishihar
a
T
.
Implementa
t
ion of
ge
netic
alg
o
rithm for
distrib
u
tio
n
s
y
stems loss
minimum re-c
o
n
figur
ation.
IE
EE T
r
ansactio
n
s on Pow
e
r Systems
. 19
92; 7(3): 104
4-5
1
.
[7]
Arand
ian B, H
ooshm
and
RA, Gholip
our E.
Decre
a
sin
g
act
i
vit
y
c
o
st of a distrib
u
tion s
y
s
t
em compa
n
y
b
y
r
e
confi
gur
ation a
nd p
o
w
e
r
gener
atio
n co
ntrol of
DGs b
a
sed o
n
sh
uffled frog l
eap
in
g alg
o
rithm.
Internatio
na
l Journ
a
l of Electr
ical Pow
e
r & Energy Syste
m
s
. 2014; 61: 48-
55.
[8]
Aman MM, J
a
smon GB,
Bakar AH, M
o
khlis
H. Optimum n
e
t
w
ork
reconfi
gur
atio
n bas
ed
on
maximizati
on
of s
y
stem lo
a
dab
ilit
y
usi
ng
contin
uati
on p
o
w
e
r flo
w
t
h
e
o
rem. Internati
ona
l jo
urna
l of
electric
al po
w
e
r & energ
y
s
y
stems. 2014; 5
4
: 123-3
3
.
[9]
Rao RS, Ravindra K,
Satish K
,
Narasimh
am
SV. Po
w
e
r los
s
minimiz
a
tion
in distri
buti
on s
y
stem us
in
g
net
w
o
rk r
e
co
n
f
igurati
on
in t
he pr
es
e
n
ce of
distrib
u
ted gen
eratio
n.
IEEE Transactions on Power
System
s
. 20
13
; 28(1): 317-2
5
.
[10]
Rao RS, R
a
vi
n
d
ra K, Satish K
,
Narasimh
am
SV.
Po
w
e
r los
s
minimiz
a
tion
in distri
buti
on s
y
stem us
in
g
net
w
o
rk r
e
co
n
f
igurati
on
in t
he pr
es
e
n
ce of
distrib
u
ted gen
eratio
n.
IEEE Transactions on Power
System
s
. 20
13
; 28(1): 317-2
5
.
[11]
Seda
gh
at M, Rokrok E, Ba
k
h
s
h
ip
our M. A N
o
vel Meth
od B
a
sed o
n
Bi
oge
o
g
rap
h
y
-Base
d
Optimizatio
n
for DG Pla
nni
n
g
in
Distrib
utio
n S
y
stem.
T
E
L
K
OMNIKA Indones
ian
Jour
n
a
l
of El
ectrica
l
Engi
neer
in
g.
201
5; 15(1): 1-
3.
[12]
Kothari
DP
. Power system
optim
i
z
a
t
i
on
. In Computati
o
n
a
l Intelli
ge
nce a
n
d
Signa
l Proces
sing (CISP
)
.
201
2: 18-2
1
.
[13]
Simon D. Bio
g
eogr
aph
y-
bas
e
d
optimiz
ation.
IEEE Transactions on Evolut
ionary Com
p
utation
. 20
08;
12(6): 70
2-1
3
.
[14]
Baran ME, W
u
F
F
.
Net
w
o
r
k reconfi
gur
ati
on
i
n
d
i
stributi
on s
y
stems fo
r loss re
ducti
on a
nd
lo
ad
bal
anci
ng.
IEEE Transactions
on Power Delivery
. 1989; 4(
2
)
: 1401-7.
[15]
Chia
ng H
D
, Jean-J
u
mea
u
R
.
Optimal net
w
o
rk rec
onfi
gur
ations i
n
distri
butio
n s
y
stem
s. II. Solution
alg
o
rithms an
d
numeric
al res
u
lts.
IEEE Transactions on Power Delivery
. 1
990; 5(3): 1
568
-74.
Evaluation Warning : The document was created with Spire.PDF for Python.