Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
Vol.
5, No. 6, Decem
ber
2015, pp. 1311~
1
318
I
S
SN
: 208
8-8
7
0
8
1
311
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Application of ANFIS for Dist
ance Relay Protection in
Transmission Line
A
z
riy
e
nni*
,
M
o
hd Wa
zir
M
u
st
af
a**
*Faculty
of Engineering
,
Univ
ersitas Riau, Indonesia
**Faculty
of Electrical
En
g
i
neer
ing, Universiti Te
knologi Malay
s
ia, Malay
s
ia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Feb 17, 2015
Rev
i
sed
Au
g 9, 201
5
Accepted Aug 22, 2015
The t
echniqu
es
h
y
brid
int
e
ll
igen
t was introduc
e
d
in transm
ission protec
tion
that us
ag
e in
ele
c
tri
c
power s
y
s
t
em
s
.
There was
appli
e
d ANF
IS
for dis
t
an
c
e
relay
pro
t
ection
particular
ly
for transm
ission line.
If a fau
lt o
ccurs
during th
e
transmission line iden
tification
caused b
y
unw
anted f
a
ult thus the power
deliv
er
y
to the consumer becomes not
going well. Th
erefor
e, it
would need
to provide
an
al
terna
tive
solutio
n to fix
this pro
b
lem
.
Th
e obj
ec
tive of
thi
s
paper uses impedance
tr
ansmission line to deter
m
ine how long
the ch
annel
spacing will be
protec
ted b
y
d
i
st
ance re
la
y. I
t
ha
s been distance r
e
la
ys when
fault o
ccurs
in t
r
ansm
ission line
with
th
e
application Sugeno
ANFIS. The
sim
u
lation shows it excel
len
t
test
ing results can b
e
contribu
ted to
an alt
e
rna
t
e
algorithm th
at
it has good perfor
m
ance to
pro
t
ecting s
y
stem in tr
ansmission
line
.
Th
is app
lic
ation
used b
y
usi
ng software
Mat
l
ab.
Keyword:
AN
FIS
Detectio
n
Distance Relay
Fau
lt
Transm
ission Line
Copyright ©
201
5 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
A
z
r
i
yenn
i,
Lect
ure
r
at
Fac
u
l
t
y
of
E
ngi
ne
eri
n
g,
Uni
v
ersitas Ri
au,
Indonesia.
Jl. Subr
an
tas Si
m
p
. Pan
a
m
Km
. 1
2
.
5, Pek
a
nb
aru
,
In
don
esi
a
Em
a
il: azriyen
n
i
@lecturer.unri.ac.id
1.
INTRODUCTION
Fau
lt o
ccurs in th
e tran
sm
issi
o
n
lin
e is ex
p
e
cted
to
av
o
i
d
e
d
,
u
tility p
r
o
b
l
e
m
s an
d
eq
u
i
pmen
t d
a
m
a
g
e
fro
m
effect o
f
th
e arc and
so
o
n
. Th
ese failu
res are
d
i
srupted
th
e reliab
ility o
p
e
ration
of th
e power sy
ste
m
.
The di
f
f
ere
n
t
r
e
searche
r
s t
o
o
v
erc
o
m
e
i
n
t
h
is pr
obl
em
have sug
g
est
e
d m
a
ny
vari
ous sc
hem
e
s and al
gori
t
h
m
s
.
There a
r
e seve
ral techni
ques
to detect fault
in the tr
ansm
ission system
,
they are:
time dom
ain, fre
quency
d
o
m
ain
,
and
wav
e
let tran
sfo
r
m
an
d
h
y
b
r
i
d
in
tellig
en
t tech
n
i
q
u
e
. Lin et.al p
r
esen
ted
a d
e
tection
o
f
fau
l
t
in
po
wer system b
y
u
s
ing
Ad
ap
tiv
e
Prob
ab
ilistic Neu
r
al
Netwo
r
k
architectu
r
e [1
-3
].
Th
e fau
lt d
e
tectio
n
b
y
u
s
ag
e Prob
ab
ilistic Neu
r
al
Network
ob
tain
s i
n
fo
rm
atio
n
from
p
r
i
m
ary an
d b
a
ck
up
p
r
o
t
ectiv
e d
e
v
i
ce to
create
th
e train
i
n
g
set. Also,
Om
e
r
et.al presen
t
e
d
to u
s
e of
Artificial Neural Ne
twor
k w
ith
b
ackpr
opag
a
tio
n
structure as an
alternative
m
e
tho
d
f
o
r
det
ect
i
ng
faul
t
an
d fa
ul
t
cl
assi
fication in transm
ission system
s [4]. The
pape
r ca
n be
cl
assi
fi
ed a fe
w fa
ul
t
s
i
n
u
n
s
ym
m
e
t
r
i
cal faults. Meanwhile, Chen &
Agga
rwal pres
e
n
ted a
cl
assi
fi
cat
i
on o
f
fa
ul
t
and
faul
t
det
ect
i
on sc
h
e
m
e
i
s
appl
y
from
dat
a
st
ream
si
gnal
i
n
t
o
t
r
ansm
i
ssi
on sy
st
em
.
The decaying flow
si
gnals
m
e
asure
d
usi
n
g
wave
let to
obtain the
re
qu
ired freque
ncy [5].
The t
y
pes of
faul
t
s
are i
d
en
t
i
f
i
e
d by
pro
p
o
se
d schem
e
of w
h
i
c
h i
s
cho
o
si
ng ne
u
r
al
net
w
o
r
k t
o
especi
al
l
y
di
st
ing
u
i
s
h i
n
t
e
r
n
al
di
st
ur
ba
nce and e
x
t
e
r
n
al
di
s
t
ur
bances
. It
c
a
n be
used t
h
e
sam
e
pat
t
e
rn i
n
t
h
e
feat
ure
s
by
e
x
t
r
act
ed ea
rl
i
e
r [
6
]
,
[
7
]
.
Thi
s
pa
per
desc
ri
bes
h
o
w t
o
de
si
g
n
a
nd
de
vel
o
pm
ent
t
h
e
new t
e
c
h
ni
q
u
es
that can detect and classify the type
of error by using a Hybrid Intellige
n
t Techniques. It is also introduce
d
t
h
e nam
e
of
N
e
ural
Net
w
or
k
and
F
u
zzy
Sy
s
t
em
s.
The
faul
t area location
has
b
ecom
e
a necessary step in t
h
e
fau
lt lo
cation
of
d
i
stribu
tio
n network [2
1
]
. Th
e
o
u
t
p
u
t
lin
e
v
o
ltag
e
s at lo
ad
term
in
als are u
s
ed
as th
e med
i
u
m
fo
r fa
ul
t
det
ect
i
on.
A l
i
n
e t
o
g
r
o
u
nd fa
ul
t
i
s
defi
ned as a si
ngl
e c
o
n
n
ect
i
o
n [
22]
.
Th
ese
r
e
su
lts of
study ar
e
ex
p
ected
g
ood
ab
ility o
f
th
e meth
od
th
at
h
a
s
b
een used previo
u
s
ly.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1311 –
1318
1
312
In
th
is p
a
p
e
r, Th
e ANFIS
is p
r
esen
ted
b
y
usin
g
Fu
zzy If-t
h
e
n-ru
les i
n
to
Neural
Netwo
r
k
co
nstru
c
ti
on
usi
n
g a
p
pr
o
p
ri
at
e l
earni
ng
an
d
red
u
ce
t
h
e
o
u
t
c
om
e of
fa
ul
t
base
d
o
n
t
h
e
d
a
t
a
of
t
h
e
t
r
ai
ni
ng
set
.
A
N
F
I
S
of
fer
s
a t
echni
q
u
e fo
r t
h
e fuzzy
m
odel
i
n
g t
o
st
u
d
y
t
h
e out
put
of t
h
e dat
a
set
.
It
i
s
obt
ai
ne
d param
e
t
e
rs of t
h
e
me
m
b
ersh
ip
fun
c
tio
ns asso
ciated
with
th
e fuzzy in
feren
ce syste
m
to
treat i
n
pu
t o
r
o
u
t
p
u
t
o
f
d
a
ta. Sin
g
l
e lin
e
di
ag
ram
t
r
ansm
i
ssi
on sy
st
em
i
s
m
odel
wi
t
h
a v
o
l
t
a
ge
of
1
5
0
kV
an
d
1
1
7
km
l
ong.
Fi
gu
re 1.
Tra
n
s
m
ission Line
S
y
ste
m
Zone1
80%
∗
GS
D
I
∗Z
∗R
a
s
i
o
(
1
)
Zone2
GS
D
I
DI
D
U
∗
50%
∗
Z
∗R
a
s
i
o
(
2
)
Zone3
GS
D
I
DI
D
U
∗
120%
∗
Z
∗R
a
s
i
o
(
3
)
2.
R
E
SEARC
H M
ETHOD
Fuzzy
I
n
fere
nc
e Sy
st
em
(FIS
)
ap
pl
i
e
d
wi
t
h
m
odel
i
ng o
f
t
h
e sy
st
em
s i
s
not
cl
ear.
Thi
s
s
y
st
em
has a
st
ruct
u
r
e o
f
r
u
l
e
s defi
ne
d by
use
d
i
n
t
e
rp
ret
a
t
i
on feat
u
r
es o
f
t
h
e m
odel
vari
abl
e
s. T
h
e m
e
m
b
ershi
p
f
u
n
c
t
i
ons
are sel
ect
ed i
n
som
e
si
t
u
at
i
ons m
odel
i
ng a
n
d
i
ndi
st
i
ngu
ish
a
b
l
e wh
ere th
e
me
m
b
ersh
ip
fun
c
tio
n sh
owed
it d
a
ta
[
3
],
[6
],
[10
]
.
ANFIS is a m
u
lti-layer
m
e
thod t
o
update network, it
is utilized neural networ
k learning algorithm
s
an
d fu
zzy reaso
n
i
n
g
th
at are
represen
ted
i
n
p
u
t
t
o
o
u
t
p
u
t
.
Th
e
v
e
rb
al ab
ility to
u
n
i
fy
v
a
lu
e of a
fuzzy syste
m
wi
t
h
t
h
e
n
u
m
e
ri
c val
u
e
of
ne
ural
net
w
or
k a
d
apt
i
v
e
,
AN
FI
S has
bee
n
sh
ow
n
g
o
o
d
per
f
o
rm
ance i
n
m
odel
i
n
g
plenty activity, excellent c
o
mpetence
l
earni
ng
an
d cl
assi
f
y
i
ng t
h
at
i
t
co
ul
d
u
pdat
e
m
a
ny
sy
st
em
s. It
has t
h
e
adva
ntage
of allowing th
e
extraction Fuz
z
y rules from
num
erical
d
a
ta o
r
exp
e
r
t
kn
ow
ledg
e an
d b
a
se
ad
ap
tiv
ely constru
c
ts a ru
le base. In
add
itio
n
a
l, it can
b
e
tu
n
e
d
i
fficu
lt co
nv
ersion
of hu
m
a
n
in
tellig
e
n
ce t
o
fuzzy
Sy
stem
s [1
1]
-[
1
2
]
,
[
1
7]
–[
2
0
]
.
Fi
gu
re
2.
FI
S
m
odel
i
ng sy
st
em
for v
o
l
t
a
ge
a
n
d
cu
rre
nt
Layer 1:
Fuzzi
fication
The
node
of this layer carry
out
m
e
m
b
ership de
gree
,
which is incl
ude
d
t
o
each of
c
o
m
p
atible
fuz
z
y,
sets
b
y
u
s
ing
me
m
b
ersh
ip
fun
c
tio
ns.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Application of
ANFIS for
Dist
ance
Relay
Pr
otection i
n
Transmissi
on Line
(Azriyenni)
1
313
O
,
μ
x
f
o
ri
1,2
(
4
)
O
,
μ
y
f
o
ri
3,4
(
5
)
Whe
r
e:
x,
y
ar
e cr
is
p inpu
ts
to
n
o
d
e
i
an
d
A
,B
are
th
e lingu
istic lab
e
ls
b
e
co
m
i
n
g
m
e
m
b
ersh
i
p
fu
n
c
tion
s
μ
,
μ
, in e
v
e
r
y each
othe
r.
Fi
gu
re
3.
Tri
a
n
gul
a
r
M
e
m
b
ershi
p
Fu
nct
i
o
n
The m
e
m
b
ership function can be a
n
y a
p
propriate functio
n s
u
ch
as;
Gaus
si
an, t
r
apez
oi
dal
,
ge
neral
i
zed
be
l
l
an
d triangu
lar
[14
]
. Th
e m
e
mb
ersh
ip fun
c
tion
in th
is
p
a
p
e
r
sh
ows in
Figu
re 3
.
μ
1
ifa
a
u
a
1
ifa
u
a
a
0
otherwise
(
6
)
μ
1
ifb
b
v
b
1
ifb
v
b
b
0
otherwise
(
7
)
Whe
r
e:
a
,b
,
c
are
param
e
ter set of t
h
e m
e
m
b
ersh
ip fun
c
tio
n
s
i
n
p
r
em
ise p
a
rt of
fu
zzy if-then
ru
les th
at
changes
the s
h
apes
of th
e m
e
m
b
ershi
p
fu
nct
i
on.
Lay
e
r
2: If
-T
h
e
n rule
Th
e
AN
D op
erato
r
u
s
ed
up
to on
e
ou
tpu
t
wh
ich ind
i
cates
the re
sult
of the an
teced
e
n
t
fo
r ru
le,
i.
e.
,
fi
ri
n
g
st
re
n
g
t
h
.
Fi
ri
n
g
st
re
ngt
h m
eans i
s
t
h
e deg
r
ee a
n
t
ece
dent
part
of a
fuzzy
r
u
l
e
,
w
h
i
c
h sat
i
s
fi
ed,
a
nd i
t
sha
p
es out
put
fu
nct
i
o
n fo
r t
h
e
r
u
l
e
. In
t
h
e ot
he
r
ha
n
d
,
t
h
e o
u
t
p
ut
s
O
,
, of th
is layer are
p
r
od
u
c
ts of
appropriate de
grees
.
O
,
w
μ
x
∗
μ
y
(
8
)
i
1,2;
j
1
,2;
k
2
i1
j
Lay
e
r 3:
N
o
rm
al
i
zat
i
on
Th
e m
a
in
ob
j
e
ctiv
e is to
m
a
n
a
g
e
n
u
m
erate th
e ratio
o
f
each
ru
le’s firi
ng
streng
th
t
o
th
e su
m
o
f
al
l
rules firing stre
ngt
h. In
e
v
ery each
s
p
ecial
chase,
w
is tak
e
n as th
e
no
rm
alize
streng
th.
O
,
w
i
1
,2,3,4
(
9
)
Layer 4
:
Defu
zzificatio
n
The
n
ode
f
unct
i
on
o
f
t
h
e
f
o
u
r
t
h
l
a
y
e
r c
o
u
n
t
up
t
h
e e
ffect
of each
ru
le toward th
e t
o
tal ou
tpu
t
and
establish
it
as fo
llo
w:
O
,
w
z
w
p
xq
yr
i
1,
…
,
4
(
1
0
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1311 –
1318
1
314
Layer
5
:
Neu
r
on
Add
itio
n
The n
o
d
e cal
c
u
l
a
t
e
s al
l
of t
h
e o
u
t
p
ut
by
s
u
m
m
i
ng al
l
t
h
e i
n
com
i
ng si
gnal
s
. I
n
co
ns
eque
nce, t
h
e
defuzzification
create process each
rule’s fuz
z
y results i
n
to
a cr
isp output i
n
this
layer.
O
,
∑
w
z
(
1
1
)
Fi
gu
re 4.
St
r
u
c
t
ure of
A
N
F
I
S M
odel
The structure of
ANFIS is s
h
owe
d
in
Figure 4
,
in
wh
ich
a circle sh
ows a it
to
fix
e
d
no
d
e
wh
ile a
square a
d
duce
s an a
d
aptive
node
[1
5]. T
o
ward easine
ss, the fuzzy infe
rence sy
st
em
cont
em
pl
at
i
on h
a
s t
w
o
i
n
p
u
t
s
,
x a
n
d
y
and
o
n
e
o
u
t
put
z
.
T
h
i
s
net
w
o
r
k
i
s
t
r
ai
ne
d c
once
r
n
of
obs
er
ve l
ear
ni
ng
. T
h
e
p
u
r
p
o
s
e t
r
ai
n
ad
ap
tiv
e n
e
t
w
or
k
s
to
p
r
ov
id
e
u
nkn
own
fun
c
t
i
o
n
s
w
ith appoin
t
ed
tr
ai
n
i
ng
d
a
ta and
calcu
l
a
te th
e r
e
su
lt ob
tain
of t
h
e
val
u
e ab
ove
param
e
t
e
rs. The di
f
f
e
r
ent
of t
y
pi
cal
of t
h
e ap
pr
oac
h
A
N
FI
S ha
ve a h
y
b
ri
d al
g
o
ri
t
h
m
,
t
h
e
gra
d
i
e
nt
desce
n
t
m
e
t
hod a
n
d t
h
e
l
east
-
s
q
uares
m
e
t
hod,
t
o
up
dat
e
p
a
ram
e
t
e
rs. Th
e g
r
adi
e
nt
de
scent
classification i
s
used to tune
prem
ise
non-linear param
e
ters
i
,i
,i
, At
t
h
e
t
i
m
e, l
east
s
qua
res
t
echni
que
u
tilize to
recog
n
i
ze con
s
equen
t
lin
ear
p
a
ra
m
e
ters
i
,i
,i
.
Thus, t
h
e
f
a
u
l
t
comes to
t
h
ro
ugh
si
gn
als
pr
o
p
agat
e
bac
k
wa
r
d
.
Gra
d
i
e
nt
desce
n
t
t
a
k
e
s a ne
w t
e
c
h
nique the
pre
m
ise param
e
te
rs.
When t
h
e
process
con
s
eq
ue
nt
pa
ram
e
t
e
rs im
prove
can
be
re
duce
t
h
e
ove
r
a
l
l
qua
drat
i
c
o
p
erat
i
o
nal
co
st
. The a
n
al
y
z
i
n
g an
d
math
e
m
atica
l
o
b
s
erv
e
con
d
ition
o
f
th
e
h
ybrid-learn
i
n
g
algo
rith
m
will sh
ow it in
[14
]
, [15
]
.
3.
R
E
SU
LTS AN
D ANA
LY
SIS
Th
e ap
p
lication
o
f
t
h
e adap
ti
v
e
techn
i
qu
es
will d
e
term
in
e zon
e
settin
g of d
i
stan
ce relay p
r
o
t
ectin
g
th
e tr
an
sm
issio
n
system
il
lu
str
a
ted
in
Figu
r
e
1
.
Th
e test syst
e
m
co
n
s
ists
o
f
2
4
d
i
stan
ce r
e
l
a
ys o
p
e
r
a
ting
at 150
kV levels. The power syste
m
analy
zed are setting and checki
ng t
o
cove
ra
ge provi
d
ed
by each z
one
of
distance relay.
Fi
gu
re
5 s
h
o
w
s t
h
e i
m
pedan
ce dat
a
at
a
n
y
di
st
ance
fr
om
t
h
e s
ubst
a
t
i
o
n t
o
t
h
e
next
sub
s
t
a
t
i
o
n
.
Train
i
ng
o
f
t
h
is
m
o
d
e
l m
a
y
o
p
tim
ize p
a
rameter as a trai
nin
g
er
ro
r. T
h
e variatio
n o
f
the trainin
g
e
r
ro
r is
respect
t
o
t
h
e
num
ber
of i
t
e
r
a
t
i
on f
o
r t
h
i
s
m
odel
.
The va
ri
at
i
on
of i
m
pedance
wi
t
h
di
ffe
rent
z
one i
n
put
i
s
sho
w
n i
n
Fi
gu
r
e
8 a
n
d
9.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Application of
ANFIS for
Dist
ance
Relay
Pr
otection i
n
Transmissi
on Line
(Azriyenni)
1
315
Fi
gu
re 5.
Fl
o
w
chart
Su
ge
no
AN
FIS
Fi
gu
re
6.
Trai
n
i
ng
o
f
Dat
a
wi
t
h
e
poc
hs
f
o
r
f
(
u
)
There
are
9
rul
e
s,
whi
c
h a
r
e s
u
f
f
i
c
i
e
nt
t
o
i
d
e
n
t
i
f
y
co
ve
rage
zone
usi
n
g
AN
FIS.
S
o
m
e
of t
h
ese
rul
e
s
are as
follows:
1)
If
(
voltage
is in
1m
f1) a
n
d
(c
ur
rent is i
n2m
f1)
then
(im
p
edan
ce is o
u
t1m
f
1)
(1
)
2)
If
(
voltage
is in
1m
f1) a
n
d
(c
ur
rent is i
n2m
f2)
then
(im
p
edan
ce is o
u
t1m
f
2)
(1
)
3)
If
(
voltage
is in
1m
f3) a
n
d
(c
ur
rent is i
n2m
f3)
then
(im
p
edan
ce is o
u
t1m
f
3)
(1
)
4)
If
(
voltage
is in
1m
f2) a
n
d
(c
ur
rent is i
n2m
f1)
then
(im
p
edan
ce is o
u
t1m
f
4)
(1
)
5)
If
(
voltage
is in
1m
f2) a
n
d
(c
ur
rent is i
n2m
f2)
then
(im
p
edan
ce is o
u
t1m
f
5)
(1
)
6)
If
(
voltage
is in
1m
f3) a
n
d
(c
ur
rent is i
n2m
f3)
then
(im
p
edan
ce is o
u
t1m
f
6)
(1
)
7)
If
(
voltage
is in
1m
f3) a
n
d
(c
ur
rent is i
n2m
f1)
then
(im
p
edan
ce is o
u
t1m
f
7)
(1
)
8)
If
(
voltage
is in
1m
f3) a
n
d
(c
ur
rent is i
n2m
f2)
then
(im
p
edan
ce is o
u
t1m
f
8)
(1
)
9)
If
(
voltage
is in
1m
f3) a
n
d
(c
ur
rent is i
n2m
f3)
then
(im
p
edan
ce is o
u
t1m
f
9)
(1
)
Th
e system
si
m
u
la
tio
n
MATLAB resu
lts are sho
w
n
i
n
Figu
re
6
un
til Figu
re
10
. Th
e i
n
p
u
t
en
tered
t
o
th
e ANFIS th
en
vo
ltag
e
and
cu
rren
t are testin
g
d
a
ta fo
r t
h
e t
r
ai
ni
n
g
p
r
o
cess. Al
l
val
u
e
s
obt
ai
ne
d f
r
o
m
t
h
e
co
rresp
ond
ing
fau
lt wh
ere
g
i
ven
as i
n
pu
t for th
e ANFIS and
also will b
e
g
o
t
b
e
tter
resu
l
t
s with
th
e v
a
l
u
es as
su
ppo
sed
.
Th
e
b
a
sed
ru
les ANFIS used
t
o
op
ti
m
i
ze train
i
n
g
m
o
d
e
led
w
h
i
ch
is shown
i
n
Fig
u
re
8
un
til 1
0
. Th
e
tech
n
i
qu
e prop
o
s
ed
uses sho
r
t circu
it fault to
d
e
te
r
m
ine the appa
rent im
pedance seen by zone relay
pr
ot
ect
i
o
n
.
When t
h
e fa
ul
t
s
are gi
vi
ng
ri
se t
o
far
fr
o
m
buses a
nd
faul
t
are i
n
cl
u
d
i
n
g cu
rre
nt
e
ffect
t
o
corres
ponding
zone
cove
rage
. T
h
e
require
d
zone
reache
s
are
c
o
m
puted for
all netw
ork topologies t
h
at is
con
s
i
d
eri
n
g
co
nt
i
nge
nci
e
s a
n
d
out
put
s
o
f
ge
nerat
i
o
n s
o
urce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1311 –
1318
1
316
Fi
gu
re 7.
De
fu
zzi
fi
cat
i
on fo
r set
t
i
ng of
z
o
ne pr
ot
ect
i
o
n
Fi
gu
re
8.
R
e
l
a
t
i
on c
u
rre
nt
an
d
v
o
l
t
a
ge t
o
m
a
in z
one
cove
ra
ge
Fig
u
re
9
.
Relatio
n
curren
t
and vo
ltag
e
to th
e
first
back
u
p
z
one
c
ove
ra
ge
Fi
gu
re
1
0
. R
e
l
a
t
i
on c
u
r
r
ent
a
n
d
v
o
l
t
a
ge t
o
t
h
e sec
o
nd
bac
k
up
zo
ne c
o
vera
ge
From
t
h
e Ta
bl
e 1, i
t
ca
n c
o
n
c
l
ude t
h
at
A
N
F
IS m
odel
i
s
s
upe
ri
o
r
t
o
A
N
N
[
3
]
i
n
am
ou
nt
o
f
a
d
j
u
st
e
d
param
e
t
e
rs, scal
e of t
r
ai
ni
n
g
dat
a
, an
d t
e
st
i
ng e
r
r
o
r
.
Tr
ai
nin
g
erro
r satisfies th
e requ
iremen
ts. It is cle
a
r th
at
ANFIS is m
o
re effective su
bj
ect to
sm
all scal
e sam
p
le d
a
ta.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Ap
pl
i
c
at
i
o
n
of
AN
FIS
f
o
r
Di
st
ance
Rel
a
y
Pr
ot
ect
i
on i
n
Tra
n
smi
ssi
on
Li
n
e
(
A
zri
yenni
)
1
317
Tabl
e 1.
C
o
m
p
ari
s
o
n
of
t
h
e pr
op
ose
d
AN
FI
S
an
d AN
N
MET
HOD
ANN
ANFIS
I
nput L
a
y
e
r
20
2
Hidden/Rules
3
9
Output L
a
y
e
r
24
1
E
r
r
o
r of T
e
sting
0.
001
0
T
r
aining 48,
370
3,
000
4.
CO
NCL
USI
O
N
ANFIS h
a
s serv
ed in
t
h
is p
a
per in
tegrated by u
s
ing
Neural
Netwo
r
k
with
Fuzzy
In
fere
n
ce
Sy
stem
.
Whe
r
e
Fuzzy
syste
m
serves
as the
Fuzzy Infe
rence
Sy
st
em
. The
pr
o
pos
ed
A
N
F
I
S
i
s
a
new
sc
h
e
m
e
t
o
det
e
rm
i
n
e set
t
i
ng
of z
one
di
s
t
ance rel
a
y
s
. The zo
ne o
f
a d
i
st
ance pr
ot
ect
i
on sc
hem
e
was reco
gni
ze
d o
n
e o
f
cont
ributing c
a
uses of blac
kouts. There
f
ore, dete
rm
ining the accurate
zone setting
of
distance re
lay is
con
s
i
d
ere
d
nec
e
ssary
. T
h
e
ba
cku
p
pr
ot
ect
i
o
n ca
n be
co
ve
rage
by
usi
n
g
pr
o
pose
d
t
e
c
h
ni
q
u
e i
n
c
o
m
p
ari
s
o
n
wi
t
h
t
h
e c
o
n
v
e
nt
i
onal
t
e
c
hni
que
. T
h
e ap
pr
oach
schem
e
can be
use
d
i
n
n
o
rm
al
con
d
i
t
i
on a
nd
d
u
r
i
ng
on
p
r
o
cessfailu
res. Lastly, test resu
lts can h
e
l
p
o
n
e
of t
h
e in
tellig
en
t altern
ati
v
e techn
i
qu
es th
at h
a
v
e
v
e
ry
g
ood
perform
a
nce to im
prove t
h
e transm
ission line.
REFERE
NC
ES
[1]
Lin W. M,
et al
., “Adapt
ive
m
u
ltiple f
a
ult d
e
tection
and alarm
processing for loop s
y
st
em
with probabi
lis
ti
c
network,”
IEEE Transactions
on Power
Deliver
y
,
Vol. 19
, No. 1, p
p
. 64-69
, 2004
.
[2]
S
ouza J
.
C.
S
,
et al.
, “
F
ault
Loc
a
tion
in E
l
ec
tri
c
al Power S
y
st
e
m
s Using Intell
i
g
ent S
y
s
t
em
s T
echniqu
es,”
IE
EE
Transactions on
Power Deliver
y
,
Vol. 16
, No. 1, p
p
. 59-67
, 2001
.
[3]
Azriy
e
nni,
et al
., “Application of Backpropag
a
tion Neural Netw
ork for Fault Location in Trans
m
ission Line 150
kV,”
IJ
E
E
E
, Vol. 2
,
No. 4, pp. 21
-30, 2013
.
[4]
O
m
e
r
E
.
B
.
M
.
T
,
“
T
r
a
n
s
mission Line Faults Detecti
on,
C
l
assification and Locati
on using
Artificial Neur
al
Network,”
IEEE
, pp
. 1-5
,
2012
.
[5]
Chen J,
and Aggarwal R.
K,
“A
New Approach to EHV
Transm
ission Line Fault
Classific
a
tion
an
d Fault Dete
ct
io
n
Based on
the
W
a
v
ele
t
Tr
ansform
and Artif
ici
a
l
Int
e
llig
enc
e
,
”
I
EEE
, pp
. 1-8
,
2012
.
[6]
Yu W,
a
nd L
i
X,
“Fuz
zy
i
d
ent
i
f
i
c
ation using fuzzy
neur
al n
e
tworks
with stable l
earn
i
ng al
gorithm
s
,”
IE
EE
Transaction on
Fuzzy System
, V
o
l. 12
, No
. 3
,
pp
. 411-419, 2004.
[7]
Negnevitsk
y
M,
and Pavlovsk
y
V, “
N
eural
Ne
twork Approach
to Onlin
e Id
en
tific
at
ion of Mu
ltipl
e
Fai
l
ures o
f
P
r
otection
S
y
s
t
e
m
s
,
”
IEEE Transactions on
Pow
e
r Delivery
, Vol. 20, No. 2
,
pp
. 5
88-594, 2005
.
[8]
Zhang J. F,
et a
l
.
, “Morphological Undecim
a
t
e
d
Wavelet D
ecom
positi
on for Faul
t Location On P
o
wer Transm
ission
Lines,
”
I
EEE Transactions on C
i
rcuits and S
y
stems
, Vol. 53
, No
. 6, pp. 1395-140
2, 2006
.
[9]
Zang H, and Z
h
ao Y, “
I
ntell
i
g
ent Ident
i
fi
cat
i
on S
y
s
t
em
of Power Quality
Disturbance in Global Congrress
S
y
ste
m
s,
”
I
E
EE Computer
Society
, pp
. 258-2
61, 2
009.
[10]
Yusuff A.
A,
et al.
, “A Novel
Fault Features Extraction Scheme fo
r Power Transmission
Li
ne Fault Diagnosis,”
IEEE
Afri
con
, p
p
. 1-4
,
2011
.
[11]
Shira
z
i
S,
et a
l
.,
“
P
redi
ction
o
f
Failur
e
in
Pi
n-Joint Using
H
y
brid
Adaptiv
e Neuro-Fuz
z
y
Approach,
”
IEEE
International Co
nference on
Fuz
z
y Systems
, pp. 6
71-677, 2006
.
[12]
Nay
a
k C,
e
t
al
.,
“
P
rediction of Cutting and Fee
d
Forces for Co
nvention
a
l Milli
ng Process using Adaptive Neuro
Fuzzy
Inf
e
ren
c
e Sy
stem (ANFIS),”
IA
ES Interna
tional Journal o
f
Artifi
cia
l
Intell
igence (
I
J-AI)
,
Vol.
3,
No.
1,
pp.
24-35, 2011
.
[13]
Zhang M
,
“
A
n
On-Line Art
e
ri
al
Route
Tr
avel
T
i
m
e
Predict
i
on A
pplic
ation
Using ANFIS”,
IEEE
,
pp. 1-4
,
2009
.
[14]
Bahram
ifar A,
et
al
., “An ANFI
S
-based approach for Predicting
the Manning R
oughness Coefficien
t in Alluvial
Chanels
at
the
B
a
nk-F
u
ll S
t
age,
”
IJE Transaction
s
B: App
l
ication
, Vol. 26, No. 2
,
pp. 177-186
, 20
13.
[15]
Sarikay
a
N,
et
al.
,
“
A
daptive Neu
r
o-F
u
zz
y
Inf
e
ren
ce S
y
s
t
em
F
o
r T
h
e Computating
of The Character
i
stic Impedan
c
e
and Eff
e
ctive Perm
ittivit
y
of
The Micro-Copl
anar
Strip
Line,” E
l
ectrom
agnet
i
cs R
e
search
, pp
. 225
-237, 2008
.
[16]
Silva, A.
P. A,
et
al.
,
“
N
eural Ne
tworks for Fault
Loc
a
tion
in
Substations”
,
IEEE Transactions
on Power Delivery,
Vol. 11
, No. 1, 1
996.
[17]
Dash P.
K,
et al
.
,
“A Novel Fuzzy
Neur
al Networ
k
Ba
se
d Dista
n
ce
Re
lay
i
ng Scheme,”
IEEE Transactions on Pow
e
r
Deliv
ery
, Vol. 1
5
, No. 3, pp. 902
-907, 2000
.
[18]
Anis Ibrahim W. R,
et al
.,
“An Adaptive Fuzzy
Self-Learning Techniqu
e for Prediction o
f
Abnormal Operation
o
f
Abnormal Operation of Electrical Sy
stems”,
IEEE Transactions on
Power Delivery,
Vol. 21, No. 4, pp. 1770-1777,
2006.
[19]
Buckle
y J
.
J
,
Ha
yas
h
i
Y,
“
N
eural
Nets
for
F
u
zz
y
S
y
s
t
em
s
,
”
Fuzzy Sets and
Systems
, Vol. 71, pp. 2
65-276, 1995
.
[20]
Nauck D, Kruse R, “A Neuro-
Fuzzy
Method to Learn Fuzzy
Classification
R
u
les from Data,” Fuzzy
Sets an
d
S
y
stems, Vol. 8
9
, pp
. 277-288
,
1997.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJECE
Vol. 5, No. 6, D
ecem
ber
2015 :
1311 –
1318
1
318
[21]
Zhongjian
K, Aina T, Zh
e B
,
“A Fault Ar
ea
Lo
cation
Method in Distribu
tion Network With DG,”
Telkomnika
, V
o
l.
11, no
. 11
, pp
.
6
870-6878, 2013
.
[22]
Sheelav
ant V. R, Vijay
a
C, Shiral
ashetti S. C, “Wavelet Based Fau
lt Detectio
n Method for Ungrounded Power
S
y
s
t
em
with
Bal
a
nced
and
Unbal
a
nced
Lo
ad,
”
IJECE
, Vol. 1, No. 1
,
pp
. 2088-87
08, 2011
.
BIOGRAP
HI
ES OF
AUTH
ORS
Azriy
e
nni r
eceiv
ed the Bachelor
degree from Univ
ersitas BungHatta, Padang, Indo
nesia, in 1998
and th
e Master
Degree
form Univ
ersiti Teknolo
g
i Malay
s
ia, Joh
o
r
Bahru, in
200
7. Now, She is
doing Ph.D in Universiti Tekno
logi
Malay
s
ia, All in Electrical
En
gineer
ing. Sin
ce
1999, She was
Lecturer in Department Electr
ical, Universitas of Riau, Peka
nbaru, Indonesia. Her research
inter
e
sts include
Power
Sy
st
em
Protection
,
Tran
smission
& Dist
ribution S
y
ste
m
s a
nd Artific
ia
l
Intell
igen
ce
.
M.W Mustafa receiv
e
d his B.En
g degree (1988)
, M.
Sc (1993) an
d Ph.D (1997) from University
of S
t
rath
cl
yde
,
Glas
gow. He
is
a P
r
ofes
s
o
r in
F
a
cult
y
of E
l
e
c
tri
c
al
Engine
ering
,
UTM
.
He
is
als
o
current
l
y
Deput
y
Dean (Academ
i
c) in Facult
y
of
Electri
cal
En
gineer
ing, Univ
ersiti T
e
knologi
M
a
la
y
s
ia
. His
r
e
s
earch
int
e
res
t
includ
es
P
o
wer S
y
s
t
em
S
t
abil
it
y, Deregu
lat
e
d P
o
wer S
y
s
t
em
,
FACTS, Power Qualit
y,
Power
S
y
stem
Distribut
ion Autom
a
tion
and Artif
ici
a
l
Int
e
llig
enc
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.