In
te
r
n
ation
a
l Jou
rn
al
o
f Po
we
r
Elec
tron
ic
s an
d
D
r
ive S
y
stem
(IJ
PED
S
)
V
o
l.
11, N
o.
1, Mar
ch 20
20,
p
p.
505~
5
1
4
IS
S
N
: 2088-
86
94,
D
O
I
:
10.11
59
1
/ij
ped
s
.
v11
.
i
1.pp
5
05-
51
4
505
Jou
rn
a
l
h
o
me
pa
ge
:
ht
tp:
//i
j
p
eds.i
a
esco
re
.com
Improvement of protection rel
ay with a single phase auto-
reclosin
g mechanism based on
a
rtificial neural network
Z
o
zan
S
aad
al
l
a
h
Hu
ssain
1
,
A
h
med
J.
A
li
2
,
Ah
med
A
. Al
l
u
3
, Rak
an
Kh
a
lil An
t
ar
4
1,
3
Tech
nical
Ins
ti
tute, N
o
rt
hern
Tech
nical
Un
i
v
e
rsit
y
,
Iraq
2,
4
Techn
i
cal
Coll
e
ge,
Northern
Tech
ni
cal U
niv
e
rs
i
t
y, Iraq
Art
i
cl
e In
fo
ABSTRACT
A
r
tic
le hist
o
r
y
:
R
e
ce
i
v
e
d
Mar
26,
2
0
1
9
R
e
v
i
s
e
d
Jul
8
,
2
019
Ac
ce
p
t
ed
Oc
t
2
2
,
2
019
Th
is
p
ap
er
p
res
e
n
t
s
a
dev
e
loped
lo
gical
t
ripp
in
g
s
c
hem
e
t
o
i
m
pr
ov
e
con
v
en
tio
nal
p
r
ot
e
c
t
i
o
n
p
erf
o
rm
an
ce.
A
dap
tive
s
i
n
g
l
e
p
ole
aut
o
-
recl
osu
r
e
(ASPAR)
s
y
s
te
m
is
p
roposed
that
c
on
si
ders,
auto
m
a
ti
cally
t
ri
pp
i
ng
a
nd
reclosi
ng
o
f
a
m
ulti-shot
i
ndepen
dent
p
ole
techni
que
of
a
c
i
r
cu
it
breaker
a
t
a
pred
eterm
i
n
e
d
sequen
c
e,
w
h
i
ch
can
b
e
u
s
ed
t
o
boo
st
t
h
e
s
y
n
chro
n
i
zatio
n
of
th
e
p
o
wer
gri
d
u
n
d
er
t
he
t
rans
ien
t
f
ault
co
nd
itio
ns.
M
o
reo
v
er,
the
ASP
A
R
can
b
e
ut
ilized
t
o
enhance
t
h
e
elect
r
i
cal
s
ys
t
e
m
stability
a
nd
rel
i
a
b
ili
t
y
a
t
the
sam
e
o
p
e
ratin
g
con
d
i
t
io
ns
.
Ba
s
e
d
on
t
he
t
h
r
ee-phas
e
s
ys
tem,
t
he
A
rtificial
neu
r
al
n
et
wo
rk
(
AN
N)
i
n
t
h
is
w
o
r
k
h
a
s
been
d
o
n
e
in
o
rd
er
t
o
dia
gn
ose
a
n
d
det
ect
h
ealt
h
y
and
f
a
u
l
t
e
d
p
h
as
es.
The
pro
p
o
s
ed
A
N
N
-f
aul
t
c
las
s
i
f
ier
method
con
s
i
s
t
s
o
f
the
log
i
c
g
a
tes,
r
ou
te
r
circuit
s
,
timers,
a
nd
p
osi
t
ive
and
negative
seq
u
en
ce
an
aly
s
es
c
i
r
cui
t
.
In
a
dd
iti
on,
i
t
is
u
s
e
d
t
o
g
ive
t
h
e
abil
i
t
y
to
recog
n
i
z
e
a
f
a
ul
t
t
y
p
e
,
which
by
trai
n
i
ng
on
th
e
seq
u
en
ce
a
ng
le
v
al
ues
and
coo
r
di
nati
on
o
f
th
e
tran
sm
is
si
on
li
n
e. Three-p
h
ase ov
erh
e
ad t
ra
ns
m
i
ssion line
inc
l
ud
in
g
the
p
r
op
ose
d
A
S
P
AR
i
s
built
i
n
M
ATL
A
B\SIMU
LI
NK
env
i
ro
nm
ent.
T
hu
s
th
e
p
e
rf
orm
a
nce
AN
N-f
a
ul
t
clas
s
i
fi
ed
i
s
t
e
st
e
d
un
der
diff
erent
f
a
ult
con
d
i
t
i
ons.
Sim
u
lati
on
r
es
ults
s
ho
w
that
t
he
p
r
op
ose
d
A
S
P
AR
bas
e
d
on
A
NN
i
s
accurat
e
a
n
d
w
el
l
perf
orm
a
n
ce.
W
hereas
r
es
ul
ta
n
t
tr
ipp
i
ng
and
recl
os
in
g
sign
a
l
s
of
A
SP
AR
a
re
s
uccess
f
ully
p
rov
i
ded
that
e
nhan
ces
t
h
e
circu
it break
e
r
m
ech
ani
s
m u
n
d
e
r
th
e
s
e o
p
erati
ng con
d
i
t
io
n
.
K
eyw
ord
s
:
A
r
tificia
l ne
ural
n
etw
o
rk
Fa
ul
t
cla
ssi
fica
ti
o
n
Ma
tla
b
\
S
i
mul
i
nk pr
og.
Power
system
S
i
ng
le
pha
se a
ut
o-rec
l
osur
e
Th
is
is a
n
o
p
en acces
s a
r
ti
cle u
n
d
e
r t
h
e
CC
B
Y
-S
A
li
cens
e
.
Corres
pon
d
i
n
g
Au
th
or:
Zoza
n Sa
adalla
h
Hussai
n
,
Te
chn
i
cal
In
s
tit
ut
e,
N
o
rther
n
Tec
h
n
i
c
a
l
U
ni
ver
s
it
y,
Mos
u
l,
Iraq,
Em
ail:
zoza
n.h
u
ss
i
a
n
@
gma
i
l.
com
1.
I
N
TR
OD
U
C
TI
O
N
Det
e
c
t
i
o
n
an
d
cl
a
ssi
fi
ca
tio
n
o
f
f
a
ul
t
s
o
ccu
rred
a
t
t
ran
s
m
i
ss
io
n
l
i
n
es
a
re
c
onsi
d
er
ed
a
n
importa
n
t
fa
ct
or
i
n
c
o
rr
ectl
y
,
safe
ly
a
nd relia
ble
o
p
era
t
ion
o
f
prote
ct
i
o
n
re
la
ys.
Inte
l
lige
n
t
tec
h
n
o
l
o
g
i
es
h
a
d
b
ee
n
w
i
de
l
y
use
d
i
n
va
ri
o
u
s
a
rea
s
o
f
elec
t
r
i
c
a
l
p
ow
er
a
pp
lica
t
i
o
ns
t
ha
t
req
u
i
r
e
a
c
o
n
s
c
i
o
u
s
l
y
m
o
n
i
t
o
r
i
n
g
t
o
e
n
s
u
r
e
t
h
e
relia
bi
l
ity
o
f
th
e
syst
e
m
t
o
e
q
uip
c
o
nsum
er
s
w
i
t
h
u
n
i
n
t
err
u
p
t
e
d
p
ow
er
[
1].
D
i
ffer
ent
t
y
pe
s
of
t
hese
m
e
t
h
o
d
s
have
b
ee
n
ha
ndle
d
t
o
de
t
e
c
t
a
nd
ide
n
t
i
fy
t
h
e
t
ransie
n
t
a
n
d
pe
rm
a
n
ent
fa
ul
ts
t
ha
t
oc
c
u
r
beca
use
o
f
e
x
t
e
r
nal
in
flue
nce
s
s
uc
h
a
s
n
a
t
ura
l
d
is
aster
s
o
r
d
u
e
t
o
i
ncre
ase
l
o
ad
s
m
ore
tha
n
t
h
e
c
apac
ity
o
f
t
h
e
gene
ra
ti
o
n
s
ys
tem
.
A
f
u
zz
y
sy
st
e
m
s
t
e
ch
ni
q
u
e
w
a
s
u
se
d
t
o
d
et
ec
t
and
c
l
as
sify
f
au
lt
s
b
e
c
a
u
s
e
o
f
i
t
h
a
s
a
g
r
e
a
t
a
b
i
l
i
t
y
t
o
a
n
a
l
y
z
e
t
h
e
c
h
ang
e
s
t
h
a
t
o
cc
u
r
i
n
th
e
e
l
emen
t
s
o
f
t
h
e
po
wer
s
y
stem
a
t
t
he
i
ns
t
a
n
t
o
f
occ
u
rrin
g
f
ail
u
re
[
2-3]
.
H
o
w
e
ver
,
th
i
s
t
ec
h
n
iq
ue
s
u
ffe
rs
f
r
o
m
the
pr
o
b
le
m
o
f
s
e
l
ec
ti
ng
t
he
t
ype
o
f
m
e
m
b
ersh
ip
f
unc
t
i
ons
o
f
t
h
e
c
o
nt
r
o
ller
,
w
h
ic
h
ha
d
bee
n
s
ol
ve
d
b
y
u
s
i
n
g
i
n
t
e
l
l
i
ge
nt
n
eur
a
l
ne
tw
or
k
s
.
It
is
f
u
n
ct
ion
is
t
o
de
c
i
de
a
nd
c
h
oo
se
t
he
s
ha
pe
,
numbe
r,
a
nd
ty
pe
of
the
s
e f
u
n
c
tio
ns
[
4].
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2088-
8694
I
nt
J
P
ow
Elec
& Dr
i
S
y
st V
ol.
11,
N
o.
1
, Ma
r
202
0
:
505
–
51
4
50
6
Co
nv
en
ti
o
n
a
l
Ne
u
r
al
N
e
t
wo
rk
i
s
a
sui
t
abl
e
t
ool
u
se
d
in
t
he
a
pp
l
i
c
a
t
i
o
n
s
o
f
pow
er
s
ys
te
m
.
A
rtificia
l
Neu
r
al
N
et
wo
r
k
a
lgo
r
i
t
h
m
(
ANN)
i
s
ap
p
l
i
e
d
accu
rat
e
l
y
t
o
id
en
ti
f
y
an
d
c
l
a
s
sify
d
i
f
fere
nt
t
y
p
es
o
f
gri
d
f
aul
t
s.
A
n
a
lg
ori
t
hm
o
f
the
Mar
q
uardt
Le
ve
nbe
rg
t
ype
e
m
p
lo
ye
d
w
i
t
h
d
ou
b
l
e
c
ir
cui
t
s
truct
u
re
.
Three
-
phase
v
ol
ta
ges
an
d
six
-
li
n
e
c
urr
e
n
t
s
are
u
s
ed
a
s
i
nput
p
att
e
rn
s
f
o
r
ANN.
It
c
an
b
e
use
d
t
o
dete
c
t
a
n
d
c
la
ssi
f
y
al
l
l
i
ne-
t
o-
gro
u
n
d
f
a
u
lt
t
ype
s,
w
hi
c
h
o
ccur
r
ed
i
n
t
h
ree
p
h
ases
o
f
bo
th
t
ra
n
smiss
i
o
n
l
ine
c
i
r
c
u
i
ts
w
hic
h
r
e
s
u
l
t
i
n
g
i
n
brea
k
dow
n
t
h
e
ins
u
la
tor
[
5
-1
0].
The
m
u
tua
l
e
ffec
t
pro
b
le
m
that
oc
c
u
rs
i
n
do
u
b
l
e
c
ir
cui
t
pow
e
r
s
y
s
te
m
transm
i
s
s
i
on
l
i
ne
s
ca
use
s
i
na
ccur
a
te
ope
ra
ti
on
of
t
he
t
ra
di
tio
n
al
p
ro
t
ecti
o
n
d
e
vi
ce
.
C
N
N
h
a
s
be
en
t
ra
ine
d
t
o
overc
ome
th
is
p
rob
l
em
d
ep
end
i
ng
on
c
u
rre
nt
m
a
g
n
itu
d
e
s
at
s
en
di
n
g
end
to
d
e
t
ec
t
fa
ul
ts
[
1
1
-1
2].
To
overc
ome
th
is
p
ro
bl
e
m
a
n
d
d
e
t
e
c
t
the
fa
ul
t
t
ype
s,
it
i
s
i
mp
or
t
a
n
t
t
o
re
al
i
ze
t
h
e
fa
ult
re
sis
t
a
n
ce
a
nd
h
a
r
mo
n
i
cs
con
t
a
i
ne
d
i
n
t
he
fu
ndam
e
nt c
om
po
ne
nt o
f l
i
ne
c
urre
nts.
H
ave
a
d
eep
k
nowl
e
dg
e
a
bou
t
th
e
f
a
ul
t
re
si
st
a
n
c
e
a
nd
harm
on
ics
co
n
t
ai
ne
d
i
n
t
he
f
unda
me
n
t
c
omp
one
n
t
o
f
line
c
u
rre
nt
s
.
A
noth
e
r
neural
n
e
t
w
o
rk
s
truc
t
u
re
c
alle
d
a
radia
l
b
ase
d
f
u
n
ct
i
o
n
a
l
gor
ith
m
w
ith
t
hre
e
-p
ha
se
c
urre
nts
as
a
n
i
np
ut
w
a
s
i
nt
ro
du
c
e
d
to
g
iv
e
a
n
i
n
d
i
ca
ti
on
a
b
out
f
a
u
lt
l
o
c
a
t
ion
.
T
wo
s
epa
r
at
e
n
e
u
r
al
n
et
wo
rk
s
a
r
e
u
s
ed
,
t
h
e
firs
t
mode
l
for
l
i
n
e-grou
n
d
f
a
u
l
t
s
w
h
ile
t
h
e
other
f
o
r
p
h
ase angle fault
s
[
13].
A
sin
g
le-p
has
e
a
uto
re
cl
osu
r
e
S
P
A
R
b
a
se
d
on
a
d
a
pt
iv
e
t
e
c
hni
qu
e
use
d
t
o
di
st
i
ngui
sh
b
e
t
wee
n
re
q
u
i
r
ed
t
i
m
e
fo
r
ext
i
ngu
ish
i
n
g
t
h
e
arc
a
nd
n
a
tu
re
o
f
f
a
ult
th
a
t
o
c
curr
ed
on
t
h
e
tr
a
n
smiss
i
o
n
line
.
T
h
u
s,
t
o
ave
r
t
a
ny
ad
v
e
rse
effec
t
s
of
c
on
ve
n
tio
na
l
S
P
A
R
d
u
e
t
o
t
he
se
t
ra
nsi
e
n
t
s,
it
is
i
m
por
t
a
nt
t
o
i
m
pro
v
e
t
he
c
o
nv
en
tio
n
a
l
SP
AR
t
ec
hn
ique
[
14
-1
5
]
.
M
o
reo
v
er,
t
h
i
s
t
ech
niq
u
e
i
s
co
nv
eni
e
n
t
f
o
r
s
up
pl
y
cont
inu
i
t
y
o
f
a
transm
i
s
s
i
on
s
ys
tem
d
u
ri
ng
a
n
d
a
f
ter
a
tr
a
n
s
i
e
n
t
fa
ul
t
a
nd
s
i
nc
e
po
w
e
r
c
a
n
b
e
t
ransmi
tt
ed
t
h
r
o
ugh
t
he
rem
a
ini
n
g
hea
l
th
y
t
w
o
p
h
a
s
e
s
e
ve
n
dur
in
g
d
ead
time
,
t
here
by
i
nc
rea
s
i
n
g
of
t
ransm
i
ss
io
n
pow
er
c
ap
ac
ity
a
n
d
the
m
a
x
i
m
u
m
transm
issi
o
n
p
o
w
e
r
is
r
estric
ted
by
sys
t
e
m
s
t
a
b
i
l
i
t
y
.
A
l
s
o
,
v
a
r
i
o
u
s
m
e
t
h
o
d
s
h
a
v
e
b
e
e
n
prese
n
t
e
d
for
trans
i
en
t
sta
b
i
lit
y
i
m
pro
v
em
ent
of
t
he
r
e
c
e
nt
p
ow
e
r
s
ys
tem
s
[
16].
A
w
a
ve
l
e
t
transform
appr
oa
ch
h
a
s
b
een
u
se
d
to
d
i
s
c
r
imi
n
a
t
e
t
h
e
sta
b
i
lit
y
of
pow
er
s
ystem
e
nha
nce
d
i
n
bot
h
unc
ompe
nsa
t
e
d
a
n
d
com
p
en
sate
d
s
y
s
t
em
s
[1
7].
Als
o
w
ave
l
e
t
b
a
s
e
d
o
n
Clar
ke
’s
t
r
a
ns
f
orm
a
ti
o
n
i
s
use
d
t
o
o
b
t
a
i
n
the
fau
l
t
c
u
rrent
as a ne
w
a
lgor
ithm
for
fa
u
l
t
lo
c
a
ti
on a
n
d
cla
s
sifica
ti
o
n
[1
8
].
Wh
ile
in
[19]
, di
s
cr
ete wave
l
e
t
t
r
ansform
me
t
h
o
d
and
F
a
st
F
o
u
r
i
e
r
T
ransform
a
lg
ori
t
h
m
are
c
o
mpa
r
ed
t
o
de
te
ct
t
he
f
a
u
l
t
l
o
c
a
t
i
o
n
o
f
d
o
u
b
l
e
c
i
r
c
u
i
t
t
r
a
n
s
m
i
s
s
i
o
n
line
.
A
f
uzz
y
l
og
i
c
m
e
t
h
o
d
w
as
u
se
d
t
o
i
mp
rove
t
he
p
e
r
for
m
anc
e
o
f
t
rad
i
tio
na
l
a
u
t
o
r
e
c
l
o
s
ur
e.
I
t
is
u
s
e
d
to
detec
t
a
nd
c
l
as
sify
s
ym
m
e
trica
l
a
n
d
uns
y
m
m
e
trica
l
f
a
u
l
t
s
a
t
s
in
gl
e
and
d
oub
l
e
c
i
r
cu
it
t
ransmi
ssi
on
l
i
n
es.
Thi
s
t
y
p
e
o
f
in
t
e
ll
ig
en
t
cl
ass
i
fi
er
d
i
s
cri
m
in
at
es
f
a
u
lt
s
b
a
se
d
o
n
an
g
l
es
b
e
t
w
e
en
pos
iti
ve
a
nd
n
e
ga
t
i
ve
s
eque
nc
e
c
o
mp
on
en
ts [
20
].
I
n
t
h
i
s
pa
per
,
a
c
on
ven
t
i
o
na
l
a
u
t
o
-
r
ec
losur
e
w
ith
t
he
a
da
pt
i
v
e
S
P
A
R
s
yste
m
is
i
m
p
rove
d
depe
nd
in
g
o
n
ANN
t
o
e
n
h
a
n
ce
t
r
an
si
en
t
st
abil
i
t
y
.
T
h
e
A
NN-f
a
u
l
t
cl
assif
i
er
m
e
t
ho
d
tha
t
e
m
p
loye
d
for
line
fa
u
lt
is
implem
e
n
te
d t
o
be a
b
le t
o
re
c
o
g
n
i
ze d
i
f
f
er
en
t fau
l
t
t
ypes. In c
ase of t
he
l
i
n
e to gro
u
nd fa
u
l
t,
her
e t
h
e
re
s
u
lta
nt
tri
ppi
ng
sig
n
a
l
s
is
o
nl
y
as
sig
n
e
d
t
o
fau
l
ted
pha
se.
In
m
ea
nw
hil
e,
t
he
pow
e
r
f
l
o
w
s
i
n
o
t
h
e
r
t
w
o
hea
l
thy
pha
ses
u
n
d
e
r
t
h
e
t
r
an
si
ent
fau
lt
c
ondi
tio
n
s.
C
on
s
i
d
e
ri
ng
t
h
a
t
,
i
t
is
a
p
o
t
en
t
i
al
i
ssu
e
t
o
im
pro
v
e
t
h
e
tra
n
s
i
e
n
t
s
t
a
b
i
l
i
t
y
of
t
he
p
ow
er
s
ys
t
e
m.
T
hat
c
a
n
be
a
c
h
i
e
ved
by
a
v
o
i
di
n
g
u
n
n
e
c
e
s
sa
ry
t
r
i
pp
i
ng
sig
n
a
l
s
to
t
he
h
ea
l
t
h
y
pha
se
c
i
rcui
t
b
r
ea
k
e
r
.
T
h
e
p
ropo
s
e
d
al
go
rit
h
m
i
s
s
i
m
ul
a
t
ed
a
nd
t
est
e
d
w
i
t
h
t
h
r
ee
-p
h
a
s
e
e
l
ect
ri
c
a
l
po
we
r
sy
st
em
us
i
n
g
MA
TLA
B/S
i
m
u
l
i
nk.
2.
POWER
SY
STEM
S
IMUL
ATIO
N
The
p
o
w
e
r
sys
t
em
i
s
des
i
g
n
e
d
,
simula
te
d
an
d
mode
l
e
d
by Ma
tla
b
\
S
imul
i
nk
to
e
s
t
i
m
ate
fa
u
lte
d
li
nes,
and
pro
t
ec
t tr
a
n
smiss
i
on
line
thr
o
u
g
h
tri
p
/
au
to-re
cl
os
i
n
g
t
h
e
f
a
u
lt base
d
on
ANN t
echn
i
qu
e.
T
h
e
s
t
u
dy
sy
s
t
e
m
con
s
is
t
s
o
f
a
(5
00
kV
)
ove
r
h
e
a
d
t
ra
nsm
i
ssio
n
l
ine
a
s
a
s
i
n
g
l
e
c
irc
u
i
t
.
The
para
me
t
e
rs
o
f
l
i
ne
s
pec
i
m
e
ns
(
R,
L
,
and
C
per
km
)
ar
e
defi
ned
in
pos
i
tive
a
n
d
z
e
r
o-se
que
nce
c
o
mpone
n
t
s.
A
S
ync
hro
n
ous
g
e
n
e
r
at
or
i
s
c
o
n
n
e
cted
to
t
he
t
r
a
nsmis
s
i
on
gr
id
t
hrou
gh
a
13.
8
kV
/5
00
kV
∆
Y
⁄
t
ransf
o
r
m
er
a
s
illus
t
ra
t
e
d
in
F
ig
ure
1.
I
n
order
to
t
e
s
t
d
i
f
f
e
r
e
n
t
f
a
u
l
t
s
a
t
t
h
e
p
o
w
e
r
s
y
s
t
e
m
,
a
3
-
p
h
a
s
e
f
a
u
l
t
b
l
o
c
k
is
a
ll
i
e
d
a
t
t
he
t
ra
nsm
i
s
s
io
n
l
i
ne
t
o
s
i
m
u
la
te
f
a
u
l
t
s
al
ong
t
he l
i
n
e
.
D
i
ffere
n
t
fau
l
ts
s
uch
as
pha
se-to
gr
ou
n
d
(
A
G
)
,
pha
se-to
-
pha
se
(
AB
),
p
h
a
se
-t
o
-
p
h
a
se
t
o
g
r
ound
(ABG)
and
three
phases-to-ground
fa
u
lts
(
A
B
CG
)
are
tested.
Li
ne
v
olta
ge
s
an
d
cur
r
en
ts
s
i
gna
ls
a
re
u
se
d
as
in
put
s
of
t
he
p
r
o
t
e
ct
i
on
s
y
s
t
em
,
w
h
ich
co
ns
ists
o
f
defin
ite
t
im
e
o
v
er
c
ur
rent
r
ela
y
(
D
T
O
C
R),
pos
i
tive
,
nega
tive
a
n
d
zer
o
seque
nce
c
a
l
c
u
l
a
t
i
o
n
,
fau
l
t
s
c
las
s
i
f
ier
us
i
n
g
ANN,
a
n
d
ASPAR
sy
st
em
o
f
t
h
e
mult
i
-
sh
o
t
in
de
pen
d
e
n
t
po
le
m
echa
n
i
s
m
bl
oc
ks
a
s show
n in
F
ig
ure
1.
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
&
D
ri S
yst
IS
S
N
:
2088-
86
94
I
m
prov
em
e
n
t
o
f
pr
ot
e
c
t
io
n
rel
a
y
wit
h
a si
ng
l
e
pha
se
au
t
o
-reclo
s
i
n
g … (Z
ozan S
a
a
d
a
lla
h
H
u
ssa
i
n)
50
7
F
i
gure
1
.
Elec
trical
p
ow
er
s
y
s
t
e
m
mode
l
I
n
t
h
i
s
wor
k
,
a
cha
r
ac
t
e
ris
tic
o
f
t
h
e
co
nve
n
t
io
nal
defi
ni
t
e
t
i
m
e
over-
current
r
ela
y
i
s
u
til
i
z
e
d
t
o
de
te
c
t
the
fa
u
lt occ
u
rrenc
es. The
co
n
v
en
t
i
o
n
a
l
fa
u
lt
detec
t
i
on ba
se
d o
n
th
is r
ela
y
gi
v
e
s
t
ri
ppi
n
g
a
nd
re
c
l
o
sin
g
s
i
g
na
l
s
to
a
ll
p
o
l
es
o
f
t
h
e
c
i
r
c
u
it
bre
a
k
e
r
.
That
i
s
on
ly
ach
ie
ve
d
a
p
ar
t
from
s
e
l
ec
ti
ng
a
pole
o
f
t
h
e
c
i
r
cu
i
t
b
re
ak
er
t
h
a
t
bel
o
ng
s
to
h
e
a
lth
y
or
f
au
l
t
e
d
pha
ses.
H
o
w
ever
,
the
pro
p
o
se
d
A
S
P
A
R
b
a
s
e
d
o
n
A
N
N
i
s
e
m
p
l
o
y
e
d
t
o
s
p
e
c
i
f
y
the
tr
ip
lin
g
s
i
ngle
req
u
ired
f
o
r
a
c
urta
in
p
ol
e
of
t
h
e
c
irc
u
it
bre
a
ker
.
T
h
i
s
can
a
l
l
ow
i
so
l
a
tin
g
on
l
y
t
he
f
aul
t
e
d
pha
ses.
O
ther
w
i
se
,
un
nec
e
ss
ary
tri
p
li
ng
si
n
g
le
s
are
bl
oc
ke
d
fr
o
m
se
ndi
ng
t
o
o
t
h
er
c
i
r
c
u
it
b
re
a
k
er
pol
es
o
f
t
h
e
h
e
al
th
y
ph
as
es.
3.
ARTIFI
C
IAL N
E
URAL
N
E
T
WO
R
K
S
TRUCTURE
A
N
N
i
s
c
onsi
d
er
s
o
n
e
o
f
t
h
e
c
omm
on
me
tho
d
s
u
s
ed
t
o
detec
t
a
nd
d
i
s
ti
ngu
i
s
h
di
ff
ere
n
t
ty
p
e
s
of
fa
ul
ts
o
c
c
u
rr
ed
on
pow
e
r
s
y
s
t
e
m
.
T
his
due
c
hara
cter
i
s
t
i
c
o
f
A
N
N
w
h
e
re
t
h
e
y
do
not
n
eed
a
w
i
d
e
i
n
fo
rmat
i
on
base
a
b
o
u
t
f
a
u
lts
c
omp
a
re
w
ith
c
lass
ifica
t
i
o
n
m
e
th
od
s.
A
N
N
c
on
s
is
ts
o
f
m
a
ny
la
ye
rs
a
nd
e
ach
o
f
the
m
c
o
nt
ai
ns
s
e
v
eral
n
eu
ron
s
, wh
i
c
h
a
re
i
nt
erc
onn
ec
t
e
d
t
o
e
a
c
h
o
t
h
e
r.
A
syst
e
m st
r
u
ct
ure
ch
oos
i
ng de
pe
nds on t
h
e
ca
t
e
g
o
riza
t
i
o
n
p
ro
blem
t
ha
t
c
o
n
t
a
i
n
s
f
a
u
lt
s
a
m
ple
s
[
2
1
].
D
etect
i
o
n
o
f
f
a
u
l
t
s
i
s
n
o
n
l
i
n
e
a
r
p
r
o
c
e
s
s
,
t
h
e
r
e
f
o
r
e
i
t
requ
ire
to
a
non
l
i
nea
r
s
o
l
ve
r
like
A
N
N
to
h
a
n
d
l
e
i
t
.
AN
N
has
t
h
e
a
b
il
it
y
t
o
d
e
a
l
w
it
h
a
n
y
in
co
m
p
lete
ma
them
at
i
c
al
p
r
oble
m
s
l
i
ke
m
issed
or
c
orr
u
p
t
e
d
d
a
t
a.
I
n
ad
d
iti
o
n,
it
c
a
n
s
o
lve
a
n
y
m
a
the
m
a
tica
l
m
att
e
r
w
itho
u
t
r
e
quir
e
me
nts
f
o
r
ap
p
r
ox
ima
tio
n
t
h
e
m
odel.
A
ll
of
m
e
n
t
i
o
ned
a
b
ove
m
a
k
e
A
N
N
i
s
the
preferre
d
w
a
y
to
t
re
at
t
h
e
f
a
u
l
t
s
of
t
ransm
i
ss
io
n
li
ne
o
v
e
r
other
m
e
t
h
ods.
A
l
s
o
,
it
di
sp
la
ys
a
g
o
od
fau
l
t
l
o
ca
t
i
on
a
n
d
c
l
a
s
si
fi
c
a
t
i
on
i
n
c
a
s
e
of
e
x
i
st
en
c
e
r
esi
s
t
a
n
ce
o
f
t
h
e
f
au
lt
s
an
d
var
i
a
t
ion
of
p
ow
e
r
s
ys
te
m
fac
t
ors.
T
he
refore
ANN
i
s
w
i
d
ely
u
s
ed
i
n
po
wer
sy
st
em
a
p
p
li
cat
ion
s
[
22
-23
]
.
Th
en
i
t
i
s
d
e
s
igne
d
to
d
e
t
ect
a
nd
c
l
ass
i
fy
t
h
e
f
a
u
lt
depe
n
d
i
n
g on
the
fol
l
ow
in
g st
eps [24]
:
a)
Trai
ni
ng
th
e
ANN sy
st
em,
d
e
p
e
n
d
i
n
g
on
t
h
e
p
ro
p
e
r d
a
t
a
.
b)
Ch
oi
ce th
e
s
u
i
tab
l
e
ANN
s
t
ru
ct
u
r
e
f
o
r
a g
i
v
e
n
i
m
p
l
emen
t
a
tio
n
.
c)
Trai
ni
ng
th
e
ANN.
d)
A
ppraisa
l
o
f
th
e
t
raine
d
A
N
N
us
i
n
g
tes
t pa
t
t
e
r
ns un
t
i
l
its
acc
o
mp
li
sh
me
nt
i
s c
o
g
e
nt
.
3.1.
T
r
a
i
n
i
ng
o
f
ANN
B
a
sed Fa
u
l
t
Cl
a
s
si
fi
er
Scheme
The
a
l
g
o
r
ithm
of
B
ac
k
P
r
opa
ga
ti
on
A
l
g
o
r
i
t
h
m
(
B
P
A
)
i
s
c
e
n
tra
l
t
o
m
u
c
h
c
u
rre
nt
w
ork
o
n
l
e
arn
i
ng
i
n
neura
l
n
etw
o
rk
s.
T
his
lea
r
ni
n
g
r
u
l
e,
a
ls
o
k
n
o
w
n
a
s
a
G
e
ne
r
a
lize
d
D
e
l
t
a
R
u
l
e
,
the
n
i
t
g
i
ve
s
a
descri
p
t
io
n
for
vary
in
g
the
w
e
ig
h
t
s
in
a
ny
fee
d
-forw
a
rd
s
ys
tem
.
I
n
BP
A
c
a
se,
a
super
v
i
s
e
d
l
e
a
rni
n
g
is
u
se
d
a
s
t
he
n
et
w
o
rk
w
ill
be
t
ra
ine
d
u
s
i
n
g
t
he
d
a
t
a
c
r
ea
ted
from
t
he
s
im
u
l
a
t
i
o
n
m
o
de
l
of
t
ra
nsm
i
ss
ion.
V
a
r
io
us
n
e
u
ral
ne
tw
ork
arc
h
i
t
e
c
t
u
re,
train
i
ng
a
l
g
o
ri
th
m
s
a
nd
trans
f
e
r
f
unc
t
i
o
n
s
w
e
r
e
s
tud
i
e
d
t
o
d
e
cide
u
p
o
n
t
he
f
in
al
n
eur
a
l
n
e
tw
ork
mode
l
f
o
r
f
a
u
l
t
de
te
c
tio
n
s
y
s
t
e
m
[
2
5
].
A
lso
i
t
h
as
b
ee
n
pr
ove
n
i
s
t
h
e
p
re
ferred
l
e
arni
ng
m
e
t
h
od
amo
ng
ot
he
rs
due
t
o
it
i
s
si
m
p
le
s
truc
ture
,
ve
ry
f
as
t
N
N
t
ra
ini
n
g
pr
o
cess
a
n
d
ea
s
y
t
o
imp
l
em
en
t
i
n
d
i
f
fere
n
t
l
ear
ning
app
l
ica
t
i
o
ns
l
i
k
e
spee
c
h
r
e
c
og
n
iti
on,
d
e
t
ec
ti
o
n
a
n
d
c
las
s
i
f
ica
t
io
n
o
f
f
a
u
l
t
s,
i
m
a
ge
p
r
o
cessi
ng
a
nd
a
d
apt
i
v
e
control [26
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N: 2
0
8
8
-
86
94
I
nt
J
P
o
w
E
l
e
c
&
D
r
i
S
yst
V
o
l.
11,
N
o.
1
,
Mar
202
0
:
505
–
51
4
50
8
The
ne
ur
al
n
et
s
use
d
i
n
th
i
s
s
tu
dy
a
r
e
tr
a
i
ne
d
b
y
t
he
B
P
A
p
r
o
po
se
d
b
y
R
u
m
e
l
ha
r
t
.
A
t
t
he
b
e
g
i
n
nin
g
of
t
he
t
r
a
in
ing
pr
oc
ess
the
ge
ner
a
te
d
w
e
igh
t
s
ar
e
se
lec
t
e
d
r
and
om
ly
a
nd
t
h
e
ou
t
p
u
t
s
i
g
n
a
l
is
c
a
l
c
u
l
a
t
e
d
usin
g
e
qua
t
i
o
n
1
a
s
show
n
bel
o
w
.
A
t
the
e
n
d
of
t
h
i
s
sa
ge,
the
e
r
r
o
r
s
igna
l
(
d
if
f
e
r
e
nce
be
tw
ee
n
ou
t
p
u
t
a
n
d
t
a
r
g
e
t
si
gna
ls)
f
o
r
a
l
l
iter
a
t
i
on
is
t
a
k
en
i
n
t
o
co
nsi
d
er
a
t
i
o
n
w
h
er
e
it
s
en
t
bac
k
w
a
r
d
t
o
t
h
e
w
e
igh
t
s,
a
nd
the
le
ar
ni
ng
pr
o
c
ess
w
i
l
l
c
on
t
i
n
u
e
u
n
t
i
l
t
he
e
r
r
o
r
value
e
qua
l
t
o
a
c
c
e
p
ta
ble
v
a
l
u
e
d
e
p
e
n
d
i
n
g
o
n
t
h
e
a
c
t
u
a
l
p
r
o
b
l
e
m
.
T
h
e
da
t
a
r
ece
i
v
e
d
f
r
o
m
outer
e
n
v
ir
onme
n
t
(
o
r
ot
her
neur
o
n
s)
a
r
e
t
r
a
nsf
e
r
r
ed
t
o
the
ne
ur
o
n
s
in
t
he
h
idde
n
lay
e
r
(
s
)
thr
o
u
g
h
w
eigh
ts
w
hic
h
adj
ust
us
i
n
g the sum
m
a
tio
n
fu
nc
t
i
o
n
and
t
he
n use
s
an
a
c
tivat
i
on
fu
nc
t
i
o
n
;
i
t
i
s use
d
t
o
ex
p
r
ess
th
e
ANN,
w
h
i
ch
c
on
sist
o
f
b
a
s
i
call
y
t
h
e
i
npu
t
d
a
ta
t
o
t
he
n
e
t
w
o
r
k
,
som
e
h
idde
n
la
yer
s
a
nd
a
n
o
u
t
p
u
t
laye
r
a
r
e
conn
e
c
ted
t
o
f
or
m
it
as
s
ho
w
n
i
n
F
i
gur
e
2.
[
13]
.
The
we
i
g
h
t
s
of
n
eur
o
n
ar
e
(
W
,
W
,
..
,
W
)
ar
e
use
d
t
o
ca
l
c
u
l
a
t
e
the
va
lue
o
f
i
n
p
u
t
pa
r
a
m
e
te
r
s
X
,
w
her
e
X
=(
X
,
X
,
…,
X
)
.
T
h
e
n
t
h
e
o
u
t
p
u
t
(
y
)
=
(
Y
,
Y
,
…,
Y
)
of
e
ac
h
ne
ur
o
n
i
s
ca
l
c
u
l
a
t
ed
b
y
e
q
u
a
t
i
o
n
1
.
The
y
a
r
e
t
r
a
ine
d
in
o
rder
t
o
implem
en
t
a
specific
m
i
ssio
n
by
ad
j
u
sti
n
g
b
o
t
h
w
e
i
gh
t
and
ba
se
o
f
the
ne
ur
o
n
s
f
o
r
a
l
l
l
a
yer
s
[
2
3
]
.
F
i
gur
e
2.
P
e
r
cept
r
on
r
e
p
r
e
sen
t
ati
o
n
∑
.
(1
)
Whe
r
e
(
b
)
kno
w
n
a
s
a
t
h
r
e
sh
ol
d
or
b
ia
s
va
lue
,
w
he
r
e
t
he
n
e
u
r
ons
p
r
odu
c
e
a
n
ou
t
p
ut
i
f
y
is
g
r
e
a
t
e
r
tha
n
z
e
r
o.
I
f
the
num
b
e
r
o
f
n
e
u
r
ons
a
n
d
h
i
dde
n
la
ye
r
s
i
n
c
r
ease
,
t
h
e
d
etec
t
i
o
n
and
cl
assi
fi
c
a
ti
on
f
eatu
r
e
s
o
f
ANN
w
ill
rise,
and
the
n
a
r
e
u
pda
t
ed
c
ontinuously
t
hrough
tr
aini
ng
pr
oce
s
s
of
t
he
n
e
u
r
a
l
ne
tw
or
k.
S
e
v
er
a
l
pr
o
c
e
dur
e
s
(
a
l
gor
i
t
hm
s)
h
a
d
b
ee
n
e
s
ta
bl
is
he
d
to
t
r
a
in
n
e
u
r
a
l
ne
t
w
or
k
ac
co
r
d
i
ng
to
t
ype
o
f
pr
ob
lem
[2
4]
.
I
n
c
ur
r
e
nt
w
or
k,
t
he
d
etec
t
i
o
n
a
nd
clas
sif
i
cati
o
n
of
l
i
n
e
-
gr
o
u
n
d
fau
l
ts
i
s
im
ple
m
e
n
t
e
d
by
usi
n
g
ANN-f
a
u
lt
classifier
i
n
a
5
00
k
V
t
ran
s
missio
n
li
n
e
.
A
feed-
f
o
r
w
a
r
d
w
i
t
h
back pr
o
pa
ga
ti
on l
e
ar
ni
ng
a
l
g
o
r
i
t
h
m
ha
d
be
en
u
sed
as
a
f
ault
det
ector
a
n
d
c
las
s
ifier.
A
s
ill
ustr
ate
i
n
Figu
re
3
,
t
h
e
ANN
s
c
h
e
me
d
ep
en
ded
on
i
n
p
u
t
s
a
n
g
l
e
s
a
r
e
(
An
g_A
,
A
ng
_B
,
An
g
_
C
)
r
e
lati
ve
t
o
pha
se
(
A
,
B
,
C)
r
espec
t
i
v
e
l
y,
w
hi
c
h
a
r
e
c
a
l
cu
la
te
d
fr
om
3-
pha
se
s
e
que
nce
a
n
al
yze
r
c
i
r
c
u
it.
T
hese
a
n
g
le
s
ba
sed
o
n
z
e
r
o,
pos
i
tive
,
a
nd
n
e
g
a
t
i
v
e
s
e
que
nce
c
o
mp
one
nts
a
r
e
ca
l
c
ul
a
t
ed
u
s
i
n
g
t
he
f
o
llo
w
i
n
g
e
qua
t
i
o
n
s
[26]
.
c
b
a
a
I
a
I
a
I
I
2
1
.
(2
)
c
b
a
a
I
a
I
a
I
I
.
.
2
2
(
3
)
2
1
a
a
I
I
A
Ang
l
e
(4)
2
1
2
.
a
a
I
a
I
a
B
Angle
(5)
2
2
1
..
.
a
a
I
a
I
a
C
A
ngl
e
(6)
The
r
e
su
l
t
s
o
f
t
he
se
r
ela
tio
ns
a
r
e
s
how
n
in
t
he
i
n
Ta
ble
1.
T
he
s
y
m
bo
l
(a)
is
(
1
∠
12
0°
)
and
(
an
d
)
r
e
pr
e
s
ent
the
pos
iti
ve
a
nd
ne
ga
ti
ve
s
eque
nc
e
c
o
mp
o
n
e
n
t
s
o
f
t
h
e
cu
rren
t
s
after
fau
l
t
ref
e
r
to
p
hase
“a
”
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
&
D
ri S
yst
IS
S
N
:
2088-
86
94
I
m
prov
em
e
n
t
o
f
pr
ot
e
c
t
io
n
rel
a
y
wit
h
a si
ng
l
e
pha
se
au
t
o
-reclo
s
i
n
g … (Z
ozan S
a
a
d
a
lla
h
H
u
ssa
i
n)
50
9
Ta
b
l
e
.
T
yp
i
c
al
v
a
l
ue
s o
f
A
N
N
a
ngle
s
duri
n
g de
t
e
c
t
i
on an
d
rec
l
a
i
m
tim
e
at di
f
fere
n
t
f
a
u
lt t
ype
s
F
a
u
l
t
T
y
p
e
A
N
N
An
g
l
es
D
u
r
in
g
D
e
t
e
c
tio
n
Ti
m
e
ANN
An
g
l
e
s
D
u
r
in
g
R
ecl
ai
m
Ti
m
e
A
ngle
(A
)
Angle
(
B
)
A
ngle
(C
)
A
ngle
(
A
)
A
ngle
(B
)
A
ngle
(
C
)
a
-
g
38
1
58
8
2
121
1
119
b-
g
82
3
8
158
119
121
1
c
-
g
158
8
2
38
1
119
121
a
-
b
40
8
0
160
120
120
0
a
-
c
80
1
60
4
0
120
0
120
b-
c
160
4
0
80
0
120
120
a
-
b-
c
67
5
3
173
2
0
140
100
He
a
r
lt
y
31
1
51
8
9
26
146
9
4
4.
PRINCIPLE
OF THE PROPOSED N
E
W LOG
I
C
ASP
A
R WIT
H
M
ULT
I
-S
H
O
T S
C
H
E
M
E
Ma
jori
t
y
o
f
th
e
fa
ul
ts
t
hat
o
c
c
u
r
o
n
t
he
o
v
e
rhead
t
ra
nsm
i
ssion
l
i
ne
a
re
t
ra
ns
i
e
n
t
i
n
na
ture.
The
s
e
fa
ul
ts
a
r
e
c
a
u
s
e
d
b
y
t
he
b
re
akd
o
w
n
o
f
a
i
r
surrou
n
d
i
ng
t
h
e
ins
u
l
a
t
or
d
ue
t
o
a
b
norm
a
l
tr
a
n
sie
n
t
o
v
e
r-v
ol
ta
ges
or
by
pa
ssin
g
o
f
o
b
j
ec
t
s
n
ea
r
or
t
hrou
gh
li
nes
(bird
s
,
vi
n
e
s,
t
re
e
bra
n
ches
e
tc
.).
These
situa
t
i
o
ns
r
esul
t
i
n
arc
i
n
g
f
au
l
t
s
can
b
e
ex
t
i
n
g
u
ishe
d
by
e
n
er
giz
i
n
g
s
im
u
lta
ne
ous
o
p
e
n
ing
o
f
c
irc
u
it
br
eake
r
(
C.B)
on
bot
h
e
n
d
s
of
t
he
line
or
o
n
o
n
e
e
n
d
o
f
t
he
l
i
n
e.
S
ince
t
h
e
c
a
use
of
t
rans
ien
t
f
a
u
lts
d
isa
ppe
ars
after
a
short
t
i
m
e
,
t
h
e
C.
B
ca
n
be
r
e
c
l
o
se
d
the
m
o
me
nt
t
he
a
rc
i
n
fa
u
l
t
ha
s
bee
n
e
x
t
i
n
g
u
i
sh
e
d
a
n
d
t
he
p
at
h
ha
s
re
gai
n
e
d
i
ts
d
ie
le
ctric
st
r
e
n
g
t
h
[
1
4
].
H
e
n
ce,
a
ut
o-re
clo
s
in
g
is
a
tt
e
m
pted
f
o
r
t
he
pur
po
se
o
f
rest
ori
ng
the
tra
n
s
m
ission
l
i
ne
t
o
se
rvic
e
and
sui
t
a
b
le
i
n
impro
v
i
n
g the
con
t
in
uit
y
o
f servic
e.
F
ur
t
h
e
r
be
ne
fits o
f
au
t
o
-re
clos
ing
are
[27]
:
Im
prove
me
nt
s
in
t
r
a
nsie
n
t
sta
bi
lit
y.
Im
prove
me
nt
s
in
s
ystem
r
e
lia
bi
lit
y.
M
i
n
i
m
i
z
i
n
g
o
f
swit
c
h
es o
ver
vo
l
t
age.
M
i
n
i
m
i
z
i
n
g
o
f
shaft t
o
rsi
ona
l vi
bra
t
i
on
of bu
l
k
y
t
herm
al
u
n
its.
M
i
n
i
m
i
ze
d
uns
ucce
ssfu
l
r
ec
losi
ng us
i
ng va
r
i
a
b
le de
a
d
t
im
e.
Whe
n
t
hree
-p
h
a
se
a
uto-r
ecl
os
in
g
is
a
pp
l
i
ed
t
o
si
n
g
le
c
irc
u
i
t
i
n
t
e
rcon
ne
c
t
i
o
ns
b
e
t
w
e
en
t
w
o
pow
er
systems,
t
he
t
hree
phases
are
tr
i
p
p
e
d
w
h
et
he
r
sing
le
-p
hase
f
au
l
t
or
m
ul
t
i
-
pha
se
f
a
u
l
t
a
re
o
cc
ur
ring
t
h
is
m
ay
ca
use
the
tw
o
syste
m
s
t
o
d
ri
ft
a
pa
r
t
i
n
pha
s
e
.
I
f
onl
y
the
fa
ul
t
y
p
hase
t
r
i
ppe
d
,
s
y
n
c
h
ro
niz
i
ng
p
o
w
e
r
can
s
ti
ll
be
i
nte
r
cha
nge
d
thr
o
u
g
h
t
he
h
e
a
lth
y
p
h
ases,
a
l
s
o
s
hor
ter
de
a
d
t
ime
c
a
n
be
s
e
t
i
n
th
is
m
od
e
w
h
en
c
om
pa
re
d
t
o
the s
i
n
g
l
e pha
se
a
ut
o-
rec
l
ose
. The
ref
o
re,
it is not ne
c
e
ssar
y
t
o i
n
terr
upt
s
ervice
s
on t
h
e
o
t
he
r tw
o
phase
s
[27].
I
n
t
his
rese
arc
h
A
S
P
A
R
m
etho
d
has
bee
n
p
r
o
p
o
se
d
to
a
utom
atic
a
l
l
y
op
erat
e
th
e
C
.
B
acco
rdi
n
g
t
o
a
prede
t
erm
i
ne
d se
que
nce
of o
p
e
n a
n
d cl
ose o
p
era
t
i
o
ns
; w
h
i
c
h co
ns
ist
of t
h
e
3-pha
se
s a
u
t
o
r
ec
l
o
s
i
ng. Ea
c
h u
n
it
of
t
he
a
u
t
o-re
clo
s
i
n
g
circ
u
i
t
opera
te
s
i
n
de
pe
nde
ntl
y
o
n
the
o
t
h
er
t
w
o
phase
s
for
trip
p
i
n
g
a
nd
re
clo
s
in
g
me
cha
n
ism
a
s
show
n in F
igur
e 4,
w
hich
i
s
th
e sim
p
lifie
d
schem
e
log
i
c
for
the
sin
g
l
e-po
le
a
uto-re
cl
os
ure.
Th
e
l
o
g
i
cal
el
emen
t
s
i
n
c
lu
d
e
S
-R
f
l
i
p
-fl
op
s,
AND,
N
A
N
D
,
N
OR
,
Ti
me
d
ela
y
,
dea
d
t
im
e
bloc
k
for
ea
ch
s
ho
t,
a
n
d
c
o
u
n
t
er
f
or
t
he
t
hre
e
-sh
o
t
aut
o
-re
clo
s
ure
after
t
h
at
d
e
s
i
g
n
re
se
t
s
i
f
re
cl
osu
r
e
i
s
s
ucc
e
s
s
fu
l
w
ithi
n
t
he
c
ho
sen
num
ber
of
s
ho
ts.
When
a
f
a
u
lt
occ
u
r
s
o
n
a
t
r
a
nsm
i
ss
io
n
li
ne,
the
s
i
ng
l
e
p
o
l
e
au
t
o
-re
clo
s
ure
at
e
a
c
h
en
d
o
p
e
n
t
o
se
pa
rat
e
t
he
f
a
u
lte
d
l
i
ne,
re
ma
in
o
p
e
n
f
o
r
a
spec
ified
t
i
me
(
dela
y
t
i
m
e
)
)
t
h
e
n
.
If
t
h
e
transm
i
s
s
i
on
l
i
n
e
fau
l
t
ha
s
c
l
e
a
re
d,
t
he
n
t
h
e A
S
P
A
R
re
ma
ins
c
l
o
sed
an
d the
tra
n
sm
issi
o
n
s
ys
tem
retur
n
s
to
its
F
i
g
u
r
e
3
.
S
i
mu
l
a
tio
n
sch
e
me
o
f
ANN-b
a
sed
fault
cl
assi
fi
er
u
ti
l
i
zing
MATLAB/
SIMULI
NK p
r
o
g
r
am
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2088-
8694
I
nt
J
P
ow
Elec
& Dr
i
S
y
st V
ol.
11,
N
o.
1
, Ma
r
202
0
:
505
–
51
4
51
0
pre-
fau
lt
c
o
n
d
iti
o
n
.
If
the
fa
u
l
t
st
ill
e
x
ist
s
,
t
h
e
n
i
n
th
is
c
a
se
t
he
A
S
P
A
R
is
d
e
s
ig
ne
d
t
o
h
a
v
e
u
p
t
o
t
h
re
e
ope
n-
clo
s
e
o
p
e
r
ati
o
ns
a
n
d
a
fter
t
h
e
se
a
f
ina
l
o
p
e
n
o
p
er
at
i
o
n
to
l
oc
k
o
u
t
t
he
a
ll
thre
e
pole
s
o
f
th
e
circu
it
bre
a
ke
r.
The
c
o
u
n
t
i
n
g
m
echa
n
ism
s
r
e
g
ister
pr
oce
s
se
s
of
t
he
p
ha
se
o
r
e
a
rt
h
fau
l
t
u
n
its
w
hic
h
c
a
n
a
l
s
o
be
i
n
i
t
i
at
ed
b
y
exter
n
al
ly
c
on
t
r
ol
l
e
d de
vic
e
s w
h
e
n
ap
p
r
opr
ia
te c
omm
u
n
i
ca
ti
o
n
m
e
an
s
a
r
e
av
a
i
l
a
bl
e
.
F
i
gure
4.
S
ing
l
e-sho
t
a
uto-
rec
l
o
s
e
sche
me
fo
r
t
ransien
t
a
n
d
per
manent fault
s
5.
SIMULATIO
N
A
ND R
E
S
ULTS
The
tr
ip
p
i
n
g
/
r
eclo
s
i
ng
sys
t
e
m
a
nd
t
h
e
prop
ose
d
A
N
N
-
base
d
fa
ult
cla
ssif
i
e
r
/d
et
ect
or
a
t
d
i
f
f
e
r
e
n
t
fa
ul
ts
a
re simu
l
ate
d
to in
ves
tiga
t
e
t
h
e a
b
il
ity
of t
h
is
s
c
h
e
m
e
for
f
a
u
l
t
t
y
p
e
de
t
e
c
t
i
on, t
r
i
p
p
i
n
g
t
he fa
u
lte
d pha
s
e
and
after
the
fa
ul
t e
x
ti
n
g
u
i
sh
t
he
fau
l
t
e
d
p
has
e
a
uto-r
ecl
os
ure
.
A
phase
-
t
o-gr
ou
n
d
(
A
G
)
f
a
u
lt
i
s
ap
p
l
ied
at
t
ime
=
1
sec
fr
om
t
he
s
t
a
r
t
o
f
t
he
s
imu
l
a
t
i
o
n.
A
fter
t
he
fa
ul
t
start,
t
he
c
urre
nt
i
n
re
sp
e
c
ti
ve
p
hase
(
A
)
w
ill
i
ncr
eas
e
and
t
h
e
pro
t
e
c
ti
o
n
s
yste
m
de
t
e
ct
s
t
h
e
fau
l
t
based
on
the
A
N
N
-
fa
ult
c
l
ass
i
fier
a
t
the
fa
u
lte
d
p
h
a
se
(
A
)
.
The
n
,
th
e
o
p
e
n
i
n
g
s
i
g
n
a
l
i
s
s
e
n
t
t
o
t
h
e
C
.
B
o
f
p
h
a
s
e
(
A
)
a
t
t
=
1
.
2
s
e
c
a
n
d
a
f
t
e
r
t
h
i
s
t
i
m
e
t
h
e
f
a
u
l
t
e
d
p
h
a
s
e
(
A
)
i
s
o
p
e
ne
d
a
n
d
the
C.B
of
p
hase
(
A
)
i
s
o
p
e
n
ed
a
nd
t
he
C.
B
of
p
hase
(
A
)
i
s
rec
l
ose
d
a
t
t
=
1.
5sec.
T
he
w
a
v
e
f
or
ms
o
f
3-
ph
a
s
e
(c
u
rren
t
s,
vol
t
a
ges,
a
ct
i
v
e
po
w
e
r,
a
n
d
rea
c
ti
ve
p
ow
er
)
duri
ng
fau
l
t
sit
u
a
t
i
o
n
i
s
s
h
o
w
n
in
F
i
g
ure
5
.
T
h
e
r
e
is
a
n
o
b
v
i
ou
s
dis
t
ort
i
on
i
n
t
he
c
ur
rent
a
n
d
vo
lta
ge
w
a
v
ef
orms
i
n
pha
se
A
w
he
n
t
h
e
se
c
onda
ry
a
rc
e
x
t
i
n
gu
is
h
e
s.
I
t
c
a
n
be
o
bse
r
ve
d
fr
om
F
igure
5
(
3)
t
ha
t
un
ba
lanc
e
o
p
e
r
ationa
l
c
o
nd
i
tio
n
lea
d
s
to
a
f
luc
t
ua
ti
on
i
n
t
h
e
a
c
t
i
v
e
a
nd
rea
c
t
i
v
e
po
w
e
r
s
duri
n
g
t
h
e
fa
ult
occ
u
rrenc
e.
T
h
e
ope
rat
i
on
of
t
ri
pp
ing
/
re
cl
os
ing
of
a
u
t
o-re
cl
osu
re
a
t
t
h
e
s
i
n
g
le
l
i
n
e-
to-
g
ro
un
d
be
fore
a
nd
a
fter
fa
ul
t oc
curr
ing
are diss
i
p
a
t
e
d
in F
i
gur
e 6
.
(a)
(b)
0.
9
1
1.
1
1.
2
1.
3
1.
4
1.
5
1.
6
-1
0
0
-5
0
0
50
10
0
Ti
m
e
(
s
e
c
)
V
o
l
t
ag
e (
V
)
Va
Vb
Vc
0.
9
1
1.1
1.
2
1.
3
1.
4
1.
5
1.
6
1.
7
1.8
-2
0
-1
0
0
10
20
30
Ti
m
e
(
sec
)
C
u
r
r
e
n
t
(A
)
Ia
Ib
Ic
T
r
i
p
o
f
pha
s
e
(
A
)
R
ecl
o
s
er
o
f
C
.
B
(A
)
Fau
l
t
i
n
cep
t
i
on
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
&
D
ri S
yst
IS
S
N
:
2088-
86
94
I
m
prov
em
e
n
t
o
f
pr
ot
e
c
t
io
n
rel
a
y
wit
h
a si
ng
l
e
pha
se
au
t
o
-reclo
s
i
n
g … (Z
ozan S
a
a
d
a
lla
h
H
u
ssa
i
n)
51
1
(c)
F
i
gur
e 5.
S
i
m
u
lat
i
on re
su
l
t
s o
f
a
s
ing
l
e
p
h
as
e-to
g
ro
un
d fa
ult (
A
G
)
a
t ge
nera
t
i
on b
u
s
b
ar
for
3-p
ha
se
w
a
vefor
m
s: (
a)
V
olt
a
ges,
(
b) Curr
e
nts a
nd
(c)
A
c
t
i
ve
and
rea
c
ti
ve tra
n
s
mi
ssion
p
o
w
e
r
be
f
o
r
e
af
t
er
the fault
c
l
ear
ing
(fa
u
lt
i
n
c
e
p
tio
n
tim
e 1se
c
)
(a)
(b)
(c)
F
i
gure
6.
S
e
qu
ence
s of
o
pe
n/
r
e
c
los
u
re
opera
t
i
o
n
f
or C.
B
i
n
3-
p
h
ases
a
t si
n
g
l
e
-
phase
-
t
o gr
ou
n
d
fau
l
t
(
A
G
)
:
(a) P
h
ase
A,
(
b
)
Phase
B and
(c)
P
h
ase
C
I
n
t
he
c
a
s
e
of
t
he
p
e
r
m
a
nen
t
f
a
u
lt,
it
is
c
o
n
s
i
dere
d
tha
t
t
he
s
in
gle
l
i
ne
t
o
gro
u
n
d
f
a
u
l
t
o
c
c
urs
in
t
he
pha
se
A
a
n
d
a
t
t
=
0.
4
sec.
I
t
de
mon
s
tra
t
e
s
t
hat
a
pe
rm
anent
fa
ul
t
is
c
l
e
ar
e
d
a
t
t
=
0.6
se
c,
t
he
a
u
t
o-re
cl
osure
on
t
he
f
a
u
lte
d
l
i
ne
i
s
o
p
e
n
e
d
a
t
t
=
0.6s,
an
d
the
C
.
B
is
r
e
c
lo
s
e
d
a
t
t
=
0
.
8
s
.
w
h
i
l
e
t
h
e
f
a
u
l
t
i
s
s
t
a
y
i
n
g
a
f
t
e
r
0
0.2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
-40
0
0
-20
0
0
0
20
00
40
00
60
00
80
00
T
i
me (sec)
P
&
Q
(
W
&
VA
R)
P
Q
Fa
ul
t
occuri
ng
C.
B
ope
ne
d
C.
B
re
cl
ose
r
b
e
f
o
r
e
f
a
u
lt
occurri
ng
0
0.
2
0.4
0.6
0.
8
1
1.2
1.4
1.6
1.8
0
0.
2
0.
4
0.
6
0.
8
1
T
i
m
e
(
s
ec)
op
e
n
/
c
l
o
s
e
ope
r
a
t
i
ons
Ta
Op
e
n
i
n
g
ti
m
e
C
l
os
i
ng
t
i
me
Au
t
o
r
e
c
l
o
s
e
0
0.2
0.4
0.6
0.
8
1
1.
2
1.
4
1.
6
1.8
0
0.
5
1
T
i
m
e
(sec)
o
p
en
/
c
l
o
se
o
p
er
at
i
o
n
s
Tb
0
0.
2
0.4
0.
6
0.8
1
1.2
1.
4
1.6
1.
8
0
0.5
1
T
i
m
e
(
s
ec)
o
p
en
/
c
lo
se
o
p
e
r
at
i
o
n
s
Tc
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2088-
8694
I
nt
J
P
ow
Elec
& Dr
i
S
y
st V
ol.
11,
N
o.
1
, Ma
r
202
0
:
505
–
51
4
51
2
clo
s
in
g
t
h
e
fa
u
lte
d
phase.
Th
e
r
efore,
t
he
C
.
B
i
s
reope
n
e
d
a
t
1
s
a
n
d
t
h
e
C.B
i
s
r
ec
l
o
se
d
a
t
t
=1.
2
s.
I
n
F
i
gur
e
7,
the
per
f
orm
a
nce
of
t
hree
-p
has
e
v
o
l
tage
s,
c
urr
e
nts
a
nd
ac
tiv
e
a
n
d
rea
c
t
i
v
e
pow
er
i
s
s
h
ow
n
u
n
d
er
t
he
lin
e–t
o
-
gro
u
n
d
f
a
u
lt
o
f
t
he
p
erm
a
ne
n
t
c
o
n
d
it
io
n.
T
he
r
esu
l
ta
nt
t
ri
pp
in
g
a
n
d
r
eclo
s
i
ng
s
i
g
n
a
l
s
di
a
g
nose
d
b
y
th
e
pro
pose
d
A
S
P
A
S
i
s i
l
l
u
s
t
rate
d in F
ig
ure
8.
I
n
th
i
s
ca
s
e
,
it c
a
n
be no
t
e
d tha
t
t
he
a
ll p
o
le
s o
f
the
C.B
a
re l
ock
e
d
ou
t af
t
e
r th
ird
tr
i
p
pi
n
g
sig
na
ls
.
(a)
(b)
(c)
F
i
gure
7.
S
i
mula
ti
on r
e
su
lts o
f a
single
l
i
ne
–
t
o-
grou
n
d
fa
u
lt
(
A
G
)
a
t ge
n
era
tio
n
busba
r w
i
t
h
3-
s
h
o
t
s a
u
to-
rec
l
os
ure
in c
ase
perm
ane
n
t
faul
t f
o
r 3-pha
s
e
w
a
v
efo
r
ms: (a)
V
o
l
t
ag
e
s
, (b
) C
u
rre
n
t
s
and
(c)
Act
i
v
e an
d
re
acti
v
e
t
r
ansmiss
i
o
n
p
ower
before
afte
r t
h
e
fau
l
t oc
curri
ng
(
f
ault i
n
ce
p
t
i
o
n
t
i
m
e
(a)
0.
4
0.6
0.
8
1
1.
2
1.
4
1.
6
-10
0
-5
0
0
50
10
0
T
i
me
(
se
c)
Vo
l
t
a
g
e
(
V
)
Va
Vb
Vc
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
-40
-30
-20
-10
0
10
20
30
40
T
i
m
e
(s
ec)
Cu
r
r
e
n
t
(
A
)
Ia
Ib
Ic
C
.
B
o
p
e
n
&
l
o
c
k
s
o
u
t
I
n
st
an
t
of
f
au
l
t
Cl
os
e
d
C.
B
I
s
ol
a
t
i
n
g
of
pha
s
e
(
A
)
0
0.2
0.
4
0.
6
0.
8
1
1.
2
1.4
1.
6
1.
8
-4
00
0
-2
00
0
0
20
00
40
00
60
00
80
00
T
i
me
(
se
c)
P
&
Q
(
W
&
V
A
R
)
P
Q
0
0.2
0.4
0.6
0.
8
1
1.
2
1.
4
1.6
1.
8
0
0.
2
0.
4
0.
6
0.
8
1
T
i
m
e
(
s
ec)
ope
n/
c
l
os
e
ope
r
a
t
i
ons
Ta
Tr
i
p
-
1
Tr
i
p
-
2
Tr
i
p
-
3
l
o
cks
o
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
Int J
P
o
w
E
l
e
c
&
D
ri S
yst
IS
S
N
:
2088-
86
94
I
m
prov
em
e
n
t
o
f
pr
ot
e
c
t
io
n
rel
a
y
wit
h
a si
ng
l
e
pha
se
au
t
o
-reclo
s
i
n
g … (Z
ozan S
a
a
d
a
lla
h
H
u
ssa
i
n)
51
3
(b)
(c)
F
i
gure
8.
S
e
quence
s
of
o
p
e
n
/
re
clos
ure
opera
t
i
o
n
for C.B i
n
3-p
hases
at (AG)
fault w
i
t
h
3-sh
ot
s a
u
t
o
-
re
clos
ure
i
n
c
a
s
e
pe
rm
anent f
a
ul
t (1)
Phase
A, (
2)
P
hase (
B) an
d (3)
P
h
a
s
e
(C)
6.
CO
NC
L
U
S
I
O
N
I
n
t
his
p
a
per,
a
n
ew
s
che
m
e
mult
i
s
h
ot
3
-
p
hase
t
ri
pp
ing
a
nd
r
e
c
los
i
n
g
s
y
s
tem
ha
s
be
e
n
d
es
ig
ne
d
for
sy
st
em
s
t
a
bil
i
t
y
.
A
l
s
o
,
ANN
i
s
t
rai
n
ed
a
nd
t
est
e
d
b
a
sed
on
Back
P
ropa
ga
ti
on
t
o
d
e
t
e
c
t
f
a
u
l
t
p
er
phase.
A
s
a
w
h
o
l
e,
t
he
s
ug
ge
sted
a
ppro
a
c
h
h
a
s
b
ee
n
m
o
dele
d
in
M
at
la
b/S
i
m
u
l
i
n
k
p
ro
gr
am
a
n
d
b
ase
d
o
n
t
h
e
ana
l
y
z
i
n
g
fu
n
d
am
enta
l
3
-
phase
c
urre
nts
at
t
he
g
e
n
e
r
ati
o
n
b
u
s-ba
r
dur
in
g
t
he
p
e
r
io
d
be
t
w
ee
n
t
h
e
occ
u
rrenc
e
and
t
h
e
clea
ranc
e
of
t
he
f
a
u
lt.
T
he
o
b
t
a
i
ne
d
r
e
s
u
lts
on
a
s
i
m
u
l
a
ted
sys
t
e
m
s
how
t
ha
t
t
h
e
A
N
N
faul
t
cl
assifi
e
r
t
e
c
hni
qu
e
i
s
accu
ra
t
e
,
e
f
f
e
c
t
i
v
e,
a
n
d
c
o
rre
c
t
ly
c
l
a
ssifi
e
s
t
h
e
fa
u
l
t.
I
n
ad
di
tio
n,
it
is
a
bl
e
t
o
m
ake
corr
ect
t
rip
d
e
ci
sion
u
n
d
e
r
d
i
f
f
ere
n
t
f
a
ult
s
.
Th
i
s
m
ean
s
t
hat
ANN
can
b
e
u
s
e
d as
a
n
e
ffec
tive
m
e
a
n
s i
n
t
he des
i
g
n
of aut
o-
rec
l
osur
e
sche
me
s.
A
lso,
i
t
show
s
t
h
a
t
a
u
t
o-r
ecl
os
ure
sc
hem
e
h
a
v
e
b
en
efit
s
su
c
h
a
s
t
r
an
si
en
t
st
a
b
i
l
i
t
y
enha
nc
em
en
t,
f
a
s
t
e
r
rec
l
osur
e
afte
r
t
r
an
sie
n
t
fau
l
t
c
l
e
a
ra
nce,
minimiz
i
n
g
p
ow
er
s
ys
te
m
o
s
c
i
l
l
a
t
io
n
t
h
ro
ug
h
g
ood
r
e
c
l
o
su
re
t
i
m
e
,
t
ri
p
p
in
g
a
l
l
ph
a
s
e
s
a
nd
l
o
c
k
out
i
n
th
e
ca
s
e
of
a
p
erm
a
ne
nt
f
au
l
t
,
and
a
r
e
d
u
ct
i
on
i
n
syste
m
e
qui
pme
n
t
dam
a
g
e
un
d
er
a
p
erm
a
ne
nt
f
a
u
lt.
REFE
RENCES
[1]
M
a
yad
e
vi
N
,
e
y
.
al.
,
"A
R
eview
on
Ex
pert
S
ys
te
m
Appli
cati
ons
i
n
P
o
w
e
r
P
l
a
n
t
s
,
"
I
n
te
r
n
a
t
io
na
l J
o
ur
na
l
of
E
l
ectrica
l a
n
d
Com
put
er Engin
eeri
ng (
I
J
E
CE)
,
Vol.
4,
No
.
1
, pp
.
1
16
~1
26
, Feb
ru
ary 2
0
1
4
.
[2]
A
r
c
k
araki
t
C
h
a
ithan
a
ku
lwat
,
"
Track
t
h
e
m
axim
um
pow
er
o
f
a
ph
o
t
o
v
o
l
t
a
ic
t
o
contro
l
a
cascad
e
f
i
ve-lev
el
i
nv
erter
a
s
i
n
g
le-p
has
e
g
ri
d-con
n
ect
ed
w
i
t
h
a
f
u
zzy
l
o
g
i
c
c
on
trol
,"
Int
e
rnati
o
n
a
l
Jou
r
n
a
l of
Power
El
ectron
i
cs
and
Drive
S
y
stem
(
I
JPE
D
S)
, Vol.
10
, No. 4, p
p
. 1
86
3
~
18
74
, D
ecemb
e
r 20
19
.
[3]
A
h
m
e
d
J.
A
li
,
Zi
yad
K
.
F
arej
,
and
N
a
shw
a
n
S.
S
ultan
,
"
P
e
rf
orman
ce
ev
aluat
i
o
n
o
f
a
hy
brid
f
uzzy
l
og
i
c
c
ont
rolle
r
b
a
sed
on
g
eneti
c
a
l
g
o
r
ith
m
f
o
r
t
h
r
ee
p
h
as
e
i
ndu
cti
o
n
m
o
t
o
r
dri
v
e,"
In
ter
n
a
tio
nal J
o
u
r
na
l o
f
P
o
wer El
ectr
oni
cs an
d
D
r
ive Sys
t
em
(
I
JP
EDS)
,
Vo
l
. 1
0
,
N
o
.
1
, pp
.
1
17
~1
27
,
Ma
rch 2
0
1
9
.
[4]
A
z
ri
yen
n
i
an
d
Mo
hd
W
azir
M
u
st
a
f
a,
"
Ap
pl
ication
o
f
A
N
F
IS
f
o
r
D
is
ta
nc
e
Re
la
y
Prote
c
t
io
n
in
T
r
a
n
s
mi
ss
io
n
L
i
n
e
,"
In
tern
atio
nal Jou
r
na
l of
El
ectrical
and
Co
mp
ut
e
r
En
g
i
neer
ing
(
I
JECE)
,
V
o
l
.
5
,
No
.
6
,
p
p.
1
31
1
~
13
18
,
Decem
b
e
r
2
015
.
[5]
N.
K
.
S
i
ng
h
a
n
d
S.
S
.
Ba
d
g
e
,
"
A
Nov
e
l
Fa
ult
De
te
c
t
io
n
a
n
d
Cla
s
s
ification
Tech
ni
qu
e
f
o
r
Dou
b
l
e
C
i
r
cui
t
T
r
an
smi
s
s
i
on
L
i
n
e
Us
in
g
Artifi
cial
N
eural
Netw
ork,
"
Int
e
rnat
iona
l
Conf
ere
n
ce
on Intelligent
C
o
mputing,
In
str
u
m
e
nt
a
t
i
on a
nd
Contro
l
T
ech
no
log
i
es (
I
CICICT
)
,
Kan
n
u
r
,
pp.
1
33
8-13
42
,
2
01
7
.
[6]
A
n
am
ik
a
Jai
n
,
A.
S
.
Th
ok
e,
a
nd
R.
N
.
P
a
t
e
l,
"
Cl
assif
i
catio
n
of
Sing
le
L
in
e
to
G
ro
un
d
Fa
ults
o
n
Double
C
i
r
cu
it
T
r
ansmi
s
s
i
on
Line
usi
n
g
A
NN,"
Int
e
rn
ation
a
l
Jo
ur
na
l
o
f
Compu
t
er
a
n
d
Electrica
l En
gi
neer
i
n
g
,
V
o
l
.
1
,
No.
2
,
p
p
.
1
793
-81
6
3
,
J
u
n
e
200
9.
[7]
G
.
V
e
n
u
M
a
d
h
a
v
a
n
d
Y
.
P
.
O
b
u
l
e
s
u
,
"
A
N
e
w
H
y
b
r
i
d
A
r
t
i
f
i
c
i
a
l
N
e
u
r
al
N
etw
o
rk
B
as
ed
C
o
n
t
r
ol
o
f
Do
ub
ly
F
ed
In
duct
i
o
n
G
enerato
r
,"
Inter
n
a
t
i
o
n
a
l
J
o
ur
na
l o
f
Electrica
l a
n
d
Comp
ut
er E
n
g
i
neer
in
g
(
I
JECE
)
,
V
o
l.
5
,
No
.
3,
p
p.
3
79~
3
9
0
,
Ju
n
e
201
5.
[8]
N
.
F
.
F
a
d
z
a
i
l
a
n
d
S
.
M
a
t
Zali
,
"F
aul
t
d
etection
and
clas
sifica
ti
on
i
n
wi
nd
t
u
r
bi
ne
b
y
u
s
i
n
g
arti
ficia
l
n
eu
ra
l
ne
twork
,
"
In
ter
n
a
t
i
onal
Jou
r
n
a
l of
P
o
wer El
ectron
i
cs
an
d
Drive S
y
st
em (
I
J
P
ED
S
)
,
Vo
l.
1
0
,
N
o.
3
,
pp
.
168
7~
1
693,
S
e
p
20
19.
0
0.2
0.
4
0.6
0.8
1
1.2
1.4
1.6
1.
8
0
0.2
0.4
0.6
0.8
1
T
i
m
e
(
sec
)
o
p
e
n
/c
lo
s
e
o
p
e
r
a
tio
n
s
Tb
0
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
0
0.
5
1
T
i
m
e
(
sec)
o
p
e
n
/c
lo
s
e
o
p
e
r
a
tio
n
s
Tc
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2088-
8694
I
nt
J
P
ow
Elec
& Dr
i
S
y
st V
ol.
11,
N
o.
1
, Ma
r
202
0
:
505
–
51
4
51
4
[9]
A
m
ri
nd
er
K
aur,
Y
adw
i
n
d
er
S
ingh
B
rar,
a
n
d
L
ee
n
a
G
.3
,
"Fau
lt
d
et
e
ctio
n
i
n
p
o
w
er
t
ra
n
s
f
o
rm
e
r
s
usin
g
rand
o
m
n
e
ural
n
etw
o
rk
s,"
I
n
te
r
n
a
t
io
na
l
J
o
ur
na
l
o
f
E
l
e
c
tr
ic
a
l
a
n
d
C
o
m
p
ut
e
r
E
n
g
i
n
e
e
r
in
g
(I
J
E
C
E
)
,
Vol
.
9
,
No.
1,
p
p.
78
~8
4,
F
e
b
rua
r
y
20
19
.
[10]
O
l
alek
an
K
abi
r
u
Kareem
,
et
.
a
l
.,
"
P
o
w
e
r
d
i
strib
u
t
i
o
n
s
ystem
fa
u
lt
m
o
n
it
ori
ng
dev
i
ce
f
o
r
s
uppl
y
n
e
tw
ork
s
i
n
Nig
e
r
i
a
,
"
Inter
n
a
t
i
onal
Jo
urn
a
l of
El
e
c
tri
c
a
l
an
d
Com
puter E
ngineer
in
g (
I
JECE,
)
,
V
o
l
.
9,
N
o
.
4
,
pp
.
28
03~
2
8
1
2
,
Au
g
u
st 2
01
9.
[11]
A
nku
sh
K
.
and
Lu
mes
h
K
.
S.,
"
A
n
Accu
r
a
te
F
aul
t
D
et
ecti
on
an
d
Cl
a
s
s
i
f
i
c
a
t
io
n
A
l
go
r
i
t
h
m
for
D
o
ub
le
C
ir
c
u
it
T
r
an
smi
s
s
i
on
Lin
e
s
U
s
i
n
g
Artif
i
cial
N
eural
N
e
two
r
k,"
In
ter
nationa
l
Jour
nal for R
e
sear
ch
in Appl
ied
Sc
i
e
nce
&
E
ngi
neer
in
g
T
e
c
h
nol
og
y
(
I
JRA
S
ET)
, Vo
l
.
5,
Issu
e VI, No
. ,
pp
.
2
06
0
-2
0
6
5
,
J
un
e
20
17
.
[12]
A
n
am
ik
a
J
a
in,
A.
S
.
Tho
k
e,
a
nd
R.
N
.
P
a
tel,
"
F
a
ul
t
Clas
s
i
ficati
on
o
f
D
o
u
b
l
e
C
i
r
cuit
T
ransm
i
s
s
io
n
Li
ne
U
s
i
ng
Ar
t
i
f
ic
i
a
l
N
e
u
r
a
l
N
e
t
w
o
r
k
,
"
Wo
rld
Ac
ad
e
m
y
of Sc
ie
n
c
e
,
En
gin
e
e
r
ing
an
d
Te
c
h
n
o
lo
gy
In
te
rna
t
io
na
l J
o
u
r
n
a
l
of
E
l
ectrica
l a
n
d
Com
put
er Engin
eeri
n
g
, Vo
l
.
2
,
No
.
1
0,
p
p.
23
6
3
-
23
6
8
, 20
0
8
.
[13]
P
r
eeti
Gu
pta
and
R.
N
.
M
a
hanty,
"
Artif
i
c
i
al
N
eural
Netw
ork
base
d
F
a
ult
Clas
sifier
a
nd
L
ocat
or
f
or
T
ransm
i
ssion
L
i
ne
P
rot
ecti
on,
"
J
o
urna
l of Ele
c
tric
a
l
an
d
Ele
c
t
ro
nic
s
En
g
i
n
e
e
r
i
n
g
(
I
OS
R-J
E
EE
)
,
Vol.
11,
I
ss
ue
1
V
er.
I,
PP.
4
1-
5
3
,
(
J
a
n
–
F
e
b)
2
01
6.
[14]
H
.
M
ey
ar-Naimi
,
S.
H
asan
za
d
e
h,
a
n
d
M
.
S
a
n
a
ye-P
asan
d,
"
Dis
c
rimi
n
ati
o
n
o
f
A
rcin
g
F
a
ults
on
Overh
ead
T
r
an
smi
s
s
i
on
lines
f
or
S
in
g
l
e-P
o
le
A
u
t
o
r
eclos
ure,
"
Int
.
Tr
an
sa
ctions
on El
ectrica
l Ener
gy Sys
t
em
s
,
v
o
l
.
2
3
,
N
o
.
8
,
p
p
.
1
52
3–
153
5,
20
1
3
.
[15]
M
.
K
ho
dad
a
di,
M
.
R
.
Noo
r
i,
a
nd
S
.
M.
S
hahrt
a
sh
,
"A
N
on
C
o
m
m
unic
atio
n
Adap
ti
ve
S
i
ngl
e-Pol
e
A
u
t
o
r
ecl
os
ure
S
c
hem
e
B
as
ed
o
n
T
h
e A
c
us
um A
lg
orit
h
m
,
"
IE
EE
T
r
a
n
s.
P
o
w
e
r
D
e
l
.,
vol.
2
8
,
N
o.
4
,
p
p
.
2
52
6–2
5
3
3,
O
c
t
o
b
er
2
013.
[16]
Z
a
id
H
.
Al-T
a
m
eem
i
,
H
ayder
H
.
E
naw
i
,
Karrar
M.
A
l
-
A
n
bar,
a
nd
Hu
ss
am
M
.
Al
m
u
k
h
t
a
r,
"
Tran
si
ent
Stab
ilit
y
Im
pro
v
em
ent
of
t
h
e
P
ow
e
r
S
ystem
s
,
"
In
don
esi
a
n
Jou
r
n
a
l
of
El
ectri
ca
l
En
g
i
neeri
n
g
an
d Co
mp
ute
r
S
c
ience
,
Vo
l
.
1
2
, N
o
.
3
,
pp
.
916–
92
3, D
ecemb
e
r 20
18
.
[17]
A
v
agad
di
P
ras
a
d
and
J.
B
elwin
Edw
a
rd,
"A
p
p
l
i
cati
o
n
of
W
avel
e
t
T
ech
ni
qu
e
f
o
r
F
a
u
l
t
Cl
assif
i
catio
n
in
T
r
ansmi
s
s
i
on Systems
,
"
Procedia C
o
mpu
t
er
Scie
nce
,
p
p
.7
8–
83
,
2
0
16
.
[18]
Ma
k
m
ur
S
a
i
n
i
,
A.
A
.
Mo
hd
Z
i
n
,
M.
W
.
Must
a
f
a
,
A
.
R.
S
u
l
t
a
n
,
a
nd
Rusdi
Nur,
"
Algor
i
th
m
f
o
r
F
a
ul
t
Locat
i
o
n
and
Clas
s
i
f
i
cation
o
n
P
aral
lel
Tran
s
m
is
s
i
on
L
i
n
e
usi
n
g
W
a
v
e
let
bas
e
d
on
C
larke’s
Transf
or
mation
,
"
Internat
iona
l
Jo
urn
a
l
o
f
Electr
i
ca
l a
n
d
Co
m
p
u
t
er Eng
i
n
eeri
ng (
I
J
E
CE)
,
Vo
l
.
8,
No
.
2,
pp
. 6
99
-7
10
,
Apr
i
l
2
0
1
8
.
[19]
A
bdu
l
H
a
d
i
B
i
n
M
ustaph
a,
R
.
Ham
d
an
,
F.
H
.
Mo
hd
N
oh,
N
.
A.
Z
am
br
i
,
M
.
H
.
A
.
J
a
l
i
l
,
M
a
r
l
i
a
M
o
r
s
i
n
,
a
n
d
M
.
F
.
Basar,
"
F
a
ul
t
Locat
io
n
Ident
i
fi
ca
tion
o
f
D
ouble
C
i
rcui
t
Transmi
s
s
i
on
Li
ne
Usin
g
D
is
crete
W
a
vel
e
t
Transf
orm,"
In
don
esi
a
n
Jou
r
n
a
l
o
f
El
ec
t
r
i
c
al En
g
i
neeri
n
g
and
Co
mp
ut
er S
c
ien
c
e
,
Vol.
1
5
,
N
o.
3
,
p
p
.
13
56
-1
36
5,
S
e
p
te
mb
e
r
20
1
9
.
[20]
Al
i,
A
.
J.,
Allu
A
.
A
.
,
and
An
tar
R.
K
.,
"Fu
zzy
L
o
g
i
c
T
ech
niq
u
e
B
as
ed
S
i
n
g
l
e
P
h
ase
Au
to
reclos
in
g
Protectio
n
Syste
m
o
f
a
Double
C
ircu
it
T
ransmi
ssion
L
ine,"
P
r
o
ceedin
gs
of In
tern
a
tio
nal
Conferen
ce
o
n
El
ectr
i
cal
,
Co
mmu
ni
ca
ti
on, Co
mputer
,
P
o
wer,
an
dContr
o
l
En
gi
neer
ing
,
pp
.
3
1
–
3
6
, 20
1
3
.
[21]
H
a
gh
M.
T.,
Razi
K
.,
and
Tag
h
i
z
ad
eh
H
.,
"F
aul
t
C
l
a
ssif
i
catio
n
an
d
Lo
c
a
t
io
n
o
f
P
owe
r
T
ra
n
s
missi
on
L
in
e
s
Using
Ar
t
i
f
ic
i
a
l
N
e
ur
a
l
N
e
t
w
o
r
k
,
"
Co
nf.
Pr
oc.
IPEC
, V
ol.
1
–
3
,
pp
.
1
1
0
9
–1
11
4, 2
00
7.
[22]
P
r
anav
.
D.
R
ava
l
,
"AN
N
b
ased
C
lassif
i
cati
on
and
Lo
catio
n
of
F
au
lt
s
in
E
HV
T
rans
m
i
s
s
i
on
L
i
n
e
,
"
P
r
oceed
in
gs of
t
h
e Int
e
rna
t
i
o
n
a
l
M
u
lti
C
onf
erence
o
f
En
gi
neer
s
an
d Com
p
u
t
er
S
c
ient
is
ts
,
Ho
ng
K
o
n
g
,
Vo
l.
I
,
ISBN:
9
7
8
-9
88
-
9
867
1-8
-
8,
pp.
1
9
–
2
1
,
M
arch
2008
.
[23]
S
.
K
.
P
a
d
h
y
,
B
.
K
.
P
a
n
i
g
r
a
h
i
,
P
.
K
.
R
a
y
,
A
.
K
.
S
a
t
p
a
t
h
y
,
R
.
P
.
N
a
nd
a
an
d
A.
N
ayak,
"Clas
s
ifi
cati
o
n
o
f
F
au
lts
in
a
T
r
an
smi
s
s
i
on
L
in
e
u
s
i
ng
Artifi
cial
N
eu
ral
Netw
o
r
k
,
"
Internation
a
l
Conf
erence on In
forma
tion T
echnology (
I
CI
T
)
,
Bhu
b
a
ne
swa
r
,
In
dia
, pp
.
2
39
-24
3
, 2
01
8.
[24]
A
v
agad
di
P
ras
a
d
,
J
.
Be
lwin
E
d
w
ard
,
a
nd
K.
R
av
i,
"
A
Re
v
i
ew
o
n
F
a
ul
t
Classifi
cati
on
M
e
th
odol
ogi
es
i
n
P
o
wer
Tr
a
n
s
m
is
sio
n
sy
ste
m
s
:
Pa
r
t—
I
,
"
Jo
urn
a
l
o
f
Electr
i
ca
l S
y
st
ems
and In
forma
ti
on T
e
chn
o
lo
gy
,
pp.
48-60
,
M
ay
201
8
.
[25]
N
.
S
a
r
a
v
a
n
a
n
a
n
d
A
.
R
a
t
h
i
n
am
,
"
A
C
o
m
p
a
r
i
t
i
v
e
S
t
u
d
y
o
n
A
N
N
B
a
s
e
d
F
aul
t
Locatio
n
an
d Classif
i
c
a
tio
n Techn
i
q
u
e
f
o
r
D
ouble
Circu
i
t
T
r
ansm
issio
n
L
i
n
e,
"
Fou
r
th
Int
e
rn
ati
o
n
a
l Co
nf
erence
on
Co
mp
ut
ati
o
n
a
l In
tel
l
i
g
ence
an
d
Co
mmu
ni
ca
ti
on Netwo
r
ks
,
pp
.
8
2
4
-
83
0,
N
ov
20
12.
[26]
M
u
rari
M
oh
an
S
aha,
J
an
I
zyko
ws
ki,
and
Eu
ge
ni
us
z
Ros
o
l
o
wski
,
"F
a
ult
Location
on
P
ower
N
e
t
works,”
Wro
c
la
w
U
n
iver
s
i
ty
o
f
T
echn
o
l
ogy,
S
p
r
i
n
g
er
L
o
n
don
D
o
rdr
e
c
h
t
Hei
d
elb
e
rg
New
York
,
ISS
N
1
6
1
2
-
128
7
e-IS
SN
1
86
0-4
676
,
D
O
I
1
0
.
1
00
7/9
7
8
-
1-8
488
2-8
8
6
-
5,
2
01
0.
[27]
S
.
J
a
m
ali
and
A
.
P
arham
,
"
New
Ap
pro
ach
t
o
A
d
ap
tiv
e
Sing
le
P
o
l
e
Aut
o
-recl
osi
n
g
of
P
ower
T
ransmi
ssion
L
in
es,
"
P
ublished in
IE
T
Ge
ner
a
tion, Tr
a
n
smissi
on &
Distribution
, Vo
l
. 4
,
Issu
e 1
, pp
.
1
15
–
1
2
2
,
Ap
ril
20
1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.