TELKOM
NIKA Indonesia
n
Journal of
Electrical En
gineering
Vol. 14, No. 3, June 20
15, pp. 381 ~ 3
8
7
DOI: 10.115
9
1
/telkomni
ka.
v
14i3.784
9
381
Re
cei
v
ed Ma
rch 1, 2
015;
Re
vised Ap
ril
18, 2015; Accepte
d
May 1
0
, 2015
Wavelet Transforms Ba
sed Power Transformer
Protection from Magnetic Inrush Current
P. Soundiraraju*, N. Loga
nath
an
K.S.Rangas
am
y Co
lle
ge of T
e
chno
log
y
, K.S.R Kalvi N
agar,
T
i
ruchengo
de,
Namakk
al, T
a
milna
du, Indi
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:soun
dirar
a
ju
e
ee@
gmai
l.com
A
b
st
r
a
ct
The Power transformer is the very
essential elem
ent in the power
systems. For
wealthy running
oper
ation
of
po
w
e
r trans
missi
on
an
d d
i
stribu
tion, the
transf
o
rmer s
e
t-up
is
in
disp
ens
abl
e. Otherw
i
se, u
n
d
e
r
fault situati
on the w
hol
e trans
miss
ion syste
m
w
ill l
ead to
p
o
w
e
r loss. The protectio
n
is n
e
c
essary for po
w
e
r
transformers
i.e. mostly
ag
ai
nst inter
nal
fa
ults a
n
d magnetic inrush cur
r
ents.E
xciting
ma
gn
etic
i
n
rus
h
current w
ill tide
, w
hen the transformer is o
per
ated w
i
t
hout ke
epi
ng lo
ad or i
n
the Volta
ge r
e
covery after th
e
fault b
e
in
g se
parate
d
. T
h
is
w
o
rk sugg
ests deve
l
o
p
in
g tr
ansi
ent d
e
tecti
on tec
hni
qu
es
usin
g w
a
vel
e
t
transform for
al
l these fa
ults. T
he w
a
velet transfor
m
h
a
s a
ben
efit of r
epre
s
entatio
n of cu
rrent an
d Volta
g
e
sign
als. E
m
p
a
thy of tra
n
si
ent
s is v
e
ry fast
and
pr
ec
ise
th
e res
earch
pr
o
poses
to
expa
nd
a
new
w
a
vele
t
meth
od to id
en
tify inrush curr
ents to classify
it
from pow
er system distur
b
ances. The su
ggeste
d practi
ce
extract faults and inr
u
sh g
ene
rated transi
ent
sign
als usi
ng w
a
vel
e
t transform proc
ess.
Ke
y
w
ords
:
inrush current, in
ternal fault, MATLAB, power tr
ansform
er, wavelet transform.
Copy
right
©
2015 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Powe
r tran
sforme
r i
s
one
of the m
o
st
i
m
por
ta
nt ele
m
ents in th
e
power sy
ste
m
. Powe
r
transfo
rme
r
p
r
ofessio
n
s a
r
e a
s
a
n
ode
to co
nne
ct two di
ssi
milar voltage l
e
vels.
Co
nsequ
entl
y
,
the contin
uity of the transforme
r
set
-
up
is of
vigorou
s importan
c
e i
n
maintainin
g
the con
s
iste
ncy
o
f
p
o
w
e
r
sys
te
m. Va
r
i
ou
s
un
sc
he
du
le
d
pr
es
e
r
v
a
tion, espe
cially repla
c
e
m
ent of fau
l
ty
transfo
rme
r
i
s
time
co
nsu
m
ing a
nd ve
ry luxuriou
s.
I
n
order to di
scover faults,
high
spe
edin
e
ss,
highly se
nsiti
v
e and relia
bl
e prote
c
tive relays are re
q
u
isite.
For this pu
rp
ose, differe
ntial prote
c
tion
has
be
en wo
rkin
g as the
prima
r
y prote
c
tion for
most of the power tran
sf
orme
rs. Diffe
rential
protection sch
eme
is based
on the fact that any
fault within
el
ectri
c
al
equi
p
m
ent
would
cause the
current ente
r
in
g i
t, to be
dissi
m
ilar f
r
om
th
a
t
leaving it. He
nce,
we can
match the t
w
o cu
rre
nt
s
which
e
ver in
magnitud
e
or in pha
se o
r
both
and i
s
sue
a trip o
u
tput if the vari
an
ce
excee
d
s
a
p
r
edetermine
d static
valu
e. This pra
c
tice
is
very attractive whe
n
both the end
s of t
he appa
ratu
s a
r
e physi
cally
nearby each other.
Conve
n
tional
tran
sform
e
r prote
c
tion scheme
s
u
s
e seco
nd
h
a
rm
o
n
ic comp
one
nt
as
the
discrimi
nator
factor bet
wee
n
an
in
ru
sh
a
nd inte
r
nal
fa
ult cu
rrent [1]
.
The
main
di
sadva
n
tage
of
this ap
proa
ch is
duri
ng
CT
saturatio
n
, the secon
d
ha
rmoni
c
comp
one
nt could li
ke
wise
be
gene
rated d
u
r
ing internal faults and the
new low-
lo
ss amorpho
us materials in
mode
rn Power
transfo
rme
r
s may prod
uce low second h
a
rmo
n
ic
cont
ent in inru
sh
curre
n
t [2].
A numbe
r of microprocessor ba
se
d algo
rithms
h
a
ve b
een p
r
opo
se
d
in the anci
e
n
t
[3, 4].
Wavelet tech
nique
s have
been im
plem
ented for
co
nsi
s
tent protection [5, 6]. Artificial Ne
ural
netwo
rks h
a
v
e been a
p
p
lied to sin
g
l
e
pha
se
po
we
r t
r
an
sformer p
r
ote
c
tio
n
to discri
mi
nate
internal
fault
s
fromm
agn
etizing
in
ru
sh
current
s
[7,
8]
.
Ho
weve
r the overhe
ad tech
niqu
es are
based on either
time
o
r
f
r
eque
ncy dom
ain sign
al
a
n
d
not
both
time a
nd frequ
ency fe
atures of
the sign
al.
In the lite
r
at
ure
of p
o
wer tran
sformer
prote
c
tion, th
e key i
s
sue l
i
es i
n
discri
minating
betwe
en tran
sform
e
r ma
g
netizin
g inru
sh curre
n
t and
internal fault curre
n
t. It is
natural that relay
sho
u
ld b
e
i
n
itiated in resp
on
se to
internal fa
ult but not
to inru
sh
curre
n
t or
over-
excitation/external fa
ult cu
rre
nt. Early method
s were ba
sed
on
desen
sitizing
or del
aying
the
relay to ove
r
come t
he tra
n
sie
n
ts. The
s
e method
s a
r
e un
sat
i
sf
a
c
t
o
ry
sin
c
e t
h
e
t
r
an
sf
orm
e
r
may
be expo
sed f
o
r a long
unp
rotected time.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 3, June 20
15 : 381 – 38
7
382
The
wavelet t
r
an
sform
is a
relatively n
e
w
a
nd
po
werf
ul tool i
n
the
analysi
s
of th
e po
we
r
transfo
rme
r
transi
ent phen
omeno
n be
ca
use of its abili
ty to
extract informatio
n from the transi
ent
sign
als sim
u
l
t
aneou
sly in
the time
and
freq
uen
cy d
o
main.
Re
ce
ntly, the wav
e
let tra
n
sfo
r
ms
have b
een
a
pplied t
o
an
alyse th
e po
wer sy
stem
t
r
an
sient
s, po
wer qu
ality, as
well
as fault
locatio
n
and
detectio
n
pro
b
lems. Th
e wavelet transfo
rm for analy
s
i
ng the tran
si
ent pheno
me
na
in a
po
wer
transfo
rme
r
unde
r
con
d
itions of fault
s
a
nd m
agn
etizing
inrush current
s
wa
s
pre
s
ente
d
, and simul
a
ted
result
s hav
e sho
w
n tha
t
it is possi
ble to use
certai
n wave
let
comp
one
nts to discrimi
nat
e betwe
en int
e
rnal f
ault
s
a
nd magn
etizi
ng inru
sh
cu
rrents.
2. Magnetic I
n
rush Curre
nt
At the time
of transformer ene
rgi
z
ation, a high current wi
ll be hagga
rd by the
transfo
rme
r
. The sp
ecifie
d
current is termed tran
si
ent
inrush cu
rre
n
t is sho
w
n i
n
Figure 1 an
d it
may up
surg
e to ten tim
e
s the
nomi
nal full loa
d
cu
rre
nt of
transfo
rme
r
durin
g p
r
o
c
e
ss.
Tran
sfo
r
mer
magneti
c
inru
sh
curre
n
ts can be
sep
a
ra
ted into thre
e
types, they are: ene
rgi
z
ati
on
inru
sh
cu
rre
n
t, recove
ry in
rush
current
and
sympath
e
tic in
ru
sh
cu
rre
nt. The e
a
rly ene
rgi
z
ati
o
n
inru
sh
re
sults from
rea
ppli
c
ation
of sy
stem Vo
ltage
to a tran
sformer
whi
c
h
h
a
s
been
form
erly
de-e
n
e
r
gi
zed.
Figure 1. Magnetic Inrush Curre
n
t
The
se
con
d
recove
ry inrush o
ccu
rs
whe
n
tran
sfo
r
me
r voltage i
s
re
store
d
afte
r h
a
ving
been redu
ce
d by a nearb
y
short ci
rcuit on power
sy
stem. The thi
r
d symp
atheti
c
inrush
cu
rrent
can
ari
s
e
wh
en two o
r
ext
r
a tran
sform
e
rs
are op
er
ated in
pa
rallel.
The l
e
vel of
the tran
sfo
r
m
e
r
inru
sh
cu
rren
t is a
fun
c
tio
n
of n
u
me
ro
us fa
cto
r
s,
such
a
s
the
switchi
ng
angl
e of the
term
inal
voltage, the resid
ual flux of the core, th
e pow
er
syst
em imped
an
ce, the transfo
rmer
de
sign
and
others. Hol
c
o
m
b [9] sugge
sts an im
prov
ed analytical equatio
n for the inru
sh.
ω
t
sin
0
t
t
core
-
air
L
e
w
R
ω
t
sin
core
-
air
2
L
2
ω
w
2
R
2U
t
i
(1)
Whe
r
e,
w
R
core
air
L
.
1
tan
(
2
)
Whe
r
e U is the applied
voltage;
R
is the windin
g
resi
stan
ce;
L
is the air-core
indu
ctan
ce of
windin
g
; and
is the time when the core
begin
s
to sat
u
rate
Bt
B
.
2.1. Cause
s
of Inrush Cu
rrent
The mai
n
so
urces of tra
n
s
ient in
ru
sh
current
a
r
e a
s
moment of
switchi
ng, resi
dual flux.
Tran
sfo
r
mer
is highly ind
u
ctive in nature. As
fig.1 after the transformer a
r
e
energi
ze
d at its
positive pe
ak value of inst
ant volt
age i.e. at 90° at this in
stant
th
e rate of current and flux are
zero. In orde
r for the tran
sform
e
r to m
a
ke a
n
opp
o
s
ing Voltag
e
drop to b
a
la
nce a
gain
s
t this
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Wavelet Tran
sform
s
Based
Power Trans
form
er
Protec
tion fro
m
Magn
etic…
(
P
.
S
o
u
n
d
i
r
a
r
a
ju
)
383
applie
d so
urce Voltage, a magneti
c
flux of rapidl
y swelling value
must be p
r
od
uce
d
. The inrush
curre
n
t interrupts whole p
o
we
r sy
stem. It mainly
influen
ce
s on protection
syst
em, transfo
rmer,
equipm
ents
couple
d
to system and incre
a
se
s po
we
r q
uality issu
es [
2
, 4, 7].
2.2.
Effec
t
s o
f
Inrush Cur
r
en
t
The inru
sh
current affects entire powe
r
sy
stem. It
mostly disturbs on tran
sf
orme
r,
prote
c
tion sy
stem, equip
m
ents conn
ect
ed to system
and up
su
rge
s
powe
r
qualit
y issue
s
[4, 7].
As the in
rush
curre
n
t incre
a
se
s the tem
perat
u
r
e in
creases
by the theory of influ
ence of
bad tempe
r
at
ure. As a re
sult the temperatur
e of bushing
s and th
e windin
g
s u
p
turn
s [7]. The
s
h
ort c
i
rc
uit
of trans
former is
always
les
s
than
inr
u
s
h
cu
rre
nt [4]. Thus in
ru
sh
cur
r
e
n
t rea
s
o
n
s
malfunctio
n
o
f
protective
syste
m [1]. As transfo
rme
r
gets i
s
olate
d
due to effe
ct on prote
c
tio
n
system
the
r
e
is di
sru
p
tion
of
system.T
he atte
n
dan
ce
of harmoni
c conte
n
t
in
inru
sh
curren
t
cau
s
e’
s incorrect an
alysi
s
with medica
l equipme
n
ts. It also affects to heat increa
se of the
s
y
s
t
em [1].
3. Proposed
Metho
d
- Wa
v
e
let Transform
The wavelet tran
sform i
s
a
novel and
po
werful
to
ol which
can
extract informatio
n from
the tra
n
sie
n
t
sign
als sim
u
l
t
aneou
sly in
both the
time
and
fre
quen
cy do
main
s
unlike Fo
uri
e
r
Tran
sfo
r
mwhi
ch
ca
n only
give the info
rmation in
the
frequ
en
cy d
o
main.
Wave
let tran
sform
s
have b
een
widely u
s
ed
fo
r an
alyzin
g t
he tra
n
si
ent
phen
omen
a i
n
a Po
we
r transfo
rme
r
fo
r
differentiating
internal fault curre
n
ts
from inrush c
u
rrents
[9-10].
-
k
k
2n
g
k
x
n
g
x
n
y
(
3
)
Whe
r
e x is the signal in di
screte time function,
the se
quen
ce is d
e
noted by x[n]. n is an integer,
g[n] is the impulse re
spo
n
s
e of the low
pass f
ilter an
d y[n] is the output of the filter.
k
k
2n
g
k
x
n
g
x
n
low
y
(
4
)
3.1. Flo
w
c
h
a
r
t of th
e Pro
posed
Algorithm
The flo
w
cha
r
t of the
sug
g
e
s
ted
algo
rith
m is
sh
own i
n
Figu
re
2 a
n
d
is explain
e
d in follo
wing
st
ep
s.
F
i
gure
2
.
F
l
owchar
t o
f
propose
d
algor
ith
m
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 3, June 20
15 : 381 – 38
7
384
Step 1:
The
cu
rre
nt a
nd Voltag
e
si
gnal
s a
r
e
attained
from
th
e thre
e p
h
a
s
e tran
sfo
r
mer
usin
g MATLA
B
software for dissimila
r ki
nds of fault a
nd inru
sh
cu
rrents.
Step 2
: The d
i
fferential cu
rrents of the po
wer tran
sform
e
r are cal
c
ul
a
t
ed.
Step 3:
Thre
e p
h
a
s
e
differential
cu
rre
nts
of wavelet tra
n
sfo
r
m
are a
c
q
u
ire
d
by mean
s of
MATLAB software.
Step 4:
The DWT inv
e
stigatio
n of the differential
current is
stu
d
ied.
Step 5:
The detail co
efficients of the sig
nal are obtaine
d.
Step 6:
Nature of the
transi
ent ba
rbs in the
DWT analysi
s
an
d the value o
f
the Detail
coeffici
ents a
nd
spe
c
tral e
nergy de
cide
s whethe
r th
e cu
rrent is
an in
ru
sh o
r
internal fault
c
u
rrent.
Step 7:
T
he
d1 co
efficient
s of different fault and inrush curre
n
ts are fed and trai
ned.
Step 8:
Wav
e
let based rel
a
y discrimin
a
tes inter
nal fault c
u
rrent from inrus
h
current.
3.2. Implementa
tion of Wav
e
let Transform
The
Wavelet
transfo
rm h
a
s b
een
use
d
to
investig
ate the tran
sient
s in the
power
transfo
rme
r
s. The
data
o
b
tained
from
the
simul
a
tions a
r
e
give
n to the
wav
e
let tra
n
sfo
r
m for
comp
ute DWT coefficie
n
ts of the
signal
s. DWT analy
s
is of filter is
sho
w
n in the
Figure 3. The
r
e
are many cat
egori
e
s
of wavelets su
ch as Dau
b
e
c
hi
es,
Haa
r
, Coi
f
let and
sym
m
let wavel
e
ts. In
this pa
pe
r, a
s
we a
r
e i
n
te
reste
d
in
se
n
s
ing
and
ana
lyzing
small
amplitude, litt
l
e du
ration, f
a
st
oscillating an
d decaying type of
high fre
quen
cy cu
rre
n
t signal
s.
F
i
gure
3
.
Block dia
g
ram
of filter
a
nalys
is
Dau
b
e
c
hie
s
wavelet
of type 6
(DB6) suite
d
well
to thi
s
typ
e
of
high
freque
ncy
cu
rrent.
Con
s
e
quently
DB6 wa
s used as the m
o
ther wavelet
.
Wavelet decomp
o
sitio
n
is com
p
leted
on
the sign
al an
d the DWT co
efficients of
le
vel 1of the signal are attained.
3.3. Wav
e
let Trans
f
orm o
f
Inrush Cu
rr
ent
The m
agneti
z
ing in
ru
sh
cu
rrent u
nde
r
steady
st
ate op
erating
ci
rcu
m
stan
ce i
s
o
n
ly 1-2%
of the tran
sforme
r rate
d
curre
n
t. However when t
he prim
ary o
f
an unloa
de
d tran
sform
e
r is
energized, th
e p
r
ima
r
y wi
n
d
ing
s
of th
e t
r
an
sfor
m
e
r e
n
ticeme
nts a
hug
e m
agn
e
t
izing
cu
rrent
is
ten times g
r
e
a
ter than th
e rated
cu
rre
nt. Due to t
he d
e
liberate atte
nuation of thi
s
current, it may
take a
r
ou
nd
10 cy
cle
s
to
settle do
wn.
This
cu
rre
nt
pretexts li
ke
a fault cu
rren
t to the differential
relay and the
relay malo
perates.
4. Simulatio
n
Resul
t
s
Differential
currents have different
b
e
h
a
viours
b
e
lo
w fault a
nd in
rush
curre
n
t situation
s
.
From th
e time when th
e mag
netizin
g inrush cu
rrent
corre
s
p
ond
s to the
tran
sform
e
r co
re
saturation, th
e inru
sh
cu
rre
n
t has
a co
ni
cal
sha
pe (no
n
-si
n
u
s
oid
a
l);
in other
wo
rd
s inrush curre
n
t
at the switchi
ng time incre
a
se
s very slo
w
ly; as
time passe
s, its sl
ope
ri
se
s. Ho
wever, when
a
fault arise
s
,
the differential curre
n
t angle
increases
comp
aring to the
opening of
the
inru
sh
cu
rre
nt. Con
s
eq
uentl
y
these topo
grap
hie
s
coul
d be u
s
ed a
s
the basi
s
of
discerning t
he
faults
from the inrus
h
current. Es
s
entia
ll
y two princi
pl
es are used i
n
pra
c
tice:
1) The
differential cu
rrent
instigated from
faults be
gins
withup
p
e
r sl
ope a
n
d
then its
slop
e dimi
nut
ions. B
u
t differential
curre
n
t initiated f
r
om in
ru
sh
cu
rre
nt be
gin
s
with a
lo
w
sl
ope
and then its
slope in
cre
a
se
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Wavelet Tran
sform
s
Based
Power Trans
form
er
Protec
tion fro
m
Magn
etic…
(
P
.
S
o
u
n
d
i
r
a
r
a
ju
)
385
2) A highe
r slope in the time domain
shows t
hat there a
r
e high
er freq
uen
cie
s
in the
freque
ncy do
main.
Based
o
n
th
e ab
ove
prin
ciple
s
, it i
s
li
kely that
the
amplitu
de
o
f
the hi
gh f
r
eque
ncy
comp
one
nts
at the initial time ha
s swelli
ng tren
d
in in
rush current. It means that
their amplitu
d
e
s
increa
se from
a low value to a high valu
e. The diffe
re
ntial curre
n
t owing to the in
rush cu
rrent at t
= 50 ms a
nd the re
sultant freque
ncy (D1
–
D5
) from
WT are present
ed in Figure 5. The overh
ead
features at frequency (up to 4 kH
z) l
e
vel D3 are
clearly notice
able.It is also likely that the
amplitude
of high frequenc
ies
(D1–
D5) at the init
ial ins
t
ants
of time has
a reduc
i
ng trend
followin
g
inte
rnalfault
s
.
Whereve
r
th
e d
i
fferential
cu
rrent
due
to t
he ABC–G
in
ternal
fault at
t =
50 ms an
d fre
quen
cy levels from the WT
are de
mon
s
trated.
Figure 4. Wavelet c
oeffic
i
ent for inrus
h
current within f
r
equ
en
cy of 4 kHz
Figure 5. Wavelet c
oeffic
i
ent for fault c
u
rrent within 4
KHz
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NI
KA
Vol. 14, No. 3, June 20
15 : 381 – 38
7
386
Analyzing p
r
ofoundly the
first two co
efficients
D1
and D2 yield
s
a be
neficia
l tool of
discrimi
nation
betwee
n
the
two dissimil
ar ca
se
s of inru
sh an
d fault current. Fi
gure 4 p
r
e
s
e
n
ts
the sh
ape
s o
f
D1 an
d D2
comp
one
nts
(first 5
0
el
e
m
ents) for the
different
ca
se
s of in
ru
sh a
nd
internal fa
ult con
d
ition
s
. Figure
5 do
wri
e
s the fi
rst fo
ur ele
m
ent
s
of the wavel
e
t comp
onent
s D1
and
D2
which sho
w
s min
o
r d
e
viation
betwe
en th
e
tw
o d
e
libe
r
at
ed
ca
se
s. It is
clea
r that it
is
difficult to discrimi
nate bet
wee
n
the inru
sh curr
ent a
n
d
the intern
al
fault founde
d on the D1, D2
coeffici
ents v
a
lue
s
only h
o
w
ever if D1,
D2
coeffi
ci
ent
s can b
e
d
r
a
w
n a
gain
s
t th
e time an
d which
pre
s
ent
s the first four el
em
ents of the wavelet comp
o
nents
D1 an
d
D2 versus ti
me.
Figure 6. D1a
nd D2
wavele
t coefficient
s for
the two
ca
se
s within fre
quen
cy of 50 Hz
Figure 6 do
wrie
s the first
four elem
ent
s of the wave
let comp
one
n
t
s D1 a
nd
D2
which
s
h
ows
minor deviation bet
ween the two deliberated ca
s
e
s
.It is
c
l
ear from the above figures
that
the value
s
of
D1,
D2
coefficients
occu
r i
n
a
small
time
for inte
rnal
fa
ult co
ndition
comp
ared to
i
t
s
time in case of inrush cu
rrent state.
5. Conclusio
n
The results
clea
rly sh
ow that the pro
pos
ed
Wavel
e
t com
b
ined
neu
ral net
work rel
a
y
accurately di
stingui
sh
es i
n
ternal
fault
and
mag
n
e
tizing
inrush current
s i
n
three
pha
se
transfo
rme
r
s. The A
N
N effectively distin
guishe
s
an
d
gives tri
p
si
g
nal withi
n
1/8
t
h of cycl
e
which
is co
nsi
d
e
r
ed
to be very fast. The rel
a
y also p
r
ovide
s
high sen
s
itivity for interna
l
fault current
s
and hi
gh sta
b
ility for inru
sh
curre
n
ts.T
he cl
assifi
cation ability of the ANN in
combinatio
n with
advan
ced
sig
nal p
r
o
c
e
ssi
n
g
techniqu
e
open
s th
e do
or fo
r
sma
r
t relays fo
r p
o
w
er tra
n
sfo
r
mer
prote
c
tion
with very less o
peratin
g time and with d
e
si
rable a
c
cu
ra
cy. The results clea
rly sho
w
ed
that the prop
ose
d
Wavelet
combin
ed ne
ural net
work
relay accurate
ly distinguish
es internal fa
ult
and mag
netizing inru
sh currents in three
phase tr
an
sforme
rs. Th
e ANN effectiv
ely distingui
shes
and
gives t
r
i
p
si
gnal
withi
n
1/8th
of cy
cle
whi
c
h
is
con
s
id
ere
d
to
be ve
ry fast.
The
rel
a
y al
so
provides high sensitivity for in
ternal fault currents and high
stability for inrush currents.
Ackn
o
w
l
e
dg
ements
The a
u
thors woul
d like t
o
t
han
kM
s. K. M. Priyadharshini
,
M
r
.
S. Manoj A
r
un, PG
schola
r
, K.S.R.C.T for thei
r co
ntri
butio
n
s
in the devel
opment of th
e
powe
r
tra
n
sf
orme
r protect
i
on
for MATLAB
simulatio
n
s.
And also
we
woul
d
like to than
k
Mr. C. Srini
v
asan, A
ssi
stant
Professo
r, De
partme
n
t of Electri
c
al an
d Electro
n
ics Engine
erin
g, K.S.R.C.
T for their supp
ort a
nd
coo
peration i
n
developin
g
the system.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
Wavelet Tran
sform
s
Based
Power Trans
form
er
Protec
tion fro
m
Magn
etic…
(
P
.
S
o
u
n
d
i
r
a
r
a
ju
)
387
Referen
ces
[1]
Ja
w
a
d F
a
iz, S
Lotfi-F
ard. A
Novel
w
a
v
e
let
– bas
ed
alg
o
rit
h
m for discr
imi
natio
n of i
n
ter
nal fa
ults for
magn
etizin
g inr
u
sh curre
nts in po
w
e
r transfor
m
ers.
IEEE
Tr
ans. On power delivery
. 20
06
; 21(3): 543
-
549.
[2]
Kasztenn
y B
R
o
sol
o
w
ski E, S
aha MM, Hi
llstr
om B.
A
self-or
gan
i
z
i
n
g
fu
zz
y
logic bas
ed pro
t
ective
rel
a
y
- an app
licati
o
n
to pow
er transformer pr
otecti
on
. IEEE PES'96 Summer M
eetin
g in D
env
er. 1996.
[3]
Liu
P, Mal
i
k O
P
, Che
n
D,
Ho
pe. Stud
y
of
no
n-op
er
atio
n for
intern
al fa
ults
o
f
secon
d
-h
armonic
restrai
n
t
differenti
a
l pr
otection
of po
w
e
r transformers.
T
r
ansactio
n
s of the Eng
i
ne
e
r
ing a
nd Op
er
ating D
i
visi
o
n
of the Can
adi
a
n
Electrica
l
Associati
o
n
.1
989;
28(3): 1-23.
[4]
PL Mao, RK A
ggar
w
a
l. A n
o
v
el a
ppro
a
ch t
o
t
he cl
assifica
tion of th
e tran
sient p
h
e
nome
na i
n
po
w
e
r
transformers u
s
ing com
b
in
ed
w
a
vel
e
t transform and
neur
al net
w
o
rk.
IEEE Transactions on Pow
e
r
Deliv
ery
. 200
1; 16(4): 655-
66
0.
[5]
M Nag
pal, MS
Sachd
e
v, Kao
Ning,
LM W
e
d
eph
ol.
Usi
ng
a
neur
al
netw
o
rk for transfor
m
er protecti
on
.
IEEE Proc. Oh
EMPD Int.confer
enc
e. 199
5; 2(3): 674-
67
9.
[6]
Okan Ozg
one
nel, Guv
e
n
Onbil
g
i
n
, C
agrik
ocama
n
. W
a
v
e
let
bas
ed tr
a
n
sformer
prot
ection.
IE
EE
Melec
o
n
. 20
04
; 3(5): 11-16.
[7]
Okan Ozgon
e
nel. W
a
ve
let b
a
sed A
NN a
p
p
roac
h for tra
n
sformer pr
ote
c
tion
. Internati
ona
l jo
urn
a
l of
computati
o
n
a
l i
n
telli
ge
nce
. 20
05; 2(3): 31-3
9
.
[8]
MA Rahma
n
, Je
yasur
y
a. A state-of the-
art revie
w
of trans
former protecti
on al
gorithms.
IEEE Trans.
Pow
e
r Deliv
ery
. 1998; 3(4): 53
4-54
4.
[9]
T
S
Sidhu, MS
Sachd
e
v
.
On
li
ne
ide
n
tificati
o
n
of m
a
g
netizi
ng
inrus
h
an
d i
n
terna
l
fa
ults i
n
thre
e
pha
s
e
transformers.
IEEE Trans. Power Del
. 1992;
7(4), 188
5-1
8
9
0
.
[10]
S Sudha, Dr A Ebenez
er Je
yakumar. W
a
vel
e
t based re
la
yi
ng for po
w
e
r tr
ansformer pr
otection.
Gests
Internatio
na
l T
r
ansacti
on o
n
computer Sci
e
n
c
e and En
gi
ne
erin
g
. 200
7; 38
(4):12-1
6
.
[11]
JS T
horp, AG Phadk
e. A microproc
essor
bas
e
d
thre
e p
hase tra
n
sfor
mer differe
ntia
l rela
y.
IE
EE
T
r
ansactio
n
on
Pow
e
r appar
atus and Syste
m
s
. 1982; 10(
3): 426-
432.
[12]
Wisznie
w
s
k
i
A,
Kasztenny
B.
A multicriter
i
a
t
r
ansformer
differenti
a
l r
e
la
y
b
a
sed
on
fuzz
y
logic.
IE
EE
Trans. On Power Deliv
ery
. 19
95; 10(4): 1
786
-179
2.
[13]
Yuxue
W
ang,
Huaz
ho
ng,
Xian
gge
n Yi
n,
Xi
an
gge
n, Yi
n
Z
hen
xi
ng
Li.
Phas
e curr
e
n
t differe
ntial
protectio
n
for
transformers
in
w
y
e
d
e
l
ta
mode
. T
E
LK
OMNIKA Indo
nesi
an Jo
urn
a
l
of Electric
al
Engi
neer
in
g
. 2012; 10(
6): 128
0-12
86.
[14]
MR Z
a
man, MA Rahma
n
. Exp
e
rime
ntal te
sting of
the ar
tificial n
eura
l
n
e
t
w
o
r
k bas
ed
protectio
n
of
po
w
e
r transfor
m
ers.
IEEE Tr
ans.On Power Deliv
ery
. 199
8; 13(2): 510-
51
7.
[15]
Z
H
OU Bin, YING Limi
ng, Yo
ngg
an
g, Z
HU. Differenti
a
l Pro
t
ection for
Dist
r
ibute
d
Micro
Grid Base
d o
n
Agent
.
T
E
LKOMNIKA Indone
sian Jo
urna
l of Electrical E
ngi
neer
ing
. 2
013;
11(5): 26
34-
26
40.
Evaluation Warning : The document was created with Spire.PDF for Python.