Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 2
,
A
p
r
il
201
6, p
p
.
82
7
~
83
4
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
2.9
530
8
27
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
Identification of
Acoustic Signal
s of Int
e
rnal El
ect
ric Dis
c
h
a
rges
on Glass Insulator under Variable Applied Voltage
Nasir
A.
Al
-ge
e
lani,
M.
A
f
en
di M.
Pia
h
,
I
b
rahim Sae
h
, Nordiana
Az
lin
Othm
an
,
F
a
t
i
n
Liya
na
M
u
ha
med
i
n
,
N
.
F
.
K
a
s
r
i
Institute of High
Voltag
e
and Hi
gh Current
, Facu
lt
y
of
El
ectrical
Engineering,
Un
iv
ersiti
T
e
knolo
g
i Mal
a
y
s
ia,
81310 Johor, Malay
s
ia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Sep 22, 2015
Rev
i
sed
D
ec 13
, 20
15
Accepte
d Ja
n
4, 2016
A Partial Discharge (PD) is an
unwanted
phenomenon in electr
i
cal
equipm
ent. Th
er
efore it is
of great im
portance to identif
y
d
i
ffer
e
nt ty
p
e
s of
PD a
nd a
sse
ss the
i
r se
ve
rity
.
This pa
pe
r
inves
t
ig
ates
th
e
acous
ti
c
em
is
s
i
ons
as
s
o
ciat
ed with I
n
terna
l
Dis
c
harg
e (ID)
from different ty
p
e
s of sou
r
ces in
th
e
time-domain. A
n
experimental set
up was arr
a
nged in th
e h
i
gh voltag
e
labora
t
or
y,
a ch
am
ber with an
ele
c
trode
config
uration
att
ach
ed
to it
was
connected to a high voltage tr
an
sforme
r for generating vario
u
s ty
pes of PD.
A laborator
y
exp
e
riment was done b
y
ma
king
the
models of these discharges
.
The test equipm
ent in
cluding
an
tennas as
a me
ans of de
t
e
ction and
digital
processing tech
niques for sig
n
al an
al
ysis w
e
re used
. W
a
v
e
le
t signal
processing was used to recover
the in
ternal d
i
scharge acoustic signal
b
y
elim
inat
ing
the
n
o
ises of m
a
n
y
n
a
tures.
Keyword:
A
c
ou
stic sign
al
Cav
ity
Glass in
su
lator
Int
e
r
n
al
di
scha
rge
Wav
e
let tran
sfo
r
m
Copyright ©
201
6 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
:
M. Af
end
i
M.
Piah
,
In
stitu
te o
f
Hi
gh
Vo
ltag
e
and
High
C
u
rren
t,
Faculty of Elec
trical Engineering
Un
i
v
ersiti Tekn
o
l
o
g
i
Malaysia,
8
131
0 Joho
r,
Malaysia.
Em
a
il: fen
d
i
@fk
e
.u
tm
.
m
y
1.
INTRODUCTION
Out
d
o
o
r
i
n
s
u
l
a
t
i
on re
pre
s
en
t
s
an i
m
port
a
nt
com
pone
nt
of el
ect
ri
c
po
we
r t
r
an
sm
i
ssi
on a
n
d
d
i
stribu
tio
n sy
ste
m
s, g
i
v
e
n
t
h
at a
sing
le insu
lato
r failure can resu
lt in
an
ex
cessi
v
e
ou
tag
e
of t
h
e
po
wer
sy
st
em
. Di
ffer
e
nt
i
n
sul
a
t
o
r d
e
si
gns a
nd m
a
teri
al
s are em
pl
oy
ed by
p
o
w
er
cor
p
o
r
at
i
o
ns and t
h
ei
r be
hav
i
ou
r
s
are in
v
e
stig
ated
and
tested
in
lab
o
ratories an
d
fiel
d
tests as well as d
u
r
ing
serv
ice co
nd
itio
ns. Sp
eci
m
en
s
(rods a
n
d plat
es) are also te
sted when
res
earche
r
s foc
u
s
on investigati
ng ce
rtain phe
nom
ena of s
u
rface
act
i
v
i
t
y
or m
a
teri
al
pe
rf
orm
a
nce
wi
t
h
o
u
t
be
i
ng i
n
fl
uence
d
by
t
h
e
i
n
s
u
l
a
t
o
r
desi
gn
[
1
]
.
The
per
f
o
rm
ance o
f
in
su
lators is strong
ly lin
k
e
d
t
o
lo
cal cond
itio
n
s
, esp
ecially to
t
h
e accu
m
u
l
atio
n
o
f
po
llu
t
a
n
t
s an
d th
e wettin
g
mechanism
s
present.
Th
e
weak
p
a
rt
s of t
h
e in
su
lat
i
o
n
are t
h
e cavities, sin
ce th
e
g
a
s
b
r
eakdo
wn streng
t
h
is l
o
wer th
an
th
at
o
f
t
h
e so
lid
i
n
su
latio
n. On
the o
t
h
e
r h
a
nd
, t
h
e electric field
in
th
e cav
ities is
m
u
ch
stro
ng
er th
an
in
t
h
e b
i
g
in
su
lation
p
a
rts d
u
e
t
o
lower g
a
s p
e
rm
it
tiv
i
t
y [2
]. Hen
ce,
th
e PD is limit
ed
in
sid
e
t
h
e cav
ities an
d
d
o
es n
o
t
penet
r
ate through t
h
e solid material
to reach the electrodes. Initiation of
a PD in
a cavi
t
y needs two
maj
o
r
co
nd
itio
ns; essen
tially th
e cav
ity electric fi
eld
sh
ou
ld
b
e
m
o
re th
an th
at
of the
g
a
s, a
co
nd
itio
n called
t
h
e
in
cep
tion
vo
ltag
e
lev
e
l. To
st
art an
electron avalanc
h
e, a free electron m
u
st
b
e
p
r
esen
t in
th
e cav
ity [3]. Th
e
ex
tin
ction
vo
lt
ag
e lev
e
l m
a
y
d
e
p
e
nd
on
the actu
a
l vo
lta
ge at wh
ich
a disch
a
rg
e st
arts
, since
pres
umably a
h
i
gh
er in
cep
tio
n
v
o
ltag
e
yield
s
a h
i
g
h
e
r
in
itial
te
m
p
er
atu
r
e in
th
e strea
m
er ch
ann
e
l. In
th
is study, th
e
ad
ap
tab
ility o
f
th
e Daub
ech
i
e
s wav
e
lets of ord
e
rs
2
h
a
s b
e
en
ev
al
u
a
ted
,
an
d
resu
lts h
a
v
e
shown
th
e
su
periority. It is b
e
fittin
g
to
select a su
itab
l
e n
u
m
b
e
r o
f
b
r
eaku
p
lev
e
ls b
a
sed
on
th
e
n
a
ture o
f
th
e sig
n
a
l.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
82
7 – 8
3
4
82
8
B
a
sed o
n
aco
u
s
t
i
c
si
gnal
feat
ures
, i
t
i
s
seen
t
h
at
si
x l
e
vel
s
of d
ecom
posi
t
i
on i
s
t
h
e best
c
hoi
ce,
beca
use i
t
has
d
e
scri
b
e
d
t
h
e
SD aco
u
s
tic si
g
n
a
l in
a m
o
re
min
d
f
u
l
and
sym
p
to
m
a
tic way. Th
is d
ecision
is m
a
in
ly d
u
e to
th
e
lo
w frequ
en
cy b
a
nd
(ap
p
rox
i
matio
n
)
, wh
ich
is
th
e
m
o
st val
u
able
pa
rt of t
h
e acoustic signal [4].
2.
ACOUSTIC SIGNAL
ANALYSIS
USING WAVELET
TRANSFORM
Wavel
e
t
T
r
a
n
s
f
o
r
m
s
(
W
T
)
h
a
s bee
n
i
n
t
r
od
uced
i
n
t
o
si
gn
al
pr
ocessi
ng
by
M
a
l
l
e
t
.
WTs em
pl
oy
ort
h
o
n
o
rm
al
basi
s f
unct
i
o
n
wi
t
h
fi
ni
t
e
s
u
pp
o
r
t
(l
o
cal in tim
e).
W
T
s
can prov
i
d
e
nearl
y
di
st
ort
i
onl
es
s
reco
nst
r
uct
i
o
n
of
si
g
n
al
s t
r
a
n
s
i
t
i
ons. T
h
e m
u
l
t
i
-resol
u
t
i
o
n c
once
p
t
un
de
rl
y
i
ng
WTs
ha
s i
n
gene
ral
l
e
d
t
o
ne
w
so
lu
tion
s
and
i
n
sigh
ts in
d
i
v
e
rse ap
p
licatio
ns su
ch
as
sp
eech
p
r
o
cessing
,
aco
ustic targ
et
id
en
tification
,
weak
si
gnal
det
ect
i
o
n a
nd
n
u
m
e
ri
cal
anal
y
s
i
s
. Th
e i
n
t
e
nt
he
re i
s
t
o
dem
onst
r
at
e t
h
e val
u
e
of
t
h
e
W
T
as a t
ool
t
o
gene
rat
e
a
n
d
e
x
t
r
act
uni
que
f
eat
ure
vect
o
r
s
of
aco
ust
i
c
si
g
n
al
s [
5
,
6]
.
2.
1. T
h
e W
a
ve
l
e
t T
r
an
sf
orm
Wavel
e
t
s
are
use
d
ext
e
nsi
v
e
l
y
i
n
m
a
ny
vari
ed
technical fields. They
ar
e us
ual
l
y
pr
esent
e
d i
n
math
e
m
atica
l
fo
rm
u
l
ae, bu
t can
actu
a
lly be
u
n
d
e
r-st
o
od
i
n
term
s o
f
sim
p
le co
m
p
ariso
n
s o
r
correlatio
ns with
t
h
e si
gnal
bei
n
g anal
y
s
ed
. A
wavel
e
t
i
s
a wavef
o
rm
of
limited duration that has an a
v
era
g
e val
u
e of zero.
Unl
i
k
e si
nus
oi
ds t
h
at
t
h
eo
ret
i
cal
l
y
ext
e
nd fr
om
m
i
nus t
o
p
l
us i
n
fi
ni
t
y
, wa
vel
e
t
s
have a b
e
gi
n
n
i
n
g an
d an en
d
[7]
.
Si
n
u
soi
d
s
are sm
oot
h a
n
d
pre
d
i
c
t
a
bl
e a
n
d
are
g
o
o
d
at
descri
bi
n
g
c
o
n
s
t
a
nt
-f
req
u
e
n
c
y
(st
a
t
i
onary
) s
i
gnal
s
.
W
a
v
e
lets ar
e i
r
r
e
g
u
l
ar
[
8
],
of
limited
du
r
a
t
i
o
n
,
and of
ten no
n-
symm
e
t
r
i
cal. Th
ey
ar
e
b
e
tter
at
d
e
scrib
i
ng
an
o
m
alies, p
u
l
ses, an
d o
t
h
e
r
ev
en
ts th
at start an
d stop
within
th
e
sign
al.
Fi
gu
re
1.
The
i
n
fi
ni
t
e
l
y
l
ong
s
i
nus
oi
d
i
s
st
ret
c
hed
(scal
e
d
i
n
wa
vel
e
t
t
e
rm
inol
ogy
) a
n
d
i
s
no
w at
l
o
we
r
fre
que
ncy
[7]
As can
be seen
from
Fi
gure
1,
wavel
e
t
s
com
e
i
n
vari
o
u
s s
h
apes an
d si
zes.
B
y
st
ret
c
hi
ng and s
h
i
f
t
i
n
g
(“d
ilatin
g and
tran
slatin
g“) t
h
e wav
e
let, it is po
ssib
l
e t
o
“match
” it to
the h
i
dd
en
ev
en
t
and
thu
s
d
i
sco
v
e
r its
fre
que
ncy
a
n
d
l
o
cat
i
on i
n
t
i
m
e [9]
.
I
n
a
d
d
i
t
i
on, a
par
ticular wav
e
let shap
e m
a
y
matc
h
th
e ev
en
t unu
sually
wel
l
(
w
he
n st
r
e
t
c
hed a
n
d s
h
i
f
t
e
d ap
p
r
o
p
ri
at
e
l
y
)
.
Fi
gu
re
2.
Exa
m
pl
es of t
y
pes
of
wa
vel
e
t
s
[
7
]
Som
e
exam
pl
es of wa
vel
e
t
f
a
m
i
ly
are pres
ent
e
d i
n
Fi
g
u
r
e 2. It
pr
o
b
abl
y
l
ooks l
i
k
e t
h
e wavel
e
t
t
o
obt
ai
n s
u
c
h
a g
o
o
d
m
a
t
c
h or cor
r
el
at
i
on.
Fo
r
exam
pl
e, t
h
e Haar wa
vel
e
t
wo
ul
d m
a
t
c
h an abr
u
pt
di
sco
n
t
i
n
ui
t
y
w
h
ile th
e Db20
w
o
u
l
d m
a
tch
a ch
i
r
p sign
al
[
1
0
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Id
en
tifica
tio
n
of Acou
stic S
i
g
n
a
l
s
o
f
In
terna
l
Electric Discha
rges
o
n
Gl
a
s
s In
su
l
a
to
r und
er … (M.A.M. Pia
h
)
82
9
2.2. Suitable Selection of
the
Mother
Wavelet
To
best cha
r
ac
terize the PD
s
p
ikes
i
n
a no
isy sig
n
a
l, carefu
l selectio
n
o
f
th
e m
o
th
er wav
e
let is v
e
ry
im
portant to
better approximate and cap
t
u
re th
e tran
sien
t
sp
ik
es
o
f
the orig
in
al sign
al.
Th
e m
o
th
er wav
e
let
will n
o
t
on
ly determin
e ho
w
well th
e
o
r
ig
i
n
al sig
n
a
l is es
ti
mated
in
term
s
o
f
th
e sh
ap
e
of th
e
PD
sp
i
k
es, bu
t
al
so af
fect
t
h
e
fre
que
ncy
s
p
ec
t
r
um
of t
h
e
de
-n
oi
sed
si
g
n
al
[1
1]
. T
h
e c
h
oi
ce o
f
m
o
t
h
er
wavel
e
t
ca
n
be
base
d
on ey
ebal
l
i
n
s
p
ect
i
on
of t
h
e
PD spi
k
es
or
i
t
can be sel
e
ct
ed base
d on
cor
r
el
at
i
on
γ
between the signal of
in
terest an
d the wav
e
let d
e
-no
i
sed sign
al [12
]
,
o
r
b
a
sed
on
th
e cu
m
u
lativ
e en
erg
y
o
v
e
r so
m
e
in
terv
al
wh
ere
PD s
p
i
k
es
occ
u
r. Being wel
l
aware
of t
h
is issue, a
num
b
er of
m
o
ther wavelets ha
ve been
exam
ined
and
it
was
fin
a
lly fo
un
d th
at th
e Daub
ech
ies wav
e
let was t
h
e m
o
st su
itab
l
e for treatin
g
PDs.
3.
METHO
D
OL
OGY
Th
e test set
u
p
co
nsists of two m
a
in
p
a
rts: the ci
rcui
t
l
o
o
p
(
A
C
s
o
u
r
ce, t
r
a
n
sf
orm
e
r, c
o
n
n
ect
i
ons a
n
d
in
su
lator), an
d th
e m
easu
r
emen
t and
acqu
isitio
n
system
(eart
h
i
n
g
re
si
st
or,
wi
deba
nd
ant
e
n
n
as)
.
Fi
gu
re
3
sho
w
s
t
h
e e
x
p
e
ri
m
e
nt
al
set
-
up
fo
r
gene
rat
i
n
g
vari
ous
t
y
pe
s
o
f
di
scha
rges
as wel
l
as
det
e
ct
i
ng t
h
e c
onse
que
nt
acoustic signal due t
o
the
PDs. T
h
e
test cell
was
connected ac
ros
s
a
high
v
o
ltage so
urce. Th
e test cell was
desi
g
n
e
d
t
o
ge
nerat
e
va
ri
o
u
s
t
y
pes o
f
PD
.
I
t
i
s
an
ai
rt
i
g
ht
cy
l
i
ndri
cal
c
h
am
ber m
a
de o
f
gl
ass i
n
whi
c
h t
h
e
electro
d
e
con
f
i
g
uration
is fitted
.
Th
e electrod
e
s are m
a
d
e
of stain
l
ess steel in
a p
o
i
n
t
-to-p
l
an
e co
nfigu
r
atio
n
.
Fi
gu
re 3 s
h
o
w
s a pi
ct
u
r
e
of t
h
e c
h
am
ber t
a
ke
n i
n
t
h
e hi
g
h
v
o
l
t
a
g
e
l
a
bor
at
ory
.
Fi
gu
re 3
.
1 s
h
ows t
h
e
expe
ri
m
e
nt
al
set
up
fo
r
gene
r
a
t
i
ng P
D
. T
h
e
cham
ber i
s
m
a
de ai
rt
i
g
ht
i
n
or
der
t
o
pr
o
d
u
ce t
h
e c
o
nt
am
i
n
at
ed
h
u
m
id
ity an
d
at th
e sam
e
ti
me to
lower th
e
effect
o
f
fi
el
d
noi
se
, w
h
i
c
h c
a
n i
n
t
e
rr
upt
t
h
e o
r
i
g
i
n
al
PD
si
gnal
.
Thi
s
cham
ber i
s
desi
g
n
ed
fo
r
m
u
lt
i
pur
pose
use, where it can be used to te
st the PD unde
r gas. T
h
e test cell
co
nsists of a
p
l
ex
ig
lass cylin
drical tub
e
attach
ed
to two
alu
m
in
iu
m
t
o
p and
bo
ttom
flan
g
e
s which
are
con
n
ect
ed t
o
t
h
e hi
gh
v
o
l
t
a
g
e
sup
p
l
y
an
d g
r
o
u
nd
pot
e
n
t
i
a
l
respect
i
v
el
y
[
13]
. T
h
ese
fl
ange
s can
be re
m
oved
easily to posit
ion t
h
e electrode
co
nfi
g
u
r
at
i
on.
The
t
o
p
and
b
o
t
t
o
m
fl
ange
s are c
o
n
n
ect
ed
wi
t
h
a
sem
i
-
cylin
d
r
ical p
i
ece o
f
in
su
lating
m
a
terial
m
a
d
e
o
f
PVC to
wh
ich
th
e sp
eci
men
is attach
ed
to
g
e
n
e
rate th
e PD.
The eart
h
t
e
rm
i
n
al
i
s
al
so pr
o
v
i
d
e
d
i
n
t
h
e
bo
t
t
o
m
fl
ange. T
h
e t
e
st
cel
l
has a hei
g
ht
of
80
0 m
m
and wei
ghs
1
2
k
g
;
it can
with
stan
d
AC
vo
ltage up
to 100
k
V
and
a
D.C
im
p
u
lse
v
o
ltag
e
up to
1
4
5
kV.
Fi
gu
re 3.
Ex
pe
ri
m
e
nt
al
set
up fo
r gene
rat
i
n
g ID
4.
INTER
NAL
DIS
C
H
A
R
G
E
GENE
RATI
O
N
To ge
nerate a cavity or internal discha
rge
,
a plas
tic container wa
s use
d
to place the spe
c
im
en inside
it an
d
th
e con
t
ain
e
r was
filled
with
tran
sform
e
r o
il in
o
r
d
e
r to
av
o
i
d
an
y o
t
h
e
r typ
e
o
f
PDs and
to fo
cu
s
excl
usi
v
el
y
o
n
i
n
t
e
rnal
di
sc
ha
rge
[
14]
.
Fi
g
u
r
e
4
re
po
rt
s t
h
e
set
u
p
f
o
r
ge
ner
a
t
i
on
of
i
n
t
e
r
n
a
l
di
scha
rge
i
n
oi
l
.
A
plane
-
to-pla
ne
electrode c
o
nfiguration
was
use
d
t
o
c
r
eate cavity disc
harge
in
oil. T
h
e specim
e
n is
placed
within the t
w
o electrodes
pla
n
es and
presse
d firm
ly in
order to avoid a
ny air between
the electrode s
u
rface
and the
glass.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
82
7 – 8
3
4
83
0
Fi
gu
re
4.
Set
u
p
f
o
r t
h
e
gene
rat
i
on
o
f
i
n
t
e
r
n
al
di
scha
rge
5.
CA
VIT
Y
FO
RM
ATIO
N I
N
GL
ASS
In orde
r t
o
ge
nerate a
n
i
n
ternal disc
ha
rge
,
three
pieces of
circula
r
glass of
equal dim
e
nsions we
re
use
d
; a hole
dia
m
eter of
0.5
mm and a dept
h
4 mm
was
drilled in one
of
the pieces a
n
d
the three piece
s we
re
attached firm
ly togethe
r
and glued,
sa
ndwic
h
ing the
piece with the hole as
can be see
n
in Figure
5. A
cross
-
sect
i
on al
on
g t
h
e sy
m
m
et
ry
axi
s
of t
h
e ca
vi
t
y
i
s
sh
ow
n i
n
F
i
gu
re
6.
Fi
gu
re
5.
C
a
vi
t
y
i
n
t
h
e
gl
ass t
o
c
r
eat
e i
n
t
e
r
n
al
di
scha
rge
Fi
gu
re
6.
Test
ob
ject
with cyl
i
ndrical cavity
6.
RESULTS
A
N
D
DI
SC
US
S
I
ON
Measurem
ents were ca
rrie
d
out
using a PD detector
whi
c
h is conside
r
e
d
to
be a relia
ble electrical
t
echni
q
u
e t
o
ca
pt
u
r
e t
h
e
PD
.
Thi
s
e
qui
pm
ent
i
s
not
gr
eat
l
y
affected in elec
trically
noi
sy
envi
ro
nm
ent
s
[1
5]
, s
o
th
e p
a
ttern
s it cap
tu
res are
co
nsid
ered
to
b
e
g
e
n
u
i
n
e
, reliab
l
e an
d
free o
f
no
ise. Co
n
s
equ
e
n
tly, all th
e
com
p
arisons are m
a
de based on the
s
e
patterns.
Int
e
r
n
al
di
scha
rges
are
ge
ner
a
t
e
d usi
n
g
a pl
ane-t
o
-
p
l
a
ne
oi
l
-
di
p
p
e
d
el
ect
r
ode
co
nfi
g
u
r
at
i
on
wi
t
h
i
n
a
sam
p
l
e
of gl
as
s consi
s
t
i
n
g of
a cavi
t
y
[14]
. Thi
s
i
s
a ty
pi
cal
voi
d f
o
u
n
d
i
n
sol
i
d
sheet
m
a
t
e
ri
al
and so
m
e
cast
com
pone
nts. T
h
e disc
ha
rges
occur in
adva
nce of t
h
e
vo
lta
ge peaks
on
both t
h
e po
sitive and ne
gative
halves
of the
wave
form
,
as seen in Figure 7. Disc
harges ar
e
of t
h
e sam
e
am
p
l
it
ude a
nd sam
e
num
ber o
n
b
o
t
h
si
des
o
f
t
h
e ellip
se,
alth
o
ugh
a
d
i
fferen
c
e is sam
e
in
m
a
g
n
itu
de
fro
m
o
n
e
si
d
e
o
f
t
h
e
d
i
sp
lay
to
th
e
o
t
h
e
r
wh
ich
is
n
o
rm
al. It is
n
o
r
m
a
l to
n
o
tice a d
e
g
r
ee
o
f
ran
d
o
m
v
a
riatio
n in
a
m
p
litu
d
e
o
r
lo
cation
wit
h
ti
m
e
. Th
ere is litt
le
Evaluation Warning : The document was created with Spire.PDF for Python.
IJECE
ISS
N
:
2088-8708
Id
en
tifica
tio
n
of Acou
stic S
i
g
n
a
l
s
o
f
In
terna
l
Electric Discha
rges
o
n
Gl
a
s
s In
su
l
a
to
r und
er … (M.A.M. Pia
h
)
83
1
o
r
no
variatio
n in
m
a
g
n
itu
d
e
with
in
creases in
v
o
ltag
e
, and
th
e d
i
sch
a
rge ex
tin
ctio
n
vo
ltag
e
is eq
u
a
l
to
o
r
slig
h
tly b
e
low
th
e d
i
sch
a
rg
e i
n
cep
tion
vo
ltag
e
[15
]
.
Fi
gu
re
7.
Ty
pi
cal
di
scha
rg
es
fr
om
a pl
ane-t
o
-
p
l
a
ne
co
nfi
g
urat
i
o
n
(i
nt
er
n
a
l
di
scha
rge
)
Fi
gu
re
8
sh
o
w
s t
h
e ca
pt
u
r
ed i
n
t
e
rnal
di
schar
g
e
P
D
si
gnal
u
n
d
er
ap
pl
i
e
d
v
o
l
t
a
ge
of
5
kV;
n
o
cont
am
i
n
at
i
on l
e
vel
s
are use
d
for t
h
i
s
t
y
pe
of
di
scha
rge si
nce i
t
i
s
di
ppe
d i
n
oi
l
.
He
re t
w
o t
y
pes o
f
s
e
ns
ors
were
use
d
, the
acoustic and t
h
e piez
o-electric sensor. T
h
e
real
PD si
g
n
a
l
due t
o
i
n
t
e
rn
al
di
schar
g
e i
s
very
noi
sy
,
an
d i
t
i
s
di
f
f
i
c
ul
t
t
o
di
s
t
i
ngui
s
h
bet
w
e
e
n
part
i
a
l
di
sc
har
g
e
p
u
l
s
es a
n
d
n
o
i
s
e,
eve
n
by
vi
sual
i
n
sp
ect
i
o
n
and
k
n
o
wl
e
dge
of
i
n
t
e
r
n
al
di
s
c
har
g
e pat
t
e
r
n
s
.
T
hus
P
D
due
t
o
i
n
t
e
r
n
al
di
sc
har
g
e
ca
n
not
b
e
di
st
i
n
gui
s
h
ed
d
u
e
to
th
e h
e
av
y
no
ise
as show
n in
Figur
e 9
and
1
0
, w
h
er
e
t
h
e
p
a
ttern
cap
t
ur
ed
und
er
op
eratin
g
v
o
ltag
e
of
1
0
kV
an
d
15
kV
r
e
spectiv
ely.
Fi
gu
re
8.
I
n
t
e
r
n
al
di
sc
har
g
es
at
5
kV
Fi
gu
re
9.
I
n
t
e
r
n
al
di
sc
har
g
es
at
10
k
V
Fi
gu
re 1
0
.
I
n
t
e
rnal
di
scha
r
g
e
s
at
1
5
k
V
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
82
7 – 8
3
4
83
2
In
tern
al d
i
sch
a
rg
es
o
c
cu
r
with
in
th
e i
n
su
lati
o
n
g
r
ou
nd-wal
l
, in
sid
e
sm
all
v
o
i
d
s
, si
n
ce they are cav
ity
di
scha
rge
s
so n
o
co
nt
am
i
n
at
i
on i
s
used i
n
t
h
i
s
di
schar
g
e
.
Int
e
rnal
pa
rt
i
a
l
di
schar
g
e
activity is characterized by
sy
mm
e
t
ry in
t
h
e m
a
x
i
m
u
m a
m
p
l
itu
d
e
and
in
th
e nu
m
b
er o
f
d
i
sch
a
rg
e
pu
lses, as id
en
tified
wh
en
th
e activ
ity
occu
rri
ng i
n
b
o
t
h
vol
t
a
ge
hal
f-cy
c
l
e
s i
s
co
m
p
ared. Di
sc
h
a
rges
occ
u
r i
n
adva
nce
of t
h
e
vol
t
a
ge
pea
k
s
and a
r
e
oft
e
n si
m
i
l
a
r i
n
n
u
m
b
er a
nd
m
a
gni
t
ude
, as
sho
w
n
by
t
h
e r
e
d i
n
di
cat
or
s i
n
Fi
g
u
r
es
8,
9
and
1
0
.
It
can
be se
e
n
fro
m
fig
u
r
e th
at th
e in
tern
al
d
i
sch
a
rg
es o
c
cu
r ex
actly in
advan
ce
o
f
th
e
po
sitiv
e and
n
e
g
a
t
i
v
e
h
a
l
f-cycle
o
f
th
e
si
nus
oi
dal
wa
v
e
, w
h
i
c
h
has
r
a
t
h
er t
h
e
sam
e
ph
ase i
n
fo
rm
at
i
on as t
h
e re
fere
nce
pat
t
e
rn
sh
ow
n i
n
Fi
g
u
re
7
.
This confi
r
m
s
that the detected acous
tic sign
al is an
in
ternal d
i
sch
a
rg
e si
g
n
a
l, althou
gh
a lo
t o
f
b
a
ckg
r
o
und
n
o
i
se is m
a
sk
in
g th
e
orig
in
al
PD si
g
n
a
l
.
Aft
e
r th
e pre-
p
r
o
c
essin
g
pro
cedure, t
h
is will be
furth
e
r co
nfirmed
.
6.
1. I
n
ter
n
al
Di
scha
rge
De
-
n
oi
sed
Si
gn
al
After
d
e
-no
i
sing
th
e acou
s
tic in
tern
al d
i
sch
a
rg
e PD
signal,
the res
u
lt can be re
vealed in
Figures 11,
12 a
n
d 13.
The DWT succe
s
s
fully elim
inated the
noise
, k
eeping the m
a
in structure of
th
e in
tern
al d
i
sch
a
rge
si
gnal
,
w
h
ere
t
h
e l
o
w e
n
er
g
y
noi
sy
si
gnal
s
we
re s
u
pp
re
ssed.
Di
sc
har
g
es occ
u
r
i
n
ad
vance
o
f
t
h
e
v
o
l
t
a
ge
peaks a
n
d are
ofte
n sim
i
lar in num
b
er and m
a
gnitude.
Fi
gu
re
1
1
.
De
-
noi
se
d i
n
t
e
rn
al
di
sc
har
g
e P
D
si
gnal
at
5
k
V
Fi
gu
re
1
2
.
De
-
noi
se
d i
n
t
e
rn
al
di
sc
har
g
e P
D
si
gnal
at
1
0
k
V
Fi
gu
re
1
3
.
De
-
noi
se
d i
n
t
e
rn
al
di
sc
har
g
e P
D
si
gnal
at
1
5
k
V
7.
CO
NCL
USI
O
N
From
t
h
e resul
t
s obt
ai
ne
d, i
t
was sh
o
w
n t
h
a
t
t
h
e st
udy
of
i
n
t
e
rnal
di
sc
ha
rge
pr
ovi
des i
n
f
o
rm
at
i
on
l
eadi
ng t
o
a b
e
t
t
e
r un
derst
a
n
d
i
n
g t
h
e be
ha
vi
o
r
of t
h
e i
n
su
lato
rs sub
j
ect
ed
to
d
i
fferen
t
lev
e
ls o
f
op
eratin
g
vol
t
a
ge
. D
u
e t
o
t
h
e pre
-
pr
oc
essi
ng
of t
h
e
dat
a
by
usi
n
g
wavel
e
t
t
r
ans
f
o
r
m
gave t
h
e
m
odel
hi
gh r
a
t
e
of
classification accuracy.
Henc
e, if the da
ta contains a lot of noise it
may
l
ead to false detection. So
we can
i
d
ent
i
f
y
faul
t
s
bef
o
re
t
h
e
de
v
e
l
opm
ent
of
fa
i
l
u
res,
beca
use
o
n
ce
pr
esent
t
h
e
dam
a
ge cause
d
by
P
D
al
way
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Id
en
tifica
tio
n
of Acou
stic S
i
g
n
a
l
s
o
f
In
terna
l
Electric Discha
rges
o
n
Gl
a
s
s In
su
l
a
to
r und
er … (M.A.M. Pia
h
)
83
3
i
n
creases l
e
a
d
i
ng t
o
asset
fai
l
u
res
,
o
u
t
a
ges a
nd
pr
ot
ect
i
on s
y
st
em
fai
l
,
di
sast
er an
d a h
u
g
e am
ount
o
f
ener
g
y
lo
ss.
In
terpretin
g of th
e signal p
r
o
p
e
rties co
u
l
d
b
e
u
s
ed
for early
failure d
e
tection
.
Th
e m
o
d
e
l is
q
u
ite
v
e
rsatile for a
wid
e
rang
e of
ap
p
lication
s
in
th
e field
of
power system
an
alyses. As fu
ture wo
rk
three typ
e
s o
f
p
a
rtial d
i
scharg
es will b
e
g
e
nerated
in
a
h
i
gh
-vo
ltag
e
labo
rato
ry,
n
a
m
e
l
y
co
ron
a
d
i
sch
a
rg
e in
th
e ai
r, fl
o
a
ting
d
i
sch
a
rg
e in o
i
l, and
surface
d
i
sch
a
rg
e in the air, at
d
i
fferen
t
ap
p
lied vo
ltag
e
s
will b
e
reco
rd
ed
an
d a
featu
r
e
v
ector will
b
e
ex
tracted
b
y
u
s
in
g
wav
e
let
tran
sfo
r
m
,
wh
ich
will b
e
u
s
ed
to
tr
ain
th
e Reg
P
SO-RBF-NN.
AC
KN
OWLE
DG
MENT
Th
e au
t
h
ors
wo
u
l
d
lik
e to
th
an
k
t
h
e Min
i
stry o
f
Hi
g
h
e
r Edu
catio
n
(MOHE) Malaysia an
d
Un
iv
ersiti
Tek
nol
ogi
M
a
l
a
y
s
i
a
(UTM
)
f
o
r s
p
on
so
ri
n
g
t
h
i
s
w
o
r
k
un
d
e
r p
r
o
j
ect
Vot
e
num
bers
Q.J
1
3
0
0
0
0
.
2
1A
2.
02
E5
5
an
d R.J1
300
00.78
23
.4
F75
1
.
Our
g
r
atitud
e
also
go
es t
o
all m
e
m
b
ers o
f
t
h
e In
stitu
te
o
f
High
Vo
ltag
e
& Hi
g
h
Cu
rren
t, Facu
lty o
f
Electrical Eng
i
n
eeri
n
g,
Un
iv
er
siti Tekn
olo
g
i
Malaysia (UTM)
fo
r t
h
eir coo
p
e
ration
.
REFERE
NC
ES
[1]
A.K. La
zar
evich
,
"P
artia
l dis
c
h
a
r
g
e det
ect
ion and
loca
liz
ati
on
in high voltag
e
transf
ormers
using an optical acoustic
sensor," Virginia Poly
technic In
stitute and
State U
n
iversity
, 2003
.
[2]
G.C. Crichton
,
P. Karlsson, an
d A.
Pedersen
,
"Partia
l disch
a
rges in
ellipsoidal
and spheroidal voids
,"
IEEE
Transactions on
Electrica
l
Insula
tion,
vol. 24
, pp
. 335-342, 1989.
[3]
C. Forssén, "Partial discharg
es in
cy
lindrical
cavities at va
r
i
able fr
equency
of
th
e
ap
plied
voltage," 2
005.
[4]
N.A. Al-geel
ani
,
M
.
A.M
.
P
i
ah, a
nd R.Q.
S
h
addad, "Charac
t
eri
z
a
t
ion of acous
tic s
i
gnals
due to s
u
rface dis
c
h
a
rges
on HV glass insulators using wavelet radi
al basis function neural networks,"
Ap
plied Soft Computing,
vol. 12, p
p
.
1239-1246, 201
2.
[5]
N.A. Al-geel
ani
and M
.
A.M
.
P
i
a
h
, "Identif
ica
tio
n and
Extra
c
tion
of S
u
rface Dis
c
harge Acous
tic
Em
is
s
i
on S
i
gnals
Using Wavelet n
e
ural network
,
"
I
n
ternational Jou
r
nal of Comput
er &
Electrica
l
Engineering
,
vol. 4,
2012
.
[6]
N.
A.
Al-geelani,
M.
A.
M.
Piah, Z.
Adzis,
and
M.
A.
Al
geelani, "Hy
b
r
i
d regrou
ping PSO based
wavelet neur
al
networks
for ch
arac
teri
za
tion of
acoustic sign
als due to
surface
discharges on
HV glass insulato
rs,"
Applied So
ft
Computing,
vo
l.
13, pp
. 4622-46
32, 2013
.
[7]
D.L. Fugal,
Conceptua
l
wavelets in digital sig
nal proce
ssing: an in-depth, p
r
actical approa
ch for the non-
mathematic
ian
:
Space
& Sign
als Te
chnic
a
l
Pub.,
2009.
[8]
A. Graps, "An
in
troduction
to
wavelets,"
Computational S
c
ience
&
E
ngineering,
I
EEE,
vol. 2
,
pp
.
50-61, 1995
.
[9]
C.K. Chui,
An
in
troduction
to wa
velets
vol. 1
:
Academic pr
ess, 19
92.
[10]
I. Daub
echi
e
s
,
T
e
n l
ectur
es
on w
a
vel
ets
vol. 61
:
SIAM, 1992.
[11]
I. Shim
, J. Sorag
h
an, and W. Siew, "Detection of
PD u
tilizing dig
ital signal processing
m
e
thods. Part 3: Open-loo
p
noise redu
ction
,
"
Electrical Insulation Magazin
e,
IEEE,
vo
l. 17, p
p
. 6-13
, 2001
.
[12]
X. Ma, C. Zhou, and I. Kemp, "A
utomated wavelet selection
and
thresholding for PD detection
,
"
Ele
c
trica
l
Insulation Maga
zine, I
E
EE,
vol.
18, pp
. 37-45
, 2
002.
[13]
A.J. Reid, M. Judd, R. Fouracr
e, B.
Stewar
t, and D. Hepburn, "Simultaneous
measurement of pa
rtial discharg
es
using IEC60270 and radio
-fr
equency
techniques
,
"
Diele
c
trics and
Ele
c
trica
l
Insula
tion, I
EEE Tran
sactions on,
vol.
18, pp
. 444-455
, 2011.
[14]
A. Gaouda, A. El-Hag, T. Abdel-
Galil, M.
Salam
a
, and R
.
Bartn
i
k
a
s, "On-line
detection and measu
r
ement of partial
discharge signals in a noisy
environment,"
Diele
c
trics and Elec
trical Insulation
,
IEEE Transactio
ns on,
vol. 15, pp.
1162-1173, 200
8.
[15]
D.A. Nattrass, "Partial disch
a
rge
m
eas
urem
ent and interpre
tat
i
on
,"
Electrical Ins
u
lation Magazin
e, IEEE,
vol. 4,
pp. 10-23
, 1988
.
BIOGRAP
HI
ES OF
AUTH
ORS
Dr. Nas
i
r Ahm
e
d Algeelan
i re
ce
ived the B
.
E
.
de
gree in e
l
e
c
tri
cal
power s
y
s
t
em
from
Univers
i
t
y
of
Aden, Yemen,
Aden, in 1997
, the M.E. d
e
gr
ee in
electr
i
cal power s
y
stem
engineering fr
om
University
Tech
nolog
y
Malay
s
ia in 2009 and the
Ph.D. degree in high voltage engineer
ing from
University
Tech
nolog
y
Malay
s
ia in 2014. H
e
was
a Lecturer
with Industr
ial Techn
i
cal
Institute
(ITI) for 25
y
e
ars, where he is currently
a senior
lectur
er of High Voltage En
gineer
ing. At th
e
present he is a postdoctoral candi
date
at hig
h
voltag
e
eng
i
n
eering d
e
par
t
ment at Univ
ers
ity
Techno
log
y
Malay
s
ia. He h
a
s published as
au
thor
ed and
co-
a
uthored more than 30 pap
e
rs in
various technical journals
and
conference proceedings. His rese
arch in
ter
e
sts in
clude h
i
gh-voltage
instrumentation, partial
d
i
scharg
e, detection
and
warning s
y
stems and
condition
monitoring of h
i
gh
power equipmen
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
82
7 – 8
3
4
83
4
Dr. M
oham
e
d
Afendi M
oham
e
d P
i
ah (M
.A
.
M
. P
i
ah)
is
an
as
s
o
ciat
e profes
s
o
r at F
acu
lt
y
of
Ele
c
tri
cal
Engin
eering
,
Universi
t
i
Tekno
logi
Ma
l
a
y
s
i
aand
a fe
llo
w m
e
m
b
er of
the Institut
e
of Hi
gh
Voltage
and Hig
h
Current (IVAT
). He is
als
o
a
S
i
gnator
y
of
High Voltage Tes
ting
accr
edit
ation
la
b
of ISO/IEC 170
25. He r
e
c
e
ived
the B
.
El
ec
t.
En
g. degr
ee f
r
om
Universiti
Tekn
ologi Mal
a
y
s
ia
i
n
1986, M.Sc
in P
o
wer S
y
stem from University
of
Strathcly
d
e, UK
in 1990
and PhD in High
Voltag
e
Engineering fro
m
Universiti T
e
knologi Mal
a
y
s
i
a
in 2004
. He was appoint
ed
as an assistant
dir
e
ctor
(Test and
Calibr
ation Div
i
sion) o
f
IVAT
fro
m 1996-2000 and Deputy
Director of
IVAT from 2007-
2009. He h
a
s b
een
involved
in
testing
and
calibra
tion o
f
high
voltag
e
equ
i
pments. His research
inter
e
sts includ
e
high voltage
insu
lation
diagnostic and
co-ordination,
electrical d
i
scharges, poly
m
er
nanocom
posites
insulating
m
a
ter
i
als and
insul
a
tor
condition
m
onit
o
ring.
Ibrahim Saeh r
e
ceiv
e
d his BSc. (Electr
i
cal & El
ectronics)
in (1
997), MSc,
(
E
lectrical Power) in
(2009) and Doctor of Philosop
h
y
(El
ect
rical Power Eng.) in (2014) from
Universiti Tekno
log
i
M
a
la
y
s
ia (UTM
). His
res
ear
ch i
n
teres
t
s
in
clude
power s
y
s
t
em
a
n
al
y
s
is
, der
e
gula
t
ed power s
y
s
t
e
m
,
power s
y
s
t
em
s
ecurit
y
and Artifi
c
ia
l int
e
ll
ig
ence
techn
i
ques
.
Dr. Ibrah
i
m
is
a dire
ctor o
f
Environmental Research & Clean
Energ
y
C
e
ntr
e
(ERCE) and T
h
e Interna
tiona
l S
o
ciet
y of Ocea
n,
M
echani
cal
and
Aeros
p
ace
–s
cie
n
ce
and
engine
er
ing (IS
OM
As
e) m
e
m
b
er.
N.
A.
Othman
was born in Johor,
Malay
s
ia on
Januar
y
19, 198
6. Sh
e r
ece
ived
B.Eng in
E
l
ec
tri
cal
Engineering fro
m
Universiti Teknologi Malay
s
i
a
(U
TM) in 201
0. She is current
l
y
pursuing Ph.D.
degree
at Institu
t
e of High Voltage and High Cu
rrent
(IVAT) in
Facult
y of E
l
e
c
t
r
ica
l
Engin
eerin
g,
UTM.
Her research interest incl
udes the detection and diagnostics
of partial dis
c
harges and space
charge in
insulation fo
r condition
monitoring.
Fatin Liy
a
na M
uhamedin was born
in Selangor
, Malay
s
ia on Februar
y
27, 1990. She received th
e
B.E.
degr
ee in electrical
pow
er engineering
fro
m
Universiti
Teknologi Mal
a
y
s
i
a
(UTM) in
201
3.
Currentl
y
, sh
e is pursuing the M.E. degr
ee a
t
Ins
titute of High Vo
ltag
e
and High Current (IVAT) i
n
F
acult
y of E
l
e
c
t
ric
a
l Eng
i
neer
i
ng, UTM
.
Her
res
earch
inte
res
t
s
include s
u
rfa
c
e
dis
c
harg
e and
materials for h
i
g
h
voltage insulation.
Nur Faiza
l
Bin
Kasri
(N. F.
Kas
r
i)
is
a
PhD
stud
ent
at Fa
cul
t
y
of
El
ectr
i
c
a
l
Engi
neering
,
Univ
ersiti
Teknologi Malay
s
ia and
a fell
ow m
e
m
b
er of th
e Institut
e
of High Voltage
and High Current
(IVAT). He r
e
c
e
i
ved th
e B
.
Me
c
h
atroni
c.
Eng.
d
e
gree
from
Univ
ersiti
Tekno
logi
Mala
y
s
ia
in 201
1
and M. Sc in
El
ect
. Eng
.
also
fr
om
Universiti
T
e
knologi M
a
lay
s
ia in
2014. He
was appoint
ed as a
part
time lecturer in UiTM Pasir
Gudang from 2014-
2015 before
pursue his PhD stud
y
.
He has b
e
en
involved in testing and calibr
a
tio
n of high voltag
e
equipment, ins
u
lation di
agnostic and consultan
c
y
project with
Tenaga Nasion
al
Berhad dan
Res
ear
ch
(TNBR).
His
res
earch
int
e
res
t
s
inc
l
ude h
i
gh
voltag
e
insulation diagnos
tic and co-ordin
atio
n,
electrical discharges,
poly
m
er nano
composites
insulating
m
a
ter
i
als and
insul
a
tor
condition
m
onit
o
ring and
food
t
echnolog
y
process.
Evaluation Warning : The document was created with Spire.PDF for Python.