Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
5
,
Octo
be
r
2020
,
pp.
4759
~
4769
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
5
.
pp
4759
-
47
69
4759
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Determi
nin
g damages in
ceramic
plates b
y using d
iscrete
wavelet
p
ac
ke
t transf
or
m and s
up
port vecto
r ma
chine
Mehme
t Yum
urt
aci
1
, Gokh
an
Gokme
n
2
,
Tahir
Cet
in
A
kinci
3
1,3
Depa
rtment
of
Elec
tr
ical and E
le
c
troni
cs
Engi
n
ee
ring
,
Af
y
on
K
oca
t
epe Unive
rsi
t
y
,
Turkey
2
Depa
rtment of
Mec
hat
ron
ic
s
En
gine
er
ing, Marmara
Univ
ersity
,
T
urke
y
3
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Ist
anbul
Te
chn
ic
a
l
Univ
e
rsit
y
,
Turkey
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
25
, 201
9
Re
vised
Ma
r 2
2
,
2020
Accepte
d
Apr
2
, 2
020
In
thi
s
stud
y
,
a
n
ana
l
y
sis
was
conduc
t
ed
b
y
u
sing
discre
te
wa
vel
e
t
pac
ke
t
tra
nsform
(DW
P
T)
and
support
v
ec
tor
m
ac
hin
e
(S
VM
)
m
et
hods
to
determ
ine
undamage
d
and
cra
ck
ed
plates
.
The
p
endul
u
m
was
used
to
la
nd
equ
al
impact
s
on
plat
es
in
thi
s
exp
er
imenta
l
stud
y
.
Sounds
,
which
emerge
from
pla
t
es
as
a
result
of
the
impact
s
appl
ie
d
to
undam
age
d
and
cr
ac
ke
d
pla
te
s,
a
r
e
sound
signal
s
us
ed
in
the
ana
l
y
s
is
and
DWPT
o
f
the
se
sound
si
gnal
s
were
obta
in
ed
with
128
dec
om
positi
ons
for
fea
ture
ext
rac
t
ion.
Th
e
first
four
components,
r
ef
le
c
ti
ng
the
cha
r
ac
t
eri
sti
cs
of
un
damage
d
and
cr
ac
ked
pl
at
e
s
withi
n
the
se
128
components,
were
sel
ec
t
ed
for
en
hancing
the
p
erf
orm
ance
of
the
cl
assifi
er
and
energ
y
val
u
es
were
used
as
fea
tur
e
ve
ct
ors.
I
n
the
st
u
d
y
,
the
SV
M
m
odel
was
cre
ated
b
y
s
el
e
ct
ing
appr
opr
ia
t
e
C
and
par
a
m
et
ers
for
the
cl
assifi
er.
U
ndamage
d
and
c
rac
ked
pl
at
es
were
see
n
to
be
succ
essful
l
y
ide
nti
f
ie
d
b
y
an
ana
l
y
sis
of
the
t
rai
ning
and
te
st
i
ng
phase
s.
Undam
age
d
and
cra
ck
ed
stat
uses
of
the
pla
t
es
th
at
are
und
amage
d
and
have
th
e
a
naly
s
is
had
ide
nti
f
ie
d
d
iffe
r
ent
cr
ac
ks.
The
bigge
st
adv
antage
of
th
is
anal
y
sis
m
et
hod
used
is
that
i
t
is
high
-
pre
ci
s
io
n,
is
r
elati
v
e
l
y
low
in
cost
reg
ard
ing
expe
riment
al e
q
uipment
and
req
uire
s ha
rdw
are
.
Ke
yw
or
d
s
:
Acousti
c em
iss
ion
Ce
ram
ic
m
at
er
ia
l
Def
ect
detect
io
n
Discrete
wav
el
et
tran
s
form
Suppor
t
v
ect
or m
achine
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Tahir
Ce
ti
n A
ki
nci,
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng,
Ele
ct
rical
an
d
Ele
ct
ro
nics
Fa
culty
,
Istan
bu
l
Tech
ni
cal
U
niv
e
rsity
,
Ayaza
ga
Ca
m
pu
s, 3
4469,
Masl
ak,
Istan
bu
l,
T
urkey
.
Em
a
il
:
akincit
c@it
u.
e
du.tr
1.
INTROD
U
CTION
Me
thods
for
de
te
rm
ining
t
he
de
form
at
ion
s
in
ce
ram
ic
m
at
erial
s
are
a
ver
y
im
po
rta
nt
sub
j
ect
of
researc
h
.
Ce
ra
m
ic
m
at
erial
s,
especial
ly
ceram
ic
plate
s
us
e
d
in
the
kitche
n,
m
ay
hav
e
vi
sible
or
m
ic
ro
-
le
vel
cracks
ori
gi
nating
from
carrying,
dr
yi
ng
a
nd
sta
ckin
g
[
1,
2].
Cracks
an
d
de
form
ation
s,
f
orm
ed
in
the
c
er
a
m
ic
plate
s,
ca
n
be
of
dif
fer
e
nt
de
pth
a
nd
le
ng
t
h
[2,
3].
T
heir
i
de
ntific
at
ion
c
a
n
be
m
ade
in
t
he
fiel
d
of
pro
du
ct
io
n
if
crac
ks
are
a
t
visible
le
vel.
On
the
ot
he
r
hand,
if
the
c
rack
s
i
n
the
c
eram
ic
plate
s
are
not
visible
or
at
m
ic
ro
-
le
vel
or
if
the
su
r
face
of
the
ceram
ic
m
at
erial
is
coated
with
glaze,
the
identific
at
ion
of
the
crac
ks
ca
n
on
ly
be
determ
ined
by
us
in
g
sp
eci
fic
m
et
ho
ds
[1,
4].
Ma
ny
diff
e
re
nt
m
eth
od
s
we
re
us
e
d
in
the
li
te
rat
ur
e
f
or
determ
ining
th
e
defor
m
at
ion
in
the
ceram
ic
m
at
erial
s.
A
m
ong
the
se
m
et
h
od
s
,
the
m
os
t
com
m
on
m
et
ho
ds
a
r
e
rad
i
ographic,
ultraso
und
m
et
hods
,
the
rm
al
te
chn
iq
ues,
therm
o
-
m
ech
anical
analy
sis
(TMA
),
dy
nam
ic
m
echan
ic
al
an
al
ysi
s
(DM
A
),
dif
fer
e
ntial
scann
i
ng
cal
ori
m
et
ry
(D
SC)
,
an
d
the
rm
o
-
gr
a
vim
et
ric
analy
sis
(TGA)
.
T
his
st
ud
y
ca
n
al
s
o
be
seen
a
s
an
a
naly
sis
of
data
obta
ined
by
a
n
ac
ou
sti
c
em
i
ssion
m
et
ho
d
(
AEM)
thr
ou
gh
data
m
ining
m
e
thod
if
the
stud
y
is
analy
sed
in
the
li
gh
t
of
a
m
et
hod.
AEM
ha
s
us
ed
as
a
powe
rful
char
act
e
rizat
ion
too
l f
or stu
dy
ing
the prope
r
ti
es o
f
te
chnol
og
ic
al
ly
i
m
po
rt
ant ceram
ic
an
d
porcelai
n
m
at
erial
s.
Acousti
c
em
issi
on
te
st
te
chn
i
qu
e
s
ha
ve
bec
om
e
a
well
-
known
no
n
-
destr
uctive
m
et
ho
d
widely
us
e
d
to
detect
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
59
-
47
69
4760
and
locat
e
fa
ul
ts
in
ceram
ic
m
at
erial
s.
Du
ri
ng
the
pro
du
ct
ion
an
d
pr
oces
sing
of
ceram
i
c
m
at
erial
s,
var
io
us
par
am
et
ers
are
firm
ly
con
tro
ll
ed
to
i
m
par
t
the
desire
d
m
echan
ic
al
an
d
chem
ic
al
sp
eci
ficat
ion
s
to
th
e
end
pro
du
ct
.
T
he
i
nfor
m
at
ion
of
therm
al
sta
bili
t
y
and
reacti
on
kin
et
ic
s
of
se
ver
al
phases
present
i
n
t
he
c
eram
ic
m
at
erial
s is o
f pr
im
e i
m
po
rtance
[5,
6].
On
e
of
the
m
os
t
prefe
rr
e
d
m
et
ho
ds
for
th
e
detect
ion
of
crack
or
dam
a
ge
in
ceram
ic
m
at
erial
s
i
s
com
pu
te
d
tom
ogra
ph
y
te
ch
niq
ue
s
or
XR
te
chn
i
qu
e
s
[7,
8].
This
m
e
tho
d
i
s
us
ed
for
ve
ry
sp
eci
al
m
at
eri
al
s
or
m
edical
m
at
eri
al
s
at
the
en
d
of
pro
duct
ion.
The
m
et
ho
d
ha
s
lim
i
ta
ti
on
s
su
c
h
as
the
siz
e
of
t
he
m
at
eri
al
and
the
ex
pe
ns
ive
ap
plica
ti
on
t
echnolo
gy.
M
any
di
ff
e
ren
t
stud
ie
s
ha
ve
been
car
ried
ou
t
on
the
te
sts
for
the
determ
inati
on
of
dam
ages
in
ceram
ic
m
at
erial
s
[7
,
9
-
12]
.
O
ver
al
l
s
tud
ie
s
are
on
non
-
dest
ru
ct
iv
e
te
sts.
Desp
it
e
t
his,
t
he
dev
el
op
m
ent
of
si
gn
al
proces
sin
g
an
d
data
a
naly
sis
m
e
tho
ds
pr
ovide
d
the
op
port
un
it
y
to
m
ake
m
or
e
detai
le
d
an
al
yse
s.
As
a
resu
lt
of
m
any
Fo
uri
er
based
a
naly
ses,
low
a
nd
high
fr
e
qu
e
ncy
c
ompone
nts
in
da
m
aged
m
at
eria
ls
wer
e
determ
ined.
I
nfrar
e
d
therm
og
ra
ph
y
is
a
m
e
tho
d
us
ed
i
n
the
char
act
eri
zat
ion
an
d
da
m
age
determ
i
nation
of
cer
a
m
ic
m
at
erials.
Since
com
po
sit
e
m
at
eria
ls
hav
e
relat
ively
hig
h
e
m
issi
on
s
depend
i
ng
on
the
pro
per
ti
es
an
d
ty
pe
of
m
at
erial
us
ed,
they
are
ver
y
su
it
a
bl
e
fo
r
infr
a
re
d
te
sts
with
or
without
surface
tr
ea
t
m
ents
[1
3,
14
]
.
This
te
ch
niq
ue
is
base
d
on
t
he
ap
pl
ic
a
ti
on
of
heati
ng
a
ce
ra
m
ic
sa
m
ple
with
a
s
hort
-
te
rm
energy
pu
lse
.
The
te
c
hn
i
qu
e
is
diff
ic
ult
an
d
ex
pen
si
ve
to
app
ly
.
In
a
ddit
ion
,
di
ff
e
ren
t
ultra
-
s
ound
m
et
ho
ds
can
be
perfor
m
ed
especial
ly
for
ex
pe
rim
ental
pur
poses
or
f
or
te
sti
ng
sensiti
ve
m
at
erial
s
[8
,
9
]
.
O
ne
of
the m
os
t
widely
us
ed
m
e
tho
ds
is
the
sou
nd
em
iss
ion
m
et
ho
d.
D
ur
i
ng
var
i
ou
s
d
ef
or
m
at
ion
s o
cc
urrin
g
i
n
ce
ram
ic
m
at
erial
s,
ultras
onic
em
issi
on
ra
te
s
an
d
i
ntensit
y
can b
e
ass
ociat
ed
with
the
ty
pe
and
sprea
d
rat
es
of
crac
ks
.
S
ign
al
s
from
the
te
st
sa
m
pl
e
can
be
detect
ed
us
in
g
a
piezo
el
ect
ric
trans
du
ce
r.
I
n
this
stu
dy,
th
e
analy
sis
of
the
sou
nds
pr
oduce
d
by
ap
plyi
ng
a
blow
on
the
m
at
erial
was
m
ade.
T
he
pro
po
s
ed
m
et
ho
d
is
m
or
e
ec
onom
ic
a
l
and
si
m
pler
to
im
pl
e
m
ent
than
th
e
m
entioned
m
et
ho
ds.
In
Ta
ble
1
[15
]
,
gen
e
ral
featur
es
a
bout
c
om
m
on
ly
us
ed
ND
T
te
c
hn
ic
al
us
age
a
re
gi
ve
n
[
8
-
11]
.
Here,
w
hen
the
gi
ven
pro
pe
rtie
s
are
e
xam
ined,
t
he
m
et
ho
d
c
hose
n
f
or
this
stu
dy;
it
will
be
seen
t
o
be
econom
ic
al
,
e
asy
to
us
e a
nd ef
fecti
ve.
Ce
ram
ic
plates
that
are
undam
aged
an
d
hav
e
diff
e
re
nt
cracks
we
re
us
e
d
in
this
exp
e
rim
ental
stud
y.
A
co
ns
t
ant
i
m
pact
was
app
li
ed
to
a
po
int
s
pecifie
d
on
the
plate
s
and
the
or
i
gi
nated
sou
nds
wer
e
recorde
d
.
Wh
il
e
the
plate
data
wer
e
analy
se
d
by
D
WPT
a
nd
t
he
si
gn
al
e
nergy,
the
extr
act
ed
feat
ur
e
ve
ct
or
was
ap
plied
to
SV
M
m
e
tho
d
as
an
input.
T
he
refor
e
,
the
cl
assifi
cat
ion
pr
oc
ess
is
carried
ou
t.
In
the
a
na
ly
sis,
qu
it
e
su
cces
sf
ul
resu
lt
s
we
r
e
ob
ta
ine
d
in
disti
nguish
in
g
the
crac
ked
(d
am
aged
)
c
eram
ic
m
a
te
rial
s
and
undam
aged
ce
r
a
m
ic
m
at
eria
ls.
Table
1
.
C
omm
on
ly
u
sed N
DT
te
c
hn
ic
al
f
eat
ur
es
[
15
]
Test
Ap
p
licatio
n
s
Ad
v
an
tag
es
Li
m
itia
tio
n
s
Ultraso
n
ic T
estin
g
Cracks
,
la
m
in
atio
n
s, po
ros
ity
,
lack
of
f
u
si
o
n
,
in
clu
sio
n
s,
th
ick
n
ess
.
Usef
u
l to v
ery
thic
k
sp
eci
m
en
s,
can g
iv
e locatio
n
and
size of
d
ef
ects,
sen
sitiv
ity
,
p
o
rtable.
Exp
en
siv
e,
th
in
com
p
l
ex
sh
ap
es
are
d
if
f
icu
lt,
requ
ir
ed
exp
ert
o
p
erators.
Aco
u
stic E
m
iss
io
n
Crack in
itiatio
n
and
gro
wth
,
leak
s, bo
ilin
g
and
cavitatio
n
,
p
h
ase chan
g
es.
Re
m
o
te and
con
tin
u
o
u
s
su
rveillan
ce,
lo
cati
o
n
,
sev
erity
,
p
er
m
an
en
t
r
ecord
.
Tests an
entire
v
ess
el or sy
ste
m
.
Co
n
tact with syste
m
,
need
m
an
y
co
n
tact po
in
ts, com
p
l
ex
in
terpretation
,
so
m
e sy
ste
m
s a
re
co
m
p
lex
.
Ter
m
al
Testin
g
Vo
id
or lack of
bo
n
d
or
co
n
tin
u
ity
,
th
in
or thick
sectio
n
s,
lo
ss
of
ins
u
latio
n
,
h
eat so
u
rces
Detect
an
d
locate
h
o
t or cold
sp
o
ts
an
d
heat g
en
erating
def
ects.
Per
m
an
en
t
record t
h
at
m
a
y
be
q
u
an
titativ
e.
Re
m
o
te sen
sin
g
,
p
o
rtable.
Po
o
r
reso
lu
tio
n
,
o
ften slo
w,
sp
ecialized eq
u
ip
m
e
n
t can b
e
ex
p
en
siv
e and
r
eq
u
ire
h
ig
h
ly
ex
p
erienced
perso
n
n
el.
Rad
io
g
raph
ic
Testi
ng
Cracks
,
v
o
id
s, corr
o
sio
n
Per
m
an
en
t
record,
p
o
rtable,
Hazar
d
,
exp
en
siv
e,
requ
ired
ex
p
ert
o
p
erators
2.
THE
ORETI
C
AL BA
CKGR
OUND
AN
D MAT
HEM
AT
ICA
L
M
ET
H
ODS
Ther
e
are
t
wo
dif
fer
e
nt
ap
proach
e
s
of
the
m
at
he
m
at
ic
a
l
m
et
ho
ds
us
e
d
in
this
e
xp
e
rim
ental
stud
y
.
These
a
re
D
W
PT a
nd S
VM tec
hn
i
qu
e
.
2.1.
Discrete
w
av
e
le
t
p
acket tr
ansf
orm
D
WPT
is
a
sign
al
analy
sis
m
et
hod
that
was
dev
el
op
e
d
as
a
n
al
te
rn
at
ive
to
discrete
Four
i
er
trans
form
and
is
us
e
d
i
n
m
any
fiel
ds
of
stud
y
t
od
ay
[
16
–
18
]
.
D
WPT
is
the
se
par
a
ti
on
of
t
he
sa
m
pled
sign
al
into
le
s
s
sam
pled
co
m
po
ne
nts
li
ke
the b
ra
nc
hes
of
a tree.
This an
al
y
sis
syst
e
m
is
c
arr
ie
d ou
t wit
h t
he
help
of
h
i
gh
-
pass
(
h
(
n
)) an
d
l
ow
-
pass fil
te
rs
(
g
(
n
)
)
a
nd
da
w
n
s
a
m
pling
. T
he
fi
rst level analy
sis
is ex
p
res
se
d
as
su
c
h for s
howing
(
s
(
n
)
) discret
e
sign
al
[17
–
19]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Determi
ning
dam
ages i
n
cer
amic
p
l
ates
by u
sing discret
e w
avelet
packet
t
ra
nsf
orm
... (
M
ehm
et
Yum
ur
ta
ci
)
4761
1
1
(
)
=
∑
(
)
ℎ
(
2
−
)
(
1
)
2
1
(
)
=
∑
(
)
(
2
−
)
(
2
)
Lo
w
-
pass
a
nd
high
-
pass
filt
er
re
spo
nd
s
of
1
1
(
)
an
d
2
1
(
)
co
ns
ti
tut
es
first
le
vel
c
om
po
ne
nts
or
deco
m
po
sit
io
ns.
Wh
il
e the
sa
m
pl
ing
nu
m
ber dec
reases,
the
nu
m
ber
of com
pone
nts inc
rea
ses
[8
-
10]
,
3
2
(
)
=
∑
1
1
(
)
ℎ
(
2
−
)
4
2
(
)
=
∑
1
1
(
)
(
2
−
)
5
2
(
)
=
∑
2
1
(
)
ℎ
(
2
−
)
6
2
(
)
=
∑
2
1
(
)
(
2
−
)
(
3
)
We
ca
n
get
t
he
seco
nd
le
vel
com
ponen
ts
(
3
2
(
)
,
4
2
(
)
,
5
2
(
)
,
6
2
(
)
)
w
he
n
t
he
sam
e
app
r
oach
is
app
li
ed
to t
he f
irst l
evel c
om
po
ne
nts
[17
-
19
]:
7
3
(
)
=
∑
3
2
(
)
ℎ
(
2
−
)
8
3
(
)
=
∑
3
2
(
)
(
2
−
)
9
3
(
)
=
∑
4
2
(
)
ℎ
(
2
−
)
10
3
(
)
=
∑
4
2
(
)
ℎ
(
2
−
)
11
3
(
)
=
∑
5
2
(
)
(
2
−
)
12
3
(
)
=
∑
5
2
(
)
ℎ
(
2
−
)
13
3
(
)
=
∑
6
2
(
)
(
2
−
)
14
3
(
)
=
∑
6
2
(
)
(
2
−
)
(
4
)
Wh
e
n
the
sam
e
proce
dure
is
app
li
ed
t
o
the
second
le
vel
com
po
ne
nts,
t
he
thir
d
le
vel
com
po
ne
nts
(
7
3
(
)
….
14
3
(
)
)
a
re
obta
ined
f
r
om
(
4
)
.
T
her
e
ar
e
2
a
m
ou
nt
of
com
pone
nt
at
le
ve
l.
12
8
com
ponen
ts
(
127
7
(
)
….
254
7
(
)
)
are
obta
ined
w
it
h disc
re
te
w
avelet
pac
ket tra
ns
f
or
m
fro
m
the level
7
[
17
-
19]
,
{
(
)
}
{
1
1
(
)
,
2
1
(
)
}
{
3
2
(
)
,
4
2
(
)
,
5
2
(
)
,
6
2
(
)
}
{
7
3
(
)
,
8
3
(
)
,
9
3
(
)
,
10
3
(
)
,
11
3
(
)
,
12
3
(
)
,
13
3
(
)
,
14
3
(
)
}
{
127
7
(
)
,
128
7
(
)
,
129
7
(
)
,
130
7
(
)
,
.
.
.
.
.
.
.
.
.
.
,
251
7
(
)
,
25
2
7
(
)
,
253
7
(
)
,
254
7
(
)
}
(
5
)
2.2.
Support
v
e
ct
or
ma
c
hine
SV
M,
de
velo
pe
d
by
V
apnic
a
nd
his
f
rien
ds,
is
a
m
achine
le
arn
i
ng
al
gorith
m
widely
us
ed
in
va
rio
us
app
li
cat
io
ns
s
uch
as
patte
r
n
r
eco
gnit
ion
,
re
gr
essi
on
a
nd
cl
assifi
cat
ion
since
it
ha
s
the
a
bili
ty
of
hi
gh
gen
e
rali
zat
ion
thr
ough
str
uctur
al
ris
k
re
duc
ti
on
[
20]
.
The
purpose
of
S
VM
al
go
rithm
is
to
form
the
m
os
t
appr
opriat
e
de
ci
sion
functi
on
for
re
duci
ng
cl
assifi
cat
ion
e
rror
s
t
o
a
m
inim
u
m
as
well
as
to
use
it
in
fu
ll
y
est
i
m
ating
the
unkn
own
dat
a
in
th
e
tw
o
cl
asses
[
21
]
.
Wh
il
e
x
i
is
t
he
unkn
own
i
nput
data
a
ppli
ed
to
SV
M belo
ng
i
ng
to
Cl
ass
1
or Cl
ass
2
a
nd
it
is
n
-
siz
ed
i
nput v
ect
or
if
t
he
outpu
t data
(
y
i
)
is
1
the
n
x
i
belo
ngs
t
o
Cl
ass
1
and
al
so
y
i
is
-
1
then
x
i
belon
gs
to
Cl
ass
2.
If
the
data
are
li
near
ly
separ
at
ed
,
then
f
(
x
)
ex
pre
ssion
,
pro
vid
e
d
in
se
par
at
in
g hype
r
-
plane (
6)
,
is
de
te
rm
ined
to
b
e
zero [
22
]
.
(
)
=
+
=
∑
+
=
0
=
1
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
59
-
47
69
4762
Her
e
,
is
an
n
-
siz
e
d
weig
ht
vecto
r
a
nd
is
the
scal
e,
th
ey
identify
th
e
posit
ion
of
separ
a
t
or
hype
r
-
plane.
I
f
(
)
≥
1
,
then
=
+
1
and
if
(
)
≤
1
,
then
=
−
1
[22].
Th
e
case
in
wh
ic
h
SV
M
a
nd
two
cl
asses
are
li
near
ly
separa
te
d
is
sh
ow
n
in
Fig
ure
1
[23]
.
In
t
he
case
of
no
t
li
near
s
epar
a
ble
two
c
la
sse
s
,
sla
ck
va
riable
an
d
C
pe
nalt
y
facto
r
a
re
a
dd
e
d
to
the
obj
ect
iv
e
functi
on.
In
this
cas
e,
if
≥
0
,
the
n
unknow
n
data
are
cl
assi
fied
a
s
.
+
≥
+
1
−
f
or
Cl
ass
1
and
.
+
≤
−
1
+
f
or
Cl
ass
2
a
nd
the
ob
j
ect
ive
f
un
ct
io
n
is
for
m
ed
li
ke
i
n (7)
[
24
]
.
,
,
{
‖
‖
+
(
∑
)
}
(
7
)
Her
e
,
wh
il
e
C
is
a
par
am
eter
determ
ined
by
the
us
er
and
co
ntr
ols
t
he
cl
assifi
cat
ion
er
rors
of
edu
cat
io
n
data,
k
is
a
po
sit
ive
intege
r
[
15]
.
I
nput
s
pace
that
is
trans
ferre
d
to
hi
gh
-
dim
ension
al
s
pace
(HD
H
)
with
nonlinea
r
(
.
)
trans
f
or
m
at
ion
cal
le
d
feat
ure
sp
ace
to
fin
d
th
e
m
os
t
ap
propriat
e
se
par
at
or
hy
per
pla
ne
as
sh
ow
n
in
Fig
ure
2
.
T
he
cal
culat
ion
of
the
inn
e
r
pro
du
ct
is
cha
nged
int
o
(
,
)
=
(
)
.
(
)
du
e
t
o
this co
nversi
on.
A
A
A
A
A
A
A
A
A
A
A
A
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
A
S
u
p
p
o
r
t
V
e
c
t
o
r
s
S
u
p
p
o
r
t
V
e
c
t
o
r
s
M
a
r
g
i
n
0
T
w
x
b
1
T
w
x
b
1
T
w
x
b
11
y
C
la
s
s
12
y
C
lass
H
y
p
e
r
p
l
a
n
e
s
Figure
1. O
ptim
u
m
h
yper
-
pla
ne
in
the li
near sepa
rab
le
data
set [
23
]
B
B
B
B
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
B
B
B
B
B
A
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
B
A
B
A
A
A
A
A
A
A
A
A
A
A
A
A
I
n
p
u
t
S
p
a
c
e
F
e
a
t
u
r
e
S
p
a
c
e
O
p
t
i
m
a
l
H
y
p
e
r
p
l
a
n
e
K
e
r
n
e
l
F
u
n
c
t
i
o
n
Figure
2.
Se
parat
ion
of d
at
a t
ha
t cannot
be
li
ne
arly
sep
a
rated
with
ke
rn
el
functi
on in H
DH [
23
]
(
,
)
is
t
he
ke
rn
el
f
unct
ion
a
nd
re
pr
e
sents
th
e
in
ner
pro
duct
in
H
D
H.
Ra
dial
basis
ke
rn
e
l
functi
on
(RBF
)
is
us
ed
to
si
m
pl
ify
the
m
o
del
in
this
stu
dy,
it
s
hig
h
c
om
pu
ta
ti
on
al
eff
ic
ie
ncy
an
d
easy
i
m
ple
m
entat
io
n
a
dv
a
ntage
s [25
]
.
Th
e
m
at
hem
at
ic
al
ex
pr
es
sion o
f
RB
F
is
g
ive
n
in
(8).
(
,
)
=
(
−
‖
−
‖
2
)
(
8
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Determi
ning
dam
ages i
n
cer
amic
p
l
ates
by u
sing discret
e w
avelet
packet
t
ra
nsf
orm
... (
M
ehm
et
Yum
ur
ta
ci
)
4763
Her
e
,
γ
is
the
width
par
am
eter
of
RB
F
ke
r
nel.
The
s
ucce
ss
of
S
VM
cl
assifi
er
de
pends
on
the
widt
h
of
C
a
nd
γ
pa
ram
et
ers.
If
th
e
values
of
C
and
γ
pa
ram
eter
s
are
sel
ect
e
d
la
r
ger
,
t
he
ge
ner
al
iz
at
ion
a
bili
ty
of
the
cl
assifi
e
r
weak
e
ns
sinc
e
there
would
be
extrem
e
le
a
rn
i
ng
on
the
tr
ai
nin
g
data
an
d
the
perf
or
m
ance
in
te
st
data
get
to
be
lowe
r
w
hile
ob
ta
ini
ng
high
su
cce
ss
in
tr
ai
nin
g
data
[
26]
.
The
valu
es
of
C
and
γ
pa
ra
m
et
ers
sh
oul
d
be
i
n
1
0
−
5
,
.
.
.
,
1
0
5
conve
rsion
ra
nge
for
both
of
them
[27].
T
he
final
ve
rsion
of
the
cl
assi
ficat
ion
functi
on
becom
es as in (9
) [
28
].
(
)
=
(
∑
(
,
)
.
+
)
(
9
)
3.
E
X
PERI
MEN
TAL STU
D
Y AND
MEA
SUREME
NT S
Y
STE
M
One
ty
pe
of
plate
was
us
e
d
in
e
xperim
ental
stu
dies.
Im
pact
pe
ndulu
m
was
us
e
d
t
o
a
pp
ly
c
onsta
nt
i
m
pacts
on
th
e
ceram
ic
plates.
Im
pact
Pendulu
m
is
a
p
endulum
m
od
el
dev
el
ope
d
to
pro
duce
i
m
pact
at
a
con
sta
nt
siz
e
[4
,
29]
.
Im
pacts
of
eq
ual
siz
e
wer
e
pro
vid
e
d
by
a
s
m
a
ll
ha
m
m
er
attached
to
i
m
pact
pen
du
l
um
in
a
way
to
not
to
inflic
t
da
m
age
to
cera
m
ic
plate
and
the
sound
generate
d
f
r
om
the
plate
as
a
resu
lt
of
the im
pact was
trans
ferre
d
to
the
data colle
ct
ion sy
ste
m
an
d analy
sed.
PO
E 20
00
ty
pe
i
m
pact p
end
ul
um
was
us
ed
in the i
m
ple
m
e
ntati
on
.
T
he
pe
ndulu
m
was
m
ade to
apply
the
sam
e
i
m
pa
ct
intensit
y
on
ceram
ic
plate
s
from
the
sa
m
e
po
i
nt
[2,
1,
4].
As
a
resu
lt
of
the
i
m
pacts
app
li
ed
to
the
ceram
ic
m
at
erial
,
so
unds,
ge
ner
at
e
d
fr
om
the
plates
with
cracks
in
their
structur
e
a
nd
un
da
m
aged
plate
s,
wer
e
transm
it
te
d
to
data
c
ollec
ti
on
syst
em
wi
th
a
m
ic
ro
ph
on
e
an
d
we
re
then
tra
ns
m
i
tt
ed
to
the
com
pu
te
r
for
data
proce
ssing
sta
ge
from
the
data
colle
ct
ion
syst
e
m
.
One
unda
m
aged
a
nd
9
crack
e
d
ceram
ic
plate
s
wer
e
us
e
d
in
the
stud
y.
All
o
f
the
crac
ks
i
n
ceram
ic
plates
wer
e
di
ff
e
r
ent
from
each
oth
e
r.
The
outp
ut
au
dio
data
of
the
a
m
plifie
r
was
transm
itted
to
the
com
pu
te
r
a
t
a
sa
m
pling
frequ
e
ncy
of
44.
1
kH
z
via Ad
van
te
c
h 171
6L
m
ulti
functi
on
PCI
car
d
a
nd d
at
a a
nal
ysi
s
was
perf
orm
ed
us
i
ng Mat
l
ab
©,
see
Fig
ure
3
.
Figure
3. Sc
he
m
at
ic
r
epr
ese
ntati
on
of the
dat
a acq
uisit
ion
s
yst
e
m
4.
THE
PROP
OSED
CLA
SSIFI
CA
TI
O
N
METH
OD
FOR
DETE
RM
I
NI
NG
DAM
AGE
T
O
THE
C
E
RAM
IC
PL
ATE
The
first
pr
oc
ess
to
be
car
r
ie
d
out
in
det
erm
ining
w
hether
t
he
plate
is
undam
aged
or
c
rac
ked
(d
am
aged
)
by
us
in
g
sou
nd
f
r
equ
e
ncy
rec
orded
f
r
om
a
plate
is
the
infer
ence
of
featu
r
e
vector
that
will
be
giv
e
n
to
the
cl
assifi
er.
Wav
el
et
transfor
m
,
wh
ic
h
is
widel
y
us
ed
in
the
proces
sin
g
of
s
ound
sig
nals
in
recent
ye
ars,
was
m
ade
use
of
.
D
WPT,
wh
ic
h
pro
vi
des
a
detai
le
d
anal
ysi
s
of
both
high
f
reque
nc
y
and
low
-
f
reque
ncy
com
po
ne
nts
of
the
exam
ined
sig
nal,
was
pr
e
ferred
.
12
8
com
po
ne
nts
of
a
so
und
si
gn
a
l
with
D
WPT fr
om
7
th
le
vel
(2
7
)
i
n 1
72, 265
6 Hz
frequ
e
ncy
range
was ob
ta
i
ned
by
u
sin
g d
b20 w
avelet
[18
]
.
The
num
ber
of
data
in
each
le
vel
decr
eas
es
in
half
w
hile
con
duct
in
g
deco
m
po
sit
io
n
by
analy
sis
filt
ers
an
d
th
us
fr
e
quency
al
s
o
decr
eases
in
half.
The
f
requ
ency
ra
nge
of
t
he
c
om
po
ne
nts
at
the
se
ven
t
h
le
vel
was
determ
ined
to
be
bet
wee
n
0
~
2
2
0
5
0
H
z
[
1
7
,
18,
30,
3
1
]
.
Wh
il
e
s
ound
si
gn
al
,
obt
ai
ned
from
the
plate
was give
n
i
n; a
) parts
of Fig
ur
es
4
-
13,
DWP
T co
m
ponen
ts
wer
e
g
i
ven
in,
b) p
a
rts
of
t
he rel
at
ed
fi
gures
.
E
a
c
h
o
f
t
h
e
1
2
8
c
o
m
p
o
n
e
n
t
s
,
w
h
i
c
h
w
e
o
b
t
a
i
n
e
d
f
r
o
m
o
u
r
m
a
i
n
s
i
g
n
a
l
a
n
d
h
a
v
e
d
i
f
f
e
r
e
n
t
f
r
e
q
u
e
n
c
y
r
a
n
g
e
,
c
o
n
s
i
s
t
s
o
f
5
6
5
s
a
m
p
l
e
s
.
W
e
n
e
e
d
m
o
r
e
t
i
m
e
a
n
d
m
e
m
o
r
y
f
o
r
p
r
o
c
e
s
s
i
n
g
.
T
h
e
s
i
g
n
a
l
e
n
e
r
g
y
,
g
i
v
e
n
i
n
(
1
0
)
,
w
a
s
u
s
e
d
f
o
r
r
e
d
u
c
i
n
g
t
h
e
d
a
t
a
s
i
z
e
w
i
t
h
o
u
t
l
o
s
i
n
g
t
h
e
s
i
g
n
a
l
’
s
f
e
a
t
u
r
e
.
=
∑
|
[
]
|
=
1
2
(
1
0
)
In
(10
),
wh
il
e
ind
ic
at
es
a
si
gn
al
,
ind
ic
at
e
s
data
num
ber
and
ind
ic
at
e
s
the
ene
rg
y
va
lue
of
sign
al
[
32
]
.
E
ach
com
ponen
t
will
hav
e
a
da
ta
value
us
i
ng
this
and
t
he
fe
at
ur
e
ve
ct
or
is
reduce
d
to
a
to
ta
l
of
1x128
in
siz
e
at
7
th
le
vel
DW
PT.
It
is
necessary
to
rem
ov
e
featur
es
w
hic
h
ha
ve
li
t
tl
e
eff
ect
or
no
help
within
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
59
-
47
69
4764
the
featu
re
ve
ct
or
f
or
pe
rfo
rm
ing
fast
cl
assifi
cat
ion
pro
cess
an
d
kee
pi
ng
the
ge
neral
iz
at
ion
abili
ty
of
the
cl
assifi
cat
ion
[
33]
.
Wh
e
n
the
feat
ur
e
vecto
r
of
1x
128
in
siz
e
wa
s
analy
sed
in
detai
l,
the
im
pacts
of
the
first
4
c
om
p
on
ents
in
10
plate
s
are
si
m
il
ar
as
seen
in
Fig
ure
14.
The
com
pone
nt
s
with
the
f
re
qu
e
ncy
ranges
of
0
-
12
7,265
6H
z
,
12
7,2
656
-
34
4,531
2H
z
,
34
4,531
2
-
47
1,796
8
Hz
,
471,7
968
-
59
9,0
624
Hz
are
m
or
e
eff
ect
ive
t
ha
n
oth
e
r
c
om
ponen
ts
.
T
he
a
m
pl
it
ud
e
val
ue
of
these
co
m
po
nen
ts
is
gr
eat
er
tha
n
t
he
oth
e
r
com
po
ne
nts.
Ther
e
f
or
e
,
the
featu
re
vecto
r
of
1x4
i
n
siz
e,
co
ns
ist
in
g
of
t
he
e
nergy
val
ues
of
these
com
po
ne
nts,
will
be
ap
plied
as
a
n
in
pu
t
to
the
cl
assif
ie
r.
Th
ere
a
re
two
cat
e
gories
of
plate
s
in
cl
ud
i
ng
undam
aged
a
nd
c
rack
e
d
(
da
m
aged
)
.
Wh
il
e
the
la
bel
of
“
1
”
was
us
e
d
f
or
c
rack
e
d
plate
s,
la
bel
“
2
”
wa
s
us
e
d
for
the
undam
aged
plate
.
Ma
tl
ab©
-
base
d
St
at
ist
ic
al
Patt
er
n
Re
co
gnit
ion
To
olbox
(S
T
P
Rt
oo
l)
s
oft
ware
was
us
e
d
in
the im
ple
m
entat
ion
of
SV
M m
et
ho
d [
34
]
.
(a)
(b)
Figure
4. U
nda
m
aged
plate
: (
a) soun
d
si
gn
al
p
lot
a
nd
(
b)
D
WPT
d
ec
om
posit
ion
s
(a)
(b)
Figure
5. Crac
ked p
la
te
1
: (
a
)
sound si
gnal
pl
ot
an
d
(
b)
D
WPT decom
po
sit
ion
s
(a)
(b)
Figure
6. Crac
ked p
la
te
2
: (
a
)
sound si
gnal
pl
ot
an
d
(
b)
D
WPT decom
po
sit
ion
s
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Determi
ning
dam
ages i
n
cer
amic
p
l
ates
by u
sing discret
e w
avelet
packet
t
ra
nsf
orm
... (
M
ehm
et
Yum
ur
ta
ci
)
4765
Wh
e
n
t
he
gra
phs
betwee
n
F
i
gures
4
-
13
are
exam
ined,
the
crack
m
ark
of
the
ceram
ic
plate
and
it
s
D
WPT
dec
ompo
sit
io
ns
are
giv
e
n.
Her
e,
c
eram
ic
plate
s
with
dif
fere
nt
crack
c
har
act
erist
ic
s
wer
e
chose
n
especial
ly
f
or
a
naly
sis
. F
or
e
xam
ple, the
gr
a
ph
ic
of F
ig
ur
e
11
(
a
)
is ve
ry d
iffe
ren
t f
ro
m
o
ther
gr
a
ph
ic
s
be
cause
the
crack
siz
e
of
the
ceram
ic
plate
is
deep
er.
So
m
e
def
or
m
at
ion
(cr
ac
k)
he
re
can
be
de
sc
ribe
d
as
m
ic
ro
crack
.
These
m
ic
ro
c
r
acks
ca
n
only
be
detect
ed
by
su
c
h
te
sts.
I
n
t
his
stu
dy,
it
is
aim
ed
to
ide
ntify
crac
ke
d
plate
s
an
d
rob
us
t
plate
s
b
y
us
i
ng
D
W
PT
an
d
S
W
M
te
chn
i
qu
e
to
gethe
r.
Her
e
,
the
reas
on
f
or
ch
oo
si
ng
the
S
W
M
te
chn
iq
ue
besi
des
t
he
wav
el
e
t
trans
form
at
io
n
is
due
to
the
fact
that
the
pl
at
es
are
unda
m
aged
a
nd
the
cra
c
k
pro
bab
il
it
y
is
l
i
m
i
te
d
to
two.
The
featu
re
ve
ct
or
s
of
undam
aged
a
nd
d
am
a
ged
plate
s
gr
a
phic
of
this
sel
ect
ion
is give
n
i
n
Fi
gure
14.
(a)
(b)
Figure
7. Crac
ked p
la
te
3
:
(a
)
sound si
gnal
pl
ot and
(
b)
D
WPT decom
po
sit
ion
s
(a)
(b)
Figure
8. Crac
ked p
la
te
4
:
(
a
)
sound si
gnal
pl
ot
an
d
(
b)
D
WPT decom
po
sit
ion
s
(a)
(b)
Figure
9. Crac
ked p
la
te
5
:
(
a
)
sound si
gnal
pl
ot
an
d
(
b)
D
WPT decom
po
sit
ion
s
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
59
-
47
69
4766
Ou
t
of
te
n
pl
at
e
data,
wh
il
e
1
undam
aged
an
d
5
crac
ke
d
(d
am
aged)
plate
data
wer
e
us
e
d
in
the
trai
ni
ng
pr
ocess
of
the
cl
assifi
er,
t
he
re
m
ai
nin
g
4
cra
cked
plate
data
an
d
1
un
da
m
aged
plate
da
ta
were
us
e
d
in
the
te
s
ti
ng
pr
ocess
of
the
cl
assifi
er.
The
values
of
C
and
γ
pa
ram
et
ers
we
re
det
erm
ined
by
tria
l
a
nd
error
m
et
ho
d
a
nd
they
are
1000
a
nd
0.01
re
sp
ect
ively
.
As
a
resu
lt
of
the
te
sti
ng
process
,
5
plate
data
wer
e
cl
assifi
ed
a
nd
100%
su
c
cess
was
obta
ine
d
.
Bl
ock
sc
hem
e
relat
ed
to
t
he
im
ple
m
entat
i
on
of
t
he
s
ug
gested
m
et
ho
d
f
or
the
ide
ntific
at
ion
of
dam
age
a
nd
dura
bili
ty
sta
tuses
i
n
plate
s
was
giv
e
n
in
Fig
ure
15
i
n
de
ta
il
s
accor
ding t
o
th
e
orde
r
of pro
c
ess.
(a)
(b)
Figure
10. C
ra
cked plat
e
6
:
(
a
)
s
ound si
gn
al
plo
t
a
nd
(
b) DWPT
d
ec
om
posit
ion
s
(a)
(b)
Figure
11. C
ra
cked plat
e
7
:
(
a
)
s
ound si
gn
al
plo
t
a
nd
(
b) DWPT
d
ec
om
posit
ion
s
(a)
(b)
Figure
12. C
ra
cked plat
e
8
:
(
a
)
s
ound si
gn
al
plo
t
a
nd
(
b) DWPT
dec
om
posit
ion
s
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N:
20
88
-
8708
Determi
ning
dam
ages i
n
cer
amic
p
l
ates
by u
sing discret
e w
avelet
packet
t
ra
nsf
orm
... (
M
ehm
et
Yum
ur
ta
ci
)
4767
(a)
(b)
Figure
13. C
ra
cked plat
e
9
:
(
a
)
s
ound si
gn
al
plo
t
a
nd
(
b) DWPT
d
ec
om
posit
ion
s
(a)
(b)
Figure
14. Fea
ture
v
ect
or
s
of
undam
aged
a
nd
dam
aged
p
la
t
es
:
(
a)
u
ndam
a
ged p
la
te
(b)
c
r
acked plat
e
D
a
t
a
S
e
t
T
r
a
i
n
i
n
g
S
e
t
T
e
s
t
S
e
t
F
e
a
t
u
r
e
E
x
t
r
a
c
t
i
o
n
(
D
W
P
T
&
S
i
g
n
a
l
E
n
e
r
g
y
)
F
e
a
t
u
r
e
S
e
l
e
c
t
i
o
n
(
S
i
g
n
a
l
A
m
p
l
i
t
u
d
e
)
F
1
F
2
F
3
F
1
2
8
C
l
a
s
s
i
f
i
e
r
S
u
p
p
o
r
t
V
e
c
t
o
r
M
a
c
h
i
n
e
F
1
F
2
F
3
F
4
D
e
t
e
r
m
i
n
e
o
f
C
a
n
d
γ
p
a
r
a
m
e
t
e
r
t
o
c
r
e
a
t
e
S
V
M
M
o
d
e
l
F
e
a
t
u
r
e
E
x
t
r
a
c
t
i
o
n
(
D
W
P
T
&
S
i
g
n
a
l
E
n
e
r
g
y
)
F
e
a
t
u
r
e
S
e
l
e
c
t
i
o
n
(
S
i
g
n
a
l
A
m
p
l
i
t
u
d
e
)
F
1
F
2
F
3
S
V
M
M
o
d
e
l
F
1
F
2
F
3
F
4
D
e
c
i
s
i
o
n
C
l
a
s
s
/
L
a
b
e
l
F
1
2
8
.
.
.
.
.
.
.
.
.
.
Figure
15. Blo
ck
sc
hem
a
of
t
he
m
et
ho
d su
ggest
ed
for plat
e cla
ssific
at
ion
5.
RESU
LT
S
AND DI
SCUS
S
ION
S
The
te
ch
nique
of
a
naly
sing
t
he
so
un
d
com
ing
out
of
the
m
at
erial
s
by
hitt
ing
the
ce
ram
i
c
m
at
erials
giv
es
ve
ry
su
c
cessf
ul
res
ults
in
pract
ic
e
[
1
-
4,
35
-
38
]
.
Thi
s
te
chn
i
qu
e
ha
s
bee
n
us
ed
by
sci
entist
s
for
a
lo
ng
tim
e.
In
t
his
st
ud
y
,
D
WPT
a
nd
S
VM
m
et
hods
a
re
us
e
d
t
oget
her
t
o
determ
ine
the
de
f
or
m
at
ion
of
the
ce
ram
i
c
plate
s.
S
hort
-
T
i
m
e
Fo
ur
ie
r
Tr
ansfo
rm
ation
(
STFT
)
[
4,
35
]
,
Con
ti
nu
ou
s
Wav
el
et
Tr
ans
form
ation
(C
WT)
[
1]
m
et
ho
ds f
or d
e
te
ct
ing
def
or
m
at
ion
s i
n
cer
am
ic
p
la
te
s h
a
ve been
used
in pr
evio
us
st
ud
ie
s
[1,
3,
4].
In
this
stu
dy,
the
sign
al
pr
ocessin
g
te
ch
nique
an
d
arti
fici
al
intel
li
ge
nce
te
ch
nique
wer
e
use
d
tog
et
he
r.
W
it
h
this
m
et
ho
d,
m
or
e
su
ccessf
ul
resu
lt
s
are
obta
ined
i
n
te
rm
s
of
com
par
ison
a
nd
cl
assif
ic
at
ion
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
59
-
47
69
4768
Althou
gh
the
ANN
m
et
ho
d
was
us
e
d
for
c
la
ssific
at
ion
i
n
previ
ou
s
stu
di
es
[
18
]
,
the
S
VM
te
ch
nique
us
e
d
in
this
stud
y
is
m
uch
m
or
e
su
cc
essfu
l
in
the
s
olid
-
c
rack
dete
ct
ion
sect
ion.
I
n
this
co
ntext,
the
m
e
tho
d
is
strong
and sati
sfacto
r
y i
n
te
rm
s o
f b
oth
a
naly
sis an
d
cl
assifi
cat
io
n.
6.
CONCL
US
I
O
N
A
D
WPT
a
nd
SV
M
ba
sed
a
na
ly
sis
m
e
tho
d
was
s
uggested
in
this
stu
dy
for
determ
ining
cracke
d
a
nd
undam
aged
plate
s.
Sou
nds sig
nals of
undam
aged an
d
c
rack
e
d plat
es w
e
re c
reated
by an i
m
pact p
endulu
m
an
d
the
12
8
de
co
m
po
sit
ion
s
of
these
s
ound
si
gn
al
s
with
D
WPT
was
obta
ined
for
featu
re
ext
racti
on.
The
first
4
com
po
ne
nts,
wh
ic
h
ref
le
ct
the
char
act
eri
sti
cs
of
plate
s
a
m
on
g
t
hese
128
c
om
po
ne
nt
s,
wer
e
sel
ect
e
d
f
or
enh
a
ncin
g
the
perform
ance
of
the
cl
assifi
er
and
e
nergy
val
ues
we
re
use
d
as
featur
e
vect
or
s
.
SV
M
m
odel
was
c
reated b
y
sel
ect
ing
ap
pro
pr
ia
te
C
and
γ p
ara
m
et
ers
fo
r
t
he
cl
assifi
er
an
d
undam
aged
a
nd
crack
e
d
plate
s
we
r
e
ob
s
er
ved to
be i
den
ti
fied
s
ucc
essfu
ll
y afte
r
tr
ai
nin
g an
d
te
sti
ng stages
.
On
ly
the “
unda
m
aged
” a
nd
“c
rack
e
d”
sta
tuse
s o
f
the
plate
s w
ere
determ
ined
in t
his stu
dy. Mo
re
d
a
t
a
on
plate
s
a
re
ne
eded
f
or
the
cl
assifi
cat
ion
of
the
dam
ages
(cr
ac
ks
)
wit
hin
them
sel
ves.
It
can
be
possi
bl
e
to
determ
ine
the
dam
ages
of
t
he
plate
s
by
t
his
m
et
ho
d
a
f
te
r
their
pro
duct
ion
sta
ge
a
nd
be
f
or
e
t
he
y
reach
the
end
-
us
er
.
The
m
ajo
r
ad
va
ntage
of
this
m
et
ho
d
is
it
s
l
ow
-
c
os
t
hard
w
are
requirem
ent.
Re
al
-
tim
e
dam
age
assessm
ent can
be pe
rfor
m
ed by usi
ng a
n
a
ppr
opriat
e m
ic
r
ophone
, im
pact p
en
dulum
and c
om
pu
te
r.
REFERE
NCE
S
[1]
O.
Akgun,
e
t
a
l
.
,
“
The
Defe
ct
De
te
c
ti
on
in
Cera
m
ic
Mat
erials
Bas
ed
on
W
avele
t
Anal
y
sis
b
y
Us
i
ng
the
Me
thod
o
f
Im
pulse
Noise,
”
JV
E
Journal
of Vi
broengin
ee
rin
g
,
vol
.
15
,
no
.
2
,
pp.
818
-
825
,
20
13.
[2]
T.
C.
Akinci,
e
t
al
.
,
“
Applic
at
i
on
of
Artifi
ci
al
Neura
l
Networks
for
Defe
ct
Dete
c
ti
on
in
Ce
ramic
Mate
ri
al
s,
”
Archi
v
es
of
Ac
o
ustic
s
,
vo
l.
37
,
n
o.
3
,
pp
.
279
-
28
6
,
2012
.
[3]
T.
C
.
Akin
ci,
e
t
al
.
,
“
Spect
ra
l
a
nd
Stat
ist
ical
A
naly
s
is
for
C
eram
ic
Plate
Spe
cim
ens
Subjec
te
d
to
Im
pac
t
,
”
Solid
Stat
e
Phe
nom
en
a
,
vol
.
199
,
pp
.
6
21
-
626
,
2013
.
[4]
T.
C
.
Akin
ci
,
“
T
he
Defe
ct
detec
tion
in
C
era
m
ic Mat
erials
Based
on
Ti
m
e
-
Freque
nc
y
Anal
y
sis
b
y
Us
ing
the
Me
tho
d
of
Im
pulse
Noise,
”
Arch
iv
es
of A
cousti
cs
,
vol
.
36
,
no.
1
,
pp.
1
–
9
,
2
011
.
[5]
A.
V.
Dros
dov,
“
Inve
stiga
ti
on
o
n
the
Mi
cro
cr
acking
of
Cer
amic
Mate
ri
al
s
Us
ing
the
Acoust
ic
E
m
ission
Method,
”
Strengt
h
o
f
Mat
e
rials
,
vol
.
46
,
no
.
1
,
pp
.
71
-
79
,
2
014
.
[6]
S.
J.
Kow
al
ski,
a
nd
J.
Szadz
inska
,
“
Non
-
Stat
iona
r
y
Dr
y
ing
of
Ce
r
amic
-
Li
k
e
Mater
ia
ls
Control
le
d
t
hrough
Acoustic
Emiss
ion
Metho
d,
”
H
eat
Mass
T
rans
fe
r
,
vol.
48
,
no.
12
,
pp
.
2023
-
2032
,
2012
.
[7]
E.
Ma
ire
and
P.
J.
W
it
her
s
,
“
Quanti
t
at
iv
e
X
-
r
a
y
Tomograph
y
,
”
Inte
rnat
ional
Mate
rials
Re
v
iews
,
vol.
59
,
no.
1
,
pp.
1
-
43
,
2014
.
[8]
F.
Yin
,
et
al
.
,
“
Methods
for
Dete
c
ti
on
of
Subs
urfa
ce
Dam
age
:
A
Revi
ew,
”
Chini
se
Journal
of
Me
chani
c
al
Engi
ne
ering
,
vol
.
31
,
no
.
41
,
pp
.
1
-
14
,
2018
.
[
9
]
R
.
B
o
h
m
,
e
t
a
l
.
,
“
A
Q
u
a
n
t
i
t
a
t
i
v
e
C
o
m
p
a
r
i
s
o
n
o
f
t
h
e
C
a
p
a
b
i
l
i
t
i
e
s
o
f
i
n
S
i
t
u
C
o
m
p
u
t
e
d
T
o
m
o
g
r
a
p
h
y
a
n
d
C
o
n
v
e
n
t
i
o
n
a
l
C
o
m
p
u
t
e
d
T
o
m
o
g
r
a
p
h
y
f
o
r
D
a
m
a
g
e
A
n
a
l
ys
i
s
o
f
C
o
m
p
o
s
i
t
e
s
,
”
C
o
m
p
o
s
i
t
e
s
S
c
i
e
n
c
e
a
n
d
T
e
c
h
n
o
l
o
g
y
,
v
o
l
.
1
1
0
,
n
o
.
6
,
p
p
.
62
-
68
,
2
0
1
5
.
[10]
K.
Ehri
g
,
et
al
.
,
“
Com
par
is
on
o
f
Crac
k
Detect
i
on
Methods
for
Anal
y
zi
ng
Da
m
age
Proce
ss
es
in
Concre
t
e
with
Com
pute
d
Tomograph
y
,
”
In
te
rn
ati
onal
Symposi
um
on
Digit
al
Industrial
Radi
ol
ogy
and
Computed
Tom
ography
,
Berl
in
,
Germ
an
y
,
pp
.
1
-
8
,
2011
.
[11]
A.
Mohan
and
S.
Poobal,
“
Crac
k
Dete
ction
Us
ing
Im
age
Proce
ss
ing:
A
Crit
ical
R
evi
ew
and
Anal
ysis,”
Al
e
xsandri
a
Engi
ne
ering
Jou
rnal
,
vol
.
57
,
no
.
2,
pp.
787
-
798
,
2018
.
[12]
A.
Y.
Sun
and
B.
F.
Ju.
“
The
Acoustic
Micro
Inte
gra
t
ed
Detect
ion
T
ec
hniqu
e
for
Sili
con
W
af
er
Proce
ss
ing,
”
Adv
anc
ed
Ma
te
r
ial
s R
ese
arch
,
v
ol.
497
,
pp
.
151
–
155
,
2012
.
[13]
J.
Kim
,
and
P.
K
.
L
ia
w,
“
The
No
ndestruc
t
ive
Evaluat
ion
of
Adva
nce
d
C
era
m
ic
s
a
nd
Cera
m
ic
-
m
a
t
rix
Com
posite
s,”
The
Me
mber
Jou
rnal
of
The
Min
e
rals,
Me
ta
ls
&
Mate
rials So
ci
e
t
y
(
JOM)
,
vol
.
50
,
no
.
11
,
pp
.
1
-
15
,
1998
.
[14]
R.
Marti
n
,
e
t
a
l
.
,
“
Inte
rpr
et
ing
the
Result
s
of
Puls
ed
The
rmogra
ph
y
Da
ta
,
”
Mate
rials
Ev
a
luat
ion
,
vol
.
61
,
pp.
611
-
616
,
20
03
.
[15]
J.
V
y
as,
and
R.
J.
Kaz
y
s
,
“
A
Rev
ie
w
on
Nondestr
uct
iv
e
T
ec
hniqu
es
and
Ch
aract
er
isti
cs
of
Com
posite
M
at
er
ia
ls
f
o
r
the
Aerospa
ce Sy
stem,
”
M
ATEC
We
b
o
f
Con
fe
re
nce
s
,
vo
l.
233
,
n
o.
3
,
pp
.
1
-
8
,
20
18
.
[16]
F.
Janicek,
et
a
l
.
,
“
A
New
Protection
R
el
a
y
B
ase
d
on
a
Faul
t
Tra
nsi
ent
Ana
l
y
sis
Us
ing
W
av
el
e
t
Tr
ansform
,
”
Journal
of
Elec
t
rical
Engi
ne
erin
g
,
vol
.
58
,
no
.
5
,
pp.
271
-
278
,
20
07
.
[17]
G.
Gokm
en,
“
W
ave
le
t
B
ase
d
Refe
ren
ce
Cu
rre
nt
Ca
lc
ul
at
io
n
Method
for
Acti
ve
Com
pen
sati
on
S
y
st
ems
,
”
El
e
ct
ronics
and
El
e
ct
rica
l
Eng
in
ee
r
ing
,
vol
.
108
,
no.
2
,
pp.
61
-
66
,
2011
.
[18]
G.
Gokm
en,
“
Th
e
Defe
c
t
Detect
i
on
in
Glass
Mate
rials
b
y
Us
ing
Discre
te
W
avele
t
Packe
t
T
ran
sfo
rm
and
Artifi
ci
a
l
Neura
l
N
et
work,
”
J
VE J
ournal
of
Vi
broeng
ine
erin
g
,
vol
.
16
,
no
.
3
,
pp.
1434
-
1443
,
2014
.
[19]
H.
Deng,
H.
Li
ng,
“
Fast
soluti
on
of
El
e
ct
r
om
agne
ti
c
Int
e
gra
l
Equations
Us
ing
Adapti
ve
W
ave
le
t
Pack
et
Tra
nsform
,
”
I
EEE
Tr
ansacti
on
o
n
Antennas and Propagation
,
vo
l.
47
,
no
.
4
,
pp
.
6
74
-
682
,
1999
.
[20]
Y.
Li
u,
et
al
.
,
“
Fault
Diagnosis
of
Analog
Circ
u
it
Based
on
Sup
port
Vec
tor
Ma
chi
nes,
”
Proc
eedings
of
ICCT
A
,
pp.
40
-
43
,
2009
.
[21]
P
G.
V.
Axelbe
rg,
e
t
al
.
,
“
Support
Vec
tor
Mac
hi
ne
for
Cla
ss
ifi
cation
of
Volta
ge
Disturbanc
es,
”
IEE
E
Tr
ansacti
o
ns
on
Powe
r D
el
i
ver
y
,
vol
.
22
,
no
.
3
,
pp
.
1297
-
1303
,
2007
.
[
2
2
]
P
.
K
.
D
a
s
h
,
e
t
a
l
.
,
“
F
a
u
l
t
C
l
a
s
s
i
f
i
c
a
t
i
o
n
a
n
d
S
e
l
e
c
t
i
o
n
I
d
e
n
t
i
f
i
c
a
t
i
o
n
o
f
a
n
A
d
v
a
n
c
e
d
S
e
r
i
e
s
-
C
o
m
p
e
n
s
a
t
e
d
T
r
a
n
s
m
i
s
s
i
o
n
L
i
n
e
U
s
i
n
g
S
u
p
p
o
r
t
V
e
c
t
o
r
M
a
c
h
i
n
e
,
”
I
E
E
E
T
r
a
n
s
a
c
t
i
o
n
s
o
n
P
o
w
e
r
D
e
l
i
v
e
r
y
,
v
o
l
.
22
,
n
o
.
1
,
p
p
.
67
-
73
,
2
0
0
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.