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.
9
, No
.
3
,
J
un
e
201
9
, pp.
1669
~
16
75
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v9
i
3
.
pp1669
-
16
75
1669
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Analysis
of ha
rmonics usi
ng wavel
et techn
iqu
e
Than
gara
j.
K
1
,
Subr
am
an
i
am
.N
.
P
2
,
N
arm
ada.
R
3
,
Om
a Magesw
ari.
M
4
1,3
EE
E
,
SM
VEC,
Indi
a
2
EE
E
,
PEC
,
Indi
a
4
ICE,
SM
VEC,
I
ndia
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
ug
11, 201
8
Re
vised N
ov 20, 2
018
Accepte
d Dec
11, 201
8
Thi
s
pape
r
d
ev
el
ops
an
appr
o
ac
h
base
d
on
wave
let
techniq
ue
for
th
e
esti
m
at
ion
of
h
armonic
pre
sen
t
s
in
power
s
y
st
em
signal
s.
Th
e
proposed
te
chn
ique
divi
d
es
the
power
s
y
stem
s
igna
ls
int
o
diffe
ren
t
fre
quen
c
y
sub
-
bands
cor
re
sponding
to
the
odd
har
m
onic
components
of
the
signal.
The
al
gor
it
hm
hel
ps
to
det
ermin
e
both
the
ti
m
e
and
fre
quen
c
y
i
nform
at
ion
from
the
har
m
onic
fre
qu
ency
b
ands.
The
compara
t
ive
stud
y
wi
ll
be
do
n
e
with
th
e
inpu
t
a
nd
the
result
s
attai
n
ed
from
th
e
wave
let
tr
ansfor
m
(W
T)
for
diffe
ren
t condi
t
i
ons a
nd
Sim
ula
t
i
on
result
s
are given.
Ke
yw
or
d
s
:
Ele
ct
ric powe
r qu
al
it
y
Har
m
on
ic
distor
ti
on
Mult
iresolutio
n
a
naly
sis
Sign
al
a
nd
no
i
se
Wav
el
et
s
Copyright
©
201
9
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
:
Tha
ng
a
raj.
K
,
EEE,
SMVEC,
Pudu
c
he
rr
y
, I
ndia
.
Em
a
il
:
thang
ar
aj.
el
ect
res
@gm
ai
l.co
m
1.
INTROD
U
CTION
Nowa
days
i
m
po
rta
nce
of
ha
r
m
on
ic
stud
ie
s
play
s
a
m
ajo
r
r
ole
in
Powe
r
s
yst
e
m
netwo
r
ks.
The
powe
r
qu
al
it
y
disturbance
ca
n
be
caused
var
ia
ti
on
in
el
ect
rical
po
we
r
ser
vice,
su
c
h
as
har
m
on
ic
s,
volt
ag
e
os
ci
ll
at
ion
s,
quic
k
disruptio
ns
an
d
tra
ns
i
ents
w
hich
r
e
su
lt
s
in
fa
ulty
op
erati
on
or
fail
ur
e
of
con
ce
r
n
equ
i
pm
ent.
Th
e
har
m
on
ic
presents
in
pow
er
syst
em
dist
or
ts
th
e
volt
ag
e
an
d
cu
rr
e
nt
sign
al
wh
ic
h
i
n
tu
rn
create
s
va
rio
us
pro
blem
s.
Con
ve
ntio
nally
,
f
or
har
m
on
ic
a
naly
sis,
the
discrete
Fou
rier
trans
form
(D
F
T)
is
su
gges
te
d
w
hi
ch
pr
ov
i
des
f
reque
ncy
inf
orm
ation
of
th
e
sign
al
,
but
it
will
no
t
gi
ve
tim
e
infor
m
at
ion
require
d [1
]
. T
her
e
fore,
DFT
is n
ot a s
uitabl
e f
or
non
-
sta
ti
onary si
gn
al
.
Power
qual
it
y
can
be
im
pr
oved
by
intr
oduci
ng
W
a
velet
te
chn
i
qu
e
f
or
ha
r
m
on
ic
stud
ie
s
t
o
ov
e
rco
m
e
the
disad
van
ta
ges
i
n
th
e
c
onve
ntion
al
m
et
ho
ds
.
W
a
velet
s
a
re
a
set
of
f
un
c
ti
on
s
w
hich
ca
n
be
us
e
d
e
ff
ec
ti
vely
to r
e
pr
es
ent
na
tural,
highly
tr
ansient
phen
om
ena th
at
r
es
ul
t from
a d
il
at
ion
a
nd sh
i
ft of t
he or
igi
nal w
a
vefor
m
.
To
a
naly
ze
no
n
-
sta
ti
onary
si
gn
al
s
,
the
W
a
ve
le
t
Techn
i
qu
e
is
an
ef
fici
ent sign
al
p
r
ocessi
ng
to
ol
a
nd
al
so
it
ha
s
wide va
riet
y of ap
plica
ti
on
s
[
2].
In
po
wer
qual
it
y
analy
sis
us
ing
w
avelet
te
chn
i
qu
e
,
the
sig
nal
will
be
com
par
ed
to
wavel
et
fu
nctio
n,
and
a
set
of
c
oeffici
ents
w
hi
ch
is
obta
ine
d
giv
es
i
nfor
m
at
ion
of
c
orrelat
ion
of
wav
el
e
t
functi
on
with
the
sign
al
.
Wa
vel
et
Tran
s
f
or
m
(
W
T
)
pro
vid
e
s
good
ti
m
e
reso
luti
on
a
nd
poor
f
reque
nc
y
reso
l
ution
a
t
high
fr
e
qu
e
ncies
a
nd
good
fr
e
quency
res
olu
ti
on
a
nd
po
or
ti
m
e
reso
luti
on
at
low
f
requ
encies
.
It
is
us
ef
ul
sp
eci
fical
ly
w
hen
the
si
gn
a
l
has
high
f
r
equ
e
ncy
c
ompone
nts
f
or
s
hort
du
rati
on
s
an
d
lo
w
f
re
qu
e
ncy
com
po
ne
nts
f
or
long
du
rati
ons.
Fi
nally
,
this
pap
e
r
com
par
es
the
pe
rform
ance
of
the
r
esults
ob
ta
i
ned
us
in
g
pro
po
se
d wa
ve
le
t t
ran
sf
or
m
(WT)
for va
rio
us co
ndit
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.
9
, N
o.
3
,
June
2019
:
1669
-
1675
1670
2.
WA
VELE
TS
Wav
el
et
s
are
os
ci
ll
at
ing
wa
vefor
m
s
of
short
du
rati
on
w
hich
am
plit
ud
e
decays
quic
kl
y
to
zero
at
bo
t
h
en
ds
.
I
n
WT,
the
wa
vel
et
is
dilat
ed
and
sh
ifte
d
to
va
ry
the
fr
eq
ue
nc
y
of
os
ci
ll
at
ion
an
d
tim
e
locat
ion
.
These
dilat
ing
and
sh
i
fting
m
echan
ism
s
are
ver
y
im
po
rta
nt
to
a
naly
ze
non
-
sta
ti
on
a
ry
sign
al
s
c
om
par
ed
t
o
conve
ntion
al
m
et
ho
ds
s
uc
h
as
discrete
F
ourier
t
ran
s
f
orm
(D
FT)
a
nd
sh
ort
tim
e
Fo
uri
er
tra
ns
f
or
m
(S
TF
T).
Wav
el
et
te
ch
ni
qu
e
a
naly
ses
the
sign
al
at
diff
e
re
nt
fr
e
quencies
with
dif
fe
ren
t
res
olu
ti
on
s
.
W
a
velet
s
hav
e
i
m
po
rtant
pro
pe
rtie
s su
it
able
f
or analy
sis
of
non
-
sta
ti
on
a
ry
wav
e
f
or
m
s.
The
filt
erin
g
process
s
how
n
in
Fig
ur
e
1
is
the
m
e
tho
d
us
e
d
in
m
os
t
of
t
he
disc
re
te
wav
el
et
trans
form
s
(D
WT)
a
nd
the
f
irst
com
po
nen
t
to
m
ulti
reso
luti
on
analy
sis
is
vector
s
pace
s
[3
]
.
The
n
f
or
ever
y
vecto
r
sp
ace
, th
ere
will
b
e h
i
gh
e
r
re
so
l
ution vecto
r
sp
ace t
i
ll
you
g
et
to
th
e require
d
sig
na
l. Also, eve
ry v
ect
or
sp
ace
c
om
pr
ise
s
al
l
vecto
r
s
pa
ces
with
lo
we
r
res
ol
ution.
T
he
basis
of
eac
h
of
t
hese
vect
or
sp
a
ce
s
is
th
e
scal
e
functi
on f
or th
e w
avelet
and sign
ifie
s the
de
ta
il
ed
ver
sio
n of
the
high
-
fr
e
qu
e
ncy com
ponen
ts
of
the signal
an
d
the
ap
pro
xim
a
ti
on
ver
si
on
of
the
l
ow
-
f
re
quency
c
om
po
ne
nts
a
nd
sim
ilarly
the
rec
on
structio
n
proc
ess
of
wav
el
et
tra
nsfo
rm
w
hich
is
show
n
i
n
Fi
gure
2.
The
lo
w
pass
filt
ering
,
A
el
i
m
inate
s
the
hig
h
f
re
qu
e
nc
y
info
r
m
at
ion
an
d
high
pa
ss
filt
ering,
D
el
i
m
inate
s
low
f
re
qu
e
ncy
inform
at
ion
bu
t
scal
e
rem
ai
ns
sam
e.
Du
r
i
ng
th
e
pro
cess
of
s
ubsa
m
pl
ing
,
the
scal
e
will
get
af
fected
.
Du
e
to
filt
erin
g
op
e
rati
ons,
the
Re
s
olu
ti
on
wh
ic
h
is
relat
ed
to
the
qua
nt
it
y
of
inf
or
m
at
ion
in t
he
sig
nal als
o gets af
fected
.
Figure
1. Fil
te
ring
proces
s
Figure
2.
W
a
ve
le
t reco
ns
tr
uction
The
us
e
of
Hal
f
ba
nd
lo
w
pa
s
s
filt
er
rem
ov
e
s
half
of
t
he
f
r
equ
e
ncies
a
nd
because
of
that
half
of
t
he
inf
or
m
at
ion
re
gardin
g
si
gn
al
will
be
lo
os
e
d.
T
he
refor
e
,
on
ce
the
opera
ti
on
of
filt
er
is
com
plete
d
th
en
th
e
reso
l
ution
is
halve
d
[4
]
.
H
ow
e
ve
r,
after
filt
ering
pr
oce
ss
the
su
bs
am
pling
op
e
rati
on
will
no
t
affe
ct
the
reso
l
ution,
be
ca
us
e
a
nyway
it
re
m
ov
es
ha
lf
of
t
he
sa
m
ples
in
the
sign
al
s
withou
t
losing
i
nform
at
ion
.
The
aut
hors
(i
n
[5
]
)
pro
pose
a
m
et
ho
d
to
co
m
pen
sat
e
the
def
ect
ive
res
ponse
of
the
filt
ers
us
ed
in
the w
a
velet
-
trans
form
fil
te
r
banks.
T
he
ne
w
i
m
pr
ove
d
ap
proac
h
W
a
velet
Tra
ns
f
orm
(
WT)
was
im
pl
e
m
ented
to
overco
m
e
the
disad
van
ta
ges
of
c
onve
ntion
al
m
et
ho
ds.
I
n
t
he
WT,
th
e
detai
ls
ar
e
f
ur
t
her
dec
om
po
se
d
to
produc
e
ne
w
coeffic
ie
nts,
t
hi
s w
ay
e
nab
li
ng a
fr
e
quency
deco
m
po
sit
io
n of t
he
i
nput sig
nal to be
obta
ined
.
3.
PROP
OSE
D
ALGO
RITH
M
The
al
gorithm
pro
posed
in
this
pap
e
r
is
w
avelet
tran
sf
orm
(W
T
)
wh
ic
h
is
c
om
patible
with
th
e
fr
e
qu
e
ncy
ba
nds
of
the d
iffe
r
ent
har
m
on
ic
gro
ups
an
d
use
s
the
Daubec
hie
s
20
as
t
he
wa
velet
functi
on
and
t
he
filt
er
ba
nk
with
th
ree
le
vels
of
deco
m
po
sit
io
n
s
how
n
i
n
Fi
gure
3.
T
he
sam
pling
f
reque
nc
y
sel
ect
ed
is
1.6
kH
z
with
f
undam
ental
fr
e
qu
e
ncy
of
50
Hz
.
Th
e
deco
m
po
sit
ion
proces
s
ca
n
be
it
erate
d,
so
that
on
e
si
gn
al
is
bro
ken
do
wn
into
m
any
lo
wer
-
res
olu
ti
on
com
ponen
ts
and
hi
gher
-
res
olu
ti
on
c
om
ponen
ts
res
pecti
vely
as
sh
ow
n
in
Fi
gur
e
3
an
d
the
ou
t
pu
t
f
re
qu
e
ncy
bands
of
wav
el
et
transfor
m
sh
own
in
Fi
gure
4.
T
he
outp
ut
of
the
filt
er b
an
k
is divid
e
d
into fre
quency ba
nds (
c
oeffici
ents o
f d
1
to d4)
wh
ic
h
offer
s in
f
or
m
ation
abo
ut h
arm
on
ic
gro
up
s
pr
ese
nt
s
in
the
inp
ut
sign
al
[
6]
-
[
7].
The
flo
wc
har
t
fo
r
the
proc
e
ss
of
wa
velet
trans
form
s
wh
ic
h
is
sh
ow
n
in
Fi
gur
e
5.
Each
c
oeffici
ent
of
wa
velet
trans
form
m
ea
su
res
the
co
rrel
at
ion
be
twee
n
the
sig
nal
a
nd
the
basi
s
functi
on.
I
f
ti
t
is
la
rg
e
coe
ff
ic
ie
nts
de
no
te
good
c
orrelat
ion;
on
the
oth
e
r
hand
sm
al
l
coef
fici
ents
denot
e
poor
correla
ti
on.
Aft
er
analy
zi
ng
t
he
pr
ese
nts
of
ha
rm
on
ic
s
in
the
f
reque
ncy
su
b
-
bands,
the
su
it
able
m
it
ig
at
io
n
m
et
ho
d
will
be
ap
plied
t
o
tho
se
s
ub
-
ba
nd
s
with
out
aff
ect
in
g
the
c
oeff
ic
ie
nts
w
hi
ch
co
ntains
or
i
gin
al
sign
al
[
10
]
.
S
o
accor
ding
to
the
thres
hold
va
lue
set
,
the
w
avelet
te
chn
iq
ue
el
i
m
inate
s
t
he
lowe
r
m
agn
it
ud
e
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
An
alysis of
har
monics
us
in
g w
avelet
techn
i
qu
e
(
Tha
ngara
j.
K
)
1671
coeffic
ie
nts
re
ta
ins
on
ly
th
e
sign
ific
ant
coeffic
ie
nts.
Af
te
r
al
te
rin
g
the
coef
fici
e
nts,
the
dec
om
po
sed
com
po
ne
nts
will
b
e assem
bled back
to get t
he ori
gin
al
s
i
gnal
w
it
ho
ut los
s of inf
or
m
at
io
n.
The
filt
er
us
e
d
in
re
co
ns
tr
uction
pro
cess
pla
ys
m
ajo
r
r
ole
i
n
at
ta
inin
g
or
i
gin
al
si
gn
al
wi
thout
loss
of
inf
or
m
at
ion
[6]
.
The
rec
onstructe
d
detai
ls
and
a
ppr
oxim
a
ti
on
s
are
t
ru
e
const
it
uen
ts
of
the
or
i
gin
al
s
ign
al
.
The
RM
S
m
a
gn
it
ude
of
the
sign
al
s
cab
be
cal
culat
ed
by
us
ing
the
s
quare
r
oot
of
th
e
m
ean
sq
uar
e
of
th
e
wav
el
et
co
ef
fici
ents.
D
ur
i
ng
dow
ns
am
pling
process
of
t
he
sign
al
c
om
po
nen
ts
t
her
e
is
chan
ce
of
a
di
stortio
n
cal
le
d
al
ia
sing
.
It
can
be
el
i
m
inate
d
by
carefu
ll
y
sel
ec
ti
ng
filt
ers
fo
r
the
deco
m
po
sit
ion
a
nd
reconstr
uctio
n process
.
Actuall
y,
it
is
ver
y
m
uch
ess
entia
l
to
get
m
axim
u
m
flat
passb
a
nd
c
ha
racteri
sti
cs
and
go
od
freq
ue
ncy
separ
at
io
n.
Ac
cordin
g
t
o
[9
]
,
wav
el
et
functi
on
s
with
a
la
rge
num
ber
of
c
oe
ff
ic
ie
nts
will
hav
e
le
sser
distor
ti
on
wh
e
n
com
par
e
d
to
lowe
r
coe
ff
ic
ie
nts.
T
he
Daubec
hies
w
avelet
is
a
go
od
proces
sin
g
too
l
f
or
powe
r
-
qu
al
it
y
m
on
it
or
ing
in
the
powe
r
syst
e
m
[8
]
.
In
order
to
m
easur
e
hi
gh
e
r
ra
ng
e
of
har
m
on
ic
orde
rs
(greate
r
t
ha
n
15
t
h
order),
the
sam
pling
fr
e
qu
enc
y
and
le
vel
of
the
dec
om
po
sit
ion
will
be
inc
reased
in
Fi
gur
e
5
accor
ding
to
the
har
m
on
ic
c
ondi
ti
on
s [7].
Figure
3. Th
re
e level
w
a
velet
d
ec
om
po
sit
ion t
ree
Figure
4. O
utput f
reque
ncy ba
nds
of
wa
velet
deco
m
po
sit
io
n
Figure
5. Flo
w
char
t
for WT
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.
9
, N
o.
3
,
June
2019
:
1669
-
1675
1672
4.
SIMULATI
O
N RESULTS
The
W
a
velet
T
ran
s
f
or
m
(W
T)
te
chn
iq
ue
f
or
analy
zi
ng
the
har
m
on
ic
s
was
i
m
ple
m
ented
by
us
in
g
th
e
so
ft
war
e
pac
ka
ge
of
M
ATL
A
B.
I
n
this
sect
ion,
a
c
om
par
at
ive
analy
sis
will
be
do
ne
with
the
in
pu
t
a
nd
th
e
resu
lt
s
obta
ine
d from
the w
av
el
et
tran
sf
or
m
(W
T
) for
dif
fere
nt m
easur
in
g condit
ions.
4.1
.
Te
s
t
sig
n
al 1
Con
si
der
the
i
nput
sig
nal
show
n
in
Fi
gure
6
w
hich
c
onta
ins
se
ven
t
h
ha
r
m
on
ic
co
m
po
ne
nt
at
ever
y
tim
e
instant
in
fun
dam
ental
com
po
ne
nt
s
ign
al
of
50
Hz.
By
us
in
g
the
a
naly
zi
ng
in
f
or
m
at
ion
fr
om
deco
m
po
se
d
si
gn
al
s
,
the
m
i
tig
at
ion
te
ch
niques
C
onsider
t
he
in
put
sig
nal
sho
wn
in
Fig
ur
e
6
w
hich
c
on
ta
in
s
seve
nth
har
m
on
ic
com
ponen
t
at
ever
y
ti
m
e
instant
in
f
undam
ental
com
po
ne
nt
sig
nal
of
50
H
z.
By
us
ing
th
e
analy
zi
ng
in
form
ation
from
deco
m
po
se
d
sign
al
s
,
the
m
it
i
gation
te
ch
niques
will
be
ap
plied.
I
f
m
it
ig
at
ion
te
chn
iq
ues
a
ppli
ed
us
in
g
that
inf
or
m
at
ion
,
then
th
e
outp
ut
sign
al
sho
wn
in
Figu
r
e
6
is
obta
ined
with
er
r
or
of
0.534
2%
w
hic
h
is
sh
ow
n
in
Table
1.
T
he
locat
ion
of
di
sturbance
sta
rti
ng
an
d
e
nd
i
ng
tim
e
are
sh
own
in
Table
2
with
pe
rcen
ta
ge
de
viati
on
an
d
from
the
info
rm
at
i
on,
the
a
m
plitu
de
of
the
gi
ve
n
distu
rb
a
nce
sign
al
are
ob
se
r
ved.
Figure
6. Ha
rm
on
ic
s
at eve
ry t
i
m
e instant
Table
1.
E
rro
r
Com
par
ison f
or D
i
ff
e
ren
t
Ha
r
m
on
ic
Conditi
on
s
RMS
v
alu
e of
r
ef
e
rence inp
u
t=0
.70
7
1
Test sig
n
al
s
RMS
v
alu
e
o
f
input
RMS
v
alu
e
o
f
ou
t
p
u
t
Er
ror
(
%)
Co
m
p
.
ti
m
e
(sec)
Har
m
o
n
ics
at ever
y
ti
m
e ins
tan
t
0
.86
3
1
0
.70
3
3
0
.53
4
2
0
.75
6
3
Har
m
o
n
ics
at ever
y
ti
m
e ins
tan
t wi
th
dif
f
erent f
requ
en
c
ies
1
.01
4
9
0
.71
5
0
1
.12
0
2
0
.77
4
8
Har
m
o
n
ics
at sp
eci
f
ied
ti
m
e
0
.80
6
4
0
.70
0
9
0
.87
8
5
0
.76
5
8
Har
m
o
n
ics
at sp
eci
f
ied
ti
m
e with d
if
f
erent f
requ
en
cies
0
.88
7
2
0
.70
1
1
0
.84
5
6
0
.87
6
0
Table
2.
Tim
e and A
m
plit
ud
e
of
Dist
urba
nc
e Sig
nal for
D
i
ff
e
ren
t
Har
m
on
ic
Co
ndit
ion
s
Test sig
n
als
Distu
rban
ce ti
m
e
(
sec)
A
m
p
litu
d
e
Theo
.
Dist.
Ti
m
e
(sec)
Act.
D
etected
(sec)
Av
g
.
D
ev
iatio
n
(%)
Theo
.A
m
p
litu
d
e
v
alu
e
Act.
D
etected
v
alu
e
Av
g
.
d
ev
iatio
n
(
%)
Har
m
o
n
ics
at ever
y
ti
m
e
in
stan
t
t1
=0
t1
=0
0
0
.7
0
.7
0
t2
=0
.2
t2
=0
.2
Har
m
o
n
ics
at
sp
eci
f
ied
ti
m
e
t1
=0
.07
5
t1
=0
.07
6
2
1
.00
0
.9
0
.91
0
5
1
.16
t2
=0
.15
t2
=0
.15
0
6
Har
m
o
n
ics
at sp
eci
f
ied
ti
m
e
with
dif
f
erent
f
requ
en
cies
t1
=0
.07
5
t1
=0
.07
3
7
1
.02
0
.7
0
.9
0
.70
5
0
.89
0
.91
t2
=0
.1
62
t2
=0
.16
3
1
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
An
alysis of
har
monics
us
in
g w
avelet
techn
i
qu
e
(
Tha
ngara
j.
K
)
1673
4.2
.
Te
s
t
sig
n
al
2
The
i
nput
si
gnal
wh
ic
h
s
how
n
in
Fi
gure
7
wh
ic
h
c
on
ta
i
ns
thir
d
a
nd
el
ev
enth
ha
rm
on
ic
com
po
ne
nt
a
t
ever
y t
im
e
instant i
n
f
undam
ental
co
m
po
ne
nt
sign
al
of 50 Hz. T
he
locat
i
on
of
distu
rban
ce sta
rting
a
nd
end
i
ng
tim
e
are
show
n
in
Ta
ble
2
with
per
ce
ntag
e
de
viati
on
an
d
from
the
inf
or
m
at
ion
,
the
a
m
plit
ud
e
of
the
gi
ve
n
disturba
nce
si
gn
al
a
re
obse
r
ved.
I
f
m
i
ti
gation
te
ch
nique
s
app
li
e
d
us
i
ng
that
in
form
at
ion
analy
ze
d
fr
o
m
deco
m
po
se
d
si
gn
al
s
,
the
n
th
e
outp
u
t
sig
nal
sh
ow
n
in
Fig
ure
7
is
obta
ine
d
with
er
r
or
of
1.1
202%
sho
wn
i
n
Table
1
f
r
om
the
f
ig
ur
e
,
it
i
s
fou
nd
that
t
he
wa
velet
transfor
m
giv
es
the
re
qu
i
red
i
nfo
rm
ation
ab
out
the
disturba
nce si
gnal
.
Figure
7. Ha
rm
on
ic
s
at eve
ry t
i
m
e instant with
diff
e
re
nt freq
uen
ci
es
4.3.
Te
s
t
sig
n
al 3
Con
si
der
the
i
nput
sig
nal
show
n
in
Fig
ure
8
wh
ic
h
co
ntains
third
harm
on
ic
co
m
po
ne
nt
at
sp
eci
fic
tim
e
instant
in
fun
dam
ental
com
ponen
t
s
ign
al
of
50
Hz.
By
us
i
ng
the
a
naly
zi
ng
in
f
or
m
at
ion
from
deco
m
po
se
d
sign
al
s
,
the
m
it
i
gation
te
ch
niques
will
be
ap
plied.
I
f
m
it
ig
at
ion
te
chn
i
ques
app
li
ed
u
sing
tha
t
inf
or
m
at
ion
,
th
en
the
ou
t
pu
t
s
ign
al
s
how
n
in
Figure
8
is
ob
ta
ined
with
e
rror
of
0.8
785%
wh
ic
h
is
sho
wn
i
n
Table
1.
T
he
locat
ion
of
dist
urba
nce
sta
rtin
g
an
d
end
i
ng
ti
m
e
are
sh
own
in
Table
2
with
per
ce
ntage
devi
at
ion
and f
ro
m
the info
rm
ation
, t
he a
m
plit
ud
e
of
t
he
g
ive
n dist
urb
ance si
gn
al
a
re
observe
d.
4.4.
Te
s
t
sig
n
al 4
The
in
pu
t
sig
na
l
wh
ic
h
s
how
n
in
Fig
ur
e
9
wh
ic
h
co
ntains
third
an
d
se
ve
nth
ha
rm
on
ic
com
po
ne
nt
at
sp
eci
fic
tim
e
i
ns
ta
nt
in
f
undam
ental
co
m
po
ne
nt
sig
nal
of
50
Hz.
T
he
locat
ion
of
dis
tur
ban
ce
sta
rti
ng
a
nd
end
i
ng
ti
m
e
ar
e
sho
wn
in
Ta
ble
2
with
perce
ntage
de
viati
on
an
d
from
the
inf
or
m
at
ion
,
the
am
plit
ud
e
of
th
e
giv
e
n
distu
r
ba
nce
sig
nal
are
ob
s
er
ved.
I
f
m
it
igati
on
te
chn
i
qu
e
s
ap
plied
usi
ng
that
in
for
m
at
ion
analy
zed
from
deco
m
po
se
d
si
gn
al
s
,
the
n
th
e
outp
ut
sig
nal
sh
ow
n
in
Fig
ure
9
is
obta
ine
d
with
er
r
or
of
0.8
456%
sho
wn
in
Table
1.
F
ro
m
the
Figure,
it
is
fo
und
that
the
wav
el
et
tr
ansfo
rm
giv
es
the
require
d
inf
or
m
at
ion
ab
ou
t
the
disturba
nce
si
gn
al
.
T
he
sig
na
l
inf
or
m
at
ion
errors
are
co
m
par
ed
us
i
ng
above
ta
bu
la
r
colum
n
f
or
dif
fer
e
nt
m
easur
in
g
c
onditi
on
s
.
F
ro
m
t
he
re
su
lt
s
ob
ta
i
ned
it
is
fou
nd
that
wa
velet
transfo
rm
has
be
tt
er
perform
ance
f
or
analy
sis o
f power
quali
ty
d
ist
urban
ce
.
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.
9
, N
o.
3
,
June
2019
:
1669
-
1675
1674
Figure
8. Ha
rm
on
ic
s
at speci
fied
ti
m
e
Figure
9. Ha
rm
on
ic
s
at speci
fied
ti
m
e w
it
h
di
ff
e
ren
t
fr
e
quen
ci
es
5.
CONCL
US
I
O
N
This
pa
per
has
presente
d
a
ne
w
m
et
ho
d
of
wav
el
et
te
c
hn
i
qu
e
ba
sed
al
gorithm
for
t
he
analy
sis
of
har
m
on
ic
s u
sin
g
db
20
wa
velet
functi
on. S
eve
ral
case
-
st
ud
ie
s,
analy
zed
wit
h
di
ff
e
ren
t
ty
pe
s
of
dist
urbance
s
i
n
el
ect
rical
powe
r
syst
em
,
hav
e
sh
ow
n
the
su
it
abili
ty
of
the
m
et
ho
d.
The
pe
rfor
m
ance
of
the
pro
pose
d
m
et
ho
d
has
bee
n
c
om
par
ed
with
the
i
nput
sig
nal
by
cal
culat
ing
R
MS
value
of
t
he
sign
al
f
or
dif
fer
e
nt
co
nd
it
io
ns
an
d
sh
owin
g
t
he w
avelet
t
ech
niqu
e analy
sis as a
n
al
te
r
native
processin
g
t
oo
l
f
or the
har
m
on
i
c estim
at
ion
.
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
An
alysis of
har
monics
us
in
g w
avelet
techn
i
qu
e
(
Tha
ngara
j.
K
)
1675
REFERE
NCE
S
[1]
Hsiung
Che
ng
Lin,
“
In
te
r
-
Harm
on
ic
Id
e
ntific
at
ion
Using
Group
-
Ha
rm
on
ic
W
ei
gh
ti
ng
Approac
h
Ba
s
ed
on the
FFT
,”
I
EEE Tr
ans
actions o
n Power
Ele
ct
ro
nics,
vo
l. 23, N
o.
3,
M
ay
2
00
8.
[2]
QU
Wei,
JI
A
Xin,
PEI
S
hib
i
ng,
WU
Jie
,
“
Non
-
sta
ti
on
a
ry
Sign
al
No
ise
Suppressi
on
B
ased
on
Wav
el
et
An
al
ysi
s
,”
C
on
gr
ess
on
Ima
ge
and Si
gnal Pr
ocessin
g.
, vol.
4,
May
2008.
[3]
Su
rya
Sa
nt
oso
,
Edw
a
r
d
J.
P
owers
,
W.
Ma
ck
G
rad
y
a
nd
P
et
er
Hofm
ann
,
“Power
Qu
al
it
y
assess
m
ent
via
Wav
el
et
T
ra
nsfo
rm
A
naly
sis
,”
IEEE
Tr
ansac
ti
on
on P
ower
D
el
iv
ery.
,
vol.
11,No
. 2, A
pr
i
l 199
6.
[4]
Kit
Po
won
g
and
Va
n
Long
Ph
am
,
“An
al
ysi
ng
Power
S
yst
e
m
wav
ef
orm
s
us
ing
W
av
el
et
Tran
sf
or
m
Approac
h
,”
Pr
oceed
i
ngs
of
the
5
th
I
ntern
at
ion
al
C
on
f.
on
A
dvances
in
Power
Syste
m
Co
ntro
l
,
ope
r.
and Ma
nage
m
ent.
, v
ol. 2, Oc
t. 2000.
[5]
V.
L.
Ph
am
and
K.
P
.
Won
g,
“
An
ti
dist
or
t
io
n
m
et
ho
d
f
or
wa
velet
tran
sform
filt
er
ban
ks
a
nd
no
n
-
sta
ti
on
ary
po
wer
syst
em
wav
e
f
or
m
har
m
on
ic
analy
si
s,”
Proc
.
I
ns
t
.
Ele
ct
r.
En
g.
—
Gen
er
.Tr
ans
m.
Distri
b.
,
vol.
148,
s
no. 2,
pp.
117
–
12
2,
Ma
r. 2
001.
[6]
Tom
asz
Taras
iuk
,
“Hy
br
i
d
W
a
velet
–
F
ourier
Me
th
od
for
Har
m
on
ic
s
an
d
Har
m
on
ic
S
ubgro
up
s
Me
asur
em
ent
—
Ca
se Stu
dy
,”
IEEE
Tr
ansac
ti
on
s
on P
owe
r D
el
iv
ery,
vol.
22, N
o.
1, Ja
n 200
7.
[7]
Enr
a
ngZ
heng,
Zhe
ng
ya
n
Liu,
Lingku
m
Ma
,
“Stud
y
on
Ha
r
m
on
ic
Detect
i
on
Me
th
od
Ba
sed
on
FFT
a
nd
Wav
el
et
Tra
nsfo
rm
,”
2
nd
In
te
r
na
ti
on
al
Con
fe
rence
on
Si
gnal
Processi
ng
Systems
(
IC
SP
S)
.
,
vo
l.
3,
Ju
l
y
2010.
[8]
Tha
ng
a
raj.K
,
“An
al
ysi
s
of
Har
m
on
ic
s
us
ing
S
-
Tra
ns
f
or
m
,”
In
te
rn
atio
na
l
c
onfe
renc
e
on
E
merg
i
ng
Trends i
n
E
ngineeri
ng,
Tec
hnology
and Scie
nce (ICET
ET
S)
, 2016.
[9]
ZHU
Sho
u
-
xia
nd
Li
uHui,
“S
i
m
ulati
on
stu
dy
of
powe
r
har
m
on
ic
bas
ed
on
D
au
bec
hies
W
a
velet
,”
E
-
Pr
oduct E
-
Se
rvi
ce a
nd E
-
E
ntertain
me
nt (I
CEEE)
, I
ntern
ational
C
on
fe
r
ence.
,
No
v.
20
10.
[10]
K.
T
ha
ng
a
raj,
“
An
al
ysi
s
an
d
Estim
at
ion
of
Har
m
on
ic
s
usi
ng
Wa
velet
Packet
Tra
ns
f
or
m
,”
In
te
rn
atio
na
l
con
fe
re
nce
on
Emer
ging T
ren
ds
i
n
E
ng
i
neeri
ng, T
ec
hnol
ogy and Scie
nce (
ICETET
S)
, 2016.
Evaluation Warning : The document was created with Spire.PDF for Python.