I
n
d
o
n
e
s
i
a
n
J
o
u
r
n
a
l
o
f
E
l
e
c
tr
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
a
n
d
C
o
m
p
u
te
r
S
c
i
e
n
c
e
V
ol
.
8,
N
o.
1,
O
c
t
ob
er
20
17
,
pp
.
101
~
10
6
D
O
I
:
10.
115
91/
i
j
eec
s
.
v
8
.i
1
.
pp
10
1
-
1
06
1
01
R
ec
ei
v
ed
J
une
1
,
2
01
7
;
R
e
v
i
s
ed
A
ugus
t
5,
2
01
7
;
A
c
c
e
pt
ed
S
ep
t
em
ber
1
,
2
017
E
n
h
an
ced
BF
G
S
Q
u
as
i
-
N
e
w
ton B
a
c
k
pr
opa
ga
ti
on
M
o
d
els
o
n
M
CC
I Dat
a
N
o
r
A
z
u
r
a
M
d
.
G
h
a
n
i
*
1
,
S
aad
i
A
h
m
ad
K
am
ar
u
d
d
i
n
2
,
N
o
r
a
z
a
n
M
o
h
a
m
m
e
d
R
a
m
l
i
3
,
Is
m
a
il M
u
s
ir
in
4
, H
is
h
a
m
u
d
d
i
n
H
a
s
h
i
m
5
1
,3
C
ent
er
f
or
S
t
at
i
s
t
i
c
al
S
t
udi
es
and D
e
c
i
s
i
o
n S
c
i
en
c
es
,
F
a
c
ul
t
y
of
C
om
p
ut
er
and
M
at
hem
at
i
c
al
S
c
i
e
nc
e
s
,
U
ni
v
er
s
i
t
i
T
ek
no
l
og
i
M
A
R
A
,
S
el
ang
or
D
ar
u
l
E
h
s
an,
M
al
ay
s
i
a
2
D
epar
t
m
ent
of
C
o
m
put
at
i
onal
and T
heor
et
i
c
al
S
c
i
e
nc
e
s
,
K
ul
l
i
y
y
ah of
S
c
i
en
c
e,
I
nt
er
nat
i
ona
l
I
s
l
a
m
i
c
U
ni
v
er
s
i
t
y
M
al
ay
s
i
a,
P
ahan
g D
ar
ul
M
ak
m
ur
,
M
al
ay
s
i
a
4
,5
C
ent
er
f
or
E
l
ec
t
r
i
c
al
P
ow
er
E
ngi
ne
er
i
n
g S
t
u
di
e
s
,
F
a
c
u
l
t
y
o
f
E
l
ec
t
r
i
c
al
E
n
gi
n
eer
i
ng,
U
ni
v
er
s
i
t
i
T
ek
no
l
og
i
M
A
R
A
,
S
el
ang
or
D
ar
u
l
E
h
s
an,
M
al
ay
s
i
a
P
i
l
i
h
s
ep
ak
a
t
S
dn.
B
hd.
,
Lot
54
31,
T
i
ngk
at
1(
B
)
,
J
a
l
an
J
9
,
F
a
s
a 6,
T
am
a
n M
el
aw
at
i
,
K
ual
a
L
um
pur
,
M
al
ay
s
i
a
*C
or
r
es
po
ndi
ng a
ut
hor
,
e
-
m
a
i
l:
az
ur
a@
t
m
s
k
.
ui
t
m
.
ed
u.
m
y
1
,
s
a
adi
a
k
@
i
i
um
.
e
du.
m
y
2
,
nor
az
an@
t
m
s
k
.
ui
t
m
.
e
du.
m
y
3
,
is
m
ai
l
bm
@
s
al
am
.
u
it
m
.
ed
u
.
m
y
4
,
ps
s
bf
6@
g
m
a
i
l
.
c
om
5
A
b
st
r
act
N
eur
oc
om
p
ut
i
n
g i
s
w
i
del
y
i
m
p
l
em
ent
ed
i
n
t
i
m
e
s
er
i
es
ar
ea,
how
ev
er
t
he
n
ear
ne
s
s
of
e
x
c
e
pt
i
on
s
t
hat
f
or
t
h
e m
os
t
par
t
h
app
en i
n i
nf
or
m
at
i
on t
i
m
e ar
r
a
ngem
e
nt
m
i
ght
be hur
t
f
ul
t
o t
he i
nf
or
m
at
i
on or
g
ani
z
e
pr
epar
i
ng.
T
hi
s
i
s
on
t
h
e
gr
ou
nds
t
h
at
t
h
e
c
a
pac
i
t
y
t
o
c
ons
e
quent
l
y
d
i
s
c
o
v
er
an
y
e
x
am
pl
e
s
w
i
t
ho
ut
ear
l
i
e
r
s
upp
os
i
t
i
o
ns
and l
o
s
s
of
a
l
l
-
inc
l
us
i
v
e
s
t
a
t
e
m
e
nt
.
I
n p
r
in
c
ip
l
e,
t
h
e
m
os
t
w
e
ll
-
k
now
n p
r
epar
i
n
g c
a
l
c
u
l
at
i
on
f
or
B
ac
k
pr
opa
gat
i
on
c
al
c
ul
at
i
ons
i
nc
l
i
ne
s
t
ow
ar
d l
es
s
en
i
ng
or
di
nar
y
l
ea
s
t
s
quar
es
es
t
i
m
at
or
(
O
LS
)
or
al
l
t
h
e
m
or
e par
t
i
c
u
l
ar
l
y
,
t
he m
ea
n s
quar
ed
er
r
or
(
M
S
E
)
.
I
n any
c
as
e,
t
hi
s
c
al
c
ul
at
i
o
n i
s
n
ot
c
o
m
pl
et
el
y
h
ear
t
y
w
hen ex
c
ep
t
i
on
s
e
x
i
s
t
i
n pr
e
par
i
ng
i
nf
or
m
at
i
on,
and i
t
w
i
l
l
pr
om
pt
f
al
s
e es
t
i
m
at
e f
u
t
ur
e
es
t
eem
.
A
l
o
n
g
t
hes
e
l
i
nes
,
i
n
t
hi
s
p
aper
,
w
e
s
how
a
not
h
er
c
al
c
ul
a
t
i
o
n
t
ha
t
c
ont
r
ol
c
al
c
ul
at
i
o
ns
f
i
r
ef
l
y
on s
l
i
g
ht
e
s
t
m
i
dd
l
e
s
quar
es
e
s
t
i
m
at
or
(
F
F
A
-
LM
edS
)
f
or
B
F
G
S
quas
i
-
new
t
on
bac
k
pr
op
agat
i
on
n
eur
a
l
net
w
or
k
non
l
i
n
ear
aut
or
e
gr
es
s
i
v
e
m
ov
i
n
g
(
B
P
N
N
-
N
A
R
M
A
)
m
odel
t
o
l
e
s
s
en
t
he
ef
f
e
c
t
of
e
x
c
ept
i
on
s
i
n
t
i
m
e
ar
r
ang
em
ent
i
nf
or
m
at
i
on.
I
n t
h
e i
n t
he
m
ean
t
i
m
e,
t
he m
ont
hl
y
dat
a of
M
al
ay
s
i
an R
oof
M
at
er
i
al
s
c
o
s
t
i
nd
ex
f
r
om
J
anu
ar
y
1980 t
o D
e
c
em
ber
20
12 (
ba
s
e y
ear
19
80=
1
00)
w
i
t
h v
ar
i
o
u
s
l
ev
el
of
e
x
c
e
pt
i
o
ns
i
s
s
u
e i
s
adj
u
s
t
ed i
n t
hi
s
ex
am
i
na
t
i
o
n.
T
ow
ar
d t
he f
i
ni
s
h of
t
hi
s
pap
er
,
i
t
w
a
s
f
o
un
d t
hat
t
he up
gr
a
d
e
d
BPN
N
-
N
A
R
M
A
m
odel
s
u
t
iliz
in
g
F
F
A
-
LM
edS
per
f
or
m
ed
ex
t
r
em
el
y
w
el
l
w
i
t
h
R
M
S
E
v
al
ue
s
j
u
s
t
abo
ut
z
er
o
er
r
or
s
.
I
t
i
s
ex
pec
t
ed
t
h
at
t
he
f
i
nd
i
ng
w
oul
d
he
l
p
t
he
s
pec
i
al
i
s
t
s
i
n
M
al
ay
s
i
an
de
v
el
opm
ent
a
c
t
i
v
i
t
i
e
s
t
o
ha
ndl
e
c
os
t
i
nd
i
c
e
s
dat
a
ac
c
or
di
n
gl
y
.
Ke
y
w
o
rd
s
:
O
ut
l
i
er
s
,
C
C
I
dat
a,
t
r
ai
n
bf
g,
r
obu
s
t
e
s
t
i
m
at
or
s
,
ev
ol
ut
i
onar
y
a
l
gor
i
t
hm
s
C
o
p
y
r
i
g
h
t
©
2
01
7
I
n
s
t
i
t
u
t
e
o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
r
i
n
g
a
n
d
S
c
i
e
n
c
e
.
A
l
l
r
i
g
h
t
s r
es
er
ved
.
1
.
I
n
tr
o
d
u
c
ti
o
n
Mone
t
ar
y
A
c
t
i
v
i
t
i
es
(
P
F
I
)
i
s
t
hes
e da
y
s
a pat
t
er
n i
n M
al
a
y
s
i
a
,
as
i
t
i
s
gene
r
a
l w
it
h
t
h
e
adm
i
ni
s
t
r
at
i
o
n'
s
pr
om
ot
i
ng m
or
e
pr
o
m
i
nent
no
n
-
ope
n s
egm
ent
'
s
i
nc
l
us
i
on i
ns
i
de
abet
m
ent
t
he
not
or
i
et
y
o
v
er
s
y
s
t
em
ad
m
i
ni
s
t
r
at
i
ons
.
T
he
m
os
t
bas
i
c
don
or
w
i
t
h
r
eg
ar
ds
t
o
P
F
I
i
s
w
or
t
h
s
i
nc
e
m
one
y
(
V
F
M)
,
w
her
e i
d
ea
l
t
r
adem
ar
k
ov
er
ad
v
anc
em
ent
ac
t
i
v
i
t
i
es
i
nc
or
por
at
i
ng
r
egar
d
as
per
c
us
t
om
er
'
s
pl
eas
ur
e
y
et
v
ent
ur
es
ar
e i
n t
hi
s
m
anner
ac
hi
e
v
ed ef
f
ec
t
i
v
e
l
y
.
I
t
i
s
k
e
y
t
o t
al
l
y
i
n
r
egar
d t
o t
ex
t
ur
e c
h
ar
ges
,
s
o m
uc
h ar
e ac
qui
r
ed
at
s
o
m
e phas
e i
n
t
h
e P
F
I
d
ev
el
o
pm
ent
s
as
per
det
er
m
i
ne as
o
v
er
s
pe
ndi
n
g
w
i
s
h
no
t
h
app
en.
S
i
nc
e
t
he b
ui
l
d
i
ng
oc
c
as
i
on t
he
n c
i
r
c
um
s
t
anc
e
c
onv
e
y
a
nc
e
i
s
t
he
adm
i
ni
s
t
r
at
i
o
n m
ot
i
v
a
t
i
o
n
i
ns
i
de t
h
e Ma
l
a
y
s
i
an
P
F
I
,
end
ea
v
o
r
s
bear
b
een
done
i
n
s
i
m
i
l
ar
i
t
y
w
i
t
h
ant
i
c
i
pat
e
t
he
pr
es
en
t
l
i
s
t
c
on
c
er
ni
ng
de
v
e
l
opm
ent
t
ex
t
u
r
e w
or
t
h
l
i
s
t
s
in
s
id
e
M
a
la
y
s
ia
.
I
t
w
as
ap
pr
opr
i
a
t
el
y
i
nt
r
od
uc
ed t
h
at
c
onc
r
et
e'
s
c
ont
r
ol
l
ed es
t
e
em
has
been
w
r
ec
k
ed
b
y
m
eans
of
t
he
Mal
a
y
s
i
an
go
v
er
nm
ent
,
b
e
gi
n
ni
n
g
c
onc
er
n
i
n
g
5
J
une
2008
[
1]
.
F
r
om
t
hat
p
oi
n
t
f
or
w
ar
d
,
s
o
w
as
a s
par
k
l
i
ng i
nt
ens
i
f
y
i
n r
eg
a
r
d t
o t
he c
onc
r
et
e go
odn
es
s
i
n J
un
e
20
08
w
hi
c
h i
s
w
i
t
h t
h
e gu
i
de of
23.
3%
am
ong P
eni
ns
ul
ar
Mal
a
y
s
i
a,
w
h
i
l
e 6.
5%
w
as
ac
c
ount
e
d f
or
am
ong S
ab
ah a
nd
5.
2%
a
m
ong S
ar
a
w
ak
[
1]
.
I
n t
he m
ean
w
hi
l
e,
Mal
a
y
s
i
an go
v
er
nm
ent
had ac
t
ual
i
z
ed Mer
c
h
and
i
s
e an
d V
e
nt
ur
es
E
x
pens
e
(
G
S
T
)
al
l
t
hr
ou
gh
t
he
c
o
unt
r
y
s
i
n
c
e
f
i
r
s
t
A
pr
i
l
201
5.
Mer
c
han
di
s
e
at
t
hat
po
i
n
t
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
10
1
–
106
102
A
dm
i
ni
s
t
r
at
i
ons
D
ut
y
(
G
S
T
)
i
s
a
m
ul
t
i
-
or
gan
i
z
e i
nt
r
i
c
ac
y
o
v
er
h
ous
eh
ol
d ut
i
l
i
z
at
i
o
n.
G
S
T
i
s
c
har
ged
on
w
h
ol
e
as
s
es
s
a
bl
e
s
egm
ent
s
ov
er
f
r
i
l
l
a
nd
app
l
i
c
at
i
ons
i
n
M
al
a
y
s
i
a
o
t
her
t
h
an
t
hes
e
es
pec
i
a
l
l
y
ex
em
pt
ed.
G
S
T
i
s
l
i
k
ew
i
s
e
c
har
g
ed
i
n
r
egar
d
t
o
i
m
por
t
at
i
on
o
v
e
r
pr
oduc
t
s
y
et
pur
pos
es
i
ns
i
de
M
al
a
y
s
i
a
[
2]
.
B
ec
a
us
e
of
t
he
us
age
of
t
hi
s
ne
w
a
ppr
o
ac
h i
n
Ma
l
a
y
s
i
a,
eng
i
ne
er
s
ar
e
m
os
t
l
y
h
i
t
b
y
t
he
c
os
t
of
c
r
ude
m
at
er
i
al
s
[
3]
.
T
he
m
o
r
e
t
er
r
i
bl
e
ef
f
ec
t
i
s
,
i
ndus
t
r
y
pl
a
y
er
s
a
nd s
p
ec
i
a
l
i
s
t
s
ex
pec
t
t
h
e c
os
t
s
of
pr
i
v
a
t
e
pr
oper
t
i
es
t
o
r
i
s
e
2%
t
o
4%
p
os
t
-
G
ST
not
w
i
t
hs
t
a
ndi
ng
t
h
e
w
a
y
t
h
at
s
uc
h
pr
oper
t
i
es
ar
e
no
t
s
ubj
ec
t
t
o
t
he
G
S
T
.
A
c
c
or
di
ng
l
y
,
w
i
t
h
t
h
e
ex
ec
ut
i
o
n
w
i
t
h
G
S
T
,
c
o
m
bi
n
ed
w
i
t
h
t
he
h
ar
der
w
o
r
k
i
ng
c
ond
i
t
i
on,
pr
o
per
t
y
des
i
g
n
er
s
ar
e
pr
oba
bl
y
g
oi
n
g t
o
m
et
hodol
ogi
es
t
o c
r
ad
l
e
an
y
ne
gat
i
v
e ef
f
ec
t
.
T
he es
t
ee
m
br
i
ng
i
s
add
i
t
i
ona
l
l
y
per
t
i
ne
nt
up i
n a
gr
eem
ent
w
i
t
h t
he r
es
t
o
f
t
he
i
m
pr
ov
em
ent
m
at
er
i
al
s
-
s
t
e
el
,
pr
e
par
ed
bl
end c
onc
r
et
e
y
e
t
v
ar
i
o
us
ot
her
s
[
4
]
.
A
s
bui
l
d
i
n
g
m
at
er
i
al
ex
pe
ns
e
s
i
n
t
o
Mal
a
y
s
i
a
ha
v
e
be
en
m
et
i
nc
l
u
di
n
g
v
ul
ner
ab
i
l
i
t
y
,
t
he
pr
ev
al
e
nt
t
ec
h
ni
que
has
be
en
ex
am
i
ned
i
n
i
m
per
s
onat
i
on
of
de
pen
d es
t
i
m
at
i
on
of
t
he
de
v
e
l
o
pm
ent
f
abr
i
c
c
os
t
s
i
n
agr
eem
ent
as
per
t
he
m
ean
ar
ea
on
Mal
a
y
s
i
a.
N
ex
t
,
t
he
l
e
gac
y
on
ac
t
u
al
i
t
i
es
c
ons
t
ant
a
m
ong
t
hat
di
r
ec
t
i
on
i
s
dep
i
c
t
ed
b
r
i
ef
l
y
i
ns
i
de
par
t
I
I
.
I
n s
har
e I
I
I
,
t
h
e t
ec
h
ni
que
d
i
agr
a
m
i
s
m
or
eov
er
pr
ov
i
de
d,
y
e
t
t
he s
t
r
at
e
g
y
ol
d as
per
br
e
ak
dow
n t
h
e i
nf
or
m
at
i
on c
l
ar
i
f
i
ed.
N
e
x
t
,
t
he f
i
ni
s
he
d
r
es
ul
t
s
a
nd
di
s
c
our
s
e
ab
ou
t
t
he
pr
om
i
s
i
ng
ga
ugi
ng
t
e
c
hni
q
ue
i
n
l
i
gh
t
of
t
h
e
f
ac
t
t
hat
as
s
es
s
i
ng
t
he m
at
er
i
al
c
har
g
e f
i
l
es
c
onc
ur
r
i
ng
i
n i
m
per
s
onat
i
on
of
Mal
a
y
s
i
an r
eg
i
o
ns
ar
e
pr
es
ent
ed
i
n
di
v
i
d
e I
V
.
A
t
l
as
t
,
d
i
v
i
de
V
c
onv
e
y
s
t
h
e
c
onc
l
us
i
o
n
on
t
he
i
n
v
es
t
i
gat
i
on
,
s
um
m
at
i
on
a
pr
opos
a
l
on ac
c
ou
nt
of
f
ut
ur
e
w
or
k
s
.
2
.
F
o
u
n
d
a
ti
o
n
o
f D
a
ta
T
he i
nf
or
m
at
i
on had b
een
f
or
m
ed
f
r
o
m
3 uni
que s
o
ur
c
es
es
pec
i
al
l
y
U
ni
t
K
er
j
as
am
a
A
w
am
S
w
as
t
a (
U
K
A
S
)
o
n H
ea
d
a
dm
i
ni
s
t
r
at
or
'
s
A
r
ea of
ex
per
t
i
s
e
,
D
e
v
e
l
op
m
ent
I
ndus
t
r
y
A
d
v
anc
em
ent
B
o
ar
d (
C
I
D
B
)
at
t
hat
p
oi
nt
Ma
l
a
y
s
i
a
n I
ns
i
gh
t
s
O
f
f
i
c
e as
had
s
ugges
t
e
d t
he
i
m
pr
ov
em
ent
c
os
t
s
l
i
s
t
s
i
n
l
i
g
ht
of
t
he
f
ac
t
t
hat
t
h
e
m
edi
um
l
oc
al
e
on
t
h
e
P
en
i
n
s
ul
ar
Mal
a
y
s
i
a
w
hi
c
h c
om
pr
i
s
es
of
t
hr
ee s
t
at
es
K
ua
l
a L
um
pur
G
ov
er
n
m
ent
D
o
m
ai
n,
S
el
ang
or
,
N
eger
i
S
em
bi
l
an
t
hen
Me
l
ak
a.
T
he gen
ui
n
e
m
oder
n
m
ont
h
-
to
-
m
ont
h m
eas
ur
em
e
nt
s
ov
er
Ma
l
a
y
s
i
a
n T
ot
al
es
t
eem
r
ec
or
ds
pas
t
J
anuar
y
198
0
i
n
i
m
per
s
onat
i
on
of
D
ec
e
m
ber
2012
(
bas
e
a
y
e
ar
1
980
=
10
0)
w
er
e
adj
us
t
ed,
w
i
t
h
anom
al
i
es
6.
1 per
c
e
nt
of
t
h
e g
en
er
al
i
nf
or
m
at
i
onal
i
n
dex
.
T
he
v
ar
i
a
bl
e
of
i
nt
r
i
gu
e
ex
per
i
e
nc
es
an
om
al
i
es
i
s
s
ue as
c
an
be
f
ou
nd
i
n F
i
gur
e 1.
F
i
gur
e 1
.
T
he d
i
s
per
s
i
o
n of
Mal
a
y
s
i
an
R
oof
t
op
Mat
er
i
a
l
s
c
os
t
s
f
i
l
es
i
nf
or
m
at
i
on
3.
M
e
th
o
d
o
l
o
g
y
T
he f
l
ow
c
har
t
o
v
er
t
he
qu
er
y
per
f
or
m
k
eep c
ons
i
der
ed i
n F
i
g
ur
e 2.
H
er
e,
t
he
c
ur
r
ent
ef
f
ec
t
i
v
e es
t
i
m
at
or
s
of
bac
k
pr
opagat
i
on
neur
a
l
s
y
s
t
e
m
hav
e be
en
ac
t
ua
l
i
z
e
d.
T
o ans
w
er
t
he
m
o
s
t
v
i
t
al
o
bj
ec
t
i
v
e
o
v
er
t
h
e
i
n
v
es
t
i
gat
i
on,
t
he
c
onc
e
i
v
abl
e
i
nt
ens
e
es
t
i
m
at
or
’
s
c
r
os
s
ov
er
i
ns
i
de
non
l
i
n
ear
a
ut
or
e
gr
es
s
i
v
e (
N
A
R
)
y
e
t
n
on
l
i
near
a
ut
or
e
gr
es
s
i
v
e m
ov
i
ng
av
er
age (
N
A
R
M
A
)
o
v
e
r
neur
a
l
s
y
s
t
em
da
y
ac
c
um
ul
at
i
o
n
h
ad b
een
pl
a
y
e
d
out
t
he ut
i
l
i
z
at
i
on
of
MA
T
LA
B
R
2012
a.
A
t
t
h
at
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
EC
S
IS
S
N
:
2
502
-
4
752
E
nh
anc
ed
B
F
G
S
Q
uas
i
-
N
e
w
t
on B
ac
k
pr
op
agat
i
o
n Mod
el
s
on
M
C
C
I
D
at
a
(
N
or
A
z
u
r
a M
d.
G
han
i
)
103
pr
ogr
es
s
i
on
,
MA
T
LA
B
c
on
t
ent
s
t
hen
c
od
i
ng
ha
v
e
b
ee
n
c
om
pos
ed
p
ar
al
l
e
l
as
pe
r
t
he
s
c
i
ent
i
f
i
c
equa
t
i
o
n t
ak
en b
ef
or
e.
F
r
om
t
hat
poi
n
t
on
w
ar
d,
t
he ex
ec
ut
i
o
n c
onc
er
n
i
ng
t
he pr
o
pos
e
d
r
obus
t
i
f
i
e
d
n
eur
a
l
s
y
s
t
em
des
i
g
ns
ha
v
e
b
een
i
n
t
er
es
t
i
ng
l
y
t
h
e
ut
i
l
i
z
at
i
o
n
of
g
en
ui
n
e c
er
t
a
i
nt
i
es
t
he
us
e of
t
h
e r
oo
t
m
eans
s
quar
e
er
r
or
(
R
M
S
E
)
.
T
he unr
i
v
a
l
ed
ne
ar
i
m
pa
c
t
s
hav
e b
een
ex
t
end
ed ap
pr
opr
i
at
e h
er
e w
her
e t
h
e v
a
l
ua
bl
e m
odel
w
as
pi
c
k
ed.
T
he
f
undam
ent
al
B
P
N
N
-
NA
R
det
a
i
l
i
ng
c
an b
e r
epr
es
ent
e
d as
:
(
)
=
t
a
nh
=
1
(
−
1
)
,
(
−
2
)
,
…
,
−
+
(
)
=
1
(
1
)
T
he f
i
ni
s
hed
B
P
N
N
-
N
A
R
M
A
def
i
n
i
t
i
on c
an b
e r
epr
es
e
nt
ed
as
:
(
)
=
t
a
nh
(
−
1
)
,
(
−
2
)
,
…
,
−
,
(
−
1
)
,
(
−
2
)
,
…
,
(
−
)
]
+
(
)
=
1
=
1
(
2
)
w
her
e
H
(x
)
i
s
t
h
e
N
A
R
MA
m
odel
,
x
(t
-
1
),
x
(t
-
2
)
,…
, x
(
t
-
n
y
)
ar
e
i
nput
l
ags
,
ε
(t
-
1
)
,
ε
(t
-
2
),
…
ε
(t
-
n
ε
)
ar
e r
es
i
d
ual
l
ags
t
er
m
s
.
H
enc
e,
ε
(t
)
ar
e t
he
w
hi
t
e n
oi
s
e.
l
i
s
t
he i
nput
neur
o
ns
w
i
t
h i
nd
ex
i
,
m
i
s
t
he
hi
dde
n ne
ur
ons
w
i
t
h
i
ndex
j
,
an
d n
i
s
t
h
e ou
t
put
neur
o
ns
w
i
t
h i
ndex
k
.
T
he near
b
y
i
m
por
t
ant
s
eg
m
ent
ov
er
t
he i
n
v
es
t
i
g
at
i
on
i
s
t
he s
c
i
ent
i
f
i
c
pl
an e
nc
ha
nc
m
ent
of
f
er
of
bac
k
pr
opagat
i
on
n
eur
al
s
y
s
t
em
c
al
c
ul
at
i
o
n ut
i
l
i
z
i
ng
f
ac
t
ual
s
o
l
i
d es
t
i
m
at
or
s
.
T
o
m
a
k
e
hear
t
y
t
he
nor
m
al
bac
k
pr
opaga
t
i
o
n c
a
l
c
ul
at
i
on
bas
ed
abs
o
l
ut
el
y
i
n
r
egar
d
t
o
t
h
e
M
-
es
t
i
m
at
or
s
t
hi
nk
i
ng f
or
br
i
ng
i
ng
do
w
n a
nom
al
y
i
m
pac
t
,
t
he
s
quar
e
d r
es
i
du
al
s
s
t
r
en
gt
h
2
am
ong t
he
s
y
s
t
em
c
at
ac
l
y
s
m
w
i
t
h t
h
e g
ui
de of
anot
her
c
om
ponent
i
n r
ega
r
d t
o t
h
e r
es
i
dua
l
s
.
(
)
=
1
2
,
(
3
)
w
hi
c
h c
om
pl
y
w
i
t
h,
(
)
=
1
(
)
,
(
4
)
w
her
e
N
i
s
t
he
s
um
as
s
or
t
m
ent
w
i
t
h
r
es
pec
t
t
o
t
es
t
s
r
eac
hab
l
e
i
n
l
i
ght
of
t
he
f
a
c
t
t
hat
s
y
s
t
em
pr
epar
i
ng.
W
e
ar
e
i
nf
er
r
i
ng
t
he
r
ef
r
es
hi
ng
o
n
t
he
s
y
s
t
em
w
ei
gh
t
s
bas
ed
a
bs
ol
ut
el
y
f
i
n
i
s
he
d
t
he
i
nc
l
i
nat
i
o
n bet
t
er
t
ha
n a
v
e
r
age l
ear
n
i
ng
c
a
l
c
ul
at
i
on.
T
o
k
eep
t
he
har
ds
h
i
p i
n r
egar
d
t
o al
l
-
i
nc
l
us
i
v
e s
t
at
em
ent
,
a f
ee
df
or
w
ar
d
ne
ur
al
s
y
s
t
em
t
o
get
h
er
w
i
t
h
one
d
ar
k
l
ev
e
l
de
l
i
ght
l
i
e
di
d
bet
w
e
en t
h
i
s
ex
am
i
nat
i
o
n.
T
he w
ei
ght
s
pas
t
t
h
e c
onc
eal
ed ne
ur
ons
i
n
i
m
per
s
onat
i
o
n of
y
i
e
l
d
neur
o
ns
,
,
ar
e c
om
m
uni
c
at
e
d s
o
.
∆
,
=
−
,
=
−
(
)
,
=
−
(
)
.
.
,
(
5
)
t
he p
l
ac
e
i
s
a c
ons
t
ant
,
ℎ
is
t
h
e
y
i
e
l
d of
t
he
ℎ
neur
on,
=
i
s
pr
oduc
e
d ov
er
t
he
ℎ
out
p
ut
n
eur
on,
=
∑
i
s
pr
oduc
e
d
at
t
h
e i
nput
on t
he
ac
t
i
v
at
i
on f
unc
t
i
o
n as
s
oc
i
a
t
ed
w
i
t
h
t
he o
ut
p
ut
n
eur
on
(
)
,
and
i
s
t
he
ac
t
i
v
at
i
o
n f
unc
t
i
o
n o
f
t
he neur
ons
i
n t
h
e o
ut
pu
t
l
a
y
er
.
A
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
10
1
–
106
104
l
i
n
ear
(
p
ur
el
i
n)
i
s
us
e
d i
n
t
he out
put
l
a
y
er
’
s
ne
ur
ons
.
T
he w
e
i
gh
t
s
f
r
o
m
t
he i
n
put
t
o hi
d
de
n
neur
o
ns
,
ar
e u
pda
t
ed
as
.
∆
,
=
−
,
=
−
(
)
,
=
−
(
)
.
.
,
.
.
,
(
6
)
F
i
gur
e 2
.
F
l
o
w
of
e
nha
nc
ed
B
F
G
S
quas
i
-
ne
w
t
on b
ac
k
p
r
opag
at
i
on
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
E
EC
S
IS
S
N
:
2
502
-
4
752
E
nh
anc
ed
B
F
G
S
Q
uas
i
-
N
e
w
t
on B
ac
k
pr
op
agat
i
o
n Mod
el
s
on
M
C
C
I
D
at
a
(
N
or
A
z
u
r
a M
d.
G
han
i
)
105
w
her
e
I
i
i
s
t
he
ent
er
af
t
er
t
he
i
t
h
ent
er
ne
ur
on,
=
∑
i
s
r
ea
l
i
z
ed
l
oc
al
s
ubj
ec
t
a
dv
anc
ed
at
t
he c
ont
r
i
but
i
o
n ov
er
t
he
enac
t
m
ent
w
or
k
c
onnec
t
ed
t
oget
h
er
w
i
t
h t
he h
i
dd
en n
eur
on
(
)
,
and
i
s
t
he
i
n
i
t
i
at
i
on
w
or
k
abo
ut
t
he
n
eur
ons
bet
w
ee
n
t
he
m
y
s
t
er
y
l
a
y
er
.
W
e
hav
e
t
h
e
g
oal
as
per
us
es
t
he
t
a
n
-
s
i
gm
oi
d t
r
a
de
m
ar
k
t
o be s
pec
i
f
i
c
t
he
i
n
i
t
i
at
i
o
n
i
nc
l
udes
on
t
he
gr
o
un
ds
t
hat
t
h
e
dar
k
l
a
y
er
'
s
neur
ons
i
n
l
i
g
ht
of
t
he f
ac
t
t
hat
on i
t
s
ada
pt
a
bi
l
i
t
y
.
T
he l
eas
t
m
edi
an s
q
uar
es
(
LMed
S
)
es
t
i
m
at
or
appr
a
i
s
es
t
he
par
am
et
er
s
b
y
m
et
hods
f
or
t
ac
k
l
i
ng t
he
no
nl
i
ne
ar
m
i
ni
m
i
z
at
i
o
n i
s
s
ue
.
[
(
2
)
]
(
7
)
T
hat
i
s
,
t
he
es
t
i
m
at
or
s
houl
d
c
r
eat
e
t
he
l
i
t
t
l
es
t
ex
c
el
l
en
c
e
f
or
t
he
m
i
ddl
e
ov
er
s
qu
ar
ed
r
es
i
dua
l
s
pr
oc
es
s
ed f
or
t
he ent
i
r
e d
at
a s
et
.
I
t
a
ppe
ar
s
t
o be s
o m
uc
h t
hi
s
a
ppr
o
ac
h i
s
c
om
pl
et
el
y
s
t
r
on
g
in
c
ongr
ui
t
y
w
i
t
h f
al
s
e
f
i
t
s
t
h
en
al
s
o
i
n
i
m
per
s
onat
i
on
of
anom
al
i
es
s
t
a
y
i
ng
i
n
i
m
per
s
onat
i
o
n of
det
er
i
or
at
i
v
e
c
onf
i
n
em
ent
[
6]
.
N
ot
s
or
t
of
t
he
M
-
es
t
i
m
at
or
s
,
non
et
he
l
es
s
,
t
he
LM
ed
S
i
nc
on
v
e
ni
e
nc
e
c
an'
t
s
t
an
d
di
m
i
ni
s
hed
b
y
a
w
ei
ght
ed
m
i
ni
m
u
m
s
quar
es
i
s
s
ue.
I
t
i
s
m
ay
be
n
ow
not
r
eas
o
nab
l
e
i
n
i
m
per
s
onat
i
on
of
s
hado
w
u
nd
er
ne
at
h
a
s
i
m
pl
e
equ
at
i
on
i
n
v
i
e
w
of
t
he
r
es
u
l
t
on
LM
ed
S
es
t
i
m
at
or
.
T
hus
,
det
er
m
i
ni
s
t
i
c
c
al
c
u
l
at
i
ons
m
ay
al
s
o
n
ot
b
e
s
k
i
l
l
ed
as
per
t
r
adem
ar
k
i
n
i
m
per
s
onat
i
on
of
l
i
m
i
t
t
h
at
es
t
i
m
at
or
.
T
he
Mon
t
e
-
C
ar
l
o
t
ec
hni
qu
e
[
7]
has
b
ee
n
w
or
k
ed
on
as
i
nd
i
c
at
e
d
b
y
c
l
ear
u
p
i
t
i
nc
onv
en
i
enc
e
am
ong
hal
f
w
a
y
n
on
-
n
eur
a
l
ap
pl
i
c
at
i
o
ns
.
Li
f
e
s
pan
S
t
oc
has
t
i
c
c
al
c
ul
at
i
ons
ar
e
al
s
o
r
ec
ogn
i
z
ed
as
t
he
s
t
r
eam
l
i
ni
n
g
c
al
c
ul
at
i
ons
as
u
t
i
l
i
z
at
i
on
f
r
eel
y
hunt
t
o har
v
es
t
an ans
w
er
.
S
t
oc
has
t
i
c
c
al
c
ul
a
t
i
o
ns
ar
e t
hus
genu
i
n
el
y
m
oder
at
e
,
ho
w
e
v
er
i
n t
ha
t
pl
ac
e
i
s
l
i
k
el
i
ho
od t
h
at
w
ant
f
i
nd t
he
w
or
l
d
w
i
d
e l
e
as
t
.
O
ne r
ea
l
l
y
w
el
l
-
k
now
n enh
anc
em
ent
c
al
c
ul
a
t
i
o
n us
ed
b
y
di
m
i
ni
s
h a LMe
dS
-
bas
ed s
y
s
t
e
m
per
pl
ex
i
t
y
i
s
m
anuf
ac
t
u
r
ed dr
i
nk
(
S
A
)
c
a
lc
u
la
t
io
n.
S
A
i
s
a
gl
or
i
o
us
c
al
c
ul
at
i
on s
i
nc
e i
t
i
s
es
p
e
c
i
al
l
y
br
oa
d,
y
et
s
uc
h
has
t
he i
nc
l
i
ni
ng
not
af
t
er
be br
ou
ght
g
ot
t
e
n of
e
i
t
her
t
he i
nc
om
pl
et
e
l
eas
t
t
hen gr
eat
es
t
[
6]
.
N
onet
he
l
e
s
s
,
[
5]
f
i
nds
s
o
i
t
er
at
ed L
Me
dS
t
en
ds
as
i
n
di
c
at
e
d b
y
o
ut
s
ai
l
t
he
S
A
-
L
Med
S
.
T
a
bl
e
1
.
S
t
o
pp
i
ng C
r
i
t
er
i
a
V
al
ues
NN T
e
r
m
s
1000
M
a
x
i
m
u
m
nu
m
ber
of
epoc
h
s
t
o
t
r
ai
n
0
P
er
f
or
m
anc
e
goal
1e
^
-
7
M
i
ni
m
u
m
per
f
or
m
an
c
e gr
adi
ent
4
.
R
e
s
u
l
ts
F
r
o
m
t
he
ex
a
m
i
nat
i
on
on
MC
C
I
dat
as
e
t
s
,
i
t
c
an
be
as
s
um
ed t
hat
t
he
pr
o
pos
ed
bac
k
pr
opagat
i
on
ne
ur
al
f
r
am
ew
or
k
t
i
m
e ar
r
angem
ent
m
odel
s
per
f
or
m
ed
w
el
l
w
h
en t
he
i
nf
or
m
at
i
on
i
n
v
ol
v
es
ex
c
ep
t
i
ons
.
S
i
nc
e
t
he
f
ac
t
of
t
he
m
at
t
er
i
s
t
o f
i
n
d
t
he
bes
t
f
i
t
t
ed
d
ec
i
d
i
ng
m
odel
s
f
or
MC
C
I
dat
as
et
s
,
t
hi
s
ex
am
i
nat
i
o
n c
an
u
nr
av
el
t
he
d
i
s
c
l
os
ur
es
as
i
n
T
ab
l
e
2.
T
he m
os
t
not
i
c
e
ab
l
y
a
w
f
ul
des
i
gn
f
or
R
oof
t
op
Mat
er
i
a
l
s
i
nf
or
m
at
i
on
i
s
5
-
5
-
20
of
B
P
N
N
-
N
A
R
M
A
s
how
w
i
t
h
R
MS
E
=
0.
8
33.
T
he
bes
t
d
es
i
gn
i
s
B
P
N
N
-
N
A
R
M
A
s
h
o
w
w
i
t
h
s
et
u
p
10
-
10
-
10
w
h
er
e
t
he
R
MS
E
=
0.
41
4.
T
hi
s
i
s
t
r
ai
l
ed b
y
B
P
N
N
-
N
A
R
d
i
s
pl
a
y
w
i
t
h d
es
i
gn 1
0
-
10
-
10 a
nd 15
-
15
-
15,
w
her
e t
he
R
MS
E
=
0.
598.
T
abl
e 2.
Y
i
el
d of
O
r
di
n
ar
y
B
P
N
N
-
NA
R
a
n
d
B
P
NN
-
N
A
R
MA
Mod
el
s
on
M
al
a
y
s
i
an R
oof
Mat
er
i
al
s
C
os
t
I
n
di
c
es
D
a
t
a
bas
ed
on D
i
f
f
er
ent
Lags
I
nput
E
rro
r
H
i
dden
R
MS
E
Lags
Lags
N
odes
BPN
N
-
NA
R
BPN
N
-
NA
RM
A
5
5
20
0.
689
0.
833
10
10
10
0.
598
0.
414
15
15
15
0.
598
0.
649
20
20
20
0.
624
0.
684
25
25
25
0.
649
0.
717
30
30
30
0.
671
0.
749
35
35
35
0.
678
0.
764
40
40
45
0.
709
0.
807
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
25
02
-
4
752
I
J
E
EC
S
V
o
l.
8
,
N
o.
1,
O
c
t
o
ber
20
17
:
10
1
–
106
106
5
.
D
i
scu
ss
i
o
n
s
T
hi
s
i
m
pl
i
es
i
f
t
he
s
y
s
t
em
i
s
gi
v
e
n
s
uf
f
i
c
i
ent
num
ber
of
i
nf
o
s
l
ac
k
s
and
m
i
s
t
a
k
e
s
l
ac
k
s
,
c
ons
ol
i
dat
ed
w
i
t
h
s
at
i
s
f
ac
t
or
y
num
ber
of
c
onc
eal
e
d
hubs
,
t
he
s
y
s
t
em
c
an
hav
e
t
he
c
apac
i
t
y
t
o
per
f
or
m
i
deal
l
y
.
F
or
t
hi
s
s
i
t
uat
i
on,
t
he
bes
t
m
odel
i
s
B
P
N
N
-
N
A
R
M
A
d
em
ons
t
r
at
e.
6
.
C
o
n
c
l
u
s
i
o
n
T
he r
eas
onab
l
e m
odel
s
u
i
t
abl
e i
n l
i
ght
of
t
he f
ac
t
t
hat
M
al
a
y
s
i
an R
oof
t
op M
a
t
er
i
a
l
s
c
har
ge f
i
l
es
m
eas
ur
em
ent
s
i
s
B
P
N
N
-
N
A
R
M
A
t
og
et
h
er
w
i
t
h s
et
up
10
-
10
-
1
0
.
I
n t
h
e f
o
l
l
ow
i
ng
ex
er
t
i
on
,
F
F
A
-
L
Med
S
m
i
gh
t
s
t
and
t
es
t
ed
c
onc
er
n
i
ng
c
er
t
i
f
i
abl
e
i
nf
or
m
at
i
on
w
ho
r
el
at
e
on
30
%
t
o ha
l
f
di
s
t
an
t
i
nf
or
m
at
i
on.
T
he pr
opos
ed
C
ol
os
s
al
c
al
c
ul
at
i
ons
s
i
nc
e
t
r
ai
ni
n
g n
eur
al
s
y
s
t
em
s
m
ay
k
eep ac
hi
ev
abl
e af
t
e
r
s
t
a
y
c
us
t
om
i
z
ed
a s
c
op
e of
f
i
el
ds
of
c
ount
er
f
ei
t
c
ons
c
i
ous
n
es
s
,
ar
r
angem
ent
r
ec
ogn
i
z
ab
l
e
pr
oof
,
s
pec
i
m
en
ac
k
now
l
e
dgm
ent
,
m
ac
hi
ne
l
ear
n
i
ng
,
qual
i
t
y
c
ont
r
o
l
and
s
t
r
eam
l
i
ni
n
g
y
e
t
l
og
i
c
al
f
ig
u
r
in
g
.
T
he pr
opos
ed
al
gor
i
t
h
m
c
a
n be f
ur
t
her
i
m
p
l
e
m
ent
ed i
n
m
any
pr
oc
e
s
s
i
ng ac
t
i
v
i
t
i
e
s
,
s
uc
h
a
s
i
m
age
pr
o
c
es
s
i
ng
[
8]
,
w
at
er
t
r
e
at
m
e
nt
pl
ant
[
9]
a
nd
p
ow
er
pl
a
nt
[
10
].
A
c
k
n
o
w
l
e
d
g
e
m
e
n
t
W
e
w
ant
t
o
ex
pr
es
s
o
ur
gr
at
ef
ul
n
es
s
and
a
ppr
ec
i
at
i
on
t
o
U
ni
t
K
er
j
as
am
a A
w
am
S
w
as
t
a
(
U
K
A
S
)
ov
er
P
r
i
m
e Mi
ni
s
t
er
'
s
O
f
f
i
c
e,
C
ons
t
r
uc
t
i
on I
ndus
t
r
y
D
ev
el
o
pm
ent
B
oar
d (
C
I
D
B
)
t
he
n
Mal
a
y
s
i
an
S
t
at
i
s
t
i
c
s
D
epar
t
m
ent
.
A
ppr
ec
i
at
i
o
n t
o U
n
i
v
er
s
i
t
i
T
e
k
nol
og
i
M
A
R
A
a
nd Ma
l
a
y
s
i
a
n
S
er
v
i
c
e
ov
er
A
d
v
anc
e
d
ed
uc
at
i
o
n
(
MO
H
E
)
as
a
r
es
ul
t
of
s
uppor
t
i
ng
i
t
i
nv
es
t
i
g
at
i
on
und
er
nea
t
h
t
he R
es
e
ar
c
h G
r
ant
N
o
.
600
-
R
MI
/
D
A
N
A
5
/
3/
P
S
I
(
197
/
2
013)
or
N
o
.
600
-
R
MI
/
F
R
G
S
5/
3
(
137/
2
014)
.
R
ef
er
en
ces
[1
]
H
M
F
oad,
A
M
ul
up
.
H
ar
g
a s
i
l
i
n
g s
i
m
en
di
m
an
s
uh
5 J
un
.
U
t
us
an,
P
ut
r
aj
a
y
a
.
2008
.
[2
]
J
G
o
h
.
D
ev
el
oper
s
s
t
r
at
e
gi
z
i
n
g t
o b
uf
f
er
G
S
T
i
m
pac
t
.
T
h
e E
dge M
al
ay
s
i
a,
M
S
N
N
ew
s
.
20
15:
27
.
[3
]
R
oy
al
M
al
ay
s
i
an C
u
s
t
om
s
.
G
o
od an
d S
ev
i
c
es
T
ax
:
G
ui
de on
C
ons
t
r
uc
t
i
on I
ndu
s
t
r
y
.
20
14:
1
-
27
.
[4
]
SBA Ka
m
ar
ud
di
n,
N
A
M
G
hani
,
NM
R
a
m
li.
B
es
t
F
or
ec
a
s
t
i
n
g M
odel
s
f
or
P
r
i
v
at
e F
i
na
nc
i
al
I
n
i
t
i
a
t
i
v
e
U
ni
t
ar
y
C
har
ge
s
D
at
a of
E
a
s
t
C
oas
t
and S
ou
t
her
n R
e
gi
ons
i
n P
e
ni
n
s
u
l
ar
M
al
ay
s
i
a
.
I
nt
er
na
t
i
o
nal
J
our
n
al
o
f
E
c
onom
i
c
s
a
nd S
t
at
i
s
t
i
c
s
.
20
14
:
2
.
[5
]
A
R
us
i
ec
k
i
.
R
obu
s
t
L
ear
n
i
ng
A
l
gor
i
t
hm
B
as
ed
on I
t
er
at
i
v
e
Leas
t
M
edi
an
of
S
q
uar
e
s
.
N
e
ur
al
P
r
oc
e
s
s
Let
t
.
2
012
;
36:
145
-
160.
[6
]
M
T E
l
-
M
el
egy
,
M
H
E
ssa
i
,
A
A Al
i
.
R
ob
us
t
t
r
ai
n
i
ng
of
ar
t
i
f
i
c
i
al
f
ee
df
or
w
ar
d neur
al
n
et
w
or
k
s
.
F
ou
nd
C
om
put
I
nt
e
l
l
1,
S
pr
i
nger
,
B
er
l
i
n,
H
ei
d
el
b
er
g
.
2
009;
1
:
217
–
2
42.
[7
]
Z
Z
hang
.
P
ar
a
m
et
er
es
t
i
m
a
t
i
o
n
t
e
c
h
ni
q
ues
:
A
t
ut
or
i
al
w
i
t
h
a
ppl
i
c
at
i
on
t
o
c
o
ni
c
f
i
t
t
i
ng
.
I
m
ag
e
and v
i
s
i
o
n
C
om
put
i
ng
.
199
7
;
15
(
1
):
59
-
76
.
[8
]
B
S
ai
c
ha
nda
na,
K
S
r
i
n
i
v
as
,
R
Ki
ra
n
Ku
m
a
r.
I
m
age f
u
s
i
on i
n
hy
per
s
pe
c
t
r
a
l
i
m
a
ge c
l
as
s
i
f
i
c
at
i
on u
s
i
ng
genet
i
c
a
l
gor
i
t
hm
.
I
n
do
n
es
i
an J
o
u
r
n
al
o
f
E
l
ec
t
r
ic
a
l
E
ngi
n
e
er
i
ng a
nd C
om
put
er
S
c
i
en
c
e
.
2017
:
2
(
3
)
:
703
-
71
1.
[9
]
MS
G
ay
a,
L
A
Y
us
uf
,
M
M
us
t
apha,
B
M
uham
m
ad,
A
S
ani
,
A
T
A
T
i
j
j
ani
,
M
T
M
K
hai
r
i
.
E
s
t
i
m
a
ti
o
n
o
f
T
ur
bi
di
t
y
i
n
W
a
t
er
T
r
eat
m
ent
P
l
ant
us
i
ng
H
am
m
er
s
t
e
i
n
-
W
i
e
ner
and
N
eur
a
l
N
et
w
or
k
T
ec
hni
qu
e
.
I
n
don
es
i
an J
ou
r
na
l
of
E
l
ec
t
r
i
c
a
l
E
n
gi
ne
er
i
n
g an
d
C
om
put
er
S
c
i
en
c
e
.
2017
;
5
(
3
):
666
-
67
2
.
[
10]
S
C
hak
r
abor
t
y
,
P
K
S
adhu
,
N
Pa
l
.
A
N
ew
A
ppr
oac
h t
ow
ar
ds
I
dea
l
Loc
at
i
o
n S
el
e
c
t
i
o
n f
or
P
V
P
o
w
er
P
l
ant
i
n
I
nd
i
a
.
I
ndo
nes
i
an
J
o
ur
nal
of
E
l
e
c
t
r
i
c
al
E
ng
i
ne
er
i
n
g an
d C
om
put
er
S
c
i
e
nc
e
.
201
4
:
12
(
11
):
7681
-
7
689.
Evaluation Warning : The document was created with Spire.PDF for Python.