Ind
o
n
es
ian Jou
r
n
al
o
f
E
le
ctric
a
l E
n
g
in
ee
r
ing
and
C
o
mp
u
t
er
S
c
ienc
e
V
ol
. 8
,
No.
3
,
Dec
em
be
r
20
17
, p
p
.
77
9
~
78
6
DO
I: 1
0.
11
5
91
/
i
j
ee
c
s
.v
8
.i
3
.
pp
7
7
9
-
78
6
7
79
Rec
ei
v
ed
A
ug
us
t
8
, 2
01
7
;
Rev
i
s
ed
O
c
to
be
r
2
6
,
2
01
7
;
A
c
c
ep
ted
Nov
e
mb
er
1
3,
20
17
Modif
i
ed BP
NN
v
i
a I
terat
ed
Lea
st Medi
an Sq
ua
res,
Par
ticle S
w
arm
O
p
tim
iz
at
ion
an
d F
iref
l
y
A
l
g
orit
hm
No
r
A
z
u
r
a M
d
. G
h
ani*
1
,
S
aa
d
i bin
A
h
m
ad Kam
ar
u
d
d
in
2
, Noraz
an M
o
h
amed Ram
li
3
,
Ismail
M
u
sir
in
4
, H
ishamu
d
d
in Hashim
5
1
,
3
Cen
t
e
r f
o
r St
a
ti
s
ti
c
a
l
St
u
d
i
e
s
a
n
d
De
c
i
s
i
o
n
S
c
i
e
n
c
e
s
,
F
a
c
u
l
ty
o
f
Co
m
p
u
t
e
r a
n
d
M
a
th
e
m
a
ti
c
a
l
S
c
i
e
n
c
e
s
,
Uni
v
e
rs
i
t
i
T
e
k
n
o
l
o
g
i
M
ARA, S
e
l
a
n
g
o
r Da
ru
l
Eh
s
a
n
,
M
a
l
a
y
s
i
a
In
te
rn
a
ti
o
n
a
l
I
s
l
a
m
i
c
Un
i
v
e
rs
i
ty
M
a
l
a
y
s
i
a
2
Ku
l
l
i
y
y
a
h
o
f
Sc
i
e
n
c
e
,
In
te
r
n
a
t
i
o
n
a
l
I
s
l
a
m
i
c
U
n
i
v
e
rs
i
ty
M
a
l
a
y
s
i
a
,
Pa
h
a
n
g
Da
ru
l
M
a
k
m
u
r,
M
a
l
a
y
s
i
a
4
Fa
c
u
l
ty
o
f
E
l
e
c
tri
c
a
l
En
g
i
n
e
e
r
i
n
g
,
Un
i
v
e
rs
i
t
i
T
e
k
n
o
l
o
g
i
M
AR
A,
Sh
a
h
A
l
a
m
,
Se
l
a
n
g
o
r Da
ru
l
Eh
s
a
n
,
M
a
l
a
y
s
i
a
5
Pi
l
i
h
s
e
p
a
k
a
t
Sd
n
.
Bh
d
.,
L
o
t
5
4
3
1
,
T
i
n
g
k
a
t
1
(B),
J
a
l
a
n
J
9
,
F
a
s
a
6
,
T
a
m
a
n
M
e
l
a
w
a
ti
,
Ku
a
l
a
L
u
m
p
u
r,
M
a
l
a
y
s
i
a
*C
o
rre
s
p
o
n
d
i
n
g
a
u
th
o
r
s
,
e
-
m
a
i
l
:
a
z
u
ra
@t
m
s
k
.u
i
tm
.e
d
u
.
m
y
1
,
m
y
s
a
a
d
i
a
k
@
i
i
u
m
.e
d
u
.
m
y
2
,
n
o
ra
z
a
n
@tm
s
k
.u
i
t
m
.e
d
u
3
,
i
s
m
a
i
l
b
m
@s
a
l
a
m
.u
i
t
m
.e
d
u
.
m
y
4
,
p
s
s
b
f
6
@gm
a
i
l
.c
o
m
5
Ab
strac
t
Th
e
re
i
s
d
o
u
b
tl
e
s
s
l
y
m
a
n
u
fa
c
t
u
re
d
a
rt
i
fi
c
i
a
l
n
e
u
r
a
l
s
y
s
t
e
m
(ANN
)
i
s
a
s
ta
n
d
o
u
t
a
m
o
n
g
s
t
th
e
m
o
s
t
a
c
c
l
a
i
m
e
d
a
l
l
-
i
n
c
l
u
s
i
v
e
a
p
p
r
o
x
i
m
a
to
rs
,
a
n
d
h
a
s
b
e
e
n
e
x
e
c
u
t
e
d
i
n
n
u
m
e
ro
u
s
fi
e
l
d
s
.
T
h
i
s
i
s
b
e
c
a
u
s
e
o
f
i
ts
c
a
p
a
c
i
t
y
t
o
n
a
tu
ra
l
l
y
ta
k
e
i
n
a
n
y
e
x
a
m
p
l
e
wit
h
n
o
e
a
rl
i
e
r
s
u
p
p
o
s
i
ti
o
n
s
a
n
d
l
o
s
s
o
f
a
l
l
i
n
c
l
u
s
i
v
e
s
ta
t
e
m
e
n
t.
ANN
s
h
a
v
e
c
o
n
tri
b
u
te
d
fu
n
d
a
m
e
n
ta
l
l
y
to
ward
s
ti
m
e
a
rra
n
g
e
m
e
n
t
e
x
p
e
c
ta
ti
o
n
fi
e
l
d
,
y
e
t
t
h
e
n
e
a
rn
e
s
s
o
f
e
x
c
e
p
ti
o
n
s
t
h
a
t
n
o
rm
a
l
l
y
h
a
p
p
e
n
i
n
th
e
ti
m
e
a
rra
n
g
e
m
e
n
t
i
n
fo
rm
a
ti
o
n
m
a
y
d
i
rt
y
th
e
s
y
s
te
m
p
re
p
a
r
i
n
g
i
n
fo
rm
a
ti
o
n
.
H
y
p
o
t
h
e
ti
c
a
l
l
y
,
t
h
e
m
o
s
t
w
i
d
e
l
y
re
c
o
g
n
i
z
e
d
c
a
l
c
u
l
a
ti
o
n
to
p
re
p
a
re
th
e
s
y
s
te
m
i
s
th
e
b
a
c
k
p
ro
p
a
g
a
t
i
o
n
(BP)
c
a
l
c
u
l
a
ti
o
n
wh
i
c
h
d
e
p
e
n
d
s
o
n
th
e
m
i
n
i
m
i
z
a
ti
o
n
o
f
th
e
c
o
m
m
o
n
o
rd
i
n
a
ry
l
e
a
s
t
s
q
u
a
r
e
s
(O
L
S)
e
s
t
i
m
a
to
r
a
s
f
a
r
a
s
m
e
a
n
s
q
u
a
re
d
e
rr
o
r
(M
SE).
Be
th
a
t
a
s
i
t
m
a
y
,
th
i
s
c
a
l
c
u
l
a
t
i
o
n
i
s
n
o
t
a
b
s
o
l
u
te
l
y
s
tr
o
n
g
wit
h
i
n
th
e
s
i
g
h
t
o
f
e
x
c
e
p
t
i
o
n
s
a
n
d
m
a
y
b
ri
n
g
a
b
o
u
t
th
e
b
o
g
u
s
fo
re
c
a
s
t
o
f
fu
tu
re
q
u
a
l
i
t
i
e
s
.
Ac
c
o
r
d
i
n
g
l
y
,
i
n
t
h
i
s
p
a
p
e
r,
we
a
c
tu
a
l
i
z
e
a
n
o
th
e
r
c
a
l
c
u
l
a
t
i
o
n
whi
c
h
e
x
p
l
o
i
t
s
f
i
re
fl
y
c
a
l
c
u
l
a
ti
o
n
o
n
th
e
m
i
n
i
m
a
l
m
i
d
d
l
e
o
f
s
q
u
a
re
s
(FA
-
L
M
e
d
S)
e
s
ti
m
a
to
r
fo
r
m
a
n
u
fa
c
tu
re
d
n
e
u
ra
l
s
y
s
te
m
n
o
n
l
i
n
e
a
r
a
u
to
re
g
r
e
s
s
i
v
e
(BPNN
-
NAR
)
a
n
d
c
o
u
n
te
rf
e
i
t
n
e
u
ra
l
s
y
s
t
e
m
n
o
n
l
i
n
e
a
r
a
u
t
o
re
g
re
s
s
i
v
e
m
o
v
i
n
g
n
o
rm
a
l
(
BPNN
-
NAR
M
A)
m
o
d
e
l
s
t
o
c
o
o
k
th
e
d
i
f
fe
re
n
t
d
e
g
re
e
s
o
f
r
e
m
o
te
i
s
s
u
e
i
n
t
i
m
e
a
rra
n
g
e
m
e
n
t
i
n
fo
rm
a
t
i
o
n
.
In
a
d
d
i
ti
o
n
,
th
e
e
x
e
c
u
ti
o
n
o
f
th
e
p
r
o
p
o
s
e
d
p
o
werf
u
l
e
s
ti
m
a
to
r
wit
h
c
o
rr
e
l
a
ti
o
n
wit
h
th
e
fi
r
s
t
M
SE
a
n
d
s
tro
n
g
i
t
e
ra
ti
v
e
s
l
i
g
h
te
s
t
m
i
d
d
l
e
s
q
u
a
re
s
(IL
M
e
d
S)
a
n
d
m
o
l
e
c
u
l
e
s
warm
a
d
v
a
n
c
e
m
e
n
t
o
n
m
i
n
i
m
u
m
m
i
d
d
l
e
s
q
u
a
re
s
(PSO
-
L
M
e
d
S)
e
s
ti
m
a
to
rs
u
ti
l
i
z
i
n
g
re
e
n
a
c
tm
e
n
t
i
n
f
o
rm
a
ti
o
n
,
i
n
l
i
g
h
t
o
f
ro
o
t
m
e
a
n
s
q
u
a
re
d
b
l
u
n
d
e
r
(RM
S
E)
a
re
l
i
k
e
w
i
s
e
ta
l
k
e
d
a
b
o
u
t
i
n
t
h
i
s
p
a
p
e
r.
It
w
a
s
fo
u
n
d
th
a
t
t
h
e
ro
b
u
s
t
i
fi
e
d
b
a
c
k
p
ro
p
a
g
a
ti
o
n
l
e
a
r
n
i
n
g
c
a
l
c
u
l
a
ti
o
n
u
ti
l
i
z
i
n
g
FA
-
L
M
e
d
S
b
e
a
t
th
e
fi
rs
t
a
n
d
o
t
h
e
r
p
o
werf
u
l
e
s
t
i
m
a
to
rs
o
f
IL
M
e
d
S
a
n
d
PSO
-
L
M
e
d
S.
As
a
c
o
n
c
l
u
s
i
o
n
,
d
e
v
e
l
o
p
m
e
n
ta
l
c
a
l
c
u
l
a
ti
o
n
s
b
e
a
t
th
e
f
i
rs
t
M
SE
m
i
s
ta
k
e
c
a
p
a
c
i
ty
i
n
g
i
v
i
n
g
h
e
a
rty
p
re
p
a
ri
n
g
o
f
c
o
u
n
te
rfe
i
t
n
e
u
ra
l
s
y
s
t
e
m
s
.
Key
w
ords
:
An
o
m
a
l
i
e
s
;
ti
m
e
s
e
ri
e
s
,
l
e
a
rn
i
n
g
a
l
g
o
r
i
th
m
,
ro
b
u
s
t
e
s
ti
m
a
to
r
s
,
e
v
o
l
u
ti
o
n
a
r
y
a
l
g
o
ri
th
m
s
.
Copy
righ
t
©
2
0
1
7
I
ns
titu
t
e
o
f
Adv
a
nc
e
d
Eng
i
ne
e
ring
a
nd
Sc
ie
nc
e
.
All
righ
t
s
re
s
e
rve
d.
1.
Int
r
o
d
u
ctio
n
T
he
ba
c
k
propag
ati
o
n
c
al
c
u
l
at
i
on
de
p
en
ds
o
n
the
f
ee
d
f
orw
ard
m
ul
ti
l
a
y
er
ne
ur
al
s
y
s
t
em
f
or
an
arr
an
ge
m
en
t
of
i
np
u
ts
w
i
t
h
de
t
erm
i
ne
d
k
no
w
n
order
s
.
T
he
c
al
c
ul
at
i
on
p
erm
i
ts
m
ul
ti
l
a
y
er
f
ee
df
orw
ard
ne
ura
l
s
y
s
t
em
s
to
tak
e
i
n
i
nf
o
y
i
e
l
d
m
ap
p
i
ng
s
f
r
om
prepari
ng
t
es
ts
[1]
.
O
nc
e
e
v
er
y
s
ec
ti
on
of
the
s
pe
c
i
m
en
s
et
i
s
di
s
p
l
a
y
e
d
to
the
s
y
s
tem
,
i
ts
y
i
e
l
d
r
e
ac
ti
o
n
wi
l
l
b
e
an
al
y
z
e
d
b
y
t
he
s
y
s
t
em
as
f
or
the
ex
am
pl
e
i
nf
orm
ati
on
d
es
i
g
n.
T
he
y
i
el
d
r
e
ac
ti
o
n
i
s
th
en
c
o
ntra
s
ted
wi
th
the
k
no
w
n
an
d
s
ou
gh
t
y
i
e
l
d
an
d
th
e
b
l
u
nd
er
i
ng
q
ua
l
i
t
y
i
s
f
i
gu
r
ed
,
w
h
ere
t
he
as
s
oc
i
at
i
on
wei
gh
ts
are
ba
l
an
c
ed
.
T
he
ba
c
k
propag
ati
o
n
c
al
c
ul
ati
on
de
pe
n
ds
on
W
i
drow
-
Hof
f
de
l
ta
l
e
arn
i
ng
pri
nc
i
p
l
e
i
n
whi
c
h
th
e
wei
gh
t
c
ha
ng
e
i
s
do
ne
thro
ug
h
m
ea
n
s
qu
ar
e
err
or
(
MS
E
)
of
the
y
i
el
d
r
ea
c
ti
o
n
to
the
ex
am
pl
e
i
nf
o
[2]
.
T
he
arr
an
ge
m
en
t
of
the
s
e
s
pe
c
i
m
en
ex
am
pl
es
i
s
ov
er
an
d
ag
ai
n
i
ntro
du
c
ed
to
the
s
y
s
t
em
un
ti
l
t
he
m
i
s
tak
e
q
ua
l
i
t
y
i
s
m
i
ni
m
i
z
ed
.
De
s
pi
te
the
f
ac
t
tha
t
A
NNs
h
av
e
ef
f
ec
ti
v
e
l
y
c
au
gh
t
the
pre
m
i
um
an
d
wor
r
y
of
nu
m
erous
s
pe
c
i
a
l
i
s
ts
i
n
n
um
erous
f
i
el
ds
b
ec
au
s
e
of
i
ts
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N:
25
02
-
4
75
2
IJE
E
CS
V
ol
.
8
,
N
o.
3
,
Dec
em
be
r
2017
:
7
7
9
–
7
8
6
780
wi
de
s
pr
ea
d
c
ap
ac
i
t
y
as
c
ap
ac
i
t
y
ap
pr
ox
i
m
ato
r
,
the
no
tab
l
e
ba
c
k
propag
at
i
o
n
l
e
arni
n
g
c
al
c
ul
ati
on
whi
c
h
de
pe
n
ds
on
th
e
m
i
ni
m
i
z
a
ti
o
n
of
the
m
ea
n
s
q
u
are
m
i
s
tak
e
(
MS
E
)
c
os
t
c
ap
ac
i
t
y
,
i
s
no
t
v
i
go
r
ou
s
w
i
thi
n
the
s
i
gh
t
o
f
an
o
m
al
i
es
tha
t
m
a
y
brin
g
ab
ou
t
bl
un
d
er
i
n
i
nf
orm
a
ti
on
pr
ep
ari
n
g
proc
es
s
[3]
.
M
S
E
i
s
a
m
i
s
tak
e
m
ea
s
ure
be
t
w
ee
n
th
e
g
en
u
i
ne
an
d
c
r
av
e
d
y
i
el
d
th
at
i
s
u
ti
l
i
z
ed
as
a
pa
r
t
of
the
m
ai
ns
tr
ea
m
ba
c
k
propag
ati
on
l
ea
r
n
i
ng
c
al
c
ul
at
i
o
n
of
m
u
l
ti
l
a
y
er
e
d
f
ee
df
orw
ard
ne
ura
l
s
y
s
tem
s
(
MFNN
s
)
prepari
ng
.
[3]
C
on
c
ur
th
at
thi
s
prom
i
ne
n
t
c
al
c
u
l
at
i
on
i
s
no
t
tot
a
l
l
y
s
tr
on
g
w
i
thi
n
th
e
s
i
g
ht
of
ex
c
ep
ti
o
ns
.
In
de
e
d,
e
v
en
a
s
ol
i
tar
y
an
om
al
y
c
an
d
e
s
tr
o
y
th
e
w
ho
l
e
ne
ura
l
s
y
s
tem
f
i
t [4
].
F
or
al
l
i
nte
n
ts
an
d
pu
r
p
os
e
s
,
ac
qu
i
r
i
n
g
great
i
nf
orm
ati
on
i
s
the
m
os
t
en
tan
gl
ed
a
po
r
ti
o
n
of
es
ti
m
ati
ng
[5]
.
In
th
i
s
wa
y
,
ac
h
i
e
v
i
ng
c
om
pl
ete
an
d
s
m
oo
th
ge
nu
i
ne
i
nf
or
m
ati
on
are
r
i
gh
t
aroun
d
z
ero
l
i
k
el
i
ho
o
d.
E
x
c
ep
ti
on
s
ar
e
i
nf
or
m
ati
on
s
erio
us
l
y
go
i
ng
am
i
s
s
f
r
o
m
th
e
ex
am
pl
e
s
e
t
of
the
do
m
i
na
nt
pa
r
t
i
nf
orm
ati
o
n.
It
ha
s
be
e
n
ac
c
ou
nt
e
d
f
or
tha
t
the
ev
e
nt
of
an
o
m
al
i
es
r
ea
c
he
s
ab
ou
t
1%
to
m
ore
tha
n
1
0
%
i
n
no
r
m
al
r
ou
ti
ne
i
nf
orm
ati
o
n
[6
]
[
7].
In
v
i
e
w
of
pa
s
t
s
tud
i
es
[8
-
10]
the
pres
en
c
e
of
the
s
e
ex
c
ep
ti
on
s
r
ep
r
es
en
ts
an
ex
tr
em
e
da
ng
er
to
the
s
tan
d
ar
d
or
c
us
to
m
ar
y
m
i
ni
m
u
m
s
qu
ares
i
nv
es
ti
g
a
ti
on
.
In
ti
m
e
arr
an
ge
m
en
t
i
n
v
es
ti
ga
ti
o
n,
th
e
ex
am
i
ne
r
s
ne
ed
to
d
ep
e
nd
o
n
i
nf
or
m
ati
on
to
r
ec
og
ni
z
e
w
h
i
c
h
p
oi
nt
i
n
t
i
m
e
are
ex
c
ep
ti
o
ns
to
ga
u
g
e
the
pro
pe
r
r
em
ed
i
al
m
ov
es
to
b
e
m
ad
e
s
o
tha
t
t
he
d
i
s
torte
d
o
c
c
as
i
on
s
c
an
b
e
as
s
es
s
ed
prec
i
s
el
y
.
H
y
p
ot
he
s
i
s
an
d
prac
t
i
c
e
are
grea
ter
pa
r
t
wor
r
i
ed
wi
th
di
r
ec
t
te
c
hn
i
q
ue
s
,
s
uc
h
A
RM
A
a
nd
A
RI
MA
m
od
el
s
[
11
].
B
e
t
ha
t
as
i
t
m
a
y
,
nu
m
erous
arr
an
ge
m
en
t
s
ho
w
de
s
i
gn
whi
c
h
c
an
no
t
be
c
l
arif
i
ed
b
y
a
l
i
ne
ar
s
y
s
t
em
w
h
i
c
h
tr
i
gg
ers
th
e
ne
ed
f
or
no
n
-
d
i
r
ec
t
m
o
de
l
s
,
f
or
i
ns
tan
c
e
bi
l
i
ne
ar
m
od
e
l
s
[12
]
a
nd
n
on
-
s
tr
ai
gh
t
A
RM
A
m
od
el
s
(
NA
RM
A
)
[1
3
].
2.
M
ateri
al
and
m
eth
o
d
In
thi
s
ex
am
i
na
ti
on
,
th
ere
wer
e
three
d
i
s
ti
nc
ti
v
e
r
ee
n
ac
tm
en
t
i
nf
or
m
ati
on
wer
e
uti
l
i
z
ed
.
F
ou
nd
ati
on
N
oi
s
e
Dat
a
f
oc
us
es
wer
e
c
h
os
en
ai
m
l
es
s
l
y
a
nd
af
ter
tha
t
s
ub
s
ti
t
ute
d
wi
th
l
i
k
el
i
ho
od
δ
w
i
t
h a
f
ou
nd
at
i
o
n c
om
m
ot
i
on
c
o
ns
i
s
ten
tl
y
a
pp
r
op
r
i
at
e
d i
n
th
e
pa
r
t
i
c
ul
ar r
ac
e.
Cas
e
1
-
W
i
th
a
s
pe
c
i
f
i
c
e
n
d
g
oa
l
to
tes
t
ou
r
c
al
c
u
l
at
i
o
n
o
n
t
he
1
-
D
ap
prox
i
m
ati
on
tas
k
,
the
c
ap
ac
i
t
y
b
y
[8]
w
as
c
o
ns
i
de
r
e
d
i
n
th
i
s
ex
am
i
na
ti
o
n,
as
l
i
k
ew
i
s
e
ut
i
l
i
z
ed
b
y
p
as
t
wor
k
s
,
f
or
ex
am
pl
e,
[14
-
1
7].
=
|
|
−
2
3
(
1
)
T
he
f
ac
ts
uti
l
i
z
ed
be
c
au
s
e
tha
t
l
oo
k
i
ng
af
ter
i
nc
orp
orate
o
v
er
N
=
4
00
f
oc
us
es
t
o
tha
t
am
ou
nt
ha
v
e
b
ee
n
c
on
s
tr
u
c
ted
b
y
us
i
ng
tr
y
i
ng
o
ut
t
he
au
to
no
m
ou
s
v
ari
ab
l
e
wi
th
i
n
the
s
c
op
e
o
n
[
-
2,
2]
wi
th
m
ea
nti
m
e 0
.01
.
Dur
ab
i
l
i
t
y
.
Cas
e
2
-
A
no
t
he
r
1
-
D
c
ap
a
bi
l
i
t
y
t
o
l
i
e
ap
prox
i
m
ate
d
was
on
c
e
as
r
eg
arde
d
of
m
uc
h
arti
c
l
es
[
14
-
18
] c
ha
r
ac
ter
i
s
ed
as
:
=
s
in
(
)
(
2
)
T
he
r
ec
ords
uti
l
i
z
ed
f
or
i
t
ev
a
l
u
ati
on
c
om
pris
e
c
on
c
e
r
ni
ng
N=
15
0
0
f
oc
us
es
tha
t
wer
e
bu
i
l
t
b
y
i
ns
pe
c
t
i
ng
t
he
s
e
l
f
s
us
tai
ni
n
g
v
ar
i
ab
l
e
i
nto
the
s
c
op
e
on
[
-
7.5
,
7
.5]
al
on
g
i
nt
erim
0.01.
Cas
e
3
-
T
he
2n
d
c
al
c
ul
at
i
o
n
was
on
c
e
as
prop
o
s
ed
throug
h
[9]
[
14
]
as
do
l
i
e
c
ha
r
ac
teri
z
ed
as
:
=
1
−
1
2
−
2
2
(
3)
T
he
r
ec
ords
f
oc
us
es
ha
v
e
be
en
c
om
m
i
tte
d
b
y
us
i
n
g
t
es
ted
c
ap
a
bi
l
i
t
y
of
the
h
on
or
16
x
16
f
r
am
ew
ork
.
Her
e,
the
i
nf
orm
ati
on
al
c
o
l
l
ec
ti
o
n
us
e
d
t
o
be
m
an
uf
ac
tured
throu
gh
c
he
c
k
i
ng
ou
t
the
f
ai
r
f
ac
tors
,
x
1,
x
2
[
-
2,
2]
i
nc
l
ud
i
n
g
m
ea
nti
m
e
0.0
1
.
In
th
e
i
nv
es
ti
g
ati
o
n
f
l
o
w
c
h
art
of
m
as
s
4,
the
ex
p
erim
en
t m
eth
od
ex
e
c
ute
s
tan
d h
o
ne
s
tl
y
ob
s
erv
ed
.
Her
e,
the
c
urr
e
nt
r
i
g
i
d
es
ti
m
ato
r
s
r
eg
ardi
n
g
ba
c
k
propag
at
i
on
ne
ura
l
s
k
el
eto
n
h
a
d
be
e
n
Evaluation Warning : The document was created with Spire.PDF for Python.
IJE
E
CS
IS
S
N:
2
50
2
-
4
75
2
Mo
d
i
fi
ed
B
P
NN v
i
a
Ite
r
at
ed
Le
as
t
Me
d
i
a
n S
qu
ares
,
P
a
r
ti
c
l
e
S
war
m
…
(
Nor
A
z
ura
Md
. G
h
an
i
)
781
c
o
m
pl
ete
d.
T
o
r
ep
l
y
t
he
m
i
dd
l
e
go
al
of
the
i
n
v
e
s
ti
ga
t
i
on
,
the
po
s
s
i
b
l
e
b
u
tte
r
s
tr
en
uo
us
es
t
i
m
ato
r
s
of
no
nl
i
n
ea
r
au
t
oregr
es
s
i
v
e
(
NA
R)
y
et
no
n
l
i
n
ea
r
au
toregr
es
s
i
v
e
tr
an
s
f
err
i
ng
r
eg
ul
ar
(
NA
RM
A
)
r
eg
ard
i
ng
t
he
n
e
ural
f
ab
r
i
c
ti
m
e
c
ou
r
s
e
r
eg
ardi
n
g
l
a
bo
r
w
ere
c
om
pl
ete
d
the
us
a
ge
of
MA
T
LA
B
.
A
t
th
i
s
pro
gres
s
i
on
,
M
A
T
LA
B
s
c
r
i
pts
or
c
od
i
n
gs
w
ere
c
om
po
s
ed
pa
r
al
l
e
l
to
th
e
s
c
i
en
ti
f
i
c
pl
a
n
do
n
e
be
f
ore.
A
f
ter
tha
t,
the
ex
ec
ut
i
on
of
the
propos
ed
r
ob
us
ti
f
i
ed
n
eu
r
a
l
s
y
s
t
em
m
od
el
s
was
tho
ug
ht
ab
o
ut
ut
i
l
i
z
i
ng
r
ec
r
ea
ti
on
i
nf
or
m
ati
on
;
1
-
D
an
d
2
-
D
uti
l
i
z
i
ng
th
e
s
tan
da
r
d
ex
ec
ut
i
on
m
ea
s
ure,
r
oo
t
m
ea
n
s
qu
are
m
i
s
tak
e
(
RMS
E
)
.
A
t
tha
t
po
i
nt
the
p
o
w
erf
ul
B
P
NN
-
N
A
R
an
d
B
P
N
N
-
NA
RM
A
t
ec
hn
i
qu
e
w
ere
tr
i
ed
on
be
nc
hm
ar
k
i
nf
or
m
ati
on
.
T
he
s
i
m
i
l
ar
r
es
ul
ts
h
av
e
att
r
ac
te
d
tho
s
e s
tr
i
d
es
.
3
.
Re
sult
s a
n
d
di
sc
u
s
sio
n
s
In
v
i
e
w
of
the
T
ab
l
es
1,
2
an
d
3,
th
e
c
on
v
e
nti
on
al
c
al
c
ul
at
i
on
d
el
i
v
ered
the
b
es
t
r
es
ul
ts
i
n
l
i
gh
t
of
the
l
i
ttl
es
t
RMS
E
v
al
u
es
f
or
the
pe
r
f
ec
t
i
nf
o
r
m
ati
on
w
i
t
ho
u
t
ex
c
ep
ti
on
s
.
T
hi
s
r
es
ul
t
i
s
pa
r
al
l
e
l
w
i
t
h
the
c
as
e
tha
t
the
MS
E
m
i
s
ta
k
e
c
ap
ac
i
t
y
i
s
i
d
ea
l
f
or
the
i
nf
orm
ati
on
w
i
tho
ut
an
om
al
i
es
b
y
[
3][
1
4][
1
9
].
In
an
y
c
as
e,
th
e
c
i
r
c
um
s
ta
nc
e
i
s
c
ha
ng
e
d
s
i
nc
e
the
i
n
f
or
m
ati
on
c
on
tai
ni
ng
m
i
s
l
ea
di
n
gl
y
produc
e
d
an
om
al
i
es
w
h
er
e
the
M
S
E
-
b
as
ed
s
tr
ate
g
y
tot
al
l
y
l
os
es
i
ts
prod
uc
ti
v
i
t
y
.
T
hi
s
c
an
be
de
m
on
s
tr
ate
d
b
y
th
e
bre
a
k
do
w
ns
of
the
t
ec
hn
i
qu
e
a
s
ap
pe
ared
i
n
T
ab
l
es
1,
2
an
d
3.
E
v
er
y
s
i
ng
l
e
he
art
y
c
al
c
u
l
at
i
o
n
of
ILM
ed
S
,
P
S
O
-
LM
ed
S
an
d
F
A
-
LM
ed
S
pe
r
f
orm
f
un
da
m
en
tal
l
y
b
ett
er
c
on
tr
as
ted
wi
th
MS
E
-
ba
s
ed
c
os
t
c
ap
ac
i
t
y
.
A
d
di
t
i
on
a
l
l
y
,
b
y
s
i
m
pl
y
l
oo
k
i
n
g
at
th
e
t
w
o
de
v
el
op
m
en
tal
i
m
prov
em
en
t
c
al
c
ul
ati
on
s
,
F
A
-
L
Me
d
S
ha
s
i
nd
i
c
ate
d
m
ore
pro
m
i
ne
nt
ex
ec
u
ti
o
n
w
i
th
th
e
m
os
t
m
i
ni
m
al
RMS
E
v
al
ue
s
i
n
ev
er
y
o
ne
of
the
f
ou
r
c
as
es
,
as
ap
pe
are
d
i
n
T
ab
l
es
1,
2
an
d
3.
T
hi
s
i
s
m
a
y
b
e
a
n
i
m
m
ed
i
ate
af
teref
f
ec
t
of
the
r
ea
s
on
th
at
f
i
r
ef
l
y
es
ti
m
ati
on
pe
r
f
or
m
be
tte
r
f
or
a
gr
ea
ter
m
ea
s
ure
of
di
s
turba
nc
e
[20
]
an
d
whe
n
i
t
j
oi
ne
d
i
n
to
a
ba
c
k
propag
a
ti
o
n
ne
ural
s
y
s
t
em
,
prepari
ng
c
al
c
ul
ati
on
,
i
t
un
i
tes
at
a
qu
i
c
k
er
r
ate
w
i
t
h
l
ea
s
t
f
ee
df
orw
ard
ne
ura
l
s
y
s
t
e
m
pl
a
n
[
21
].
A
s
s
p
ec
i
f
i
ed
b
y
[
22
],
the
f
i
r
ef
l
i
es
’
c
al
c
u
l
at
i
on
i
s
t
he
ex
c
ep
ti
on
a
l
i
ns
ta
nc
e
of
qu
i
c
k
en
ed
m
ol
ec
ul
e
s
w
arm
en
ha
nc
e
m
en
t
c
al
c
ul
ati
o
n.
B
es
i
de
s
,
c
on
tr
as
ted
w
i
t
h
F
A
-
LM
ed
S
an
d
IL
Me
d
S
,
the
P
S
O
-
LM
ed
S
c
al
c
ul
at
i
o
n
ha
s
a
te
nd
e
nc
y
to
ac
c
om
pl
i
s
h
m
ore
no
te
w
orth
y
R
S
M
E
as
the
r
at
e
of
ex
c
ep
ti
on
s
i
nc
r
em
en
t.
In
thi
s
s
i
tua
t
i
o
n,
P
S
O
-
LM
e
dS
was
s
ee
n
t
o
de
l
i
v
er
m
ore
no
t
e
w
ort
h
y
bl
u
nd
ers
f
or
the
i
nf
or
m
at
i
on
c
om
pris
i
ng
ex
c
ep
ti
o
ns
m
ore
tha
n
60
pe
r
c
en
t,
a
s
ap
pe
ared
i
n
T
ab
l
es
1,
2
an
d
3.
H
ere,
i
t
i
s
tr
us
ted
tha
t
th
e
P
S
O
-
L
Me
dS
m
a
y
pe
r
f
or
m
be
tte
r
on
t
he
of
f
c
ha
nc
e
tha
t
w
e
ex
pa
nd
the
qu
an
t
i
t
y
of
s
w
arm
s
i
z
e
an
d
c
y
c
l
e
es
tee
m
.
T
he
m
od
i
f
i
ed
ne
ural
ne
t
wor
k
c
an
be
f
ur
the
r
i
m
pl
em
en
ted
to
s
ol
v
e
m
an
y
ot
he
r
r
ea
l
-
l
i
f
e
probl
em
s
,
s
uc
h
as
i
m
ag
e
proc
es
s
i
ng
[2
3
],
wate
r
tr
e
atm
en
t p
l
a
nt
[
24
]
a
nd
p
o
w
er p
l
a
nt
[2
5
].
A
c
kno
w
ledg
ement
W
e
m
i
gh
t
w
an
t
to
d
ev
ote
ou
r
gratef
ul
n
es
s
an
d
ap
prec
i
at
i
on
to
U
ni
t
K
erj
as
am
a
A
wam
S
w
as
ta
(
U
K
A
S
)
of
P
r
i
m
e
Mi
n
i
s
ter'
s
De
pa
r
tm
en
t,
C
on
s
tr
uc
ti
o
n
In
du
s
tr
y
De
v
e
l
op
m
en
t
B
oa
r
d
(
CIDB
)
an
d
t
he
Ma
l
a
y
s
i
a
n
S
tat
i
s
ti
c
s
Depa
r
tm
en
t.
E
x
tr
ao
r
di
na
r
y
t
ha
nk
s
ad
di
ti
on
al
l
y
g
o
t
o
Uni
v
ers
i
t
i
T
ek
no
l
og
i
MA
R
A
an
d
Ma
l
a
y
s
i
an
M
i
ni
s
tr
y
of
Hi
gh
er
E
du
c
at
i
on
(
MO
HE
)
f
or
s
up
po
r
ti
n
g
thi
s
ex
a
m
i
na
t
i
on
un
d
er th
e
Res
ea
r
c
h Gr
an
t
No.
60
0
-
R
MI/
D
A
NA
5/3
/
CIF
I (6
5
/20
13
)
an
d N
o.
6
00
-
RMI/FRG
S
5/
3
(
13
7
/20
14
)
.
Ref
er
en
ce
s
[1
]
P.
Si
b
i
,
S.
A.
J
o
n
e
s
,
a
n
d
P.
Si
d
d
a
rth
.
An
a
l
y
s
i
s
O
f
Dif
fe
re
n
t
A
c
ti
v
a
t
i
o
n
Fu
n
c
ti
o
n
s
Us
i
n
g
Ba
c
k
Pro
p
a
g
a
ti
o
n
Neu
r
a
l
Ne
tw
o
rk
s
.
J
o
u
rn
a
l
O
f
T
h
e
o
re
ti
c
a
l
A
n
d
Ap
p
l
i
e
d
In
fo
rm
a
t
i
o
n
Te
c
h
n
o
l
o
g
y
.
2
0
1
3
;
47
(
3
);
1264
-
1
2
6
8
.
[2
]
A.
P.
En
g
e
l
b
re
c
h
t
.
Com
p
u
ta
t
i
o
n
a
l
i
n
te
l
l
i
g
e
n
c
e
:
An
i
n
tr
o
d
u
c
ti
o
n
.
W
i
l
e
y
.
c
o
m
.
2
0
0
7
.
[3
]
M
.
T
.
El
-
M
e
l
e
g
y
,
M
.
H.
Es
s
a
i
a
n
d
A.
A.
Al
i
.
Rob
u
s
t
tra
i
n
i
n
g
o
f
a
rti
fi
c
i
a
l
fe
e
d
f
o
rw
a
rd
n
e
u
ra
l
n
e
tw
o
rk
s
.
Fo
u
n
d
C
o
m
p
u
t
In
te
l
l
1
.
2009
;
1
:
217
–
2
4
2
.
[4
]
Hec
to
r,
M
.
Cla
u
d
i
o
,
a
n
d
S.
Rod
ri
g
o
.
Rob
u
s
t
Es
t
i
m
a
t
o
r
fo
r
t
h
e
L
e
a
r
n
i
n
g
Pro
c
e
s
s
i
n
Neu
r
a
l
Net
w
o
rk
Ap
p
l
i
e
d
i
n
T
i
m
e
Se
r
i
e
s
.
Arti
fi
c
i
a
l
Ne
u
ra
l
Ne
two
r
k
s
-
ICANN
2
0
0
2
.
2
0
0
2
;
1080
-
1
0
8
6
.
[5
]
L
.
J
.
G
i
tm
a
n
,
a
n
d
C.
M
c
Dan
i
e
l
.
T
h
e
f
u
tu
re
o
f
b
u
s
i
n
e
s
s
,
T
h
e
e
s
s
e
n
ti
a
l
s
.
So
u
th
-
W
e
s
te
rn
P
u
b
,
2
0
0
8
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N:
25
02
-
4
75
2
IJE
E
CS
V
ol
.
8
,
N
o.
3
,
Dec
em
be
r
2017
:
7
7
9
–
7
8
6
782
[6
]
Z.
Zh
a
n
g
.
P
a
ra
m
e
te
r
e
s
t
i
m
a
ti
o
n
te
c
h
n
i
q
u
e
s
:
A
tu
to
r
i
a
l
w
i
th
a
p
p
l
i
c
a
ti
o
n
to
c
o
n
i
c
fi
tt
i
n
g
.
I
m
a
g
e
a
n
d
Vi
s
i
o
n
Co
m
p
u
t
i
n
g
J
o
u
r
n
a
l
.
1
9
9
7
;
15
(
1
);
59
-
76.
[7
]
R.
K.
Pe
a
rs
o
n
.
An
o
m
a
l
i
e
s
i
n
p
ro
c
e
s
s
m
o
d
e
l
i
n
g
a
n
d
i
d
e
n
t
i
fi
c
a
ti
o
n
.
IEEE
Tra
n
s
a
c
ti
o
n
s
o
n
Con
tr
o
l
Sy
s
te
m
s
T
e
c
h
n
o
l
o
g
y
.
2
0
0
2
;
1
0
(1
):
5
5
-
63
.
[8
]
K.
L
i
a
n
o
.
Rob
u
s
t
e
rro
r
m
e
a
s
u
re
fo
r
s
u
p
e
rv
i
s
e
d
n
e
u
ra
l
n
e
two
rk
l
e
a
r
n
i
n
g
w
i
th
a
n
o
m
a
l
i
e
s
.
IEEE
Tra
n
s
Neu
ra
l
Ne
tw
o
rk
s
.
1
9
9
6
;
7
(
1
):
2
4
6
-
2
5
0
.
[9
]
P.
J
.
Hu
b
e
r
.
Ro
b
u
s
t
St
a
t
i
s
t
i
c
s
.
J
o
h
n
W
i
l
e
y
a
n
d
S
o
n
s
.
New
Y
o
rk
,
1
9
8
1
.
[1
0
]
P.
J
.
Ro
u
s
s
e
e
u
w
,
a
n
d
A.
M
.
L
e
ro
y
.
Rob
u
s
t
re
g
re
s
s
i
o
n
a
n
d
o
u
tl
i
e
r
d
e
t
e
c
t
i
o
n
.
W
i
l
e
y
,
New
Y
o
rk
,
1
9
8
7
[1
1
]
G
.
E.
Bo
x
,
G
.
M
.
J
e
n
k
i
n
s
a
n
d
G
.
C.
Rei
n
s
e
l
.
T
i
m
e
s
e
ri
e
s
a
n
a
l
y
s
i
s
,
fo
re
c
a
s
ti
n
g
a
n
d
c
o
n
tro
l
.
E
d
.
Pre
n
ti
c
e
Ha
l
l
,
1
9
9
4
.
[1
2
]
M
.
M
.
G
a
b
r
.
Rob
u
s
t
e
s
ti
m
a
t
i
o
n
o
f
b
i
l
i
n
e
a
r t
i
m
e
s
e
ri
e
s
m
o
d
e
l
s
.
Com
m
.
Sta
ti
s
t,
T
h
e
o
r
y
a
n
d
M
e
th
.
1
9
9
8
;
27
(
1
):
41
-
5
3
.
[1
3
]
J
.
T
.
Con
n
o
r
a
n
d
R.
D.
M
a
rti
n
.
Rec
u
rre
n
t
n
e
u
ra
l
n
e
tw
o
rk
s
a
n
d
ro
b
u
s
t
ti
m
e
s
e
r
i
e
s
p
re
d
i
c
t
i
o
n
.
IEEE
Tra
n
s
o
f
Neu
r
a
l
Ne
two
rk
s
.
1
9
9
4
;
5
(
2
):
240
-
253.
[1
4
]
A.
Rus
i
e
c
k
i
.
Rob
u
s
t
L
e
a
r
n
i
n
g
Al
g
o
ri
t
h
m
B
a
s
e
d
o
n
It
e
ra
t
i
v
e
L
e
a
s
t
M
e
d
i
a
n
o
f
Sq
u
a
re
s
.
Neu
r
a
l
Pro
c
e
s
s
L
e
tt
.
2
0
1
2
;
3
6
:
1
4
5
-
160
.
[1
5
]
D.
S.
Che
n
a
n
d
R.
C.
J
a
i
n
.
A
ro
b
u
s
t
b
a
c
k
p
ro
p
a
g
a
ti
o
n
l
e
a
r
n
i
n
g
a
l
g
o
ri
th
m
fo
r
f
u
n
c
ti
o
n
a
p
p
r
o
x
i
m
a
ti
o
n
.
IEEE
Tra
n
s
.
Neu
r
a
l
Ne
two
r
k
s
.
1994
;
5
(
3
):
467
–
479
.
[1
6
]
Chu
a
n
g
,
S.
Su
,
a
n
d
C.
H
s
i
a
o
.
T
h
e
a
n
n
e
a
l
i
n
g
r
o
b
u
s
t
b
a
c
k
p
ro
p
a
g
a
t
i
o
n
(ARBP)
l
e
a
r
n
i
n
g
a
l
g
o
ri
th
m
.
IEEE
Tra
n
s
Ne
u
ra
l
Ne
two
r
k
s
.
2000
;
11
(
5
):
1
0
6
7
–
1
0
7
6
.
[1
7
]
A.
L
.
Ru
s
i
e
c
k
i
.
Fa
u
l
t
to
l
e
r
a
n
t
fe
e
d
fo
rw
a
rd
n
e
u
ra
l
n
e
tw
o
rk
w
i
th
m
e
d
i
a
n
n
e
u
ro
n
i
n
p
u
t
fu
n
c
ti
o
n
.
El
e
c
tr
o
n
i
c
s
L
e
tt
e
rs
.
2005
;
41
(
10
):
603
-
6
0
5
.
[1
8
]
C.
Chu
a
n
g
C,
J
.
T
.
J
e
n
g
a
n
d
P.
T
.
L
i
n
.
An
n
e
a
l
i
n
g
ro
b
u
s
t
ra
d
i
a
l
b
a
s
i
s
fu
n
c
t
i
o
n
n
e
tw
o
rk
s
f
o
r
fu
n
c
ti
o
n
a
p
p
ro
x
i
m
a
ti
o
n
w
i
th
a
n
o
m
a
l
i
e
s
.
Neu
ro
c
o
m
p
u
ti
n
g
.
2
0
0
4
;
56
:
1
2
3
–
1
3
9
.
[1
9
]
G
.
Be
l
i
a
k
o
v
,
A.
Ke
l
a
re
v
,
a
n
d
J
.
Y
e
a
r
w
o
o
d
.
Deri
v
a
ti
v
e
-
fre
e
o
p
t
i
m
i
z
a
ti
o
n
a
n
d
n
e
u
ra
l
n
e
tw
o
rk
s
fo
r ro
b
u
s
t
re
g
re
s
s
i
o
n
.
O
p
t
i
m
i
z
a
ti
o
n
.
2
0
1
2
;
61
(
12
):
1
4
6
7
-
1
4
9
0
.
[2
0
]
S.
K.
Pa
l
,
C.
S.
Rai
,
a
n
d
A.
P.
Si
n
g
h
.
Com
p
a
ra
t
i
v
e
Stu
d
y
o
f
Fi
re
fl
y
Al
g
o
r
i
th
m
a
n
d
Pa
rt
i
c
l
e
Sw
a
rm
O
p
ti
m
i
z
a
t
i
o
n
f
o
r
Noi
s
y
Non
-
L
i
n
e
a
r
O
p
ti
m
i
z
a
ti
o
n
Pro
b
l
e
m
s
.
In
t
e
rn
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
I
n
te
l
l
i
g
e
n
t
Sy
s
t
e
m
s
a
n
d
Ap
p
l
i
c
a
t
i
o
n
s
.
2
0
1
2
;
1
0
:
50
-
57.
[2
1
]
S.
Nan
d
y
,
P
.
P.
Sa
r
k
a
r
a
n
d
A.
Da
s
.
An
a
l
y
s
i
s
o
f
a
Nat
u
r
e
In
s
p
i
r
e
d
Fi
re
f
l
y
Al
g
o
ri
th
m
b
a
s
e
d
Ba
c
k
-
p
ro
p
a
g
a
ti
o
n
Neu
ra
l
Net
w
o
rk
T
ra
i
n
i
n
g
.
In
t
e
rn
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Com
p
u
te
r
Ap
p
l
i
c
a
ti
o
n
s
.
2
0
1
2
;
42
(
22
):
8
-
16.
[2
2
]
X
.
S.
Y
a
n
g
.
Fi
re
fl
y
Al
g
o
r
i
th
m
s
fo
r
M
u
l
ti
m
o
d
a
l
O
p
ti
m
i
z
a
ti
o
n
.
Pro
c
e
e
d
i
n
g
o
f
th
e
5
th
Sy
m
p
o
s
i
u
m
o
n
Sto
c
h
a
s
ti
c
Al
g
o
ri
th
m
s
,
F
o
u
n
d
a
ti
o
n
s
a
n
d
Ap
p
l
i
c
a
ti
o
n
s
.
2
0
0
9
;
5
7
9
2
;
1
6
9
-
1
7
8
.
[2
3
]
B.
Sa
i
c
h
a
n
d
a
n
a
,
K.
Sr
i
n
i
v
a
s
a
n
d
R.
Ki
r
a
n
Ku
m
a
r,
"
I
m
a
g
e
fu
s
i
o
n
i
n
h
y
p
e
r
s
p
e
c
tra
l
i
m
a
g
e
c
l
a
s
s
i
f
i
c
a
ti
o
n
u
s
i
n
g
g
e
n
e
t
i
c
a
l
g
o
ri
t
h
m
"
,
In
d
o
n
e
s
i
a
n
J
o
u
r
n
a
l
o
f
El
e
c
tri
c
a
l
E
n
g
i
n
e
e
ri
n
g
a
n
d
Com
p
u
te
r
S
c
i
e
n
c
e
,
v
o
l
.
2
,
n
o
.
3
,
7
0
3
-
7
1
1
,
2
0
1
7
.
[2
4
]
M
.
S.
G
a
y
a
,
L
.
A.Y
u
s
u
f,
M
.
M
u
s
ta
p
h
a
,
B.
M
u
h
a
m
m
a
d
,
A
.
S
a
n
i
,
A
.
T
.
A.
T
i
j
j
a
n
i
,
a
n
d
M
.
T
.
M
.
Kh
a
i
ri
.
Es
ti
m
a
ti
o
n
o
f
T
u
rb
i
d
i
ty
i
n
W
a
te
r
T
re
a
tm
e
n
t
Pl
a
n
t
u
s
i
n
g
Ha
m
m
e
r
s
te
i
n
-
W
i
e
n
e
r
a
n
d
Neu
r
a
l
Net
w
o
rk
T
e
c
h
n
i
q
u
e
.
In
d
o
n
e
s
i
a
n
J
o
u
r
n
a
l
o
f
El
e
c
tr
i
c
a
l
En
g
i
n
e
e
ri
n
g
a
n
d
Com
p
u
te
r Sc
i
e
n
c
e
.
2017
;
5
(
3
):
666
-
6
7
2
.
[2
5
]
S.
Cha
k
ra
b
o
rty
,
P.
K.
Sa
d
h
u
a
n
d
N.
Pa
l
.
A
New
Ap
p
ro
a
c
h
to
w
a
rd
s
Id
e
a
l
L
o
c
a
ti
o
n
Se
l
e
c
ti
o
n
fo
r
PV
Po
w
e
r
Pl
a
n
t
i
n
I
n
d
i
a
.
In
d
o
n
e
s
i
a
n
J
o
u
rn
a
l
o
f
El
e
c
tr
i
c
a
l
En
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
.
2
0
1
4
;
12
(
11
):
7
6
8
1
-
7
6
8
9
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJE
E
CS
IS
S
N:
2
50
2
-
4
75
2
Mo
d
i
fi
ed
B
P
NN v
i
a
Ite
r
at
ed
Le
as
t
Me
d
i
a
n S
qu
ares
,
P
a
r
ti
c
l
e
S
war
m
…
(
Nor
A
z
ura
Md
. G
h
an
i
)
783
Fi
g
u
re
1
.
Pro
c
e
s
s
o
f
th
e
p
ro
p
o
s
e
d
BPNN
-
NAR
a
n
d
BPNN
-
NAR
M
A
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N:
25
02
-
4
75
2
IJE
E
CS
V
ol
.
8
,
N
o.
3
,
Dec
em
be
r
2017
:
7
7
9
–
7
8
6
784
T
ab
l
e 1
. T
he
R
MS
E
s
c
ores
f
or tes
t f
un
c
ti
on
i
n c
as
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IJE
E
CS
IS
S
N:
2
50
2
-
4
75
2
Mo
d
i
fi
ed
B
P
NN v
i
a
Ite
r
at
ed
Le
as
t
Me
d
i
a
n S
qu
ares
,
P
a
r
ti
c
l
e
S
war
m
…
(
Nor
A
z
ura
Md
. G
h
an
i
)
785
T
a
b
l
e
2
.
T
h
e
RM
SE s
c
o
re
s
f
o
r
t
e
s
t
fu
n
c
ti
o
n
i
n
c
a
s
e
2
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N:
25
02
-
4
75
2
IJE
E
CS
V
ol
.
8
,
N
o.
3
,
Dec
em
be
r
2017
:
7
7
9
–
7
8
6
786
T
ab
l
e 3
. T
he
R
MS
E
s
c
ores
f
or tes
t f
un
c
ti
on
i
n c
as
e
3
Evaluation Warning : The document was created with Spire.PDF for Python.