I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
E
lec
t
r
ical
an
d
Com
p
u
t
e
r
E
n
gin
e
e
r
in
g
(
I
JE
CE
)
Vol.
1
4
,
No.
5
,
Oc
tober
20
2
4
,
pp
.
5970
~
5983
I
S
S
N:
2088
-
8708
,
DO
I
:
10
.
11591/i
jec
e
.
v
1
4
i
5
.
pp
5
970
-
5983
5970
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
e
c
e
.
iaes
c
or
e
.
c
om
E
n
h
an
c
e
d
ac
c
u
r
ac
y
e
st
imat
io
n
m
od
e
l
e
n
e
r
gy i
m
p
o
r
t
i
n
o
n
-
gr
id
r
oof
t
o
p
sol
ar
p
h
ot
ovol
t
ai
c
Alf
in
S
a
h
r
in
,
I
m
am
Aba
d
i,
Ali
M
u
s
yaf
a
D
e
pa
r
tm
e
nt
of
E
ngi
ne
e
r
in
g P
hys
ic
s
, F
a
c
ul
ty
of
I
ndus
tr
ia
l
T
e
c
h
nol
ogy a
nd S
ys
te
m E
ngi
ne
e
r
in
g, I
ns
ti
tu
t
T
e
knol
ogi
S
e
pul
uh N
ope
mb
e
r
,
S
ur
a
ba
ya
, I
ndone
s
ia
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
M
a
r
21,
2024
R
e
vis
e
d
J
ul
9,
2024
Ac
c
e
pted
J
ul
17,
2024
In
s
t
al
l
i
n
g
ro
o
ft
o
p
s
o
l
ar
p
h
o
t
o
v
o
l
t
ai
c
(PV
)
w
i
t
h
an
o
n
-
g
ri
d
s
y
s
t
em
b
e
n
efi
t
s
co
n
s
u
mer
s
b
ecau
s
e
i
t
can
red
u
ce
i
mp
o
rt
s
o
f
el
ec
t
ri
ca
l
en
erg
y
fro
m
t
h
e
g
r
i
d
.
T
h
i
s
s
t
u
d
y
a
i
ms
t
o
mo
d
el
t
h
e
e
s
t
i
mat
i
o
n
o
f
en
erg
y
i
m
p
o
r
t
s
g
e
n
erat
e
d
fro
m
on
-
g
ri
d
ro
o
f
t
o
p
s
o
l
ar
PV
s
y
s
t
em
s
.
T
h
i
s
es
t
i
ma
t
i
o
n
mo
d
el
w
as
carri
e
d
o
u
t
i
n
2
0
p
r
o
v
i
n
c
i
al
cap
i
t
a
l
s
i
n
In
d
o
n
e
s
i
a.
T
h
e
p
ar
amet
er
s
u
s
e
d
are
w
eat
h
er
co
n
d
i
t
i
o
n
s
,
o
ri
e
n
t
a
t
i
o
n
a
n
g
l
e,
an
d
e
n
erg
y
g
e
n
erat
e
d
fro
m
t
h
e
o
n
-
g
r
i
d
r
o
o
f
t
o
p
s
o
l
ar
PV
s
y
s
t
em.
T
h
e
v
al
u
e
o
f
i
mp
o
r
t
ed
en
er
g
y
i
s
d
i
v
i
d
ed
i
n
t
o
t
h
ree
co
mb
i
n
a
t
i
o
n
s
b
as
e
d
o
n
t
h
e
azi
m
u
t
h
an
g
l
e
d
i
rect
i
o
n
,
w
h
i
ch
d
es
cr
i
b
e
s
t
h
e
t
y
p
e
an
d
s
h
a
p
e
o
f
t
h
e
r
o
o
f
o
f
t
h
e
b
u
i
l
d
i
n
g
(o
n
e
-
d
i
rect
i
o
n
,
t
w
o
-
d
i
rect
i
o
n
s
,
an
d
t
h
ree
-
d
i
rec
t
i
o
n
s
).
Mo
d
el
i
n
g
w
a
s
d
o
n
e
u
s
i
n
g
mac
h
i
n
e
l
earn
i
n
g
w
i
t
h
n
eu
ra
l
n
et
w
o
r
k
(N
N
),
l
i
n
ear
re
g
res
s
i
o
n
,
an
d
s
u
p
p
o
r
t
v
ect
o
r
ma
ch
i
n
e.
A
co
mp
ar
i
s
o
n
o
f
t
h
e
mach
i
n
e
l
earn
i
n
g
al
g
o
r
i
t
h
m
res
u
l
t
s
i
s
N
N
p
r
o
d
u
ces
t
h
e
s
mal
l
es
t
ro
o
t
mean
s
q
u
are
erro
r
(RMSE
)
v
a
l
u
e
o
f
t
h
e
t
h
ree.
Mo
d
el
en
h
a
n
cemen
t
u
s
e
s
a
g
ri
d
s
earch
cro
s
s
-
v
a
l
i
d
at
i
o
n
(G
SCV
)
t
o
b
eco
me
t
h
e
G
SCV
-
N
N
mo
d
el
.
T
h
e
RMSE
res
u
l
t
s
w
ere
en
h
an
ce
d
fro
m
5
3
.
1
8
4
t
o
4
4
.
3
8
9
i
n
t
h
e
o
n
e
-
d
i
rect
i
o
n
co
mb
i
n
a
t
i
o
n
,
1
4
5
.
5
6
2
t
o
1
4
1
.
2
8
6
i
n
t
h
e
t
w
o
-
d
i
r
ect
i
o
n
co
mb
i
n
a
t
i
o
n
,
a
n
d
8
1
.
4
4
2
t
o
7
6
.
3
1
3
i
n
t
h
e
t
h
ree
-
d
i
rec
t
i
o
n
c
o
mb
i
n
a
t
i
o
n
.
T
h
e
i
mp
o
rt
e
d
en
er
g
y
es
t
i
mat
i
o
n
mo
d
e
l
o
n
t
h
e
o
n
-
g
ri
d
ro
o
f
t
o
p
s
o
l
ar
PV
s
y
s
t
em
w
i
t
h
G
SCV
-
N
N
p
ro
d
u
ce
s
a
mo
re
o
p
t
i
ma
l
an
d
accu
ra
t
e
mo
d
e
l
.
K
e
y
w
o
r
d
s
:
E
ne
r
gy
im
po
r
t
Gr
id
s
e
a
r
c
h
c
r
os
s
-
va
li
da
ti
on
E
s
ti
mation
model
M
a
c
hine
lea
r
ning
On
-
gr
id
r
oof
top
s
olar
photovol
taic
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
Ali
M
us
ya
f
a
De
pa
r
tm
e
nt
of
E
nginee
r
ing
P
hys
ics
,
F
a
c
ult
y
o
f
I
nd
us
tr
ial
T
e
c
hnology
a
nd
S
ys
tem
E
nginee
r
ing
,
I
ns
ti
t
ut
T
e
knologi
S
e
puluh
Nope
mber
I
T
S
C
a
mpus
,
S
ur
a
ba
ya
60111
,
I
ndone
s
ia
E
mail:
mus
ya
f
a
@e
p.
it
s
.
a
c
.
id
1.
I
NT
RODU
C
T
I
ON
T
he
incr
e
a
s
ing
a
dopti
on
o
f
on
-
gr
id
r
oof
top
s
olar
photovol
taic
(
P
V)
tec
hnology
r
e
quir
e
s
unde
r
s
tanding
e
ne
r
gy
im
po
r
ts
in
planning
r
e
ne
wa
ble
e
ne
r
gy
in
f
r
a
s
tr
uc
tur
e
that
c
a
n
mee
t
f
utu
r
e
g
r
o
wth
a
nd
s
us
taina
bil
it
y
de
mands
.
S
olar
P
V
p
r
oduc
ti
on
p
r
od
uc
e
d
f
r
om
r
e
s
idential
a
r
e
a
s
a
nd
c
omm
e
r
c
ial
buil
d
ings
c
a
n
mee
t
pe
a
k
load
ne
e
ds
to
mi
nim
ize
im
po
r
ts
f
r
om
t
he
gr
id
[
1
]
.
S
ola
r
P
V
ins
talled
with
va
r
ious
types
of
r
oo
f
s
c
a
n
r
e
duc
e
e
lec
tr
icity
im
por
ts
by
be
twe
e
n
4
a
nd
7%
a
nd
li
m
it
pe
a
k
e
lec
tr
icity
e
xpo
r
t
to
125
%
[
2]
.
T
he
a
ppli
c
a
ti
on
of
r
oof
top
s
olar
P
V
c
ombi
ne
d
with
c
ombi
ne
d
he
a
t
a
nd
powe
r
ge
ne
r
a
ti
on
c
a
n
r
e
duc
e
im
por
ts
of
e
lec
tr
ica
l
e
ne
r
gy
a
nd
incr
e
a
s
e
the
us
e
o
f
e
lec
tr
ic
ve
hicle
s
[
3
]
.
Va
r
ious
s
ys
tem
modes
,
s
uc
h
a
s
ba
t
ter
ies
a
s
e
ne
r
gy
s
tor
a
ge
f
r
om
s
olar
P
V
whe
n
e
ne
r
gy
pr
oduc
ti
on
is
a
t
it
s
pe
a
k,
c
a
n
r
e
duc
e
e
lec
tr
ica
l
powe
r
im
p
or
ts
a
nd
e
xpor
t
e
xc
e
s
s
to
the
dis
tr
ibut
ion
gr
id
[
4]
,
[
5]
.
S
tu
dies
c
onduc
ted
[
6
]
s
how
that
T
u
r
ke
y
he
a
vil
y
de
p
e
nds
on
e
ne
r
gy
im
por
ts
,
with
inc
r
e
a
s
e
d
r
oof
top
s
olar
P
V
in
s
tallations
lowe
r
ing
powe
r
im
po
r
ts
a
nd
s
a
ving
90,
298
kW
h
f
r
om
a
s
ys
tem
that
s
uppli
e
s
13.
2%
of
a
nnua
l
e
lec
tr
ica
l
e
ne
r
gy
c
ons
umpt
ion
in
buil
dings
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
e
d
ac
c
ur
ac
y
e
s
ti
mation
mode
l
e
ne
r
gy
impor
t
in
on
-
gr
id
r
ooft
op
s
olar
photovoltai
c
(
A
lf
in
Sahr
in
)
5971
E
s
ti
mating
im
po
r
ted
e
ne
r
gy
in
on
-
gr
id
r
oof
top
s
olar
P
V
s
ys
tems
is
highl
y
de
pe
nde
nt
on
s
olar
ir
r
a
diation,
we
a
ther
c
ondit
ions
,
ge
ogr
a
phica
l
loca
t
ion,
ti
lt
a
nd
a
z
im
uth
a
ngles
,
a
nd
buil
ding
type.
T
h
e
s
ize
of
e
ne
r
gy
de
li
ve
r
y
f
r
om
buil
dings
c
onne
c
ted
to
r
oo
f
top
s
olar
P
V
with
the
gr
id
is
a
n
e
s
s
e
nti
a
l
c
ons
ider
a
ti
on
in
de
ter
mi
ning
r
e
ne
wa
ble
e
ne
r
gy
ge
ne
r
a
ti
on
c
a
pa
c
it
y
[
7]
.
S
olar
P
V
pe
ne
tr
a
ti
on
in
c
it
ies
c
a
n
r
e
duc
e
poll
uti
on
a
nd
c
a
r
bon
e
mi
s
s
ions
us
e
d
a
s
a
n
e
ne
r
gy
s
our
c
e
f
or
ve
hicle
s
a
nd
r
e
duc
e
de
pe
nde
nc
e
on
the
gr
id
f
r
om
0.
809
to
[
8]
.
On
-
gr
id
r
oo
f
top
s
olar
P
V
a
ls
o
ha
s
a
dva
ntage
s
in
e
ne
r
gy
s
tor
a
ge
i
f
s
tor
a
ge
is
s
witche
d
to
the
gr
id
with
hour
ly
ne
t
mea
s
ur
e
ments
with
a
n
incr
e
a
s
e
of
3
t
o
5
pe
r
c
e
ntage
point
s
[
9]
.
S
ola
r
P
V
in
bui
ldi
ngs
c
a
n
be
int
e
gr
a
ted
with
c
ontr
o
l
a
nd
mon
it
or
ing
s
ys
tems
thr
ough
int
e
r
ne
t
of
thi
ngs
(
I
o
T
)
f
or
e
ne
r
gy
s
upply
a
nd
c
ons
umpt
ion.
I
t
is
to
moni
tor
int
e
r
na
l
ne
e
ds
a
nd
e
xter
na
l
c
ondit
ions
of
e
ne
r
gy
s
upply
a
nd
de
mand
ne
e
ds
f
r
om
the
gr
id
[
10]
.
T
he
r
e
f
or
e
,
a
de
e
p
unde
r
s
tanding
of
e
ne
r
gy
im
por
t
e
s
ti
mation
models
in
on
-
gr
id
r
oof
top
s
olar
P
V
s
ys
tems
is
e
s
s
e
nti
a
l
f
or
mor
e
e
f
f
icie
nt
a
nd
e
nvir
onmenta
ll
y
f
r
iendly
s
us
taina
ble
e
ne
r
gy
planning.
De
e
p
lea
r
ning
s
olar
P
V
ins
tallation
e
s
ti
mation
modeling
that
opti
mi
z
e
s
the
tr
a
ini
ng
pr
oc
e
s
s
us
ing
a
non
-
d
omi
na
nt
s
or
ti
ng
ge
ne
ti
c
a
lgor
it
hm
(
NSGA
)
a
nd
c
us
tom
ize
d
da
tas
e
t
with
f
unc
ti
ona
l
da
ta
a
na
lys
is
(
F
DA
)
pr
oduc
e
s
e
s
ti
mate
s
with
e
r
r
or
s
of
6%
a
nd
18
%
in
s
ynthes
is
da
ta
a
nd
6
%
a
nd
12%
in
a
c
tual
da
ta
[
11]
.
T
he
mul
ti
s
tr
a
tegy
gr
a
dient
-
ba
s
e
d
a
lgor
it
hm
(
M
AG
B
O)
to
s
olve
the
pr
ob
lem
o
f
mul
ti
modal
a
nd
n
on
-
li
ne
a
r
c
ha
r
a
c
ter
is
ti
c
s
c
ombi
ne
d
with
a
modi
f
ied
gr
a
dien
t
s
e
a
r
c
h
r
ule
(
M
GSR
)
a
nd
nove
l
r
e
f
r
e
s
h
ope
r
a
tor
(
NR
O)
r
e
s
ult
s
in
high
pr
e
c
is
ion
a
nd
r
e
li
a
bil
it
y
[
12]
.
T
he
o
ppos
it
e
-
ba
s
e
d
e
xpone
nti
a
l
dis
tr
ibut
ion
opti
mi
z
e
r
(
OB
E
DO
)
a
lgor
it
hm
us
e
s
the
e
s
ti
mation
of
P
V
s
ys
tems
a
f
f
e
c
ti
ng
c
ur
r
e
nt
a
nd
e
ne
r
gy
.
T
he
a
lgor
it
hm
pe
r
f
or
ms
be
tt
e
r
than
other
s
to
identif
y
s
olar
P
V
pa
r
a
mete
r
s
to
im
pr
ove
pe
r
f
or
manc
e
[
13]
.
I
S
S
AE
DO
c
ombi
ne
s
a
s
pa
r
r
o
w
s
e
a
r
c
h
a
lgor
it
hm
a
nd
e
xpone
nti
a
l
dis
tr
ibu
ti
on
opt
im
iza
ti
o
n
(
E
DO
)
to
pr
oduc
e
e
f
f
e
c
ti
ve
ne
s
s
a
nd
s
pe
e
d
in
e
s
ti
mating
P
V
models
s
uc
h
a
s
s
ingl
e
-
diode
a
nd
double
-
diode
models
[
14]
.
R
a
t
s
wa
r
m
opti
m
ize
r
a
lgor
it
hm
to
s
olve
c
ompl
e
x
mathe
matica
l
model
pr
oblems
a
nd
non
-
li
ne
a
r
pr
oblems
.
T
he
a
lgo
r
it
hm
c
ompar
e
d
to
pa
r
ti
c
l
e
s
wa
r
m
opti
mi
z
a
ti
on,
a
nt
li
on
opti
mi
z
e
r
,
s
a
lp
s
wa
r
m
a
lgor
it
hm,
Ha
r
r
is
Ha
wks
opti
m
iza
ti
on
(
HH
O)
,
a
nd
g
r
a
s
s
hoppe
r
opti
mi
z
a
ti
on
ha
s
r
e
s
ult
e
d
in
f
a
s
t
a
nd
a
c
c
ur
a
te
P
V
c
e
ll
e
s
ti
mation
[
15]
.
How
e
ve
r
,
it
is
s
ti
ll
di
f
f
icult
to
f
ind
r
e
s
e
a
r
c
h
th
a
t
dis
c
us
s
e
s
e
s
ti
mating
im
por
ted
e
ne
r
gy
on
-
gr
id
r
oof
top
s
olar
P
V
.
P
r
e
vious
s
tudi
e
s
,
s
uc
h
a
s
thos
e
on
dif
f
e
r
e
nt
we
a
ther
c
ondit
ion
pa
r
a
mete
r
s
,
c
a
n
a
f
f
e
c
t
the
e
s
ti
mation
r
e
s
ult
s
of
s
olar
P
V
s
ys
tems
with
mul
t
i
-
c
r
it
e
r
ia
de
c
is
ions
c
ombi
ne
d
with
e
mpi
r
ica
l
f
o
r
mu
las
[
16]
.
T
he
e
s
ti
mation
method
to
de
ter
mi
ne
the
potential
a
r
e
a
of
r
oof
top
s
olar
P
V
c
a
n
be
ins
talled
us
ing
objec
t
-
ba
s
e
d
im
a
ge
c
las
s
if
ica
ti
on
a
nd
digi
tal
s
ur
f
a
c
e
models
with
ge
ogr
a
phica
l
li
mi
tations
[
17
]
.
I
ntegr
a
ted
3D
modeling
f
or
a
s
s
e
s
s
ing
s
olar
potential
on
buil
ding
r
oo
f
s
us
ing
li
ght
de
tec
ti
on
a
nd
r
a
nging
(
L
iDAR
)
c
a
n
be
us
e
d
to
e
s
ti
mate
r
oof
top
s
olar
P
V
ins
tallations
[
18]
.
T
ur
in
c
ombi
ne
d
the
ge
ogr
a
phic
inf
or
mation
s
ys
tem
(
GI
S
)
s
ys
tem
with
e
ne
r
gy
ge
ne
r
a
ti
on
a
nd
c
ons
umpt
ion
e
s
ti
mation
models
to
pr
oduc
e
e
ne
r
gy
powe
r
pr
of
i
les
[
1
9]
.
T
he
W
e
r
ne
r
de
c
onvolut
ion
method,
whic
h
im
pr
ove
s
t
he
de
tec
ti
on
o
f
ge
ologi
c
a
l
s
tr
uc
tur
e
s
f
r
om
magne
ti
c
f
ield
da
ta,
ha
s
s
uc
c
e
s
s
f
ull
y
de
ter
mi
ne
d
the
e
s
ti
mation
o
f
the
de
pth
pos
it
ion
o
f
ge
ologi
c
a
l
objec
ts
[
20
]
.
I
n
th
e
c
it
y
of
S
a
r
dini
a
,
the
a
lgo
r
it
hm
o
f
c
a
lcula
ti
ng
the
s
ha
dow
of
the
r
oo
f
of
the
buil
ding
a
nd
s
olar
r
a
diation
in
r
e
a
l
ti
me
ha
s
r
e
s
ult
e
d
in
a
potential
r
oof
top
s
olar
P
V
of
22
T
W
h
[
21]
.
Ana
lyze
the
a
c
c
ur
a
c
y
of
the
P
V
model
to
obtain
opti
mal
pe
r
f
or
manc
e
r
e
s
ult
s
of
the
maximum
po
we
r
point
by
r
e
c
ons
tr
uc
ti
ng
the
c
ha
r
a
c
ter
is
ti
c
c
u
r
ve
s
on
dif
f
e
r
e
nt
P
V
pa
ne
ls
[
22]
.
E
xplor
ing
mac
hine
lea
r
ning
a
lgor
it
hms
mo
r
e
pr
o
f
oundly
a
nd
the
a
tt
r
ibut
e
s
'
powe
r
c
a
n
im
p
r
ove
the
a
c
c
ur
a
c
y
of
im
por
ted
e
ne
r
gy
e
s
ti
mation
models
.
T
he
modeling
e
ne
r
gy
c
ons
umpt
ion
a
nd
dif
f
e
r
e
nc
e
s
be
twe
e
n
AH
.
264/AVC
a
nd
V
P
8
e
nc
ode
r
s
us
ing
li
ne
a
r
r
e
gr
e
s
s
ion
in
mac
hine
lea
r
ning
[
23]
.
Applying
mac
hine
lea
r
ning
with
c
r
os
s
-
gr
oup
lea
r
ning
c
a
n
make
good
pr
e
diction
models
be
twe
e
n
s
ingl
e
-
a
tom
c
opp
e
r
-
ba
s
e
d
a
ll
oys
with
dif
f
e
r
e
nt
a
ll
oying
e
leme
nts
[
24
]
.
S
upp
or
t
ve
c
to
r
mac
hine
(
S
VM
)
a
nd
na
ïve
B
a
ye
s
a
lgor
i
thm
s
f
or
im
a
ge
pr
oc
e
s
s
ing,
then
c
onti
nue
d
us
ing
P
VW
a
tt
,
P
VG
I
S
,
a
nd
A
r
c
GI
S
,
ha
ve
ge
ne
r
a
ted
s
olar
P
V
pot
e
nti
a
l
of
21.
1,
24
.
9,
a
nd
22
.
9
GW
h/yea
r
[
25]
.
M
ode
ls
of
ne
ur
a
l
ne
twor
ks
(
NN
)
,
ba
c
k
pr
opa
ga
ti
on
ne
ur
a
l
ne
twor
ks
(
B
P
NN
)
,
a
nd
c
onvolut
ion
ne
ur
a
l
ne
twor
ks
(
C
NN
)
a
r
e
us
e
d
to
e
s
ti
mate
s
ol
a
r
r
a
diation
in
a
r
e
a
s
with
li
mi
ted
da
ta.
L
im
it
e
d
mete
or
ologi
c
a
l
a
nd
r
a
diation
da
ta
a
r
e
then
s
im
ulate
d
with
e
ne
r
gy
a
nd
dies
e
l
c
on
s
um
pti
on
s
o
that
thi
s
method
c
a
n
s
olve
p
r
oblems
in
a
r
e
a
s
wi
th
low
r
a
diation
da
ta
[
26]
.
E
s
t
i
mat
e
d
P
V
p
r
o
duc
t
io
n
p
e
r
ho
ur
u
s
i
ng
f
e
e
d
f
o
r
wa
r
d
ne
ur
a
l
ne
tw
or
k
(
F
F
NN
)
us
in
g
nu
me
r
ica
l
w
e
a
the
r
da
ta
s
u
c
h
a
s
a
m
b
ien
t
tem
pe
r
a
t
ur
e
,
r
e
la
ti
ve
h
u
mi
di
ty
,
a
n
d
wi
nd
s
pe
e
d
.
T
h
e
mo
de
l
r
e
s
u
l
ted
in
a
n
r
oo
t
m
e
a
n
s
qu
a
r
e
e
r
r
o
r
(
R
M
S
E
)
o
f
6
.
4
%
,
r
e
s
u
l
ti
ng
in
s
a
vi
ng
s
a
nd
in
c
r
e
a
s
e
d
s
e
l
f
-
c
o
ns
u
med
e
ne
r
gy
[
2
7]
.
B
y
c
o
mp
a
r
i
ng
t
h
r
e
e
r
a
nd
om
f
o
r
e
s
t
(
R
F
)
a
lg
o
r
i
th
ms
,
g
r
a
di
e
n
t
b
o
os
t
r
e
g
r
e
s
s
io
n
tr
e
e
(
G
B
R
T
)
a
nd
Ada
B
o
os
t
we
r
e
u
s
e
d
f
o
r
a
nn
ua
l
r
o
of
r
a
d
ia
ti
on
e
s
t
i
ma
ti
on
.
T
he
r
e
s
ul
ts
of
th
is
r
a
ndom
f
or
e
s
t
method
p
r
ovide
a
mo
r
e
a
c
c
ur
a
te
a
nd
f
a
s
ter
e
s
ti
mation
with
a
n
a
bs
olut
e
mea
n
e
r
r
or
(
M
AE
)
of
22.
83
kW
h/m
²/yea
r
[
28
]
.
E
s
ti
mating
e
ne
r
gy
pote
nti
a
l
ge
ne
r
a
ted
f
r
om
s
olar
P
V
us
ing
high
-
de
f
ini
t
ion
map
im
a
ge
s
a
nd
de
e
p
lea
r
ning
(
DL
)
r
e
s
ult
e
d
in
a
powe
r
potential
of
442.
4
M
W
in
r
ur
a
l
C
hina
[
29]
.
T
he
gr
id
s
e
a
r
c
h
c
r
os
s
-
va
li
da
ti
on
(
GSC
V)
p
r
oc
e
s
s
c
a
n
s
igni
f
ica
ntl
y
r
e
duc
e
the
a
bil
it
y
to
p
r
oduc
e
models
that
pe
r
f
o
r
m
be
tt
e
r
a
nd
im
pr
ove
the
a
c
c
ur
a
c
y
of
im
po
r
ted
e
ne
r
gy
e
s
ti
mation
models
.
S
ome
pr
e
vious
r
e
s
e
a
r
c
h,
s
uc
h
a
s
the
a
ppli
c
a
ti
on
in
de
ter
mi
ning
th
e
pe
r
f
or
manc
e
li
m
it
s
of
R
C
s
e
is
mi
c
c
olum
ns
us
ing
mac
hine
lea
r
ning
methods
,
then
opti
mi
z
e
d
with
g
r
id
s
e
a
r
c
h
a
nd
c
r
os
s
-
va
li
da
ti
on
to
de
ter
mi
ne
hype
r
pa
r
a
mete
r
s
,
ha
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
1
4
,
No.
5
,
Oc
tober
2
02
4
:
5970
-
5983
5972
r
e
s
ult
e
d
in
e
xc
e
ll
e
nt
a
c
c
ur
a
c
y
[
30]
.
T
he
a
ppli
c
a
ti
o
n
of
pr
e
dicting
a
nd
identi
f
ying
c
hil
dr
e
n
a
nd
a
dole
s
c
e
nts
a
t
r
is
k
of
r
e
a
dmi
s
s
ion
to
the
hos
pit
a
l
us
ing
GSC
V
me
thod
c
a
n
s
igni
f
ica
ntl
y
r
e
duc
e
the
a
bil
it
y
to
pr
oduc
e
models
that
pe
r
f
or
m
be
tt
e
r
a
nd
im
p
r
ove
the
a
c
c
ur
a
c
y
of
i
mpor
ted
e
ne
r
gy
e
s
ti
mation
models
.
S
ome
pr
e
vious
r
e
s
e
a
r
c
h,
s
uc
h
a
s
the
a
ppli
c
a
ti
on
in
de
ter
mi
ning
the
pe
r
f
or
m
a
nc
e
li
mi
ts
of
R
C
s
e
is
mi
c
c
olum
ns
us
ing
mac
hine
lea
r
ning
methods
,
then
opti
mi
z
e
d
with
gr
id
s
e
a
r
c
h
a
nd
c
r
os
s
-
va
li
da
ti
on
to
de
ter
mi
ne
hype
r
pa
r
a
mete
r
s
,
ha
s
r
e
s
ult
e
d
in
e
xc
e
ll
e
nt
a
c
c
ur
a
c
y
c
hine
lea
r
ning
c
las
s
if
ica
ti
on
a
n
d
r
e
gr
e
s
s
ion
tr
e
e
(
C
AR
T
)
,
R
F
,
gr
a
dient
boos
ti
ng
mac
hine
(
GB
M
)
,
da
n
e
xtr
e
me
gr
a
dient
boos
ti
ng
(
XG
B
oos
t)
.
75%
tr
a
in
ing
da
ta
a
nd
25%
tes
ti
ng
a
nd
a
pplyi
n
g
c
r
os
s
-
va
li
da
ti
on
with
f
ive
f
olds
,
then
looki
ng
f
or
hype
r
pa
r
a
mete
r
s
us
ing
gr
id
s
e
a
r
c
h
tec
hniques
.
T
he
X
GB
oos
t
a
lgor
it
hm
ha
s
p
r
oduc
e
d
mo
r
e
a
c
c
ur
a
te
r
e
s
ult
s
than
other
s
[
31]
.
I
n
the
he
a
lt
h
s
e
c
tor
,
to
a
na
lyze
ga
s
tr
os
c
opy
im
a
ge
s
us
ing
the
E
f
f
icie
ntDe
t
method
,
a
gr
id
s
e
a
r
c
h
is
us
e
d
to
opt
im
ize
im
a
ge
r
e
s
olut
ion,
a
nd
c
r
os
s
-
va
li
da
ti
on
with
k=
5
p
r
oduc
e
s
pr
e
c
is
e
r
e
s
ult
s
[
32
]
in
the
tempe
r
a
tur
e
tes
t
be
nc
h
e
xpe
r
im
e
nt
to
s
t
udy
the
phe
nomenon
of
nuc
lea
te
boil
ing
us
ing
s
e
ve
r
a
l
ma
c
hine
lea
r
ning
methods
a
nd
gr
id
opti
m
iza
ti
on
s
e
a
r
c
h
c
r
os
s
-
va
li
da
ti
on
in
the
tr
a
ini
ng
pr
oc
e
s
s
.
T
he
s
uppor
t
ve
c
tor
mac
hine
a
lgor
it
hm
pr
oduc
e
s
the
be
s
t
pe
r
f
or
ma
nc
e
with
R
2
pe
r
f
or
manc
e
gr
e
a
ter
than
0.
95
[
33]
.
T
o
p
r
e
dict
the
pe
r
c
e
ntage
of
biodi
e
s
e
l
yield
us
ing
the
C
a
tB
oos
t
r
e
gr
e
s
s
or
method
opti
mi
z
e
d
us
ing
a
s
e
a
r
c
h
c
r
os
s
-
v
a
li
da
ti
on
gr
id
to
p
r
oduc
e
a
n
e
f
f
e
c
ti
ve
a
nd
opti
mal
biodi
e
s
e
l
pr
e
diction
model
[
34]
.
T
his
s
tudy
is
c
r
it
ica
l
be
c
a
us
e
the
e
ne
r
gy
im
po
r
ted
e
s
ti
mation
model
on
r
oof
top
s
olar
P
V
s
ys
tems
c
a
n
he
lp
plan
e
ne
r
gy
c
ons
umpt
ion
e
f
f
icie
ntl
y
by
a
c
c
ur
a
tely
e
s
ti
mating
the
a
mount
of
e
ne
r
gy
im
po
r
ted
f
r
om
the
gr
id.
I
n
a
ddit
ion,
li
m
it
e
d
s
tudi
e
s
dis
c
us
s
ing
e
s
ti
m
a
ti
ng
e
ne
r
gy
im
por
ted
a
r
e
us
e
d
a
s
a
n
indi
c
a
tor
in
planning
a
nd
ins
talli
ng
r
oof
top
s
olar
P
V
.
T
he
e
xis
tenc
e
of
a
n
e
ne
r
gy
im
por
ted
e
s
ti
mation
model
in
the
on
-
gr
id
r
oof
top
s
olar
P
V
s
ys
tem
c
a
n
a
c
c
e
ler
a
te
the
im
pleme
ntati
on
of
r
oof
top
s
olar
P
V
tec
hnology
that
c
a
n
r
e
duc
e
c
a
r
bon
e
mi
s
s
ions
in
ur
ba
n
a
r
e
a
s
.
T
he
pur
pos
e
of
th
is
s
tu
dy
is
to
model
e
ne
r
gy
im
po
r
ted
e
s
ti
mation
us
ing
mac
hine
lea
r
ning
(
M
L
)
a
lgo
r
it
hms
,
na
mely
ne
ur
a
l
ne
twor
k
(
NN
)
,
l
inea
r
r
e
gr
e
s
s
ion
(
L
R
)
,
a
nd
s
uppor
t
ve
c
tor
mac
hine
(
S
VM
)
.
T
he
da
tas
e
t
wa
s
obtaine
d
f
r
om
the
s
im
ul
a
ti
on
r
e
s
ult
s
on
the
r
oof
top
s
olar
P
V
on
-
gr
id
s
ys
tem,
a
nd
e
ne
r
gy
da
ta
that
c
a
n
mee
t
the
ne
e
ds
of
e
lec
tr
ica
l
loads
wa
s
obtaine
d.
T
he
e
ne
r
gy
im
por
t
ge
ne
r
a
ted
f
r
om
the
s
ys
tem
is
divi
de
d
ba
s
e
d
on
the
a
z
im
uth
a
ngular
dir
e
c
ti
on
that
de
s
c
r
ibes
the
model
a
nd
s
ha
pe
of
the
b
uil
ding's
r
oof
(
one
dir
e
c
ti
on
,
two
dir
e
c
ti
ons
,
a
nd
thr
e
e
dir
e
c
ti
ons
)
.
A
c
ompar
is
on
of
the
mac
hine
lea
r
ning
a
lgor
it
hm
r
e
s
ult
s
s
hows
NN
pr
oduc
e
s
the
s
malles
t
R
M
S
E
va
lue
of
the
thr
e
e
.
T
he
n,
model
im
p
r
ove
ments
a
r
e
c
a
r
r
ied
out
us
ing
GSC
V
to
obtain
mor
e
a
c
c
ur
a
te
model
r
e
s
ult
s
.
B
y
c
ombi
ning
the
GSC
V
pr
oc
e
s
s
with
NN
(
GSC
V
-
NN
)
,
the
model's
r
e
s
ult
s
c
a
n
e
nha
nc
e
the
R
M
S
E
r
e
s
ult
s
e
ve
n
mor
e
a
c
c
ur
a
tely.
Although
e
ne
r
gy
im
por
t
e
s
ti
mation
modeling
c
ompar
e
s
s
im
ple
a
lgor
it
hms
s
uc
h
a
s
NN
,
L
R
,
a
nd
S
VM
,
a
dding
the
GSC
V
pr
oc
e
s
s
c
a
n
im
pr
ove
t
he
highl
y
a
c
c
ur
a
te
e
s
ti
mation
model
on
the
GS
C
V
-
NN
c
ombi
ne
d.
T
his
s
tudy
pr
ovides
ne
w
ins
ight
s
int
o
modeling
e
ne
r
gy
im
por
ted
e
s
ti
mate
s
.
Ac
c
ur
a
te
modeling
r
e
s
ult
s
c
a
n
pr
ovide
a
n
ove
r
view
of
a
c
ompelli
ng
r
oof
top
s
olar
P
V
on
-
gr
id
ins
tallation
plan
ba
s
e
d
on
the
a
mount
of
e
ne
r
gy
im
por
ted
f
r
o
m
the
g
r
id.
F
ur
ther
mor
e
,
the
im
pleme
ntation
be
twe
e
n
the
s
im
ulation
r
e
s
ult
s
of
the
e
ne
r
gy
im
por
t
e
s
ti
mation
model
a
nd
the
a
ppli
c
a
ti
on
of
a
c
tual
c
ondit
ions
is
s
tudi
e
d
holi
s
ti
c
a
ll
y.
2.
M
AT
E
R
I
AL
AN
D
M
E
T
HO
D
C
ompr
e
he
ns
ively
c
onduc
t
a
r
e
view
o
f
va
r
ious
t
hings
to
c
ompl
e
te
th
is
r
e
s
e
a
r
c
h.
I
nit
ially,
it
will
e
xplain
the
c
oll
e
c
ti
on
o
f
da
ta
us
e
d
in
thi
s
pr
oc
e
s
s
.
B
e
f
or
e
da
ta
p
r
oc
e
s
s
ing,
a
pr
e
-
pr
oc
e
s
s
ing
s
tage
will
be
c
a
r
r
ied
out
to
pr
e
pa
r
e
the
da
ta
to
be
modelled
.
Va
r
ious
mac
hine
lea
r
ning
models
wil
l
be
tr
ied
a
nd
c
ompar
e
d,
a
nd
opti
mi
z
a
ti
ons
will
a
dd
a
c
c
ur
a
c
y
to
e
nha
nc
e
their
pe
r
f
or
manc
e
.
T
he
e
va
luation
mus
t
be
a
ppli
e
d
to
mea
s
ur
e
model
pe
r
f
o
r
manc
e
a
nd
thor
oughly
a
na
l
yz
e
the
a
r
c
hit
e
c
tur
e
.
I
t
dis
c
us
s
e
s
the
r
e
s
ult
ing
m
ode
l
f
or
s
olvi
ng
pr
oblems
a
nd
e
va
luate
s
the
r
e
gr
e
s
s
ion
mo
de
l
to
e
s
ti
mate
e
ne
r
gy
im
po
r
ted
in
the
on
-
gr
id
r
oo
f
top
s
olar
P
V
s
ys
tem.
Da
ta
a
c
c
ur
a
c
y
is
a
mea
s
ur
e
of
the
e
s
ti
mation
model
r
e
li
e
d
on
in
da
ta
-
ba
s
e
d
r
e
s
e
a
r
c
h.
I
n
ge
ne
r
a
l,
thr
ough
thr
e
e
s
tage
s
:
da
ta
s
e
lec
ti
o
n,
e
va
luation,
a
nd
c
ompar
is
on
of
the
r
e
s
ult
ing
r
e
gr
e
s
s
ion
model.
T
he
mos
t
a
c
c
ur
a
te
models
a
r
e
opti
mi
z
e
d
to
e
nha
nc
e
pe
r
f
o
r
manc
e
in
p
r
e
dicting
t
e
s
t
da
ta.
F
igur
e
1
s
hows
the
modeling
f
r
a
mew
or
k
in
thi
s
s
tudy,
s
tar
ti
ng
f
r
om
da
ta
c
oll
e
c
ti
on,
p
r
e
-
pr
oc
e
s
s
ing,
modeling
pr
oc
e
s
s
,
a
nd
model
e
va
luation.
T
he
r
e
s
ult
s
will
be
dis
c
us
s
e
d
in
the
dis
c
us
s
ion.
T
he
s
tage
s
of
c
ompl
e
ti
ng
thi
s
s
tudy
will
be
e
xplaine
d
in
mor
e
de
tail
a
s
f
ol
lows
.
2.
1.
S
t
e
p
1:
Dat
a
c
oll
e
c
t
ion
Da
ta
is
obtaine
d
f
r
o
m
ope
n
s
our
c
e
s
a
nd
pr
oc
e
s
s
e
d
on
P
Vs
ys
t
f
or
a
n
on
-
gr
id
s
olar
P
V
s
ys
tem
a
na
lys
is
.
T
his
da
tas
e
t
ha
s
a
tt
r
ibut
e
s
:
latit
ude
(
°)
,
l
ongit
ude
(
°)
,
a
i
r
pr
e
s
s
ur
e
(
k
P
a
)
,
tempe
r
a
tur
e
(
°C
)
,
r
e
lative
humi
dit
y
(
%
)
,
wind
s
pe
e
d
(
m/
s
)
,
pr
e
c
ipi
tation
(
m
m/
da
y)
,
to
tal
ins
ulation
(
W
/m
²)
,
ti
lt
a
ngle
(
°)
,
a
z
im
uth
a
ngle
(
°)
,
p
r
oduc
ti
on
e
ne
r
gy
in
Kilowa
tt
hou
r
pe
r
ye
a
r
(
k
W
h/yea
r
)
,
a
r
r
a
y
e
ne
r
gy
(
kW
h/yea
r
)
,
a
nd
ha
s
labe
ls
:
e
ne
r
gy
im
por
ts
(
kW
h/yea
r
)
.
M
e
teor
ologi
c
a
l
pa
r
a
mete
r
s
a
r
e
obtaine
d
f
r
om
NA
S
A
P
r
e
diction
o
f
W
or
ldwide
E
ne
r
gy
R
e
s
our
c
e
s
.
T
he
da
tas
e
t
is
obtaine
d
f
r
om
e
a
c
h
of
th
e
20
pr
ovincia
l
c
a
pit
a
ls
in
I
ndone
s
ia,
a
s
s
hown
in
F
igur
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
e
d
ac
c
ur
ac
y
e
s
ti
mation
mode
l
e
ne
r
gy
impor
t
in
on
-
gr
id
r
ooft
op
s
olar
photovoltai
c
(
A
lf
in
Sahr
in
)
5973
T
h
is
a
n
gl
e
d
e
s
c
r
i
be
s
t
ha
t
t
he
t
ype
a
nd
s
ha
pe
o
f
t
he
r
o
o
f
o
f
a
bu
il
di
ng
c
a
n
c
ons
is
t
o
f
o
ne
or
m
o
r
e
s
id
e
s
.
T
h
e
a
n
gu
la
r
d
i
r
e
c
ti
on
f
o
l
lo
ws
th
e
e
i
gh
t
c
a
r
di
na
l
di
r
e
c
t
io
ns
:
no
r
th
(
N
)
,
n
or
th
e
a
s
t
(
N
E
)
,
e
a
s
t
(
E
)
,
s
ou
t
he
a
s
t
(
S
E
)
,
s
ou
th
(
S
)
,
s
ou
th
we
s
t
(
S
W
)
,
we
s
t
(
W
)
,
a
n
d
n
o
r
t
hwe
s
t
(
NW
)
a
s
s
how
n
i
n
F
ig
u
r
e
3
.
T
h
e
da
tas
e
t
is
d
ivi
de
d
i
nt
o
t
h
r
e
e
c
lu
s
te
r
s
b
a
s
e
d
o
n
th
e
nu
mb
e
r
of
a
z
i
mu
th
a
ng
les
a
n
d
th
e
nu
mb
e
r
o
f
i
nv
e
r
te
r
s
:
on
e
-
d
i
r
e
c
ti
on
,
t
wo
-
dir
e
c
t
io
ns
,
a
nd
th
r
e
e
-
d
i
r
e
c
ti
ons
.
T
he
n
u
mbe
r
o
f
da
tas
e
t
s
a
mp
les
in
the
p
r
o
vi
nc
ia
l
c
a
pi
ta
l
is
di
vi
de
d
i
nt
o
t
h
r
e
e
c
om
bi
na
t
io
ns
,
a
s
s
ho
wn
in
T
a
b
le
1
.
De
ta
il
s
o
f
t
he
lo
c
a
ti
on
o
f
e
a
c
h
p
r
ov
in
c
i
a
l
c
a
p
i
ta
l
is
s
ho
wn
in
T
a
b
le
2
.
F
igur
e
1.
M
ode
li
ng
f
r
a
mew
or
k
F
igur
e
2.
P
r
ov
incia
l
c
a
pit
a
l
F
igur
e
3.
On
-
gr
id
s
olar
P
V
s
ys
tem
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
1
4
,
No.
5
,
Oc
tober
2
02
4
:
5970
-
5983
5974
T
a
ble
1
.
Da
tas
e
t
s
a
mpl
e
s
C
ombi
na
ti
on
A
mount
of
da
ta
O
ne
-
di
r
e
c
ti
on
2400
T
w
o
-
di
r
e
c
ti
ons
2800
T
hr
e
e
-
di
r
e
c
ti
ons
5300
T
a
ble
2
.
L
oc
a
ti
on
of
p
r
ovincia
l
c
a
pit
a
l
pos
it
ion
No
P
r
ovi
nc
ia
l
c
a
pi
ta
l
L
a
ti
tu
de
L
ongi
tu
de
No
P
r
ovi
nc
ia
l
c
a
pi
ta
l
L
a
ti
tu
de
L
ongi
tu
de
1
B
a
nda
A
c
e
h
5.552846°
95.319291°
11
B
a
nj
a
r
ba
r
u
-
2.636591°
115.114494°
2
M
e
da
n
3.589665°
98.673826°
12
S
a
ma
r
in
da
-
0.501780°
117.139309°
3
P
a
da
ng
-
0.924759°
100.363256°
13
P
ont
ia
na
k
-
0.022690°
109.344749°
4
P
a
le
mba
ng
-
2.988830°
104.756857°
14
M
a
ka
s
s
a
r
-
5.134296°
119.412428°
5
J
a
ka
r
ta
-
6.175394°
106.827183°
15
M
a
na
do
1.490058°
124.840871°
6
B
a
ndung
-
6.934469°
107.604954°
16
K
upa
ng
-
10.163221°
123.601776°
7
S
e
ma
r
a
ng
-
6.990399°
110.422910°
17
A
mbon
-
3.695943°
128.178785°
8
Y
ogya
ka
r
ta
-
7.977838°
110.367226°
18
S
or
ong
-
0.863410°
131.254480°
9
S
ur
a
ba
ya
-
7.245972°
112.737827°
19
J
a
ya
pur
a
-
2.538754°
140.703739°
10
D
e
npa
s
a
r
-
8.652497°
115.219117°
20
M
e
r
a
uke
-
7.792519°
140.018355°
F
igur
e
3
s
hows
the
s
ys
tem
is
a
n
on
-
gr
id
s
ys
tem
without
ba
tt
e
r
ies
.
T
he
s
ys
tem
is
buil
t
us
ing
P
Vs
ys
t
a
nd
opti
mally
s
im
ulate
d
to
a
c
hieve
ope
r
a
ti
ona
l
viabili
ty.
E
ne
r
gy
im
por
ts
a
r
e
c
a
r
r
ied
ou
t
th
r
ough
im
por
t
-
e
xpor
t
mete
r
s
(
e
xim
mete
r
s
)
.
T
he
a
mount
o
f
e
n
e
r
gy
s
e
nt
f
r
om
the
r
oo
f
top
s
olar
P
V
on
-
gr
id
s
ys
tem
a
nd
r
e
c
e
ived
f
r
om
the
gr
id
is
r
e
c
or
de
d
thr
ough
thi
s
mea
s
ur
ing
ins
tr
ument.
T
he
load
is
de
s
igned
f
or
h
ous
e
hold
c
ons
umer
s
a
nd
ha
s
be
e
n
c
ons
tant
ove
r
the
ye
a
r
s
.
L
oa
d
ope
r
a
tes
f
or
s
e
ve
n
da
ys
a
we
e
k
a
nd
24
hou
r
s
a
da
y.
T
a
ble
3
s
hows
the
a
ppli
a
nc
e
of
c
ons
umed
load
.
T
he
number
o
f
P
V
modul
e
s
us
e
d
is
24,
with
a
c
a
pa
c
it
y
of
250
W
p.
M
odule
volt
a
ge
26
V
with
polycr
ys
talli
ne
type.
Ope
n
c
ir
c
uit
volt
a
ge
(
)
37.
84
V
a
nd
s
hor
t
c
ir
c
uit
c
ur
r
e
nt
(
)
8
.
710
A
.
s
e
r
ies
a
nd
s
tr
ing
s
pli
c
ing
n
umber
of
modul
e
s
a
djus
ted
to
a
z
im
uth
di
r
e
c
ti
on
.
T
he
c
a
pa
c
it
y
of
the
inve
r
ter
us
e
d
is
a
djus
ted
in
the
a
z
i
mut
h
dir
e
c
ti
on
.
I
n
e
a
c
h
dir
e
c
ti
on
,
one
inver
ter
is
i
ns
talled,
a
nd
a
ll
inver
ter
s
ha
ve
a
f
r
e
que
nc
y
of
50
Hz
monop
ha
s
e
with
a
gr
id
volt
a
ge
of
230
V
.
T
a
ble
3
.
L
oa
d
c
ons
umpt
ion
Nb
A
ppl
ia
nc
e
P
ow
e
r
U
ni
t
D
a
il
y
us
e
(
hour
s
)
D
a
il
y
e
ne
r
gy
(
W
h)
10
L
a
mps
(
L
E
D
or
F
lu
o)
10
W
/l
a
mp
12
1200
2
T
V
/P
C
/M
obi
le
100
W
/a
pp
7
1400
7
D
ome
s
ti
c
a
ppl
ia
nc
e
s
150
W
/a
pp
24
25200
2
F
r
id
ge
/De
e
p
-
f
r
e
e
z
e
0.4
kW
/d
a
y
24
802
1
D
is
h a
nd
c
lo
th
w
a
s
h
e
r
300
W
a
ve
r
.
2
600
3
A
ir
c
ondi
ti
one
r
250
W
/a
pp
10
7500
1
H
e
a
te
r
500
W
/a
pp
5
2500
1
S
ta
nd
-
by c
ons
ume
r
s
100
W
24
2400
T
ot
a
l
da
il
y
e
ne
r
gy
41602 W
h/
da
y
M
ont
hl
y
e
ne
r
gy
1248.0 kW
h/
mt
h
2.
2.
S
t
e
p
2:
Dat
a
p
r
e
-
p
r
oc
e
s
s
in
g
E
f
f
e
c
ti
ve
de
c
is
ions
c
a
n
r
e
s
ult
f
r
om
a
c
c
ur
a
te
e
s
ti
mate
s
in
da
ta
a
na
lys
is
.
Util
izing
his
tor
ica
l
da
ta
is
one
of
the
da
ta
-
dr
iven
modeling
methods
.
E
f
f
e
c
ti
ve
modeling
r
e
quir
e
s
the
a
bil
it
y
to
us
e
the
pr
ope
r
c
omput
a
ti
ona
l
a
nd
s
tatis
ti
c
a
l
tec
hniques
.
F
or
da
ta
-
dr
iven
modeling
a
nd
a
c
c
ur
a
te
e
s
ti
mate
s
to
be
pr
o
duc
e
d,
it
is
ne
c
e
s
s
a
r
y
to
know
the
s
teps
of
the
p
r
oc
e
s
s
.
T
he
d
a
ta
c
lea
ning
identif
ies
a
nd
ha
ndl
es
mi
s
s
ing
or
incomplete
da
ta.
I
t
c
a
n
be
done
by
f
il
li
ng
in
or
de
leting
m
is
s
ing
va
lues
with
a
n
a
ve
r
a
ge
va
lue
.
I
n
the
d
a
ta
no
r
m
a
li
z
a
ti
on
,
the
pr
oc
e
s
s
of
c
ha
nging
da
tas
e
t
va
lues
s
o
that
they
ha
ve
un
if
or
m
va
lues
.
T
he
da
tas
e
t
va
lue
is
c
ha
nge
d
to
ha
ve
a
mi
nim
um
va
lue
of
0
a
nd
a
maximum
o
f
1
.
T
he
p
r
oc
e
s
s
is
pe
r
f
or
med
us
ing
(
1
)
.
=
X
−
m
i
n
(
)
m
a
x
(
)
−
m
i
n
(
)
(
1)
is
the
nor
malize
d
va
lue,
(
)
(
)
is
the
a
bs
olut
e
mi
nim
um
a
nd
maximum
va
lue,
a
nd
is
the
or
igi
na
l
va
lue
.
I
n
d
a
ta
s
tanda
r
diza
ti
on
,
the
pr
oc
e
s
s
of
c
ha
nging
the
va
lues
o
f
a
da
tas
e
t
s
o
that
it
ha
s
a
mea
n
va
lue
of
z
e
r
o
a
nd
a
s
tanda
r
d
de
viation
of
one
by
us
ing
(
2
)
.
=
−
(
2)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
e
d
ac
c
ur
ac
y
e
s
ti
mation
mode
l
e
ne
r
gy
impor
t
in
on
-
gr
id
r
ooft
op
s
olar
photovoltai
c
(
A
lf
in
Sahr
in
)
5975
is
the
s
tanda
r
dize
d
va
lue,
is
the
a
ve
r
a
ge
of
a
ll
d
a
tas
e
t
va
lues
,
is
the
de
viation
o
f
a
ll
da
tas
e
t
va
lues
,
a
nd
is
the
o
r
igi
na
l
va
lue
.
2.
3.
S
t
e
p
3:
Ne
u
r
al
n
e
t
wor
k
M
ode
li
ng
us
ing
s
upe
r
vis
e
d
M
L
to
e
s
ti
mate
is
e
s
s
e
nti
a
l
[
35]
.
M
ode
l
lea
r
ning
f
r
om
the
t
r
a
ini
ng
da
tas
e
t
a
nd
a
pply
the
known
pa
tt
e
r
ns
to
pe
r
f
or
m
e
s
ti
mate
s
tes
ted
with
da
ta
s
e
t
tes
ti
ng.
T
he
lea
r
ning
a
lgor
it
hm
in
thi
s
s
tudy
us
e
s
a
NN
a
nd
is
c
ompar
e
d
with
L
R
a
nd
S
VM
a
lgo
r
it
hms
.
T
he
ne
ur
a
l
ne
twor
k
is
s
hown
in
F
igur
e
4
a
nd
(
3
)
-
(
4
)
.
=
0
(
∑
,
∗
=
1
)
(
3)
∗
=
1
(
∑
∗
=
1
)
(
4)
NN
c
ons
is
ts
o
f
a
ne
tw
o
r
k
o
f
la
ye
r
e
d
w
e
i
gh
ts
wi
th
i
nt
e
r
c
on
ne
c
te
d
mat
he
ma
ti
c
a
l
ope
r
a
t
or
s
(
n
e
u
r
o
ns
)
.
W
he
r
e
i
s
t
he
n
um
be
r
o
f
in
pu
ts
,
,
i
s
th
e
w
e
i
gh
t
o
f
th
e
i
np
u
t
va
r
i
a
b
le
,
i
s
th
e
n
e
u
r
o
n
tha
t
ha
s
a
n
a
c
t
iv
a
t
io
n
f
u
nc
ti
on
,
i
s
t
he
ou
tp
ut
,
i
s
the
we
ig
ht
o
f
,
a
n
d
0
a
nd
1
a
r
e
a
c
t
iva
t
io
n
f
u
nc
ti
ons
wi
t
h
s
i
gm
oi
d
f
u
nc
ti
ons
a
s
i
n
(
5
)
.
(
)
=
1
1
+
−
(
5)
NN
de
s
c
r
ibes
the
human
br
a
in,
whic
h
c
ons
is
ts
of
many
c
e
ll
s
c
a
ll
e
d
ne
ur
ons
that
f
or
m
br
a
nc
he
d
ne
twor
ks
.
C
onne
c
ted
ne
ur
ons
s
e
nd
s
ignals
to
e
a
c
h
other
to
p
r
oc
e
s
s
inf
or
mation
a
nd
a
r
e
c
onne
c
ted
to
the
ne
xt
ne
ur
on.
E
s
s
e
nti
a
ll
y,
ne
ur
ons
c
onne
c
t
thr
e
e
laye
r
s
i
n
a
n
a
r
ti
f
icia
l
ne
ur
a
l
ne
twor
k
a
s
f
o
ll
ows
[
36]
.
a.
I
nput
laye
r
c
onne
c
ts
the
in
f
or
mation
r
e
c
e
ived
on
t
his
laye
r
:
latit
ude
,
longi
tude
,
a
ir
pr
e
s
s
ur
e
,
tempe
r
a
tur
e
,
r
e
lative
humi
dit
y
,
wind
s
pe
e
d,
p
r
e
c
ipi
tation,
tot
a
l
i
ns
ulation,
ti
lt
a
ngle
,
a
z
im
uth
a
ngle,
p
r
oduc
ti
on
e
ne
r
gy,
a
r
r
a
y
e
ne
r
gy
,
a
nd
e
ne
r
gy
i
mpor
ts
.
I
t
is
then
pr
oc
e
s
s
e
d,
a
na
lyze
d
ba
s
e
d
on
the
node
s
of
thi
s
laye
r
,
a
nd
moved
to
the
ne
xt
laye
r
.
b.
Hidde
n
laye
r
:
Hidde
n
laye
r
s
ge
t
input
f
r
om
othe
r
laye
r
s
or
hidden
laye
r
s
.
Hidde
n
laye
r
s
in
a
n
a
r
ti
f
icia
l
ne
ur
a
l
ne
twor
k
c
a
n
numbe
r
mor
e
than
one
.
T
he
ou
tput
f
r
o
m
the
p
r
e
vious
laye
r
is
p
r
oc
e
s
s
e
d,
a
na
lyze
d,
a
nd
c
onti
nue
d
to
the
ne
xt
laye
r
.
c.
Output
laye
r
:
All
p
r
e
vious
pr
oc
e
s
s
e
s
will
e
nd
on
t
his
node
,
a
nd
then
the
tr
a
ns
mi
s
s
ion
pr
oc
e
s
s
will
o
c
c
ur
.
T
he
output
va
lue
will
be
"
1"
or
"
0"
a
c
c
or
ding
to
th
e
a
c
ti
va
ti
on
f
unc
ti
on
.
F
igur
e
4.
Ne
ur
a
l
ne
two
r
ks
2.
4.
S
t
e
p
4:
Grid
s
e
ar
c
h
c
r
os
s
-
vali
d
at
ion
Gr
id
s
e
a
r
c
h
c
r
os
s
-
va
li
da
ti
on
(
GSC
V)
is
a
tec
hniqu
e
us
e
d
to
e
nha
nc
e
model
pe
r
f
or
manc
e
by
f
indi
ng
the
opti
mal
c
ombi
na
ti
on
of
pa
r
a
mete
r
s
.
S
ha
r
ing
t
r
a
ini
ng
a
nd
tes
ti
ng
da
ta
is
a
s
tep
that
mus
t
be
de
ter
mi
ne
d
in
a
dva
nc
e
.
T
his
s
tudy
divi
de
d
the
da
ta
ba
s
e
d
on
80%
tr
a
ini
ng
a
nd
20%
tes
ti
ng
da
ta
[
37
]
.
I
n
the
tr
a
ini
ng
pr
oc
e
s
s
,
da
ta
is
divi
de
d
int
o
10
-
f
old,
9
-
f
old
da
ta
a
s
tr
a
ini
ng,
a
nd
1
-
f
old
da
ta
a
s
va
li
da
ti
on
of
a
s
many
a
s
ten
c
r
os
s
e
s
.
T
he
a
ve
r
a
ge
va
li
da
t
ion
va
lue
is
the
be
s
t
hype
r
pa
r
a
mete
r
s
e
a
r
c
h
f
o
r
tes
ti
ng
f
o
r
a
r
ti
f
ici
a
l
ne
ur
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
1
4
,
No.
5
,
Oc
tober
2
02
4
:
5970
-
5983
5976
ne
twor
k
pa
r
a
mete
r
s
,
a
s
il
lus
tr
a
ted
in
F
igu
r
e
5.
T
h
is
c
r
os
s
-
va
li
da
ti
on
pr
oc
e
s
s
is
c
a
r
r
ied
out
r
a
ndoml
y
on
the
s
e
a
r
c
h
gr
id
with
vulne
r
a
ble
number
s
o
f
2
–
100
.
2.
5.
S
t
e
p
5:
M
od
e
l
e
valu
a
t
ion
T
o
e
va
luate
model
pe
r
f
or
manc
e
in
r
e
gr
e
s
s
ion
m
ode
li
ng
us
ing
R
M
S
E
.
I
t
c
a
n
indi
c
a
te
the
model's
a
c
c
ur
a
c
y
with
the
a
c
tual
va
lues
in
the
da
tas
e
t.
T
he
be
tt
e
r
the
a
c
c
ur
a
c
y
pr
e
diction
pe
r
f
o
r
manc
e
,
the
s
m
a
ll
e
r
the
R
M
S
E
va
lue.
T
he
R
M
S
E
is
s
hown
in
(
6
)
,
=
√
∑
(
1
−
ŷ1
)
2
=
1
(
6)
w
he
r
e
is
t
he
n
u
mbe
r
o
f
s
a
mp
le
da
ta
in
t
he
da
tas
e
t
,
i
s
the
a
c
tu
a
l
va
lu
e
o
f
th
e
d
a
t
a
,
a
nd
ŷ
i
s
th
e
va
l
ue
tha
t
t
he
mo
de
l
pr
e
d
ic
ts
f
o
r
th
e
d
a
t
a
.
T
his
s
tu
dy
c
o
m
pa
r
e
d
t
he
R
M
S
E
r
e
s
ul
ts
o
f
t
he
NN
,
L
R
,
a
n
d
S
VM
m
od
e
l
s
b
e
f
o
r
e
a
n
d
a
f
t
e
r
t
he
a
dd
it
i
on
of
G
S
C
V
in
a
c
o
mb
i
na
t
i
on
o
f
o
ne
-
d
i
r
e
c
t
io
n
,
t
wo
-
d
i
r
e
c
t
io
ns
,
a
nd
th
r
e
e
-
d
ir
e
c
t
io
ns
.
F
igur
e
5.
10
-
f
old
c
r
os
s
-
va
li
da
ti
on
diagr
a
m
3.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
T
his
s
tudy
a
im
s
to
make
a
n
e
s
ti
mate
d
model
of
im
por
ted
e
ne
r
gy
f
r
o
m
on
-
gr
id
r
oo
f
top
s
olar
P
V
s
ys
tems
.
An
e
s
s
e
nti
a
l
pa
r
t
of
thi
s
r
e
s
e
a
r
c
h
is
the
model
ge
ne
r
a
ted
f
r
om
tr
a
ini
ng
da
ta
with
va
li
da
ti
o
n
that
is
s
e
pa
r
a
ted
indepe
nde
ntl
y
without
invol
ving
tes
ti
n
g
da
ta.
T
his
method
c
a
n
e
nha
nc
e
the
a
c
c
ur
a
c
y
of
be
tt
e
r
e
s
ti
mation
models
.
T
he
r
e
s
ult
s
s
how
the
c
or
r
e
la
ti
on
be
twe
e
n
a
tt
r
ibut
e
va
r
iable
s
in
the
da
tas
e
t
u
s
ing
the
c
or
r
e
lation
matr
ix
,
a
s
s
hown
in
F
igur
e
6,
a
s
the
a
tt
r
ibut
e
pa
r
a
mete
r
o
f
the
e
s
ti
mation
model.
T
h
e
c
o
r
r
e
lat
i
on
ma
t
r
i
x
mea
s
u
r
e
s
t
he
s
t
r
e
ng
th
a
n
d
d
i
r
e
c
t
io
n
of
t
he
r
e
l
a
t
io
ns
h
ip
b
e
t
we
e
n
tw
o
v
a
r
ia
b
les
c
o
r
r
e
s
p
on
di
ng
to
v
a
l
ue
s
r
a
n
gi
ng
f
r
o
m
-
1
to
1
.
I
f
th
e
c
o
r
r
e
la
t
io
n
va
lue
is
c
l
os
e
t
o
1
,
it
in
di
c
a
t
e
s
a
pe
r
f
e
c
t
p
os
i
ti
ve
c
o
r
r
e
l
a
t
io
n
tha
t
te
nds
to
m
ove
i
n
t
he
s
a
me
d
ir
e
c
ti
o
n
.
A
c
o
r
r
e
la
ti
on
va
lue
r
e
l
a
t
iv
e
t
o
-
1
in
d
ica
tes
a
p
e
r
f
e
c
t
n
e
g
a
t
iv
e
c
o
r
r
e
la
ti
on
.
T
he
va
r
i
a
b
le
te
nd
s
to
m
ov
e
in
t
he
o
pp
os
i
te
d
i
r
e
c
ti
on
.
H
ow
e
ve
r
,
i
f
t
he
c
or
r
e
l
a
t
io
n
va
lu
e
is
c
los
e
to
0
,
it
me
a
ns
th
e
a
bs
e
nc
e
o
f
c
o
r
r
e
l
a
t
io
n
or
w
e
a
k
c
or
r
e
la
t
io
n
b
e
t
we
e
n
v
a
r
i
a
b
les
.
I
n
F
i
gu
r
e
6
,
th
e
mos
t
p
o
ten
t
p
os
it
iv
e
c
o
r
r
e
la
ti
ons
to
im
po
r
te
d
e
n
e
r
g
y
a
r
e
la
ti
tu
de
(
0
.
32
8
)
,
r
e
l
a
t
iv
e
h
um
id
i
ty
(
0
.
48
4
)
,
p
r
e
c
ip
i
ta
ti
ons
(
0
.
48
4
)
,
a
nd
t
i
lt
a
n
g
le
(
0
.
4
74
)
.
T
he
ne
ga
ti
ve
ly
c
or
r
e
la
ted
a
r
e
l
on
gi
tu
de
(
-
0
.
4
29
)
,
te
m
pe
r
a
tu
r
e
(
-
0
.
2
90
)
,
w
in
d
s
pe
e
d
(
-
0
.
4
98
)
,
to
ta
l
i
ns
ola
t
io
n
(
-
0
.
2
40
)
,
p
r
o
duc
t
io
n
e
ne
r
g
y
(
-
0
.
87
7
)
,
a
n
d
a
r
r
a
y
e
n
e
r
g
y
(
-
0
.
5
55
)
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
e
d
ac
c
ur
ac
y
e
s
ti
mation
mode
l
e
ne
r
gy
impor
t
in
on
-
gr
id
r
ooft
op
s
olar
photovoltai
c
(
A
lf
in
Sahr
in
)
5977
T
he
dis
c
us
s
ion
of
im
por
ted
e
ne
r
gy
on
the
on
-
gr
id
r
oof
top
s
olar
P
V
s
ys
tem
he
r
e
goe
s
f
ur
ther
to
us
ing
e
lec
tr
ica
l
e
ne
r
gy
that
ha
s
r
e
a
c
he
d
c
ons
umer
s
.
T
he
pr
oduc
ti
on
of
s
olar
P
V
e
ne
r
gy
s
hould
be
a
ble
to
s
upply
a
ll
the
ne
e
ds
of
c
ons
umer
loads
.
How
e
ve
r
,
with
th
e
f
luctua
ti
ng
na
tur
e
of
s
olar
P
V
due
to
the
in
f
l
ue
nc
e
of
e
xter
na
l
we
a
ther
c
ondit
ions
,
e
ne
r
gy
ne
e
ds
to
be
s
uppli
e
d
f
r
o
m
the
gr
id
.
I
mpor
ted
e
ne
r
gy
on
a
n
on
-
gr
id
r
oof
top
s
olar
P
V
s
ys
tem
s
hows
how
much
e
lec
tr
i
c
a
l
e
ne
r
gy
is
de
li
ve
r
e
d
f
r
om
the
g
r
id
s
ys
tem.
T
he
s
maller
e
ne
r
gy
pr
ovided
by
the
gr
id
s
hows
a
s
igni
f
ica
nt
e
f
f
e
c
t
of
r
e
ne
wa
ble
e
ne
r
gy
on
r
e
duc
ing
c
a
r
bon
e
mi
s
s
ions
be
c
a
us
e
the
gr
id
s
ys
tem
s
ti
ll
us
e
s
many
pr
im
a
r
y
mate
r
ials
f
r
om
f
os
s
il
s
.
T
his
s
tudy
s
hows
the
e
s
ti
mate
d
a
mount
of
i
mpor
ted
e
ne
r
gy
in
20
pr
ovincia
l
c
a
pit
a
ls
in
I
ndone
s
ia.
T
he
potential
f
or
on
-
gr
id
r
oo
f
top
s
olar
P
V
de
ve
lopm
e
nt
in
I
ndone
s
ia
is
in
big
c
it
ies
with
m
a
ny
buil
dings
a
nd
r
e
s
idential
de
ns
it
y.
E
ne
r
gy
i
m
por
ted
in
e
a
c
h
town
di
f
f
e
r
s
,
e
ve
n
wi
th
the
e
xa
c
t
P
V
c
a
pa
c
it
y,
due
to
other
c
ondit
ions
,
s
uc
h
a
s
ti
lt
a
nd
a
z
im
ut
h
a
ngles
.
T
his
s
tudy
wa
s
divi
de
d
in
to
thr
e
e
da
tas
e
ts
ba
s
e
d
on
the
number
of
a
z
im
uth
a
ngle
di
r
e
c
ti
ons
a
nd
i
nve
r
ter
s
.
E
a
c
h
a
z
im
uth
a
ngle
dir
e
c
ti
on
c
a
n
f
ind
the
s
malles
t
a
mount
of
im
por
ted
e
ne
r
gy,
a
nd
the
loca
ti
o
n
of
the
pr
ovincia
l
c
a
pit
a
ls
is
s
hown
in
T
a
ble
4.
F
igur
e
6.
C
or
r
e
lation
matr
ix
of
a
tt
r
ibut
e
on
-
gr
id
r
o
of
top
s
olar
P
V
T
a
ble
4.
T
he
pr
ovincia
l
c
a
pit
a
l
ha
s
the
lowe
s
t
e
ne
r
gy
im
por
t
C
ombi
na
ti
on
A
ngl
e
di
r
e
c
ti
on
P
r
ovi
nc
ia
l
c
a
pi
ta
l
E
ne
r
gy i
mpor
t
O
ne
-
di
r
e
c
ti
on
−
one
i
nve
r
te
r
nor
th
D
e
npa
s
a
r
5338 kW
h/
ye
a
r
−
two
in
ve
r
te
r
s
nor
th
D
e
npa
s
a
r
5339 kW
h/
ye
a
r
−
th
r
e
e
in
ve
r
te
r
s
nor
th
D
e
npa
s
a
r
5441 kW
h/
ye
a
r
T
w
o
-
di
r
e
c
ti
ons
nor
th
, nor
th
e
a
s
t
D
e
npa
s
a
r
5393 kW
h/
ye
a
r
T
hr
e
e
-
di
r
e
c
ti
ons
nor
th
, nor
th
e
a
s
t,
nor
th
w
e
s
t
D
e
npa
s
a
r
5509 kW
h/
ye
a
r
E
xplor
e
mo
r
e
p
r
of
ound
mac
hine
lea
r
ning
a
lgo
r
i
thm
s
to
e
nha
nc
e
the
a
c
c
ur
a
c
y
a
nd
a
c
c
ur
a
c
y
of
im
por
ted
e
ne
r
gy
e
s
ti
mation
models
.
Appr
oa
c
he
s
s
uc
h
a
s
NN
,
L
R
,
a
nd
S
VM
a
lgor
it
hms
c
a
n
ident
if
y
c
ompl
e
x
pa
tt
e
r
ns
in
his
tor
ica
l
da
ta
a
nd
e
nha
nc
e
the
model'
s
a
bil
it
y
to
e
s
ti
mate
f
luctua
ti
ons
in
im
por
ted
r
oof
t
op
s
olar
P
V
e
ne
r
gy.
NN
r
e
s
ult
s
a
r
e
e
nha
nc
e
d
f
or
be
tt
e
r
a
c
c
ur
a
c
y
with
GSC
V.
T
he
tec
hnology
c
a
n
a
ls
o
be
tt
e
r
a
da
pt
to
e
nvir
onmenta
l
c
ondit
ions
that
a
f
f
e
c
t
the
pe
r
f
or
manc
e
of
s
olar
P
V
s
ys
tems
.
M
or
e
p
r
e
c
is
e
a
nd
a
da
pti
ve
e
s
ti
mation
models
c
a
n
be
pr
oduc
e
d
to
make
s
mar
ter
de
c
is
ions
by
ha
r
ne
s
s
ing
the
powe
r
of
mac
hine
lea
r
ning.
F
igur
e
7
is
the
one
-
dir
e
c
ti
on
a
z
im
uth
a
ngle
of
the
e
ne
r
gy
i
mpor
t
e
s
ti
mation
model
,
F
igur
e
8
s
hows
the
two
-
dir
e
c
ti
ons
a
z
im
uth
a
ngle
o
f
the
e
ne
r
gy
im
por
t
e
s
ti
mation
model,
a
nd
F
igur
e
9
is
the
thr
e
e
-
dir
e
c
ti
ons
a
z
im
uth
a
ngle
of
the
e
ne
r
gy
im
por
t
e
s
ti
mation
model
.
T
he
f
igur
e
e
xplains
the
c
ompar
is
on
be
twe
e
n
NN
,
L
R
,
a
nd
S
VM
,
s
howing
that
NN
ge
ts
mor
e
a
c
c
ur
a
te
r
e
s
u
lt
s
.
T
he
NN
r
e
s
ult
s
we
r
e
e
nha
nc
e
d
to
ge
t
mor
e
a
c
c
ur
a
te
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
1
4
,
No.
5
,
Oc
tober
2
02
4
:
5970
-
5983
5978
r
e
s
ult
s
by
a
dding
GSC
V
s
o
that
it
be
c
a
me
the
GS
C
V
-
NN
r
e
s
ult
.
All
e
s
ti
mation
model
r
e
s
ult
s
a
r
e
il
l
us
tr
a
ted
by
c
ompar
ing
NN
,
L
R
,
S
VM
,
a
nd
GSC
V
-
NN
.
T
his
r
e
s
ult
a
ppli
e
s
to
a
ll
im
a
ge
s
in
the
e
s
ti
mation
model
in
one
-
dir
e
c
ti
on
a
z
im
uth
a
ngle,
two
-
dir
e
c
ti
ons
a
z
im
u
th
a
ngle,
a
nd
thr
e
e
-
dir
e
c
ti
ons
a
z
im
uth
a
ngle
.
T
he
va
lue
of
the
hidden
laye
r
o
f
e
a
c
h
node
a
nd
the
th
r
e
s
hold
in
the
c
ombi
na
ti
on
of
dir
e
c
ti
ons
a
r
e
s
hown
in
T
a
ble
5
.
F
igur
e
7.
One
-
dir
e
c
ti
on
a
z
im
uth
a
ngle
o
f
the
e
ne
r
g
y
im
por
t
e
s
ti
mation
model
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
e
d
ac
c
ur
ac
y
e
s
ti
mation
mode
l
e
ne
r
gy
impor
t
in
on
-
gr
id
r
ooft
op
s
olar
photovoltai
c
(
A
lf
in
Sahr
in
)
5979
F
igur
e
8.
T
wo
-
dir
e
c
ti
ons
a
z
im
uth
a
ngle
of
the
e
ne
r
gy
im
por
t
e
s
ti
mation
model
T
he
GSC
V
pr
oc
e
s
s
c
a
n
s
igni
f
ica
ntl
y
incr
e
a
s
e
the
a
bil
it
y
to
p
r
oduc
e
models
that
pe
r
f
or
m
be
tt
e
r
a
nd
e
nha
nc
e
the
a
c
c
ur
a
c
y
of
i
mpor
ted
e
ne
r
gy
e
s
ti
m
a
ti
on
models
.
GSC
V
c
a
n
s
ys
tema
ti
c
a
ll
y
f
ind
the
opti
mal
pa
r
a
mete
r
in
the
e
s
ti
mation
model
by
c
onduc
ti
ng
a
br
oa
de
r
a
nd
mor
e
c
ompl
e
x
pa
r
a
mete
r
s
e
a
r
c
h.
G
S
C
V
c
a
n
a
ls
o
r
e
duc
e
the
r
is
k
of
ove
r
f
it
ti
ng
to
model
ne
w
da
ta
pr
ope
r
ly.
T
a
ble
6
s
hows
the
r
e
s
ult
s
of
e
nha
n
c
ing
the
a
c
c
ur
a
c
y
of
the
e
ne
r
gy
im
por
t
e
s
ti
mation
model
with
the
a
ddit
ion
of
the
GSC
V
pr
oc
e
s
s
.
I
n
a
one
-
dir
e
c
ti
on
Evaluation Warning : The document was created with Spire.PDF for Python.