I
nte
rna
t
io
na
l J
o
urna
l o
f
P
o
wer
E
lect
ro
nics
a
nd
Driv
e
S
y
s
t
em
(
I
J
P
E
DS)
Vo
l.
1
7
,
No
.
1
,
Ma
r
ch
20
2
6
,
p
p
.
6
9
6
~
7
08
I
SS
N:
2
0
8
8
-
8
6
9
4
,
DOI
: 1
0
.
1
1
5
9
1
/ijp
ed
s
.
v
1
7
.
i
1
.
p
p
6
9
6
-
7
08
696
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
p
e
d
s
.
ia
esco
r
e.
co
m
Perf
o
rma
nce
enh
a
ncement of
phot
o
v
o
ltaic sy
stems us
ing
hybrid
LST
M
–
CNN so
la
r f
o
recas
ting i
nte
g
ra
ted
wit
h P
&
O
M
PP
T
Sa
ra
F
enna
ne,
H
o
ud
a
K
a
ci
m
i,
H
a
m
za
M
a
bcho
ur,
F
a
t
e
hi AL
t
a
lqi
,
Ad
il E
chc
helh
La
b
o
r
a
t
o
r
y
o
f
E
l
e
c
t
r
o
n
i
c
S
y
st
e
ms I
n
f
o
r
mat
i
o
n
P
r
o
c
e
ssi
n
g
M
e
c
h
a
n
i
c
s
a
n
d
E
n
e
r
g
e
t
i
c
s
,
I
b
n
To
f
a
i
l
U
n
i
v
e
r
si
t
y
,
K
e
n
i
t
r
a
,
M
o
r
o
c
c
o
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Ma
y
2
9
,
2
0
2
5
R
ev
is
ed
Dec
2
2
,
2
0
25
Acc
ep
ted
J
an
2
3
,
2
0
2
6
Th
e
in
c
re
a
sin
g
p
e
n
e
trati
o
n
o
f
p
h
o
t
o
v
o
lt
a
ic
(P
V)
sy
ste
m
s
in
s
m
a
rt
g
rid
s
h
ig
h
li
g
h
ts
t
h
e
n
e
e
d
f
o
r
re
li
a
b
le
so
l
u
ti
o
n
s
to
m
it
ig
a
te
th
e
i
n
h
e
re
n
t
in
te
rm
it
ten
c
y
o
f
so
lar
e
n
e
rg
y
.
S
h
o
rt
-
term
v
a
riab
il
it
y
in
s
o
lar
irrad
ian
c
e
re
m
a
in
s
a
c
rit
ica
l
c
h
a
ll
e
n
g
e
fo
r
sta
b
le
g
ri
d
o
p
e
ra
ti
o
n
a
n
d
e
fficie
n
t
P
V
e
n
e
r
g
y
m
a
n
a
g
e
m
e
n
t.
Th
is
p
a
p
e
r
p
r
o
p
o
se
s
a
n
i
n
teg
ra
ted
f
o
r
e
c
a
stin
g
–
c
o
n
tro
l
fra
m
e
wo
rk
t
h
a
t
c
o
m
b
in
e
s
sh
o
rt
-
term
g
lo
b
a
l
h
o
riz
o
n
tal
irrad
ian
c
e
(G
HI)
p
re
d
ictio
n
wit
h
a
c
o
n
v
e
n
t
io
n
a
l
P
&
O
M
P
P
T
stra
teg
y
to
e
n
h
a
n
c
e
P
V
sy
ste
m
p
e
rfo
rm
a
n
c
e
.
A
h
y
b
ri
d
LS
TM
–
CNN
a
rc
h
it
e
c
tu
re
is
d
e
v
e
lo
p
e
d
to
f
o
re
c
a
st
o
n
e
-
ste
p
-
a
h
e
a
d
G
HI
u
n
d
e
r
th
e
se
m
i
-
a
rid
c
li
m
a
ti
c
c
o
n
d
it
i
o
n
s
o
f
Da
k
h
la,
M
o
r
o
c
c
o
,
a
re
g
i
o
n
c
h
a
ra
c
teriz
e
d
b
y
h
ig
h
so
lar
p
o
te
n
ti
a
l
a
n
d
p
r
o
n
o
u
n
c
e
d
irrad
ian
c
e
flu
c
t
u
a
ti
o
n
s.
T
h
e
fo
re
c
a
stin
g
m
o
d
e
l
is
v
a
li
d
a
ted
u
si
n
g
m
e
a
su
re
d
irrad
ian
c
e
d
a
ta
fro
m
th
e
Na
ti
o
n
a
l
Re
n
e
wa
b
le
En
e
rg
y
Lab
o
ra
to
r
y
(
NREL)
v
ia
th
e
Na
ti
o
n
a
l
S
o
lar
Ra
d
iatio
n
Da
tab
a
se
(NSRDB).
P
re
d
icte
d
irrad
ian
c
e
is
th
e
n
u
se
d
to
imp
r
o
v
e
P
V
p
o
we
r
e
stim
a
ti
o
n
a
n
d
su
p
p
o
rt
p
re
d
icti
v
e
m
a
x
imu
m
p
o
we
r
p
o
in
t
trac
k
i
n
g
(
MPP
T
)
o
p
e
ra
ti
o
n
.
S
imu
lati
o
n
re
su
lt
s
o
b
t
a
in
e
d
i
n
M
ATLAB/S
im
u
li
n
k
d
e
m
o
n
stra
te
th
a
t
th
e
p
r
o
p
o
se
d
fra
m
e
wo
rk
a
c
h
i
e
v
e
s
a
c
c
u
ra
te
G
HI
fo
re
c
a
stin
g
,
fa
ste
r
M
P
P
T
c
o
n
v
e
rg
e
n
c
e
,
re
d
u
c
e
d
ste
a
d
y
-
sta
te
o
sc
il
latio
n
s,
a
n
d
imp
ro
v
e
d
P
V
p
o
we
r
sta
b
il
it
y
u
n
d
e
r
ra
p
id
l
y
c
h
a
n
g
i
n
g
i
rra
d
ia
n
c
e
.
Th
e
p
ro
p
o
se
d
a
p
p
r
o
a
c
h
p
r
o
v
i
d
e
s
a
p
ra
c
ti
c
a
l
a
n
d
c
o
m
p
u
tati
o
n
a
ll
y
e
fficie
n
t
so
lu
ti
o
n
fo
r
e
n
h
a
n
c
in
g
th
e
d
y
n
a
m
ic
re
sp
o
n
se
a
n
d
e
n
e
r
g
y
e
x
trac
ti
o
n
e
fficie
n
c
y
o
f
P
V
sy
ste
m
s
in
sm
a
rt
g
rid
a
p
p
li
c
a
ti
o
n
s.
K
ey
w
o
r
d
s
:
Dee
p
lear
n
in
g
Hy
b
r
id
L
STM
-
C
NN
m
o
d
el
P&
O
MPPT
alg
o
r
ith
m
So
lar
en
er
g
y
So
lar
ir
r
ad
ian
ce
p
r
ed
ictio
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sar
a
Fen
n
an
e
L
ab
o
r
ato
r
y
o
f
E
lectr
o
n
ic
Sy
s
t
em
s
I
n
f
o
r
m
atio
n
Pro
ce
s
s
in
g
Me
ch
an
ics an
d
E
n
e
r
g
etics,
Facu
lty
o
f
Scien
ce
s
I
b
n
T
o
f
ail
Un
iv
er
s
ity
Ken
itra
,
Mo
r
o
cc
o
E
m
ail:
s
ar
a.
f
en
n
an
e@
u
it.a
c.
m
a
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
g
r
o
win
g
d
em
a
n
d
f
o
r
cle
an
an
d
s
u
s
tain
ab
le
en
er
g
y
h
a
s
ac
ce
ler
ated
th
e
g
lo
b
al
d
e
p
lo
y
m
en
t
o
f
p
h
o
to
v
o
ltaic
(
PV)
s
y
s
tem
s
,
p
o
s
itio
n
in
g
s
o
lar
p
o
wer
as a
k
ey
co
m
p
o
n
en
t o
f
m
o
d
er
n
s
m
ar
t
g
r
id
s
[
1
]
.
Desp
ite
its
en
v
ir
o
n
m
en
tal
an
d
ec
o
n
o
m
ic
ad
v
an
tag
es,
PV g
en
er
atio
n
r
e
m
ain
s
h
ig
h
ly
v
ar
iab
le
d
u
e
to
it
s
s
tr
o
n
g
d
ep
en
d
en
ce
o
n
m
eteo
r
o
lo
g
ical
co
n
d
itio
n
s
s
u
ch
as
s
o
lar
ir
r
ad
ian
ce
,
win
d
s
p
ee
d
,
h
u
m
i
d
ity
,
an
d
am
b
ien
t
tem
p
er
atu
r
e
[
2
]
.
I
n
ad
d
itio
n
,
p
o
wer
elec
tr
o
n
ic
c
o
n
v
er
ter
s
,
th
e
n
o
n
lin
ea
r
b
eh
av
i
o
r
o
f
PV
m
o
d
u
les
,
an
d
g
r
i
d
i
n
ter
ac
tio
n
s
lead
to
f
r
eq
u
e
n
t
p
o
wer
f
lu
ctu
atio
n
s
th
at
m
ay
co
m
p
r
o
m
is
e
v
o
ltag
e
s
tab
ilit
y
,
p
o
wer
q
u
ality
,
an
d
o
v
er
all
g
r
id
r
eliab
ilit
y
if
n
o
t p
r
o
p
er
l
y
m
an
a
g
ed
[
3
]
.
Acc
u
r
ate
s
h
o
r
t
-
ter
m
PV
p
o
we
r
f
o
r
ec
asti
n
g
is
th
er
ef
o
r
e
ess
en
tial
f
o
r
r
eliab
le
g
r
id
o
p
er
atio
n
,
o
p
tim
al
en
er
g
y
s
ch
e
d
u
lin
g
,
an
d
ef
f
ici
en
t
en
er
g
y
m
an
a
g
em
en
t
[
4
]
,
[
5
]
.
Fo
r
ec
asti
n
g
also
s
u
p
p
o
r
ts
d
em
an
d
r
esp
o
n
s
e
s
tr
ateg
ies
an
d
en
er
g
y
s
to
r
ag
e
p
lan
n
in
g
,
en
a
b
lin
g
h
ig
h
er
PV
p
en
etr
atio
n
in
s
m
ar
t
g
r
id
s
.
PV
f
o
r
ec
asti
n
g
m
eth
o
d
s
ar
e
g
en
er
ally
ca
teg
o
r
ized
in
to
d
ir
ec
t
ap
p
r
o
ac
h
es,
wh
ich
esti
m
ate
p
o
wer
o
u
tp
u
t
d
ir
ec
tly
f
r
o
m
h
is
to
r
ical
d
ata,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
P
erfo
r
ma
n
ce
en
h
a
n
ce
me
n
t o
f
p
h
o
to
v
o
lta
ic
s
ystems
u
s
in
g
h
yb
r
id
LS
TM
–
C
N
N
…
(
S
a
r
a
F
en
n
a
n
e
)
697
an
d
in
d
ir
ec
t
a
p
p
r
o
ac
h
es,
wh
i
ch
f
ir
s
t
p
r
ed
ict
k
e
y
m
eteo
r
o
l
o
g
ical
v
ar
iab
les
m
o
s
t
n
o
ta
b
ly
g
lo
b
al
h
o
r
izo
n
tal
ir
r
ad
ian
ce
(
GHI
)
a
n
d
s
u
b
s
eq
u
en
tly
d
er
iv
e
PV
p
o
wer
u
s
in
g
an
aly
tical
m
o
d
els.
GHI
r
ep
r
es
en
ts
th
e
co
m
b
in
ed
d
ir
ec
t,
d
if
f
u
s
e,
an
d
r
ef
lecte
d
s
o
lar
r
ad
iatio
n
in
cid
en
t
o
n
a
h
o
r
izo
n
tal
s
u
r
f
a
ce
an
d
s
er
v
es
as
a
f
u
n
d
am
en
tal
in
d
icato
r
o
f
s
o
lar
r
eso
u
r
ce
a
v
ailab
ilit
y
[
6
]
.
H
o
wev
er
,
its
ac
cu
r
ate
esti
m
atio
n
r
em
ain
s
ch
allen
g
in
g
d
u
e
to
atm
o
s
p
h
er
ic
v
ar
ia
b
ilit
y
d
r
iv
e
n
b
y
clo
u
d
s
,
ae
r
o
s
o
ls
,
an
d
h
u
m
id
ity
.
T
r
ad
itio
n
al
p
h
y
s
ical
m
o
d
els
r
e
ly
o
n
d
etailed
atm
o
s
p
h
er
ic
f
o
r
m
u
latio
n
s
an
d
ar
e
m
o
r
e
s
u
itab
l
e
f
o
r
lo
n
g
-
ter
m
f
o
r
ec
asti
n
g
,
wh
ile
s
tatis
ti
ca
l
an
d
m
ac
h
in
e
lear
n
in
g
(
ML
)
tech
n
iq
u
es
,
s
u
ch
as
AR
I
MA
[
7
]
,
s
u
p
p
o
r
t
v
ec
to
r
r
eg
r
ess
io
n
,
r
an
d
o
m
f
o
r
ests
,
an
d
ar
tific
ial
n
eu
r
al
n
etwo
r
k
s
(
A
NNs)
,
h
av
e
d
em
o
n
s
tr
ated
ef
f
e
ctiv
en
ess
f
o
r
s
h
o
r
t
-
ter
m
h
o
r
izo
n
s
.
Nev
er
th
eless
,
th
ese
m
o
d
els
o
f
ten
s
tr
u
g
g
le
u
n
d
er
r
ap
id
ly
ch
a
n
g
in
g
wea
th
er
co
n
d
itio
n
s
.
R
ec
en
t
ad
v
an
ce
s
in
d
ee
p
lear
n
in
g
(
DL
)
,
p
ar
ticu
lar
ly
r
ec
u
r
r
e
n
t
n
eu
r
al
n
etwo
r
k
s
,
lo
n
g
s
h
o
r
t
-
ter
m
m
e
m
o
r
y
n
etwo
r
k
s
,
an
d
co
n
v
o
l
u
tio
n
al
n
eu
r
al
n
etwo
r
k
s
,
h
av
e
s
ig
n
if
ican
tly
im
p
r
o
v
ed
f
o
r
ec
asti
n
g
ac
cu
r
ac
y
b
y
ca
p
t
u
r
in
g
n
o
n
lin
ea
r
an
d
m
u
ltiv
ar
iate
tem
p
o
r
al
p
atter
n
s
[
8
]
.
T
o
f
u
r
th
er
en
h
a
n
ce
p
r
ed
ictiv
e
p
er
f
o
r
m
an
ce
,
s
ev
er
al
s
tu
d
ies
h
av
e
ex
p
lo
r
e
d
h
y
b
r
id
DL
a
r
c
h
itectu
r
es.
Atten
tio
n
-
b
ased
L
STM
m
o
d
el
s
[
9
]
,
c
o
m
p
ar
ativ
e
an
aly
s
es
o
f
r
ec
u
r
r
e
n
t
n
etwo
r
k
s
u
s
in
g
lo
n
g
-
ter
m
s
atellite
d
ata
[
1
0
]
,
a
n
d
C
NN
–
L
STM
co
m
b
in
atio
n
s
[
1
1
]
h
av
e
d
em
o
n
s
tr
ated
im
p
r
o
v
e
d
ac
cu
r
ac
y
u
n
d
er
v
ar
io
u
s
clim
atic
co
n
d
itio
n
s
.
Ad
d
itio
n
al
in
v
esti
g
atio
n
s
h
av
e
ass
ess
ed
u
n
iv
ar
iate
an
d
m
u
ltiv
ar
iate
h
y
b
r
id
m
o
d
els
[
1
2
]
,
s
p
atio
tem
p
o
r
al
ex
te
n
s
io
n
s
in
c
o
r
p
o
r
ati
n
g
n
eig
h
b
o
r
in
g
-
s
ite
d
a
ta
[
1
3
]
,
clu
s
ter
ed
C
NN
–
L
ST
M
ar
ch
itectu
r
es
[
1
4
]
,
an
d
s
ea
s
o
n
al
f
o
r
ec
asti
n
g
s
tr
ate
g
ies
[
1
5
]
.
Ov
er
all,
th
ese
s
tu
d
ie
s
co
n
f
ir
m
th
e
e
f
f
ec
tiv
en
ess
o
f
h
y
b
r
id
DL
m
o
d
els
f
o
r
s
h
o
r
t
-
ter
m
GHI
p
r
e
d
ictio
n
.
Ho
wev
er
,
th
eir
r
o
b
u
s
tn
ess
a
cr
o
s
s
d
iv
er
s
e
clim
atic
r
eg
io
n
s
,
p
ar
ticu
lar
ly
s
em
i
-
ar
id
en
v
ir
o
n
m
e
n
ts
ch
ar
ac
ter
iz
ed
b
y
a
b
r
u
p
t ir
r
ad
ian
ce
v
ar
iati
o
n
s
,
r
em
ain
s
lim
ited
[
1
6
]
.
W
h
ile
ac
cu
r
ate
f
o
r
ec
asti
n
g
en
h
an
ce
s
g
r
id
-
lev
el
p
lan
n
in
g
,
r
e
al
-
tim
e
PV
en
er
g
y
ex
tr
ac
tio
n
d
ep
en
d
s
o
n
ef
f
icien
t
m
ax
im
u
m
p
o
wer
p
o
in
t
tr
ac
k
in
g
(
MPPT)
.
T
r
a
d
itio
n
al
MPPT
tech
n
iq
u
es
s
u
ch
as
p
er
tu
r
b
an
d
o
b
s
er
v
e
(
P&
O)
,
in
cr
em
e
n
tal
co
n
d
u
ct
an
ce
,
an
d
Hill
clim
b
in
g
a
r
e
co
m
m
o
n
ly
im
p
lem
e
n
ted
b
ec
au
s
e
th
ey
a
r
e
s
tr
aig
h
tf
o
r
war
d
to
d
esig
n
an
d
r
eq
u
ir
e
m
in
im
al
co
m
p
u
tatio
n
a
l
r
eso
u
r
ce
s
.
Ho
wev
er
,
th
eir
p
e
r
f
o
r
m
a
n
ce
d
eg
r
a
d
es
u
n
d
er
r
a
p
id
ly
v
ar
y
in
g
ir
r
ad
ia
n
ce
,
lead
in
g
to
s
lo
w
co
n
v
er
g
en
c
e
an
d
s
tead
y
-
s
tate
o
s
cillatio
n
s
.
Alth
o
u
g
h
ad
v
a
n
ce
d
in
tellig
en
t
an
d
m
etah
e
u
r
is
tic
MPPT
tech
n
iq
u
es
im
p
r
o
v
e
ad
ap
tab
ilit
y
,
th
ey
o
f
te
n
in
tr
o
d
u
ce
in
cr
ea
s
ed
co
m
p
lex
ity
an
d
tu
n
in
g
r
eq
u
ir
e
m
en
ts
.
C
o
n
s
eq
u
en
tly
,
P&
O
r
em
ain
s
th
e
m
o
s
t
co
m
m
o
n
ly
im
p
lem
en
ted
m
eth
o
d
in
co
m
m
er
cial
PV
s
y
s
tem
s
.
D
esp
ite
ex
ten
s
iv
e
r
esear
ch
o
n
d
ee
p
lear
n
in
g
-
b
ased
s
o
lar
f
o
r
ec
asti
n
g
an
d
MPPT
alg
o
r
ith
m
s
,
f
ew
s
tu
d
ies
h
av
e
in
teg
r
ated
h
y
b
r
id
d
ee
p
lear
n
i
n
g
f
o
r
ec
asti
n
g
with
co
n
v
en
tio
n
al
MPPT
co
n
tr
o
l,
p
ar
ticu
lar
ly
in
s
em
i
-
ar
id
clim
a
tes.
T
h
is
d
is
co
n
n
ec
t
lim
its
th
e
p
r
ac
tical
e
x
p
lo
itatio
n
o
f
f
o
r
ec
asti
n
g
in
tel
lig
en
ce
f
o
r
r
ea
l
-
tim
e
PV c
o
n
tr
o
l.
T
o
ad
d
r
ess
th
is
g
ap
,
th
is
p
ap
er
p
r
o
p
o
s
es
an
in
teg
r
ated
f
r
am
e
wo
r
k
th
at
co
m
b
in
es
h
y
b
r
id
L
S
T
M
–
C
NN
-
b
ased
s
h
o
r
t
-
ter
m
GHI
f
o
r
ec
a
s
tin
g
with
th
e
P&
O
MPPT
alg
o
r
ith
m
to
en
h
a
n
ce
PV
s
y
s
tem
ef
f
icien
cy
an
d
s
tab
ilit
y
.
T
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
is
v
alid
ate
d
u
s
in
g
r
ea
l
ir
r
ad
ia
n
ce
d
ata
f
r
o
m
Dak
h
la,
M
o
r
o
cc
o
,
a
s
em
i
-
ar
id
r
eg
io
n
with
h
i
g
h
s
o
lar
v
ar
iab
i
lity
.
Simu
latio
n
r
esu
lts
o
b
tain
ed
in
MA
T
L
AB
/Si
m
u
lin
k
d
e
m
o
n
s
tr
ate
im
p
r
o
v
ed
f
o
r
ec
asti
n
g
ac
cu
r
ac
y
,
f
aster
MPPT
co
n
v
er
g
en
ce
,
r
e
d
u
ce
d
o
s
cillatio
n
s
,
an
d
en
h
an
ce
d
p
o
wer
s
tab
ilit
y
u
n
d
er
r
ap
id
ly
ch
a
n
g
in
g
co
n
d
itio
n
s
.
T
h
e
m
ain
co
n
tr
ib
u
tio
n
s
o
f
th
is
ar
ticle
ar
e
s
u
m
m
ar
ized
as f
o
llo
ws:
−
Dev
elo
p
m
en
t
o
f
a
h
y
b
r
i
d
L
ST
M
–
C
NN
ap
p
r
o
ac
h
f
o
r
ac
c
u
r
at
e
s
h
o
r
t
-
ter
m
GHI
f
o
r
ec
asti
n
g
.
−
I
n
teg
r
atio
n
o
f
DL
-
b
ased
ir
r
a
d
ian
ce
p
r
ed
ictio
n
with
th
e
co
n
v
en
tio
n
al
P&
O
MPPT
s
tr
a
teg
y
to
im
p
r
o
v
e
d
y
n
am
ic
tr
ac
k
in
g
p
er
f
o
r
m
an
ce
.
−
Valid
atio
n
u
s
in
g
r
ea
l
-
wo
r
ld
d
ata
f
r
o
m
a
s
em
i
-
ar
id
Mo
r
o
cc
an
clim
ate,
h
ig
h
lig
h
tin
g
m
o
d
el
r
o
b
u
s
tn
ess
.
Dem
o
n
s
tr
atio
n
o
f
im
p
r
o
v
ed
PV
p
o
wer
s
tab
ilit
y
an
d
e
n
er
g
y
ex
tr
ac
tio
n
ef
f
icie
n
cy
th
r
o
u
g
h
in
te
g
r
ated
f
o
r
ec
asti
n
g
–
co
n
tr
o
l
d
esig
n
.
−
B
y
u
n
if
y
in
g
d
ata
-
d
r
iv
en
f
o
r
ec
asti
n
g
with
r
ea
l
-
tim
e
P
V
co
n
tr
o
l,
th
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
s
u
p
p
o
r
ts
en
h
an
ce
d
o
p
er
atio
n
al
i
n
tellig
en
ce
,
c
o
n
tr
i
b
u
tin
g
t
o
im
p
r
o
v
ed
s
u
s
tain
ab
le
r
en
ewa
b
le
e
n
er
g
y
in
teg
r
ati
o
n
an
d
s
m
ar
t g
r
i
d
r
esil
ien
ce
.
2.
M
E
T
H
O
D
2
.
1
.
So
la
r
f
o
re
ca
s
t
ing
mo
dels
2
.
1
.
1
.
L
o
ng
s
ho
rt
-
t
er
m m
e
mo
ry
net
wo
r
k
L
STM
n
etwo
r
k
s
we
r
e
d
ev
elo
p
ed
to
ad
d
r
ess
th
e
v
an
is
h
in
g
an
d
ex
p
lo
d
in
g
g
r
ad
ien
t
lim
i
tatio
n
s
o
f
co
n
v
en
tio
n
al
R
NNs
[
1
7
]
.
B
y
em
p
lo
y
in
g
m
em
o
r
y
ce
lls
co
n
tr
o
lled
b
y
in
p
u
t,
f
o
r
g
et,
an
d
o
u
tp
u
t
g
ates,
L
STM
s
ef
f
ec
tiv
ely
ca
p
tu
r
e
lo
n
g
-
ter
m
tem
p
o
r
al
d
ep
en
d
en
cies
in
s
eq
u
en
tial
d
ata,
m
ak
i
n
g
t
h
em
we
ll
-
s
u
ited
f
o
r
tim
e
-
s
er
ies
f
o
r
ec
asti
n
g
.
Alth
o
u
g
h
c
o
m
p
u
tatio
n
ally
in
te
n
s
iv
e
an
d
s
en
s
itiv
e
to
h
y
p
er
p
ar
am
eter
s
elec
tio
n
,
en
h
an
ce
d
v
ar
ian
ts
s
u
ch
as
atten
tio
n
-
b
ased
an
d
b
i
d
ir
ec
tio
n
al
L
STM
s
f
u
r
th
er
im
p
r
o
v
e
f
ea
tu
r
e
r
ep
r
esen
tatio
n
an
d
lear
n
i
n
g
ca
p
ab
ilit
y
[
1
8
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
1
,
Ma
r
c
h
20
2
6
:
696
-
7
08
698
T
h
e
f
u
n
ctio
n
in
g
o
f
th
e
d
if
f
er
e
n
t
g
ates
in
a
n
eu
r
al
n
etwo
r
k
,
s
u
ch
as
th
e
in
p
u
t
g
ate
it,
f
o
r
g
et
g
ate
f
t,
in
ter
m
ed
iate
g
ate
g
t,
an
d
o
u
tp
u
t g
ate
o
t,
ca
n
b
e
m
at
h
em
atica
lly
ex
p
r
ess
ed
as
(
1
)
-
(
4
)
[
1
9
]
.
=
(
+
ℎ
ℎ
−
1
+
)
(
1
)
=
(
+
ℎ
ℎ
−
1
+
)
(
2
)
=
(
+
ℎ
ℎ
−
1
+
)
(
3
)
=
(
+
ℎ
ℎ
−
1
+
)
(
4
)
T
h
e
s
i
g
m
o
i
d
a
c
t
i
v
a
ti
o
n
f
u
n
c
t
i
o
n
i
s
d
e
n
o
t
e
d
b
y
σ
.
T
h
e
w
e
i
g
h
t
m
a
t
r
i
c
es
c
o
r
r
e
s
p
o
n
d
i
n
g
t
o
t
h
e
i
n
p
u
t
X
t
a
n
d
t
h
e
p
r
i
o
r
h
i
d
d
e
n
s
t
a
t
e
h
t
-
1
a
r
e
i
n
d
i
c
at
e
d
b
y
W
x
a
n
d
W
h
,
r
e
s
p
ec
t
i
v
el
y
.
T
h
e
b
i
a
s
te
r
m
u
n
i
q
u
e
t
o
e
a
c
h
g
a
t
e
i
s
r
e
p
r
es
e
n
t
e
d
b
y
b
.
T
h
e
f
o
r
g
et
g
ate
f
t
d
ete
r
m
in
e
s
wh
ich
in
f
o
r
m
atio
n
f
r
o
m
th
e
p
r
ev
io
u
s
m
em
o
r
y
ce
ll
m
t
-
1
s
h
o
u
ld
b
e
r
etain
ed
,
wh
ile
th
e
in
p
u
t
g
ate
it
ca
lcu
lates
th
e
in
f
o
r
m
atio
n
to
b
e
s
to
r
ed
in
th
e
cu
r
r
en
t
m
em
o
r
y
ce
ll
m
t.
T
h
e
u
p
d
ated
m
em
o
r
y
ce
ll m
t is co
m
p
u
ted
u
s
in
g
(
5
)
[
1
9
]
.
=
+
−
1
(
5
)
Her
e,
g
t
r
ep
r
esen
ts
th
e
in
ter
m
ed
iate
g
ate’
s
co
n
tr
ib
u
tio
n
,
it
co
n
tr
o
ls
th
e
in
p
u
t'
s
in
f
lu
en
ce
,
an
d
f
t
r
eg
u
lates
th
e
r
eten
tio
n
o
f
p
ast in
f
o
r
m
atio
n
.
2
.
1
.
2
.
Co
nv
o
lutio
na
l
neura
l
net
wo
rk
A
C
NN
is
a
d
ee
p
lear
n
in
g
ar
c
h
itectu
r
e
o
r
ig
in
ally
d
ev
elo
p
ed
f
o
r
im
ag
e
r
ec
o
g
n
itio
n
,
b
u
t
it
ca
n
also
b
e
ap
p
lied
to
tim
e
-
s
er
ies
an
aly
s
is
,
as
it is
ef
f
ec
tiv
e
at
ca
p
tu
r
in
g
lo
ca
l
p
atter
n
s
an
d
s
h
o
r
t
-
ter
m
d
ep
en
d
en
cies
with
in
s
eq
u
en
tial
d
ata.
C
NNs
u
s
e
co
n
v
o
lu
tio
n
al
f
ilter
s
to
ex
t
r
ac
t
lo
c
al
p
atter
n
s
an
d
co
r
r
elatio
n
s
am
o
n
g
in
p
u
t
f
ea
tu
r
es,
as
o
p
p
o
s
ed
to
r
ec
u
r
r
en
t
m
o
d
el
s
,
wh
ich
ex
p
licitly
d
escr
ib
e
tem
p
o
r
al
r
elatio
n
s
h
ip
s
.
C
NNs
a
r
e
m
o
d
if
ied
f
o
r
1
D
s
eq
u
en
tial
d
ata
in
th
e
c
o
n
tex
t
o
f
tim
e
s
er
ies
f
o
r
ec
asti
n
g
,
lik
e
GHI
.
Usu
ally
,
th
e
m
o
d
el
h
as
m
u
ltip
le
im
p
o
r
tan
t
lay
er
s
[
1
5
]
.
C
o
n
v
o
lu
tio
n
al
lay
er
: A
p
p
lies
m
u
ltip
le
1
D
f
ilter
s
o
v
er
s
u
b
s
eq
u
en
ce
s
o
f
th
e
in
p
u
t
tim
e
s
er
ie
s
to
ex
tr
ac
t
lo
ca
l f
ea
tu
r
es.
T
h
e
o
p
er
atio
n
f
o
r
th
e
k
-
th
f
ilter
is
g
iv
e
n
b
y
(
6
)
.
ℎ
,
=
(
∑
,
−
1
=
0
.
+
+
)
(
6
)
F
is
th
e
f
ilter
s
ize,
W
k
,
j
is
th
e
l
ea
r
n
ab
le
weig
h
ts
,
b
k
is
th
e
b
ias,
an
d
f
is
th
e
ac
tiv
atio
n
f
u
n
ctio
n
,
ty
p
ically
r
ec
tifie
d
lin
ea
r
u
n
it
(
R
eL
U
)
.
Po
o
lin
g
lay
e
r
:
Per
f
o
r
m
s
a
m
a
x
p
o
o
lin
g
o
p
e
r
atio
n
to
d
o
wn
s
am
p
le
th
e
f
ea
tu
r
e
m
ap
s
p
r
o
d
u
ce
d
b
y
th
e
co
n
v
o
l
u
tio
n
al
lay
er
,
th
er
eb
y
r
ed
u
cin
g
d
im
en
s
io
n
ality
an
d
c
o
m
p
u
tatio
n
al
c
o
m
p
lex
ity
wh
i
le
r
etain
in
g
s
alien
t
f
ea
tu
r
es:
=
ma
x
=
1
ℎ
,
(
7
)
f
u
lly
co
n
n
ec
ted
lay
e
r
: u
s
es a
d
en
s
e
lay
er
to
f
latten
t
h
e
p
o
o
le
d
m
ap
s
an
d
ad
d
f
ea
tu
r
es to
th
e
o
u
tp
u
t:
=
(
+
)
(
8
)
g
is
th
e
f
in
al
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
f
o
r
p
r
e
d
ictio
n
,
W
an
d
b
ar
e
th
e
weig
h
ts
an
d
b
iases
o
f
th
e
d
en
s
e
lay
er
.
2
.
1
.
3
.
H
y
brid L
ST
M
-
CNN
m
o
del
T
o
p
r
e
d
ic
t
s
h
o
r
t
-
te
r
m
v
a
r
i
ati
o
n
s
i
n
GH
I
,
a
h
y
b
r
i
d
L
STM
–
C
N
N
m
o
d
el
was
d
e
v
e
lo
p
ed
.
T
h
e
m
o
ti
v
at
io
n
b
e
h
i
n
d
t
h
is
a
r
c
h
i
tec
t
u
r
e
s
t
em
s
f
r
o
m
t
h
e
c
o
m
p
le
m
e
n
t
ar
y
s
tr
e
n
g
t
h
s
o
f
t
h
e
tw
o
d
ee
p
le
ar
n
i
n
g
co
m
p
o
n
e
n
ts
:
C
NN
lay
e
r
s
ar
e
ef
f
e
cti
v
e
at
e
x
t
r
ac
ti
n
g
lo
ca
l
s
p
a
tial
p
a
tte
r
n
s
f
r
o
m
te
m
p
o
r
al
d
at
a,
w
h
i
le
L
S
T
M
la
y
e
r
s
ca
p
t
u
r
e
l
o
n
g
-
te
r
m
tem
p
o
r
al
d
e
p
e
n
d
e
n
ci
es
a
n
d
d
y
n
a
m
ic
r
el
ati
o
n
s
h
i
p
s
w
it
h
i
n
s
eq
u
e
n
ti
al
i
n
p
u
ts
.
I
n
t
h
e
im
p
l
em
e
n
ted
m
o
d
el,
t
h
e
i
n
p
u
t
lay
e
r
r
ec
ei
v
es
a
s
e
q
u
e
n
ce
o
f
p
a
s
t ir
r
a
d
ia
n
c
e
v
al
u
es
c
o
r
r
esp
o
n
d
in
g
t
o
a
f
i
x
e
d
l
o
o
k
-
b
a
ck
w
i
n
d
o
w
(
e.
g
.
,
4
8
h
o
u
r
s
)
.
I
n
o
r
d
e
r
to
e
x
t
r
a
ct
ty
p
i
ca
l
p
at
ter
n
s
a
n
d
s
m
o
o
t
h
s
h
o
r
t
-
te
r
m
v
a
r
i
ati
o
n
s
,
t
h
e
C
N
N
c
o
m
p
o
n
e
n
t
f
i
r
s
t
a
p
p
li
es
o
n
e
-
d
i
m
e
n
s
i
o
n
al
co
n
v
o
l
u
ti
o
n
al
f
il
t
er
s
.
T
h
e
n
e
two
r
k
m
a
y
s
im
u
l
t
an
eo
u
s
l
y
l
ea
r
n
s
p
atia
l
a
n
d
te
m
p
o
r
al
co
r
r
el
ati
o
n
s
th
a
n
k
s
t
o
t
h
e
L
ST
M
l
a
y
e
r
s
,
wh
i
ch
a
r
e
f
e
d
t
h
e
e
x
t
r
a
cte
d
f
ea
t
u
r
e
m
ap
s
a
n
d
s
im
u
l
ate
t
h
e
p
att
er
n
s
'
t
em
p
o
r
al
ev
o
l
u
ti
o
n
o
v
er
ti
m
e
.
T
h
e
o
u
t
p
u
t
l
ay
er
r
ep
r
es
e
n
ts
t
h
e
ex
p
e
ct
e
d
GH
I
v
al
u
e
f
o
r
th
e
f
o
l
lo
wi
n
g
h
o
u
r
a
n
d
is
m
a
d
e
u
p
o
f
a
s
i
n
g
le
n
e
u
r
o
n
wi
th
a
li
n
e
ar
a
cti
v
ati
o
n
f
u
n
ct
io
n
.
T
h
r
o
u
g
h
it
er
ati
v
e
b
a
c
k
p
r
o
p
a
g
at
io
n
,
t
h
e
m
o
d
el
le
a
r
n
s
t
o
r
e
d
u
ce
t
h
e
p
r
e
d
ic
ti
o
n
e
r
r
o
r
b
et
wee
n
t
h
e
ac
tu
al
an
d
f
o
r
ec
as
te
d
GHI
d
u
r
in
g
tr
ai
n
i
n
g
.
T
h
e
MS
E
was
u
t
ili
ze
d
as
th
e
lo
s
s
f
u
n
ct
io
n
,
a
n
d
t
h
e
A
d
a
m
o
p
ti
m
i
ze
r
w
as
u
t
ili
ze
d
f
o
r
ef
f
e
ct
iv
e
g
r
a
d
ie
n
t
d
esc
e
n
t
wit
h
a
d
a
p
tiv
e
l
ea
r
n
i
n
g
r
a
tes.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
P
erfo
r
ma
n
ce
en
h
a
n
ce
me
n
t o
f
p
h
o
to
v
o
lta
ic
s
ystems
u
s
in
g
h
yb
r
id
LS
TM
–
C
N
N
…
(
S
a
r
a
F
en
n
a
n
e
)
699
2
.
2
.
F
o
re
ca
s
t
ing
perf
o
r
m
a
n
ce
m
et
rics
T
h
e
m
o
d
e
l
s
w
e
r
e
d
e
v
e
l
o
p
e
d
i
n
P
y
t
h
o
n
u
s
i
n
g
t
h
e
T
e
n
s
o
r
F
l
o
w
a
n
d
K
e
r
a
s
f
r
a
m
e
w
o
r
k
s
,
w
h
i
l
e
d
a
t
a
v
i
s
u
a
l
i
z
a
t
i
o
n
w
a
s
c
a
r
r
i
e
d
o
u
t
w
i
t
h
M
a
t
p
l
o
t
l
i
b
.
T
o
e
n
s
u
r
e
a
f
a
i
r
c
o
m
p
a
r
i
s
o
n
,
t
h
e
t
h
r
e
e
f
o
r
e
c
a
s
t
i
n
g
a
p
p
r
o
a
c
h
e
s
,
L
S
T
M
,
C
N
N
,
a
n
d
t
h
e
h
y
b
r
i
d
L
S
T
M
–
C
N
N
,
w
e
r
e
t
r
a
i
n
e
d
a
n
d
e
v
a
l
u
a
t
e
d
o
n
t
h
e
s
a
m
e
d
a
t
a
s
e
t
u
s
i
n
g
i
d
e
n
t
i
c
a
l
e
x
p
e
r
i
m
e
n
t
a
l
s
e
t
t
i
n
g
s
.
M
o
d
e
l
p
e
r
f
o
r
m
a
n
c
e
w
a
s
q
u
a
n
t
i
t
a
t
i
v
e
l
y
a
s
s
e
s
s
e
d
b
a
s
e
d
o
n
t
h
r
e
e
s
t
a
n
d
a
r
d
s
t
a
t
i
s
t
i
c
a
l
m
e
t
r
i
c
s
[
2
0
]
:
=
[
1
∑
(
−
)
2
=
1
]
1
2
(
9
)
=
1
∑
|
ŷ
−
|
(
1
0
)
2
=
1
−
∑
(
−
)
2
=
1
∑
(
−
̅
̅
̅
)
2
=
1
(
1
1
)
2
.
3
.
Da
t
a
s
et
des
cr
iptio
n a
nd
prepro
ce
s
s
i
ng
pro
ce
du
re
s
T
h
e
d
ataset
u
s
ed
i
n
th
is
s
tu
d
y
was so
u
r
ce
d
f
r
o
m
t
h
e
Natio
n
al
R
en
ewa
b
le
E
n
er
g
y
L
a
b
o
r
ato
r
y
(
NR
E
L
)
th
r
o
u
g
h
th
e
Natio
n
al
So
lar
R
ad
iatio
n
Data
b
ase
(
NSR
DB
)
.
I
t
p
r
o
v
id
es
h
ig
h
-
r
eso
lu
tio
n
,
g
r
o
u
n
d
-
b
ased
s
o
lar
ir
r
ad
ian
ce
m
ea
s
u
r
em
e
n
ts
f
o
r
Dak
h
la,
Mo
r
o
cc
o
a
r
eg
io
n
c
h
ar
ac
ter
ized
b
y
a
s
em
i
-
ar
id
cli
m
ate
an
d
ex
ce
p
tio
n
al
s
o
lar
p
o
ten
tial.
I
t
co
m
p
r
is
es
h
o
u
r
ly
m
ea
s
u
r
em
en
ts
o
f
GHI
,
am
b
ien
t
tem
p
er
atu
r
e,
an
d
o
t
h
er
r
elev
an
t
m
eteo
r
o
lo
g
ical
p
ar
am
eter
s
f
o
r
th
e
en
tire
y
ea
r
o
f
2
0
2
2
,
y
ield
in
g
8
,
7
6
0
d
ata
p
o
in
ts
th
at
ca
p
tu
r
e
b
o
th
s
ea
s
o
n
al
an
d
d
aily
v
ar
iatio
n
s
in
s
o
la
r
r
ad
iatio
n
.
Data
p
r
ep
r
o
ce
s
s
in
g
,
s
eq
u
e
n
ce
p
r
ep
ar
atio
n
,
m
o
d
el
tr
ain
in
g
,
a
n
d
ev
alu
atio
n
f
o
llo
wed
t
h
e
wo
r
k
f
l
o
w
s
u
m
m
ar
ized
in
Alg
o
r
ith
m
1
,
with
d
ataset
d
etails
an
d
h
y
p
er
p
ar
am
eter
s
ettin
g
s
p
r
o
v
id
e
d
in
T
a
b
le
1
.
Alg
o
r
ith
m
1
.
Ps
eu
d
o
co
d
e
f
o
r
th
e
o
p
tim
ized
d
ee
p
lear
n
in
g
f
r
a
m
ewo
r
k
f
o
r
GHI
p
r
ed
ictio
n
1.
Star
t
2.
L
o
ad
GHI
d
ataset
(
2
0
2
2
)
3.
Pre
p
r
o
ce
s
s
d
ata
˗
C
lean
an
d
s
tan
d
ar
d
ize
c
o
lu
m
n
n
am
es
˗
I
n
ter
p
o
late
m
is
s
in
g
v
al
u
es (
f
o
r
all
m
eteo
r
o
lo
g
ical
f
ea
tu
r
es)
˗
C
o
n
s
tr
u
ct
Date
tim
e
co
lu
m
n
˗
No
r
m
alize
all
in
p
u
t
f
ea
t
u
r
es (
m
in
–
m
ax
s
ca
lin
g
)
4.
Pre
p
ar
e
s
eq
u
en
ce
s
˗
C
r
ea
te
in
p
u
t seq
u
en
ce
s
X
an
d
tar
g
et
y
b
ased
o
n
s
eq
u
e
n
ce
_
le
n
g
th
=
3
0
5.
Sp
lit
d
ataset
˗
T
r
ain
in
g
(
6
4
%),
Valid
atio
n
(
1
6
%),
T
esti
n
g
(
2
0
%)
6.
Def
in
e
b
ase
m
o
d
els
˗
L
STM
→
Den
s
e
˗
C
NN
→
Ma
x
Po
o
lin
g
→
Flatt
en
→
Den
s
e
→
Den
s
e
˗
Hy
b
r
id
C
NN
–
L
STM
→
Den
s
e
7.
Op
tim
ize
h
y
p
er
p
ar
am
eter
s
˗
Def
in
e
s
ea
r
ch
s
p
ac
e
(
L
STM
u
n
its
,
C
NN
f
ilter
s
,
k
er
n
el
s
iz
e,
ac
tiv
atio
n
,
b
atch
s
ize,
lear
n
in
g
r
ate)
˗
Ap
p
ly
r
an
d
o
m
s
ea
r
c
h
to
m
in
i
m
ize
v
alid
atio
n
MSE
8.
T
r
ain
an
d
Select
B
est M
o
d
els
˗
T
r
ain
o
n
tr
ain
in
g
s
et
˗
Valid
ate
an
d
s
elec
t c
o
n
f
ig
u
r
ati
o
n
with
lo
west v
alid
atio
n
lo
s
s
9.
Fin
al
T
r
ain
in
g
˗
T
r
ain
o
p
tim
ized
m
o
d
els o
n
co
m
b
in
ed
tr
ain
i
n
g
+
v
alid
atio
n
d
ata
(
1
0
0
e
p
o
ch
s
)
10.
Pre
d
ictio
n
an
d
E
v
alu
atio
n
˗
Pre
d
ict
GHI
o
n
test
s
et
˗
C
o
m
p
u
te
R
MSE
,
MA
E
,
R
²
11.
R
esu
lts
a
n
aly
s
i
s
˗
Sto
r
e
p
r
ed
ictio
n
s
an
d
m
etr
ics
˗
B
u
ild
r
esu
lts
tab
le
(
Date
tim
e,
Actu
al
GHI
,
Pre
d
icted
GHI
)
˗
Plo
t r
eg
r
ess
io
n
an
d
tim
e
-
s
er
ie
s
g
r
ap
h
s
12.
E
x
p
o
r
t O
u
t
p
u
ts
˗
E
x
p
o
r
t
r
esu
lts
to
C
SV a
n
d
E
x
ce
l
13.
E
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
1
,
Ma
r
c
h
20
2
6
:
696
-
7
08
700
T
ab
le
1
.
Su
m
m
a
r
y
o
f
d
ataset
c
h
ar
ac
ter
is
tics
an
d
d
ee
p
lear
n
in
g
h
y
p
e
r
p
ar
a
m
eter
s
f
o
r
GHI
p
r
ed
ictio
n
P
a
r
a
me
t
e
r
/
s
e
t
t
i
n
g
V
a
l
u
e
/
r
a
n
g
e
N
o
t
e
s
S
e
q
u
e
n
c
e
l
e
n
g
t
h
30
Ti
me
st
e
p
s
p
e
r
i
n
p
u
t
s
e
q
u
e
n
c
e
Tr
a
i
n
/
v
a
l
i
d
a
t
i
o
n
/
t
e
st
sp
l
i
t
6
4
%
/
1
6
%
/
2
0
%
C
h
r
o
n
o
l
o
g
i
c
a
l
s
p
l
i
t
H
y
p
e
r
p
a
r
a
me
t
e
r
t
u
n
i
n
g
s
t
r
a
t
e
g
y
R
a
n
d
o
m
s
e
a
r
c
h
(
K
e
r
a
s T
u
n
e
r
)
U
sed
f
o
r
a
l
l
mo
d
e
l
s
LSTM
u
n
i
t
s
3
2
,
6
4
,
1
2
8
,
2
5
6
1
2
8
s
e
l
e
c
t
e
d
v
i
a
r
a
n
d
o
m
sea
r
c
h
a
s
t
h
e
b
e
s
t
c
o
n
f
i
g
u
r
a
t
i
o
n
C
N
N
f
i
l
t
e
r
s
3
2
,
6
4
,
1
2
8
O
p
t
i
mal
f
i
l
t
e
r
si
z
e
f
o
u
n
d
v
i
a
r
a
n
d
o
m
s
e
a
r
c
h
:
6
4
C
N
N
k
e
r
n
e
l
s
i
z
e
3
,
5
O
p
t
i
mal
k
e
r
n
e
l
si
z
e
d
e
t
e
r
m
i
n
e
d
b
y
t
u
n
i
n
g
:
3
A
c
t
i
v
a
t
i
o
n
f
u
n
c
t
i
o
n
s
R
e
LU
,
t
a
n
h
R
e
LU
s
e
l
e
c
t
e
d
f
o
r
b
e
st
p
e
r
f
o
r
mi
n
g
m
o
d
e
l
s
B
a
t
c
h
si
z
e
1
6
,
3
2
,
6
4
3
2
se
l
e
c
t
e
d
v
i
a
r
a
n
d
o
m se
a
r
c
h
Le
a
r
n
i
n
g
r
a
t
e
1e
-
4
,
1
e
-
3
,
1
e
-
2
O
p
t
i
mal
l
e
a
r
n
i
n
g
r
a
t
e
:
1
e
-
3
Ep
o
c
h
s (f
i
n
a
l
t
r
a
i
n
i
n
g
)
1
0
0
Tr
a
i
n
i
n
g
o
n
t
h
e
c
o
mb
i
n
e
d
t
r
a
i
n
+
v
a
l
i
d
a
t
i
o
n
se
t
O
p
t
i
mi
z
e
r
A
d
a
m
U
sed
f
o
r
a
l
l
t
u
n
e
d
m
o
d
e
l
s
H
y
b
r
i
d
mo
d
e
l
a
r
c
h
i
t
e
c
t
u
r
e
C
o
n
v
1
D
(
6
4
f
i
l
t
e
r
s)
→
LST
M
(
1
2
8
u
n
i
t
s)
→
D
e
n
se
B
e
st
p
e
r
f
o
r
m
i
n
g
h
y
b
r
i
d
c
o
n
f
i
g
u
r
a
t
i
o
n
F
e
a
t
u
r
e
s
n
o
r
ma
l
i
z
e
d
A
l
l
i
n
p
u
t
s
M
i
n
–
M
a
x
sc
a
l
i
n
g
Lo
ss f
u
n
c
t
i
o
n
d
u
r
i
n
g
t
u
n
i
n
g
M
S
E
U
sed
t
o
mi
n
i
mi
z
e
v
a
l
i
d
a
t
i
o
n
l
o
ss
2
.
4
.
M
a
t
hema
t
ica
l
m
o
delin
g
o
f
t
he
P
V
pa
nel
T
h
e
p
h
o
to
v
o
ltaic
(
PV)
s
y
s
tem
in
t
h
is
s
tu
d
y
is
d
esig
n
ed
to
s
im
u
late
th
e
elec
tr
ical
b
e
h
av
io
r
o
f
a
s
o
lar
ar
r
ay
u
n
d
er
v
ar
y
in
g
ir
r
a
d
ian
ce
.
T
h
e
PV a
r
r
ay
is
b
ased
o
n
th
e
Ky
o
ce
r
a
So
lar
KC
2
0
0
GT
m
o
d
u
le,
ar
r
a
n
g
ed
with
1
0
s
er
ies
-
co
n
n
e
cted
m
o
d
u
les
(
N
s
=
1
0
)
an
d
2
p
ar
allel
s
tr
in
g
s
(
N
p
=
2
)
,
p
r
o
v
id
i
n
g
s
u
itab
le
v
o
ltag
e
an
d
cu
r
r
en
t
f
o
r
DC
–
DC
co
n
v
er
s
io
n
an
d
lo
ad
o
p
er
atio
n
.
I
ts
m
at
h
em
atica
l
m
o
d
el
is
d
er
iv
ed
f
r
o
m
th
e
s
in
g
le
-
d
io
d
e
eq
u
iv
ale
n
t
cir
cu
it,
co
m
p
r
is
in
g
a
cu
r
r
e
n
t
s
o
u
r
ce
,
d
io
d
e
,
s
er
ies
r
esis
tan
ce
(
R
s
)
,
an
d
s
h
u
n
t
r
esis
tan
ce
(
R
sh
)
.
T
h
e
PV
m
o
d
u
le
o
u
tp
u
t c
u
r
r
en
t
(
I
PV
)
is
g
iv
en
b
y
(
1
2
)
[
2
1
]
.
=
ℎ
−
0
[
e
xp
(
(
⁄
+
⁄
)
)
−
1
]
−
⁄
+
⁄
ℎ
(
1
2
)
W
h
er
e
A
i
s
th
e
d
io
d
e
id
ea
lity
f
ac
to
r
,
T
is
th
e
ce
ll
tem
p
er
atu
r
e
in
Kelv
in
,
q
is
th
e
elec
tr
o
n
ch
ar
g
e,
I
0
is
th
e
d
io
d
e
r
ev
er
s
e
s
atu
r
atio
n
c
u
r
r
e
n
t,
k
is
th
e
B
o
ltzm
an
n
co
n
s
tan
t,
a
n
d
I
ph
is
th
e
p
h
o
t
o
cu
r
r
e
n
t
th
at
is
in
f
lu
en
ce
d
b
y
tem
p
er
atu
r
e
a
n
d
s
o
lar
ir
r
ad
iatio
n
T
h
e
Simu
lin
k
PV
ar
r
ay
is
f
ed
with
p
r
ed
icted
ir
r
a
d
ian
ce
a
n
d
a
co
n
s
tan
t
tem
p
er
atu
r
e
o
f
2
5
°
C
.
T
h
e
PV
v
o
ltag
e
(
V_
PV)
an
d
c
u
r
r
en
t
(
I
_
PV)
ar
e
co
n
tin
u
o
u
s
ly
m
ea
s
u
r
ed
an
d
p
r
o
v
i
d
ed
to
th
e
MPPT
co
n
tr
o
ller
to
e
n
s
u
r
e
m
ax
im
u
m
p
o
wer
ex
tr
ac
tio
n
u
n
d
er
v
ar
y
in
g
co
n
d
itio
n
s
.
A
D
C
–
DC
b
o
o
s
t
co
n
v
er
ter
r
aises
t
h
e
PV
v
o
ltag
e
to
th
e
d
esire
d
lo
ad
lev
el.
I
t
c
o
n
s
is
ts
o
f
an
in
d
u
cto
r
,
a
MO
SF
E
T
s
witch
,
a
d
io
d
e
,
a
n
d
a
n
o
u
tp
u
t
ca
p
ac
ito
r
,
an
d
o
p
er
ates
in
co
n
tin
u
o
u
s
co
n
d
u
ctio
n
m
o
d
e
to
e
n
s
u
r
e
a
s
tab
le
o
u
tp
u
t
v
o
ltag
e
with
lo
w
r
ip
p
le.
T
h
e
i
n
p
u
t
–
o
u
tp
u
t
v
o
ltag
e
r
elatio
n
s
h
ip
is
g
iv
en
b
y
(
1
3
)
[
2
2
]
.
=
1
−
(
1
3
)
W
h
er
e
D
d
en
o
tes th
e
d
u
ty
c
y
c
le
co
n
tr
o
lled
b
y
th
e
P&
O
-
b
ased
MPPT
alg
o
r
ith
m
to
m
ain
tain
th
e
PV a
r
r
a
y
at
its
m
ax
im
u
m
p
o
wer
p
o
in
t.
T
h
e
s
y
s
tem
d
y
n
am
ics ar
e
g
iv
en
b
y
(
1
4
)
an
d
(
1
5
)
.
=
−
(
1
−
)
(
1
4
)
=
(
1
−
)
−
(
1
5
)
W
h
er
e
R
d
en
o
tes
th
e
lo
a
d
r
esi
s
tan
ce
,
th
is
co
n
v
er
ter
en
ab
les ef
f
ec
tiv
e
p
o
we
r
d
eliv
er
y
f
r
o
m
th
e
PV
ar
r
ay
t
o
th
e
lo
ad
wh
ile
ad
ap
tin
g
to
ch
an
g
es
in
ir
r
a
d
ian
ce
an
d
tem
p
er
atu
r
e.
T
h
e
PW
M
-
d
r
iv
en
d
u
ty
cy
cle
co
n
tr
o
ls
th
e
MO
SF
E
T
,
an
d
th
e
r
esu
ltin
g
b
o
o
s
ted
v
o
ltag
e
(
)
is
m
o
n
ito
r
ed
to
ass
ess
co
n
v
er
ter
p
er
f
o
r
m
an
ce
an
d
s
tab
ilit
y
.
2
.
5
.
P
&O
M
P
P
T
a
lg
o
rit
h
m
T
h
e
MPPT
t
ec
h
n
iq
u
e
is
an
ess
en
t
ial
c
o
m
p
o
n
e
n
t
i
n
p
h
o
t
o
v
o
lt
aic
(
P
V)
e
n
er
g
y
s
y
s
t
em
s
t
o
e
n
s
u
r
e
th
at
t
h
e
PV
g
en
er
at
o
r
o
p
e
r
ates
at
i
ts
o
p
t
im
a
l
p
o
w
er
p
o
i
n
t
u
n
d
e
r
v
a
r
y
in
g
e
n
v
i
r
o
n
m
e
n
t
al
c
o
n
d
iti
o
n
s
.
A
m
o
n
g
t
h
e
n
u
m
er
o
u
s
MPPT
alg
o
r
ith
m
s
p
r
o
p
o
s
ed
i
n
th
e
liter
atu
r
e,
t
h
e
P&
O
tech
n
iq
u
e
is
am
o
n
g
th
e
m
o
s
t
co
m
m
o
n
ly
u
s
ed
tech
n
iq
u
es,
ap
p
r
ec
iated
f
o
r
its
s
tr
aig
h
tf
o
r
war
d
d
esig
n
,
s
im
p
le
im
p
lem
e
n
tatio
n
,
an
d
r
eliab
le
o
v
er
all
p
e
r
f
o
r
m
a
n
ce
[
2
3
]
.
T
h
e
P&
O
alg
o
r
ith
m
'
s
b
asic
id
ea
is
t
o
o
b
s
er
v
e
h
o
w
th
e
o
u
tp
u
t
p
o
w
er
ch
an
g
es
in
r
esp
o
n
s
e
to
m
in
o
r
p
er
tu
r
b
atio
n
s
th
at
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
P
erfo
r
ma
n
ce
en
h
a
n
ce
me
n
t o
f
p
h
o
to
v
o
lta
ic
s
ystems
u
s
in
g
h
yb
r
id
LS
TM
–
C
N
N
…
(
S
a
r
a
F
en
n
a
n
e
)
701
ar
e
p
er
i
o
d
ically
in
tr
o
d
u
ce
d
to
th
e
PV
s
y
s
tem
'
s
o
p
er
atio
n
al
v
o
ltag
e
o
r
cu
r
r
en
t.
T
h
e
o
p
er
atin
g
p
o
i
n
t
is
s
h
if
ted
i
n
th
e
s
am
e
way
in
th
e
s
u
b
s
eq
u
e
n
t
iter
atio
n
wh
en
a
p
er
t
u
r
b
ati
o
n
ca
u
s
es
th
e
p
o
wer
to
in
c
r
ea
s
e.
I
n
co
n
tr
ast,
th
e
d
ir
ec
tio
n
o
f
t
h
e
d
is
tu
r
b
an
ce
r
e
v
er
s
es
if
th
e
p
o
wer
d
r
o
p
s
.
Un
til
th
e
s
y
s
tem
r
ea
ch
es
th
e
MP
P,
wh
en
th
e
r
ate
o
f
ch
an
g
e
o
f
p
o
wer
with
r
esp
ec
t
to
v
o
ltag
e
(
d
P/d
V)
ap
p
r
o
ac
h
es
ze
r
o
,
th
is
iter
ativ
e
p
r
o
ce
s
s
k
e
ep
s
g
o
in
g
.
Ma
th
em
atica
lly
,
th
e
in
s
tan
tan
eo
u
s
PV p
o
wer
is
g
iv
en
b
y
(
1
6
)
:
P(k
)
=V
(
k
)
×I
(
k
)
(
1
6
)
a
n
d
th
e
c
h
an
g
e
i
n
p
o
wer
b
etw
ee
n
two
co
n
s
ec
u
tiv
e
s
am
p
les is
ex
p
r
ess
ed
as
(
1
7
)
.
Δ
P=P(
k
)
−P(
k
−1
)
(
1
7
)
Similar
ly
,
th
e
ch
an
g
e
in
v
o
ltag
e
is
(
1
8
)
.
Δ
V=
V(
k
)
−V
(
k
−1
)
(
1
8
)
T
h
e
o
p
e
r
atin
g
p
r
in
cip
le
ca
n
b
e
o
u
tlin
ed
as f
o
llo
ws:
−
If
Δ
>0
an
d
Δ
V>
0
,
th
e
PV a
r
r
ay
v
o
ltag
e
is
in
cr
ea
s
ed
.
−
I
f
Δ
P>0
an
d
Δ
V<
0
,
th
e
PV a
r
r
ay
v
o
ltag
e
is
d
ec
r
ea
s
ed
.
−
I
f
Δ
P<0
an
d
Δ
V>
0
,
th
e
PV a
r
r
ay
v
o
ltag
e
is
d
ec
r
ea
s
ed
.
−
I
f
Δ
P<0
an
d
Δ
V<
0
,
th
e
PV a
r
r
ay
v
o
ltag
e
is
in
cr
ea
s
ed
.
T
h
is
p
r
o
ce
s
s
k
ee
p
s
th
e
PV
s
y
s
tem
o
p
er
atin
g
n
ea
r
th
e
MPP
d
esp
ite
v
ar
iatio
n
s
in
tem
p
er
at
u
r
e
an
d
ir
r
ad
ian
ce
.
Ho
wev
er
,
th
e
P&
O
m
eth
o
d
ca
n
in
d
u
ce
s
m
all
s
tead
y
-
s
tate
o
s
cillatio
n
s
ar
o
u
n
d
th
e
MPP,
wh
ich
ca
n
b
e
r
e
d
u
ce
d
b
y
ad
a
p
tiv
ely
ad
ju
s
tin
g
t
h
e
p
e
r
tu
r
b
atio
n
s
tep
s
ize.
As
illu
s
tr
ated
in
Fig
u
r
e
1
[
2
4
]
.
T
h
e
alg
o
r
ith
m
m
ea
s
u
r
es
th
e
in
s
tan
tan
eo
u
s
PV
v
o
ltag
e
a
n
d
cu
r
r
en
t,
co
m
p
u
tes
th
e
o
u
tp
u
t
p
o
wer
,
an
d
ev
alu
ates
its
v
ar
iatio
n
f
r
o
m
t
h
e
p
r
e
v
io
u
s
s
am
p
le.
B
ased
o
n
th
e
s
ig
n
s
o
f
Δ
P
an
d
Δ
V,
th
e
r
ef
er
en
ce
v
o
ltag
e
is
in
cr
ea
s
ed
o
r
d
ec
r
ea
s
ed
t
o
m
o
v
e
to
war
d
th
e
MPP,
an
d
th
e
p
r
o
ce
d
u
r
e
is
r
ep
ea
ted
f
o
r
co
n
tin
u
o
u
s
r
ea
l
-
tim
e
tr
ac
k
in
g
[
2
5
]
.
Fig
u
r
e
1
.
Flo
wch
ar
t
o
f
th
e
P&
O
MPPT
alg
o
r
ith
m
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
1
,
Ma
r
c
h
20
2
6
:
696
-
7
08
702
2
.
6
.
I
nte
g
ra
t
io
n o
f
G
H
I
F
o
r
ec
a
s
t
ing
wit
h M
P
P
T
co
ntr
o
l
T
h
e
p
r
o
p
o
s
ed
f
r
am
ewo
r
k
(
Fi
g
u
r
e
2
)
in
teg
r
ates
s
h
o
r
t
-
ter
m
GHI
f
o
r
ec
asti
n
g
d
ir
ec
tly
in
to
th
e
MPPT
co
n
tr
o
l
l
o
o
p
to
im
p
r
o
v
e
PV
s
y
s
tem
r
esp
o
n
s
iv
en
ess
u
n
d
er
r
ap
id
ly
v
a
r
y
in
g
ir
r
a
d
ian
ce
.
A
h
y
b
r
id
L
STM
–
C
NN
m
o
d
el
g
e
n
er
ates
o
n
e
-
s
tep
-
a
h
ea
d
GHI
p
r
e
d
ictio
n
s
,
Ĝ(
t+1
)
,
f
r
o
m
r
ec
en
t
m
eteo
r
o
lo
g
ical
d
ata.
T
h
ese
f
o
r
ec
asts
ar
e
tr
an
s
lated
in
to
a
r
ef
er
e
n
ce
P
V
p
o
wer
an
d
th
e
co
r
r
esp
o
n
d
in
g
o
p
tim
al
o
p
er
atin
g
v
o
ltag
e,
̂
(
+
1
)
,
u
s
in
g
tem
p
er
atu
r
e
-
d
ep
en
d
en
t
PV
ch
ar
ac
ter
is
tics
.
Un
lik
e
th
e
co
n
v
en
tio
n
al
P&
O
alg
o
r
ith
m
,
w
h
ich
r
elies
s
o
lely
o
n
in
s
tan
tan
eo
u
s
p
o
wer
p
er
tu
r
b
atio
n
s
an
d
is
p
r
o
n
e
to
o
s
cillatio
n
s
an
d
d
elay
ed
c
o
n
v
e
r
g
en
ce
,
th
e
p
r
o
p
o
s
ed
s
ch
em
e
em
p
lo
y
s
̂
(
+
1
)
as
a
f
ee
d
f
o
r
war
d
r
e
f
er
en
ce
to
p
r
e
-
ad
ju
s
t
th
e
DC
–
DC
co
n
v
er
ter
d
u
ty
cy
cle.
T
h
e
P&
O
alg
o
r
ith
m
th
en
ac
ts
o
n
ly
as
a
lo
ca
l
r
ef
in
em
en
t
m
ec
h
an
is
m
ar
o
u
n
d
th
is
p
r
ed
icted
o
p
er
atin
g
p
o
in
t.
T
h
is
p
r
ed
ictiv
e
f
ee
d
f
o
r
war
d
–
f
ee
d
b
ac
k
s
tr
u
ct
u
r
e
en
ab
les
p
r
o
ac
tiv
e
a
d
ap
tatio
n
to
ir
r
ad
ian
ce
c
h
an
g
es,
r
e
d
u
c
in
g
tr
an
s
ien
t
p
o
wer
lo
s
s
es,
l
im
itin
g
s
tead
y
-
s
tate
o
s
cillatio
n
s
,
an
d
ac
ce
ler
atin
g
co
n
v
er
g
en
ce
to
th
e
tr
u
e
MPP.
Ov
er
all,
em
b
ed
d
in
g
d
ata
-
d
r
iv
e
n
ir
r
a
d
ian
ce
f
o
r
ec
asti
n
g
tr
an
s
f
o
r
m
s
MPPT
f
r
o
m
a
r
ea
ctiv
e
s
tr
ateg
y
in
to
a
p
r
e
d
ictiv
e
c
o
n
tr
o
l
-
en
h
an
ce
d
ap
p
r
o
ac
h
,
y
ield
in
g
im
p
r
o
v
ed
p
o
wer
s
tab
ilit
y
an
d
en
er
g
y
ca
p
t
u
r
e
u
n
d
er
h
ig
h
ly
v
a
r
iab
le
atm
o
s
p
h
er
ic
co
n
d
itio
n
s
.
Fig
u
r
e
2
.
Flo
wch
ar
t
o
f
th
e
in
te
g
r
atio
n
o
f
h
y
b
r
id
L
STM
–
C
NN
GHI
f
o
r
ec
asti
n
g
with
th
e
c
o
n
v
en
tio
n
al
P&
O
MPPT
co
n
tr
o
ller
3.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
3
.
1
.
P
re
dict
iv
e
perf
o
rma
nce
o
f
G
H
I
f
o
re
ca
s
t
ing
mo
dels
Fig
u
r
e
3
illu
s
tr
ates
h
o
w
th
e
th
r
ee
d
ee
p
-
lea
r
n
in
g
m
o
d
els
r
e
p
r
o
d
u
ce
th
e
r
ea
l
GHI
p
r
o
f
ile
o
v
er
tim
e.
W
h
ile
all
m
o
d
els
ca
p
tu
r
e
th
e
g
en
er
al
d
aily
r
h
y
th
m
o
f
s
o
lar
ir
r
ad
ian
ce
,
th
e
h
y
b
r
i
d
L
STM
–
C
NN
alig
n
s
m
u
ch
m
o
r
e
tig
h
tly
with
t
h
e
m
ea
s
u
r
e
d
cu
r
v
e
.
I
ts
p
r
ed
ictio
n
s
r
em
ain
clo
s
er
to
th
e
ac
tu
al
f
lu
ct
u
atio
n
s
,
esp
ec
ially
d
u
r
in
g
r
ap
id
ch
a
n
g
es
ty
p
ically
ca
u
s
ed
b
y
p
ass
in
g
clo
u
d
s
o
r
atm
o
s
p
h
e
r
ic
d
is
tu
r
b
an
ce
s
.
T
h
is
b
eh
av
io
r
r
ef
lects
th
e
b
en
e
f
it
o
f
co
m
b
in
in
g
tem
p
o
r
al
lear
n
i
n
g
f
r
o
m
t
h
e
L
STM
with
th
e
C
NN’
s
ab
ilit
y
to
ex
tr
ac
t
m
ea
n
i
n
g
f
u
l
p
atter
n
s
f
r
o
m
th
e
in
p
u
t
f
ea
tu
r
es.
Ov
er
all,
th
e
f
ig
u
r
e
h
ig
h
lig
h
ts
th
e
h
y
b
r
id
m
o
d
el’
s
s
tr
o
n
g
e
r
s
tab
ilit
y
an
d
ac
cu
r
ac
y
,
c
o
n
f
ir
m
in
g
its
ad
v
an
tag
e
f
o
r
s
h
o
r
t
-
ter
m
ir
r
ad
ian
ce
f
o
r
ec
asti
n
g
an
d
its
p
o
ten
tial
f
o
r
m
o
r
e
r
eliab
le
s
o
lar
-
b
ased
en
er
g
y
m
an
ag
em
en
t.
Fig
u
r
es
4
(
a
)
-
4(
c
)
d
em
o
n
s
tr
ate
th
at
all
th
r
ee
m
o
d
els
ac
h
iev
e
a
s
tr
o
n
g
co
r
r
elatio
n
b
etwe
en
p
r
e
d
icted
an
d
m
ea
s
u
r
ed
GHI
v
al
u
es;
h
o
wev
er
,
th
eir
p
r
ed
ictiv
e
r
eliab
ilit
y
d
if
f
er
s
m
ar
k
e
d
ly
.
T
h
e
L
STM
m
o
d
el
ca
p
tu
r
es
th
e
o
v
er
all
tem
p
o
r
al
e
v
o
lu
tio
n
o
f
i
r
r
ad
ian
ce
b
u
t
ex
h
ib
its
in
cr
ea
s
ed
d
is
p
er
s
io
n
at
lo
w
an
d
m
o
d
er
a
te
ir
r
ad
ian
ce
lev
els,
in
d
icatin
g
r
ed
u
ce
d
r
o
b
u
s
tn
ess
d
u
r
in
g
r
ap
id
l
y
v
ar
y
in
g
c
o
n
d
iti
o
n
s
.
T
h
e
C
NN
m
o
d
el
s
h
o
ws
i
m
p
r
o
v
e
d
alig
n
m
e
n
t
in
th
e
m
i
d
-
ir
r
a
d
ian
ce
r
an
g
e,
r
ef
lectin
g
its
ef
f
ec
tiv
en
ess
in
lear
n
in
g
lo
ca
lized
f
ea
tu
r
e
p
atter
n
s
,
y
et
its
p
er
f
o
r
m
an
ce
d
eter
io
r
ates
at
ir
r
ad
ian
ce
ex
tr
e
m
es
d
u
e
t
o
lim
it
ed
tem
p
o
r
al
m
o
d
elin
g
ca
p
ab
il
ity
.
I
n
c
o
n
tr
ast,
th
e
h
y
b
r
id
L
STM
–
C
NN
m
o
d
el
d
eliv
er
s
th
e
m
o
s
t
co
n
s
is
ten
t
an
d
u
n
b
iased
p
r
ed
ictio
n
s
,
with
d
ata
p
o
in
ts
clo
s
ely
co
n
ce
n
tr
ated
alo
n
g
th
e
i
d
ea
l
r
eg
r
ess
io
n
lin
e
ac
r
o
s
s
th
e
f
u
ll
ir
r
ad
ian
ce
r
an
g
e.
T
h
is
b
eh
a
v
io
r
co
n
f
ir
m
s
th
at
co
m
b
in
in
g
co
n
v
o
lu
tio
n
al
f
ea
tu
r
e
ex
tr
ac
tio
n
with
r
ec
u
r
r
e
n
t
tem
p
o
r
al
lear
n
in
g
ef
f
ec
ti
v
ely
m
itig
ates
th
e
in
d
iv
id
u
al
lim
itatio
n
s
o
f
s
tan
d
alo
n
e
L
STM
an
d
C
NN
ar
ch
ite
ctu
r
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
P
erfo
r
ma
n
ce
en
h
a
n
ce
me
n
t o
f
p
h
o
to
v
o
lta
ic
s
ystems
u
s
in
g
h
yb
r
id
LS
TM
–
C
N
N
…
(
S
a
r
a
F
en
n
a
n
e
)
703
Fig
u
r
e
3
.
Pre
d
ictin
g
GHI
with
th
e
th
r
ee
ar
c
h
itectu
r
es
(
a)
(
b
)
(
c)
Fig
u
r
e
4
.
R
eg
r
ess
io
n
an
al
y
s
is
o
f
(
a)
L
STM
,
(
b
)
C
NN
,
an
d
(
c
)
h
y
b
r
id
L
STM
-
C
NN
m
o
d
els f
o
r
GHI
p
r
e
d
ictio
n
3
.
2
.
P
er
f
o
r
m
a
nce
ev
a
lua
t
io
n o
f
deep
lea
rning
a
rc
hite
ct
ures f
o
r
G
H
I
predict
io
n
T
ab
le
2
co
n
f
ir
m
s
th
e
s
u
p
e
r
io
r
p
er
f
o
r
m
an
ce
o
f
th
e
h
y
b
r
id
L
STM
–
C
NN
m
o
d
el,
wh
ich
ac
h
iev
es
th
e
lo
west
p
r
ed
ictio
n
er
r
o
r
s
an
d
h
ig
h
est
g
o
o
d
n
ess
o
f
f
it
am
o
n
g
t
h
e
ev
alu
ate
d
a
p
p
r
o
ac
h
es.
T
h
e
h
y
b
r
id
ar
c
h
itectu
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
0
8
8
-
8
6
9
4
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
,
Vo
l.
1
7
,
No
.
1
,
Ma
r
c
h
20
2
6
:
696
-
7
08
704
r
ec
o
r
d
s
th
e
m
in
im
u
m
R
MSE
(
2
9
.
9
9
W
/m
²)
an
d
MA
E
(
2
3
.
9
3
W
/m
²)
,
to
g
eth
er
with
th
e
h
ig
h
est
co
ef
f
icien
t
o
f
d
eter
m
in
atio
n
(
R
²
=
0
.
9
9
3
)
,
in
d
icatin
g
en
h
an
ce
d
ac
cu
r
ac
y
an
d
r
ed
u
ce
d
p
r
e
d
ictio
n
b
ias.
T
h
e
s
tan
d
alo
n
e
L
STM
m
o
d
el
s
h
o
ws
co
m
p
etitiv
e
b
u
t
less
co
n
s
is
ten
t
p
er
f
o
r
m
a
n
ce
,
with
in
cr
ea
s
ed
er
r
o
r
lev
els
d
u
r
in
g
p
er
io
d
s
o
f
r
ap
id
ir
r
ad
ian
ce
v
ar
iatio
n
,
r
ef
lectin
g
its
lim
ited
ab
ilit
y
to
ca
p
tu
r
e
lo
ca
lized
f
ea
tu
r
e
d
y
n
am
ics.
I
n
co
n
tr
ast,
th
e
C
NN
m
o
d
el
ex
h
ib
its
th
e
lar
g
est
er
r
o
r
s
an
d
lo
west
R
²
v
alu
e,
h
ig
h
lig
h
tin
g
th
e
in
ad
eq
u
ac
y
o
f
p
u
r
ely
co
n
v
o
lu
tio
n
al
s
tr
u
ctu
r
es f
o
r
m
o
d
elin
g
th
e
te
m
p
o
r
al
v
a
r
iab
ilit
y
o
f
s
o
lar
ir
r
a
d
ian
ce
.
A
co
m
p
ar
is
o
n
with
r
ec
e
n
t
s
tu
d
ies
f
u
r
th
er
em
p
h
asizes
th
e
ef
f
ec
tiv
en
ess
o
f
t
h
e
p
r
o
p
o
s
ed
ap
p
r
o
ac
h
.
Pre
v
io
u
s
h
y
b
r
id
d
ee
p
lear
n
in
g
m
o
d
els
f
o
r
s
h
o
r
t
-
ter
m
s
o
lar
ir
r
ad
ian
ce
f
o
r
ec
asti
n
g
ty
p
ically
r
ep
o
r
t
R
MSE
v
alu
es
ab
o
v
e
3
5
W
/m
²
an
d
R
²
v
alu
es
b
elo
w
0
.
9
9
[
2
6
]
.
Dee
p
lear
n
in
g
m
o
d
els
ev
alu
ated
o
v
e
r
lo
n
g
er
f
o
r
ec
asti
n
g
h
o
r
izo
n
s
g
en
e
r
ally
ex
h
ib
it
h
ig
h
er
er
r
o
r
lev
els wh
e
n
a
p
p
lied
t
o
h
ig
h
-
r
eso
lu
tio
n
tem
p
o
r
al
d
at
a
[
2
7
]
.
Mo
r
e
r
ec
en
t
h
y
b
r
id
an
d
tr
a
n
s
f
er
-
lear
n
in
g
-
b
ased
f
o
r
ec
asti
n
g
f
r
a
m
ewo
r
k
s
r
ep
o
r
t
R
MSE
v
alu
es
in
th
e
r
a
n
g
e
o
f
3
0
–
4
0
W
/m
²
an
d
MA
E
v
al
u
es
ex
ce
ed
in
g
2
5
W
/m
²,
p
ar
ticu
lar
ly
u
n
d
er
r
a
p
id
ly
ch
a
n
g
in
g
atm
o
s
p
h
e
r
ic
c
o
n
d
itio
n
s
[
2
8
]
,
[
2
9
]
.
I
n
co
n
tr
ast,
th
e
p
r
o
p
o
s
ed
h
y
b
r
id
L
STM
–
C
NN
m
o
d
el
c
o
n
s
is
ten
tly
ac
h
iev
es
lo
wer
er
r
o
r
m
etr
ics
an
d
s
tr
o
n
g
er
g
o
o
d
n
ess
o
f
f
it,
co
n
f
ir
m
in
g
its
r
o
b
u
s
tn
ess
an
d
s
u
p
er
io
r
p
r
ed
i
ctiv
e
ca
p
ab
ilit
y
f
o
r
s
h
o
r
t
-
ter
m
GHI
f
o
r
ec
asti
n
g
.
T
ab
le
2
.
C
o
m
p
a
r
ativ
e
p
er
f
o
r
m
an
ce
m
etr
ics o
f
th
e
t
h
r
ee
p
r
ed
i
ctiv
e
m
o
d
els f
o
r
GHI
esti
m
atio
n
Ev
a
l
u
a
t
i
o
n
me
t
r
i
c
s
D
e
e
p
l
e
a
r
n
i
n
g
mo
d
e
l
s
LSTM
C
N
N
LSTM
-
C
N
N
R
M
S
E
4
0
.
0
0
9
2
4
3
6
0
.
1
1
9
8
5
0
2
9
.
9
9
8
5
1
2
M
A
E
3
1
.
9
3
5
6
6
7
4
7
.
9
5
4
8
2
5
2
3
.
9
3
2
2
6
7
R²
0
.
9
8
7
5
0
9
0
.
9
7
1
7
9
7
0
.
9
9
2
9
7
8
3
.
3
.
Ass
ess
m
ent
o
f
M
P
P
T
perf
o
rma
nce
us
ing
t
he
P
&O
co
ntr
o
ller
T
h
e
p
r
o
p
o
s
ed
m
o
d
el
f
o
r
th
is
wo
r
k
,
d
esig
n
e
d
an
d
test
ed
in
MA
T
L
AB
/S
im
u
lin
k
,
is
s
h
o
wn
in
Fig
u
r
e
5
.
Usi
n
g
th
e
p
r
ed
icted
GHI
f
r
o
m
th
e
h
y
b
r
id
L
STM
–
C
NN
m
o
d
el
as
in
p
u
t,
Fig
u
r
e
6
illu
s
tr
ates
th
e
d
y
n
am
ic
r
esp
o
n
s
e
o
f
th
e
p
h
o
t
o
v
o
ltaic
(
PV)
s
y
s
tem
u
n
d
er
th
e
P&
O
M
PP
T
co
n
tr
o
l
s
tr
ateg
y
.
T
h
is
ap
p
r
o
ac
h
allo
ws
th
e
PV
s
y
s
tem
to
ad
j
u
s
t
p
r
o
ac
tiv
ely
t
o
ir
r
ad
ian
ce
c
o
n
d
itio
n
s
f
o
r
ec
a
s
ted
b
y
th
e
d
ee
p
lear
n
in
g
m
o
d
el.
T
h
r
o
u
g
h
o
u
t
th
e
s
im
u
latio
n
,
s
o
lar
ir
r
ad
ian
ce
f
o
llo
ws
a
s
m
o
o
th
,
b
ell
-
s
h
a
p
ed
d
iu
r
n
al
p
r
o
f
ile,
r
esu
ltin
g
in
p
r
o
p
o
r
tio
n
al
v
ar
iatio
n
s
in
PV
v
o
ltag
e
(
<V
_
PV>)
,
cu
r
r
en
t
(
<I
_
PV>)
,
an
d
p
o
wer
o
u
t
p
u
t.
T
h
e
m
ea
s
u
r
em
e
n
ts
co
n
f
ir
m
th
at
th
e
s
y
s
tem
co
n
s
is
ten
tly
tr
ac
k
s
th
e
MPP
,
wh
ile
th
e
d
u
ty
cy
cle
(
<d
u
ty
>)
s
tab
ilizes
af
ter
a
b
r
ie
f
in
itial
tr
an
s
ien
t,
d
em
o
n
s
tr
atin
g
th
at
th
e
P&
O
al
g
o
r
ith
m
ef
f
ec
tiv
ely
co
n
v
er
g
es
d
esp
ite
th
e
n
o
n
-
lin
ea
r
b
eh
av
i
o
r
o
f
th
e
PV
s
y
s
tem
.
Fig
u
r
e
5
.
Simu
lin
k
m
o
d
el
o
f
P
V
s
y
s
tem
with
P&
O
MPPT
an
d
b
o
o
s
t c
o
n
v
er
ter
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J Po
w
E
lec
&
Dr
i Sy
s
t
I
SS
N:
2088
-
8
6
9
4
P
erfo
r
ma
n
ce
en
h
a
n
ce
me
n
t o
f
p
h
o
to
v
o
lta
ic
s
ystems
u
s
in
g
h
yb
r
id
LS
TM
–
C
N
N
…
(
S
a
r
a
F
en
n
a
n
e
)
705
T
h
ese
o
b
s
er
v
atio
n
s
h
i
g
h
lig
h
t
t
h
e
s
tr
o
n
g
s
y
n
e
r
g
y
b
etwe
en
ac
cu
r
ate
d
ata
-
d
r
iv
en
ir
r
ad
ian
ce
f
o
r
ec
asti
n
g
an
d
MPPT
co
n
tr
o
l.
B
y
in
teg
r
atin
g
h
y
b
r
id
L
STM
–
C
NN
p
r
e
d
ictio
n
s
with
th
e
P&
O
alg
o
r
it
h
m
,
th
e
PV
s
y
s
tem
ca
n
an
ticip
ate
ch
an
g
es
in
s
o
lar
co
n
d
itio
n
s
an
d
ad
ju
s
t
its
o
p
er
atin
g
p
o
in
t
i
n
r
ea
l
tim
e.
T
h
is
in
teg
r
atio
n
im
p
r
o
v
es
p
o
wer
s
tab
ilit
y
,
en
s
u
r
es
ef
f
icien
t
en
er
g
y
ex
tr
ac
tio
n
,
an
d
en
h
an
ce
s
th
e
s
y
s
tem
’
s
r
esil
ien
ce
t
o
r
ap
id
ir
r
ad
ian
ce
f
lu
ctu
atio
n
s
.
Mo
r
eo
v
er
,
r
eliab
l
e
s
h
o
r
t
-
ter
m
f
o
r
ec
asti
n
g
f
ac
ilit
ates
b
etter
en
er
g
y
s
ch
ed
u
lin
g
an
d
s
m
o
o
th
p
o
we
r
in
jectio
n
in
to
th
e
g
r
id
,
co
n
tr
ib
u
tin
g
to
o
v
er
all
g
r
id
s
tab
ilit
y
.
T
h
e
in
s
tan
tan
eo
u
s
PV
p
o
wer
o
u
tp
u
t
d
u
r
in
g
t
h
e
MPPT
s
i
m
u
latio
n
d
r
iv
e
n
b
y
h
y
b
r
id
L
STM
–
C
NN
p
r
ed
ictio
n
s
f
o
llo
w
th
e
ex
p
ec
te
d
b
ell
-
s
h
ap
ed
d
aily
tr
ajec
t
o
r
y
,
r
ea
ch
in
g
a
p
ea
k
o
f
a
p
p
r
o
x
im
at
ely
8
5
0
W
ar
o
u
n
d
m
id
d
ay
.
Min
o
r
o
s
cillatio
n
s
ar
e
o
b
s
er
v
e
d
,
r
e
f
lectin
g
t
h
e
n
o
r
m
al
p
er
tu
r
b
atio
n
b
eh
a
v
io
r
o
f
th
e
P&
O
m
eth
o
d
.
T
h
ese
s
m
all
d
ev
iatio
n
s
in
d
icate
a
s
u
itab
le
tr
ad
e
-
o
f
f
b
etw
ee
n
f
ast
tr
ac
k
in
g
an
d
s
tab
le
r
eg
u
latio
n
,
en
s
u
r
in
g
co
n
s
is
ten
t sy
s
tem
p
er
f
o
r
m
an
c
e
th
r
o
u
g
h
o
u
t th
e
d
a
y
.
B
y
co
m
b
in
in
g
ac
cu
r
ate
s
h
o
r
t
-
t
er
m
f
o
r
ec
asti
n
g
with
r
ea
l
-
tim
e
MPPT
co
n
tr
o
l,
th
is
f
r
a
m
ewo
r
k
n
o
t
o
n
ly
m
ax
im
izes
PV
ef
f
icien
cy
b
u
t
also
s
tr
en
g
th
en
s
g
r
id
in
teg
r
ati
o
n
.
Pre
d
ictab
le
PV
g
en
er
atio
n
allo
ws
o
p
er
ato
r
s
to
s
ch
ed
u
le
en
er
g
y
m
o
r
e
ef
f
ec
ti
v
ely
,
r
ed
u
ce
s
v
ar
iab
ilit
y
in
p
o
wer
in
jectio
n
,
an
d
en
h
a
n
ce
s
th
e
r
esil
ien
ce
o
f
PV
s
y
s
tem
s
u
n
d
er
d
y
n
am
ic
wea
t
h
er
c
o
n
d
itio
n
s
.
T
h
e
r
esu
lt
s
u
n
d
er
s
co
r
e
th
e
p
o
ten
tial
o
f
co
u
p
lin
g
in
tellig
en
t
f
o
r
ec
asti
n
g
with
class
ical
co
n
tr
o
l
s
tr
ateg
ies
to
o
p
tim
ize
PV
p
er
f
o
r
m
an
ce
,
s
u
p
p
o
r
t
r
eliab
le
en
er
g
y
s
u
p
p
ly
,
an
d
f
ac
ilit
ate
s
m
ar
ter
m
an
ag
e
m
en
t
o
f
d
is
tr
ib
u
ted
s
o
lar
r
eso
u
r
ce
s
in
s
em
i
-
ar
id
r
eg
io
n
s
lik
e
Mo
r
o
cc
o
an
d
in
b
r
o
a
d
er
s
m
ar
t
-
g
r
id
ap
p
licatio
n
s
.
Fig
u
r
e
6
.
Simu
latio
n
r
esu
lts
o
f
P&
O
MPPT
alg
o
r
ith
m
4.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
d
ev
elo
p
ed
an
d
v
alid
ated
a
h
y
b
r
id
L
STM
–
C
NN
f
o
r
ec
asti
n
g
m
o
d
el
in
teg
r
ate
d
with
a
P&
O
MPPT
co
n
tr
o
ller
f
o
r
p
h
o
to
v
o
ltaic
s
y
s
tem
s
.
T
h
e
m
o
d
el
ac
h
iev
ed
th
e
h
i
g
h
est
p
r
e
d
ictiv
e
a
cc
u
r
ac
y
am
o
n
g
th
e
test
ed
ar
ch
itectu
r
es,
with
an
R
MSE
o
f
2
9
.
9
9
W
/m
²,
an
MA
E
o
f
2
3
.
9
3
W
/m
²,
an
d
a
n
R
²
o
f
0
.
9
9
3
,
r
ep
r
esen
tin
g
a
s
ig
n
if
ican
t
im
p
r
o
v
em
en
t
o
v
e
r
s
tan
d
alo
n
e
L
STM
an
d
C
NN
n
etwo
r
k
s
.
W
h
en
ap
p
lied
to
th
e
MPPT
s
im
u
latio
n
,
f
o
r
ec
asted
GHI
en
ab
led
f
aster
d
u
ty
-
cy
cle
c
o
n
v
e
r
g
en
ce
a
n
d
r
ed
u
ce
d
o
s
cillatio
n
s
ar
o
u
n
d
th
e
MM
P,
r
esu
ltin
g
in
sm
o
o
th
er
v
o
ltag
e,
cu
r
r
en
t,
a
n
d
p
o
wer
p
r
o
f
iles
an
d
m
o
r
e
s
tab
l
e
PV e
n
er
g
y
o
u
tp
u
t.
T
h
ese
r
esu
lts
h
av
e
d
ir
ec
t
im
p
licatio
n
s
f
o
r
r
en
ewa
b
le
e
n
er
g
y
i
n
teg
r
atio
n
s
tr
ateg
ies.
I
n
Mo
r
o
c
co
,
wh
er
e
s
o
lar
r
eso
u
r
ce
s
ar
e
ab
u
n
d
an
t
b
u
t
s
u
b
ject
to
h
ig
h
ir
r
ad
ia
n
ce
v
ar
iab
ilit
y
,
ac
cu
r
ate
s
h
o
r
t
-
ter
m
f
o
r
ec
asti
n
g
co
m
b
in
ed
with
a
d
ap
tiv
e
MPPT
en
h
an
ce
s
g
r
id
s
tab
ili
ty
,
i
m
p
r
o
v
es
en
er
g
y
s
ch
e
d
u
lin
g
,
an
d
m
a
x
im
izes
PV
co
n
tr
ib
u
tio
n
to
th
e
n
atio
n
al
en
er
g
y
m
ix
.
Glo
b
ally
,
th
e
a
p
p
r
o
a
ch
d
em
o
n
s
tr
ates
a
s
ca
lab
le
p
ath
way
f
o
r
i
n
teg
r
atin
g
d
is
tr
ib
u
ted
PV
s
y
s
tem
s
in
to
s
m
ar
t
g
r
id
s
,
s
u
p
p
o
r
tin
g
ef
f
icien
t
en
er
g
y
m
an
a
g
em
en
t,
an
d
m
itig
atin
g
in
ter
m
itten
c
y
ch
allen
g
es in
v
ar
ia
b
le
clim
ates.
Evaluation Warning : The document was created with Spire.PDF for Python.