TELKOM
NIKA
, Vol. 11, No. 4, April 2013, pp. 2239
~22
4
6
ISSN: 2302-4
046
2239
Re
cei
v
ed
De
cem
ber 5, 20
12; Re
vised
Ma
rch 1, 201
3; Acce
pted
March 9, 201
3
A Method for Electric Vehicle Ownership Forecast
Considering Different Economic Factors
Han
w
u Luo
1
, Fang Li*
2
1
School of Elec
trical Eng
i
ne
eri
ng, W
uhan U
n
i
v
ersi
t
y
, W
u
h
a
n
,
43007
2, Chi
n
a, Eastern Inne
r Mongo
lia
Electric Po
w
e
r Comp
an
y, Hoh
hot, 010
020, C
h
in
a;
2
Hena
n Electri
c
Po
w
e
r Com
p
an
y, Z
hengz
ho
u, 4500
52, Ch
i
n
a
*Cores
po
ndi
ng
author, e-mai
l
: lifang
hn
ep@
1
63.com
A
b
st
r
a
ct
T
he constructi
on of electric v
ehicl
es (EVs) char
g
i
ng statio
n nee
ds to be
pla
ned acc
o
rd
i
ng to the
ow
nershi
p
of
EVs, traffic conditi
on, p
o
p
u
la
tion etc.
T
her
efore
a BP n
eura
l
n
e
tw
ork base
d
meth
od
t
o
forecast the E
V
ow
nershi
p
for a city is pre
s
ented
in the
pap
er, w
h
ich c
onsi
ders the
in
fluenc
e on th
e
EV
ow
nershi
p
cau
s
ed by
many
relate
d eco
n
o
m
y factors,
in
cludi
ng GDP
of a city, vehicle pro
ductio
n
,
per
capita cr
ude steel
production, per
c
apita generation capacit
y, road
pa
ssenger traffic, highway m
i
leage
and
the total
p
opu
l
a
tion.
A BP
n
eura
l
n
e
tw
ork i
s
set u
p
for th
e forec
a
st of E
V
ow
nersh
ip,
a
nd th
e i
n
p
u
t la
yer
contai
ns seve
n
neutro
ns, w
h
ich repres
ent d
i
fferent ec
ono
mi
c factors. T
her
e are thre
e n
e
u
r
ons in
its hid
d
e
n
layer, a
nd th
e
output
is the
EV ow
nershi
p
.
T
hen th
e
method t
o
pre
d
i
c
t the EV ow
n
e
rshi
p of a c
i
ty is
prese
n
ted, w
h
ich is based o
n
the fore
cast of the civilian ca
r ow
nership
in
a city and the country. T
he E
V
ow
nershi
p
i
n
th
e city of
Ch
ong
qin
g
fro
m
th
e y
ear
201
3 to
20
20 is
pr
edicte
d
, an
d the
acc
u
racy of th
e
mo
d
e
l
is verifie
d
firstly, then the EV ow
nership i
n
Cho
ngq
in
g
is obtai
ne
d, w
h
ich is hel
pful to mak
e
pl
ans for
th
e
deve
l
op
ment o
f
electric vehic
l
e.
Ke
y
w
ords
: ow
nersh
ip, BP ne
ural n
e
tw
ork, electric vehic
l
e, civili
an car ow
n
e
rshi
p
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Red
u
ci
ng de
pend
en
ce o
n
the crud
e
oil and emissi
on
s of carb
on dio
x
ide and
particulate
s are am
ong t
he leadi
ng reason
s that el
ectri
c
vehi
cles (EV
s
) a
r
e increa
sing
in
popul
arity. Most EVs are
plann
ed to have a fully
electri
c
ra
nge
betwe
en 10
-40 miles, whi
c
h is
within th
e
da
ily comm
ute
distan
ce
of t
he ave
r
a
ge
driver [1]. Ult
i
mately, EVs will
shift en
ergy
deman
ds fro
m
crude oil t
o
elect
r
icity for the pe
rso
nal tran
sp
ort
a
tion se
cto
r
. To pro
m
ote t
h
e
developm
ent of
EVs,
many
cha
r
gin
g
sta
t
ions
h
a
ve
be
en
built,
an
d more and mo
re will
be built
[2-5].
Mean
while
m
any stu
d
ie
s
regarding
EVs are
b
e
ing
co
ndu
cted, in
cl
uding
the
de
sign a
n
d
optimizatio
n of
ch
arging st
ations,
i
n
vesti
gation
on th
e
co
ntrol
of ve
hic
l
e-to-grid (V2G) [1]
[6-12],
analysi
s
o
n
the influe
nce
cau
s
e
d
by th
e ch
ar
gi
ng
machi
n
e
s
[1
3-16], e
nergy storage
of EV,
cha
r
gin
g
tech
nique
s [17
-
20
], and so on.
In China th
e national g
r
id
has
starte
d the co
nst
r
u
c
tion
of charging
station sin
c
e 2
009, whi
c
h ai
ms at fasting
the prom
otion
of EVs.
For
the con
s
truction of
chargi
ng station
and
other facilities rel
a
ted to EVs, a better
planni
ng will i
n
crea
se the
eco
nomy ben
efits. Theref
o
r
e it is ne
ce
ssary to get th
e own
e
rship
of
the
ele
c
tric
vehicl
es, with whi
c
h we can
make
re
ason
able pl
an
s for the co
nst
r
u
c
tion of chargin
g
station, strat
egie
s
,
etc. T
he
fo
recast
of
t
he o
w
n
e
rship
of EVs is
co
ndu
cte
d
ba
se
d o
n
the
statistical ownership of the EVs.
However the dev
elopment of
electri
c
vehicles i
s
still at
the
initial stage, h
ence there i
s
not enou
gh st
atistical d
a
ta for the fore
ca
st of the ownership of EVs.
In this stu
d
y, a m
e
thod
b
a
se
d o
n
BP
neu
ral
net
work to fo
re
ca
st the
lon
g
-t
erm EV
own
e
rship
for a city i
s
p
r
e
s
ented, which
con
s
id
ers the
influen
ce
on
the EV ownership
ca
used
by
many related
econo
my factors, inclu
d
i
ng GDP of
a
city, vehicle produ
ction,
per capita crude
steel produ
cti
on, per capit
a
gene
ration
cap
a
city
, roa
d
passe
nge
r traffic, highway mileage a
nd
the total pop
ulation. A BP neural n
e
twork i
s
se
t u
p
for the fore
ca
st of EV owne
rship, an
d the
input layer co
ntains
seven
neutro
ns, whi
c
h re
present
different econ
omic facto
r
s. There are th
ree
neuron
s in its hidden laye
r, and the outp
u
t is the EV
o
w
ne
rship. Then the metho
d
to predi
ct the
EV own
e
rshi
p of a
city i
s
pre
s
ente
d
, which
is ba
se
d
on th
e fo
re
cast of th
e
civilian
car o
w
ne
rship
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 4, April 2013 : 2239 – 2
246
2240
in a city. The EV ownersh
ip in Chon
gq
ing from
the year 201
3 to 2020 is pre
d
icted, and t
h
e
accuracy of t
he mod
e
l is v
e
rified firstly, then
the EV own
e
rship i
n
Cho
ngqin
g
is obtained,
wh
ich
is helpful to
make pl
an
s for the
develo
p
ment of electric vehicl
e.
2. Basics o
f
BP Neur
al Net
w
o
r
k
2.1. BP Neur
al Net
w
o
r
k
BP neu
ral n
e
twork [1] [6
-10] i
s
a mul
t
i-layer
hierarchi
c
al
neu
ral
network
wit
h
up
per
neuron
s fully asso
ciated
wi
th lower n
e
u
r
ons. When
a
cou
p
le of learning sample
s are supplie
d to
the network, the tran
sfe
rre
d value is
pro
pagate
d
fr
om
the input lay
e
r throug
h mi
ddle laye
r to the
output layer,
and we ca
n g
e
t neural net
work inp
u
t
re
spo
n
se from
neuron
s in ou
tput layer. Along
the dire
ction
of red
u
ci
ng
the erro
r b
e
twee
n
expe
cted o
u
tput
and a
c
tual
o
u
tput, con
n
e
c
tion
weig
hts are adju
s
ted
f
r
o
m
the output
layer to ev
ery middle layer, and ultim
a
tely to the input
layer. With th
e ongoi
ng a
m
endm
ent b
y
this back-p
r
opa
gation, t
he co
rrect rate for the net
work
respon
se to i
nput also increases
contin
uou
sly.
As BP algorithm i
m
pleme
n
ts m
i
ddle hid
den l
a
yer
and ha
s a correspon
ding
learnin
g
rul
e
s to follo
w, it has the ab
ility to
identify the non-lin
ear
pattern. Espe
cially to those learning th
at has
clea
r cal
c
ulatio
n
m
e
thod
s
and well-defin
ed steps,
BP algorithm
has mo
re ext
ensive a
pplications.
A BP neural
netwo
rk i
s
u
s
ually com
p
o
s
ed of input la
yer, hidde
n la
yer (mid
dle la
yer) an
d
output laye
r
as
sh
own
in
Figure 1. F
o
r som
e
p
r
a
c
tical
pro
b
lem
s
, many hi
dde
n layers m
a
y b
e
use
d
. Accordi
ng to the
stru
cture
of BP n
eural
net
work, the learning
informatio
n o
f
BP is forward
prop
agatio
n,
and the
erro
r is b
a
ck p
r
o
p
agation.
Hen
c
e the
main
pro
c
e
s
s of B
P
is
comp
rise
d o
f
two pa
rts: th
e forward
cal
c
ulatio
n on
the inp
u
t
info
rmation and
backward cal
c
ulatio
n
on
the
error.
Figure 1. Structure of BP Neutral Netwo
r
k
2.2. Process
of BP Neural
Net
w
o
r
k
The ne
uro
n
s
in the input la
yer coll
ect th
e informatio
n
and tra
n
sfe
r
the inform
atio
n to the
neuron
s in th
e hidd
en lay
e
rs.
Then
cal
c
ulatio
n on t
he inp
u
t information in i
n
p
u
t layer
will
be
carrie
d out in
the hidden l
a
yer. The
structure of
hid
den layer
usually is de
sig
ned a
c
cordi
n
g to
the ch
aracte
ristics of the i
nput
info
rmat
ion. Finally th
e re
sult
s obt
ained i
n
the
hidde
n layer
are
transfe
rred to
the output layer, now the f
o
rward learni
ng pro
c
e
s
s is finished [7, 8
]
.
If the error b
e
twee
n the o
u
tput value a
nd t
he o
b
je
ctive is g
r
eat,
the ba
ckwa
rd
erro
r
cal
c
ulation and amendment
will be
started. Firstly the error
will
be transferred to the output lay
e
r
then the wei
ghts for ea
ch
layer will be
chang
ed
a
c
cording to gradient de
sce
n
t method. This
process
will be repeated until the
error satisfies the
cri
t
erion.
BP neural n
e
twork le
arni
ng
rule is
a su
p
e
rvise
d
lea
r
ni
ng metho
d
, so a ce
rtain n
u
m
ber of
training
samp
les a
r
e
nee
d
ed which a
r
e
the sta
nda
rd
input a
nd o
u
t
put vectors.
Figure 2 i
s
th
e
pro
c
e
ss
of the wei
ghts
adj
ustment, h
e
re
()
k
dn
is the
obje
c
tive,
()
k
y
n
is the
output an
d
()
k
en
is
the error.
1
x
i
x
n
x
1
y
i
y
j
y
m
y
1
o
l
o
V
W
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Method for
Electri
c
Vehicle Ownership
Fore
cast Considering
Diffe
rent... (Hanwu Luo)
2241
Figure 2. The
Proce
s
s of Erro
r Co
rrectio
n
2.3. Algorith
m
of BP Neu
r
al Net
w
o
r
k
The alg
o
rith
m use
d
in
BP neural n
e
twork
c
onta
i
ns two
part
s
: the first p
a
rt is the
cal
c
ulatio
n of
error an
d a
m
endm
ent of
weig
hts
fo
r
the outp
u
t la
yer, the
se
cond p
a
rt i
s
t
h
e
cal
c
ulatio
n of
error and
am
endme
n
t of
weights in th
e
hidde
n laye
r.
To
simplify th
e an
alysi
s
, he
re
we ta
ke
the B
P
neu
ral
net
work co
ntaini
ng o
n
ly on
e h
i
dden
layer fo
r a
n
exam
ple
to formul
ate t
he
equatio
ns u
s
ed in BP network.
For the BP n
eural
network contai
ning o
ne hi
dd
en lay
e
r, the e
rro
r
E betwee
n
th
e output
and the obj
ective is as follo
ws:
22
1
11
()
(
)
22
l
kk
k
Ed
o
dO
(
1
)
Expand the e
rro
r
E
to the hidden laye
r, then we ca
n calcul
ate
E
as
follows
:
22
11
0
11
[(
)
]
[(
)
]
22
ll
m
kk
k
j
k
j
kk
j
Ed
f
n
e
t
d
f
w
y
(2)
And for the in
put layer, the error
E
is:
22
10
10
0
11
{[
(
)
]
}
=
{[
(
)
]
}
22
lm
lm
m
kj
k
j
kj
k
i
j
i
kj
kj
i
Ed
f
w
f
n
e
t
d
f
w
f
v
x
(
3
)
It can
be
co
n
c
lud
ed th
at th
e e
rro
r
E
in the input layer
is
the func
tion of
j
k
w
and
ij
v
, here
j
k
w
and
ij
v
are the weig
ht for the hidde
n laye
r and in
put la
yer re
spe
c
tively. Hence the error
E
can b
e
chan
ged by ame
n
d
ing the weig
hts. The pu
rp
ose to am
en
d the weig
hts is to decrea
s
e
the erro
r, therefore the adj
ustm
ent of the weight
s sh
ould be prop
ortional to the gradi
ent of the
error, whi
c
h a
r
e de
scrib
ed
as follo
ws:
0
,
1,
,
;
1,
2
,
,
jk
jk
E
wj
m
k
l
w
(4)
0
,
1,
,
;
1,
2
,
,
ij
ij
E
vi
n
j
m
v
(5)
Whe
r
e
is a consta
nt, rangi
ng from 0 to 1.
3. Model for
the long-ter
m
Forecas
t of EV O
w
n
e
r
s
hip
3.1 Metho
d
to Foreca
st the EV O
w
n
e
rship
The predi
ctio
n of EV owne
rshi
p nee
ds h
i
stori
c
al
an
d statistical dat
a as the sam
p
le data.
However, the development
of the electri
c
vehi
cle is
st
ill in the initial
stage, the hi
stori
c
al
data i
s
not enou
gh for an a
c
cura
te foreca
stin
g, therefore it
is difficult to forecast th
e own
e
rship
of
electri
c
vehi
cl
e.
1
x
0
1
x
2
x
n
x
0
jj
W
j
S
()
()
k
yn
()
k
dn
()
k
en
1
j
W
2
j
W
nj
W
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 4, April 2013 : 2239 – 2
246
2242
For the
con
s
t
r
uctio
n
of the
cha
r
gin
g
sta
t
ions in a
cit
y
, we need t
o
make a sh
ort-te
rm
planni
ng
and
long
-term
pl
annin
g
, which can
help
to in
cre
a
se t
he e
c
o
nomi
c
ben
efit of t
he
cha
r
gin
g
stati
ons.
Hen
c
e i
t
’s ne
ce
ssary
to fore
ca
st t
he o
w
ne
rship
of the
ele
c
tric vehi
cle. T
o
prom
ote the
developm
ent
of ele
c
tr
ic vehicl
e, the
Chine
s
e g
o
vernment h
a
s
m
ade a
serie
s
of
strategi
es
an
d plan
s for t
he develo
p
m
ent of elec
t
r
i
c
vehi
cles, in
whi
c
h the E
V
owne
rship
in
Chin
a for 201
5, 2020 are ai
med to be 2.6
6
million and
16.98 million
respe
c
tively.
The governm
ent will take many measures to
achieve the aim of EV ownership i
n
China.
Hen
c
e the fo
reca
st of the EV owne
rshi
p of a
city ca
n be condu
ct
ed ba
sed o
n
the EV owne
rship
for the
whole
cou
n
try, whi
c
h is p
r
edi
cted
by t
he government. The
p
r
ocess fo
r the
fore
ca
st of EV
own
e
rship of
a city is as fol
l
ows:
1.
Foreca
st the long-term civilian
car o
w
nership of a city
acco
rding
to its historical a
nd statistical
data based on BP neural n
e
twork.
2.
Foreca
st the long-term
civ
ilian car ownership
of the country a
c
cording
to the country’
s
historical and statistical data based on BP neural network.
3.
Calculate the EV ownershi
p
for each ye
ar in
the
futur
e
based on cubic Hermite interpolation
and the ownership plannin
g
by
the gove
r
nment.
4.
According to the long-term civilian car owner
ship of the city and
the country and planning EV
ownership of the country,
the long-term EV ownersh
ip
of
the city can be calculated as follows.
_
__
_
**
c
a
r
c
ity
E
V
c
ity
E
V
c
ountr
y
c
a
r
c
ountry
C
CC
C
(6)
Whe
r
e
_
C
E
V
c
ity
is the EV ownership of a city,
_
C
E
V
c
ountr
y
is the EV o
w
ne
rship of the co
untry,
_
c
a
r
c
ity
C
is the
civilian car owne
rshi
p of the
city and
_
c
a
r
c
oun
try
C
is the
civi
lian car
owne
rshi
p of the
cou
n
t
r
y
.
λ
is a coefficient
indicting tha
t
the differ
ence of the strategie
s
on the prom
otion of
electri
c
vehi
cl
e betwe
en
a
city and the country.
3.2. BP Neur
al Net
w
o
r
k for the For
ec
ast o
f
Car O
w
n
e
rs
hip
To forecast t
he EV ownership of a
city, t
he civilian
car ownershi
p of a
city and the
cou
n
try sho
u
l
d
be fore
ca
st firstly, wh
ich i
s
pre
d
icte
d b
y
BP neural network.
The o
w
ne
rsh
i
p of civilian car
not only depe
nd
s on
the histo
r
ical data of civilian ca
r
own
e
rship, b
u
t also o
n
other
statistical
data of
the e
c
on
omy, incl
uding G
r
o
s
s Dome
stic P
r
o
duct
(G
DP), vehi
cle p
r
od
uctio
n
, per
ca
pita
crude
st
eel prod
uctio
n
,
p
e
r capita gen
eration ca
pa
city,
road
pa
ssen
g
e
r traffic, hig
h
w
ay mile
age
and th
e to
tal
popul
ation. Hence all th
ese stati
s
tical
d
a
ta
need to be
co
nsid
ere
d
in the forecast of
civilian car o
w
ne
rship.
In this study, the input layer of BP neural net
work is comp
rised o
f
all the factors liste
d
above
and
th
e hi
stori
c
al
ci
vilian ca
r
own
e
rship, a
nd
o
n
ly one
hidd
e
n
layer is u
s
e
d
. The
co
unt
o
f
neuron
s in the hidden laye
r is
ln
m
, here n is the cou
n
t of neuro
n
s in the input layer, and
m is the cou
n
t of neuron
s in the output layer.
Statistical data in the past 20 years a
r
e u
s
ed
as
training
sam
p
les. Figu
re 3
is the BP ne
ural n
e
tw
o
r
k
f
o
r t
he f
o
re
ca
st
of
civilian car ownershi
p
.
There are th
ree neu
ron
s
in
the hidden la
yer,
the outpu
t is the civilian car o
w
n
e
rship.
Figure 3. BP
Neutral Net
w
ork F
o
r Th
e
F
o
re
ca
st Of Ci
vilian Car O
w
nership
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Method for
Electri
c
Vehicle Ownership
Fore
cast Considering
Diffe
rent... (Hanwu Luo)
2243
For the forecast of civilian
car o
w
n
e
rshi
p,
the data in the past 20 y
ears are take
n as the
training
sam
p
les, and the e
rro
r between
the output
an
d real value i
s
co
mputed a
s
follows:
0
0
()
(
)
(
)
1
00%
()
yk
y
k
ek
yk
(7)
2
1
(
(
)
/
10
0)
N
i
ek
N
(8)
whe
r
e
()
y
k
i
s
the
predi
cted
civil
i
an
car o
w
ne
rship,
0
()
y
k
is the
statistical
civili
an
ca
r o
w
ne
rship,
()
ek
is the relati
ve erro
r bet
wee
n
them,
is mea
n
sq
u
a
re e
r
ror an
d N is the y
ears to be
forecas
t
.
4. Forecas
t
of EV O
w
n
e
r
s
hip in the Cit
y
of Chong
qing
To verify the
method p
r
e
s
ente
d
in th
e pape
r, the
EV owne
rship in Chon
gqing i
s
predi
cted b
a
sed on the stat
istical d
a
ta from the
year 1
991 to 2010 a
s
listed in the
Appendix [12]
.
Firstly the
ci
vilian ca
r o
w
nership
of Chong
qing
an
d the
cou
n
try
is p
r
edi
cted,
and th
e rela
tive
error for the p
r
edi
ction is a
nalyze
d
.
4.1. Forecas
t of the Civ
ilian Car O
w
n
e
rship
Table I sho
w
s the comp
arison of civilian
ca
r own
e
rship bet
we
en the statistics an
d
cal
c
ulate
d
value in China from the year
2004 to
2
010
, and the rel
a
tive erro
r is
shown in Figu
re
4. The maximum relative
erro
r is le
ss than 2%,
hence the a
c
cu
racy of the fo
recast meth
o
d
is
accepta
b
le.
Table 1. Co
m
pari
s
on of Civ
ilian Ca
rs Bet
w
ee
n the Cal
c
ulate
d
Value
s
and Statisti
cs (Unit:
Million)
Y
e
a
r
2004
2005
2006
2007
Statistics
26.9371
31.5966
36.9735
43.5836
Calculated value
27.4092
31.8295
36.9626
42.9235
Y
e
a
r
2008
2009
2010
Statistics
50.9961
62.8061
78.0183
Calculated value
51.8120
63.2632
78.3105
Figure 4. Rel
a
tive Error of
Predi
cted Val
ue and the
Real Value
Table 2. Ownership of Civ
ilian Ca
r in China from 2
0
1
3
to 2020 (Un
i
t: million)
Y
e
a
r
2013
2014
2015
2016
Civilian car owne
rship
105.267
122.2433
141.9573
164.8506
Y
e
a
r
2017
2018
2019
2020
Civilian car owne
rship
191.4359
222.3085
258.1599
299.793
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 4, April 2013 : 2239 – 2
246
2244
Table 3. O
w
n
e
rship of Civil
i
an Ca
r in Ch
ongqi
ng from
2013 to 202
0
(Unit: million)
Y
e
a
r
2013
2014
2015
2016
Civil
i
an car
ownership
3.855
2
4.5356
5
5.3361
6.2779
Y
e
a
r
2017
2018
2019
2020
Civil
i
an car
ownership
7.385
9
8.6895
10.223
1
12.027
5
Then th
e
civilian
car o
w
ne
rship
in
Chin
a
and
in
Cho
n
gqing
from t
h
e year 20
13 t
o
20
20
is pre
d
icte
d, the re
sults a
r
e
sho
w
n in Ta
ble 2 and Ta
b
l
e 3.
4.2. Forecas
t of the EV O
w
n
e
rs
hip
Acco
rdi
ng to
the process t
o
predi
ct the
EV owne
rshi
p in a
city, the EV ownership in th
e
cou
n
try sh
ou
ld be
cal
c
ul
ated a
c
cordi
ng to the pl
annin
g
EV own
e
rship
b
y
cubi
c Hermite
interpol
ation.
Tabl
e IV
sh
ows the
EV
own
e
rship
in
Chin
a from
2
013 to
2
020
according
to
the
planni
ng EV own
e
rship 2.
66 million a
n
d
16.98 millio
n for the year 2015 an
d 20
20 re
sp
ective
ly.
Then th
e EV
own
e
rship
in
Cho
ngqin
g
from the ye
ar
2013 to
20
20
is p
r
e
d
icted,
the results a
r
e
sho
w
n in Ta
b
l
e 5.
Table 5. O
w
n
e
rship of EV in Chin
a from
2013 to 20
20
(UNIT: MILLION)
Y
e
a
r
2013
2014
2015
2016
EV
ow
n
e
rship
1.697
2.151
2.66
3.921
Y
e
a
r
2017
2018
2019
2020
EV
ow
n
e
rship
6.373
9.626
13.292
16.98
Table 6. O
w
n
e
rship of EV in Cho
ngqin
g
from 201
3 to 2020
Y
e
a
r
2013
2014
2015
2016
EV
ow
n
e
rship
10200
19700
31100
55700
Y
e
a
r
2017
2018
2019
2020
EV
ow
n
e
rship
83600
121500
154400
187400
5. Conclusio
n
A method
b
a
se
d on
BP neu
ral n
e
twork is pr
ese
n
ted in th
e
pape
r to fo
reca
st the
own
e
rship of
electri
c
vehi
cle in a
city.
The input la
yer ha
s seve
n neu
ron
s
, which
rep
r
e
s
e
n
ts
different econ
omy factor in
fluenci
ng the own
e
rship
of
electri
c
vehi
cl
e. One hidd
en is de
sign
ed
in the mo
del,
whi
c
h
contai
ns 3
neu
ro
ns. Histo
r
ical
a
nd stati
s
tical
data of fact
ors influe
nci
ng
the
own
e
rship
of
ca
rs a
r
e
coll
ected
in th
e
past
20 ye
ars, which i
s
th
e trai
ning
sa
mple fo
r the
BP
neural network.
The forecast
of the EV ownership
of a
city is
ba
se
d
on that of a
cou
n
try at the initial
stage fo
r ele
c
tric vehi
cle
s
. The EV own
e
r
shi
p
in
the ci
ty of Chongqi
ng is p
r
edi
cte
d
acco
rdin
g the
histori
c
al
an
d
statisti
cal
da
ta from
19
91
– 20
10
and
p
l
annin
g
EV o
w
ne
rship i
n
Chin
a, a
c
curacy
of the method
is verified in this in
stan
ce.
Referen
ces
[1]
M Etezad
i-Am
oli, K
Ch
oma, J
Stefani
.
Ra
pid
-
Charg
e
E
l
ectri
c
-Vehicl
e Stati
ons.
IEEE Trans. on P
o
wer
Deliv
ery
. 201
0; 25: 188
3-18
87
.
[2]
A Ashtari, E Bibea
u, S Shahi
din
e
ja
d, T Molinski
. PEV Char
gin
g
Profile Pr
edicti
on an
d Anal
y
s
is Bas
e
d
on Veh
i
cle Us
a
ge Data.
IEEE Trans. on Smart Grid.
2012; 3: 341-35
0.
[3]
Z Amjadi, S
S
Willi
amson.
Protot
ype
D
e
sig
n
a
nd
C
ontrol
l
er Impl
ementati
on fo
r a Batter
y
-
Ultraca
pacitor H
y
brid
E
l
ectric
Vehicl
e En
erg
y
Stora
ge S
y
st
em.
IEEE Trans. on Smart Grid
. 20
12; 3
:
332-
340.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
2302-4
046
A Method for
Electri
c
Vehicle Ownership
Fore
cast Considering
Diffe
rent... (Hanwu Luo)
2245
[4]
L F
angc
he
ng,
L Jinj
un, Z
.
Bi
n, Z
.
Haodo
ng,
F
a
st
chargi
ng
s
y
stem of e
l
e
c
tric vehicl
e (E
V) base
d
on
h
y
bri
d
en
erg
y
storage s
y
st
em,
IEEE Conferenc
e and
Ex
position in
Applied
P
o
wer
Electronics
(APEC)
, 2012,
211
5-21
20.
[5]
KK Lilj
a, K La
akso, J Pal
o
mki. Using th
e
Delp
hi met
hod
.
Internation
a
l Confer
ence in
T
e
chno
lo
g
y
Mana
ge
me
nt in the Ener
gy Smart W
o
rld (PI
C
MET
)
. 2011; 1-10.
[6]
JJ Jamian, H
Musa, MW Mustafa, H Mok
h
lis,
SS Ad
am
u. Combi
n
e
d
Voltag
e Stab
ili
t
y
Ind
e
x
for
Char
gin
g
Stati
on Effect o
n
D
i
stributi
on
Net
w
o
r
k.
Intern
ati
ona
l R
e
view
of
Electrica
l
En
gi
neer
ing-Ire
e
.
201
1; 6: 3175-
318
4.
[7]
DD Ras
o
lom
a
mpio
non
a, F
Maeg
ht, PY Cresson, P F
a
vier. Exp
e
rim
ental s
o
lar-b
ase
d
cha
r
gin
g
stati
o
n
for electric vehicles.
Pr
z
e
gl
ad Elektrotech
n
ic
zny.
87; 58-6
2
.
[8]
P Moha
nt
y
,
N
Dasg
upta, A S
harma. Ce
ntral
i
zed so
la
r l
ante
r
n charg
i
n
g
sta
t
ion u
nder '
l
i
ght
ing a
bil
lio
n
lives'
camp
aig
n
:
a technol
ogic
a
l evo
l
utio
n.
Progress i
n
Phot
ovolta
ics.
201
1
;
18: 516-5
34.
[9]
A Chaur
e
y
, T
C
Kand
pal. So
lar lant
erns for domesti
c lig
hting in In
dia: Vi
abil
i
t
y
of ce
ntral char
gi
n
g
station mod
e
l.
Energy Po
licy.
200
9; 37
：
4
910
-4
91
8
.
[10]
Y Ota, H
T
aniguch
i
, T
Nakaji
ma, KM Li
yana
ge,
J Baba, A Yoko
yam
a
. Au
tonomo
u
s Distr
ibute
d
V2G
(Vehicl
e-to-Gri
d) Satisf
y
i
ng S
c
hed
ule
d
Ch
ar
gin
g
.
IEEE Tra
n
s. on Smart Grid
. 201
2; 3: 55
9-56
4.
[11]
T
Sousa, H Morais, Z Vale, P
Faria, J Soar
e
s
. Intellig
ent E
nerg
y
R
e
so
urc
e
Man
agem
ent
Consi
der
ing
Vehic
l
e-to-Grid
:
A Simulated
Anne
ali
ng Ap
p
r
oach.
IEEE Trans. on S
m
art Grid
. 2012; 3: 535-
542.
[12]
E Sortomme, MA El-Sharka
w
i
. Optim
a
l Sc
hed
uli
ng
of Ve
hicle-t
o
-Grid E
nerg
y
and A
n
ci
llar
y
Serv
ices
,
IEEE Trans. on Sm
art Grid
. 20
12; 3: 351-
359.
[13]
W
Chen
ye,
H
Mohse
n
ia
n-Ra
d, H Ji
an
w
e
i.
Vehic
l
e-to-Ag
g
r
egator I
n
terac
t
ion Gam
e
.
IEEE Trans.
on
Sm
art Grid
. 20
12; 3: 434-
442.
[14]
L Panc
hi, L S
h
i
y
o
ng. Le
ar
n
i
ng a
l
gor
ithm a
nd a
ppl
icati
on
of
qua
ntum B
P
neur
al n
e
t
w
orks bas
ed o
n
univ
e
rsal q
u
a
n
tum gates.
Jour
nal of Syste
m
s
Engin
eeri
ng a
nd Electro
n
ics
.
2008; 1
9
:167-
174.
[15]
M Hu, H W
a
n
g
, G Hu, S Ya
ng. Soft fau
l
t d
i
ag
nosis f
o
r a
n
a
lo
g circu
i
ts b
a
sed
on s
l
o
pe
fault featur
e
and BP n
eura
l
net
w
o
rks.
T
s
in
ghu
a Scie
nce
and T
e
ch
no
log
y
. 2007; 12:2
6
-
31.
[16]
W
W
e
i,
F
Guorui, L Z
heng
xue
,
X Yueshe
ng.
Determin
i
stic conver
genc
e of an on
lin
e grad
i
ent metho
d
for BP neural
n
e
t
w
o
r
ks,
IEEE Trans. on Neur
al Networks
. 2005. 16: 5
33-5
40.
[17]
L Z
ilon
g
, L Gu
ozho
ng, L Ji
e. Adaptiv
e tracking
co
ntrol
l
er
usin
g BP ne
ur
al net
w
o
rks for
a class of
non
lin
ear s
y
ste
m
s.
Journal of
Systems En
gin
eeri
ng an
d Ele
c
tronics
. 200
4; 15: 598-
60
4.
[18]
http://
w
w
w
.
stat
s.gov.cn/tjsj/ndsj/
[19]
Yunp
u So
ng.
Stud
y o
n
the
Energ
y
-
R
eg
en
eratio
n-bas
ed Veloc
i
t
y
Co
ntrol of th
e H
y
d
r
aulic-
H
ybri
d
Vehic
l
e.
Te
l
k
om
ni
ka
. 20
12; 1
0
(7): 170
0-1
7
0
7
.
[20]
Xu
ni
an L
a
i, Lin
g
Li, W
e
imin
g Lia
ng, Bin
g
lo
n
g
W
ang, F
en Li
u.
T
e
lkomnik
a
.
201
2; 10(8): 21
00-2
108.
Appe
ndix- S
t
atis
tical Da
ta
The
statisti
cal
data
is from
t
he
web
s
ite
of
Natio
nal
B
u
reau
of Statics of
Chin
a [12]
, the d
a
ta from
the year 199
1
to 2010 wa
s
use
d
, here
we just list the data from the
year 2001 to
2010.
Table A-1. G
D
P of China
(unit: billion Yuan)
Y
e
a
r
2001
2002
2003
2004
2005
GDP
10806.82
11909.57
13517.40
15958.68
18361.85
Y
e
a
r
2006
2007
2008
2009
2010
GDP
21588.39
26641.10
31527.47
34140.15
40326.00
Table A-2. GDP of Chongqing (unit: billion Yuan)
Y
e
a
r
2001
2002
2003
2004
2005
GDP
197.686
223.286
255.572
303.458
346.772
Y
e
a
r
2006
2007
2008
2009
2010
GDP
390.723
467.613
579.366
653.001
792.558
Table A-3. Ve
hicle Produ
cti
on in Chin
a
Y
e
a
r
2001
2002
2003
2004
2005
Vehicle
production 2341700
3251000
4443900
5074100
5704900
Y
e
a
r
2006
2007
2008
2009
2010
Vehicle
production 7278900
8888900
9345500
13795300
18265300
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 23
02-4
046
TELKOM
NIKA
Vol. 11, No. 4, April 2013 : 2239 – 2
246
2246
Table A-4. Ve
hicle Produ
cti
on in Cho
n
g
q
ing
Y
e
a
r
2001
2002
2003
2004
2005
Vehicle production
243800
331300
404500
428900
421500
Y
e
a
r
2006
2007
2008
2009
2010
Vehicle production
519900
708000
76.6400
1186500
1615800
Table A-5. Statistical
Road passe
nger in China (unit: million)
Y
e
a
r
2001
2002
2003
2004
2005
Road
passenger
14027.98
14752.57
14643.35
16245.26
16973.81
Y
e
a
r
2006
2007
2008
2009
2010
Road
passenger
18604.87
20506.80
26821.14
27790.81
30527.38
Table A-6. Statistical
Road passe
nger in Chongqing
(unit: million)
Y
e
a
r
2001
2002
2003
2004
2005
Road
passenger
592.44
619.18
582.90
634.95
604.36
Y
e
a
r
2006
2007
2008
2009
2010
Road
passenger
612.28
771.87
1071.91
1145.98
1268.04
Table A-7. Pe
r capita
Crud
e Steel in Chi
na (unit: kg
)
Y
e
a
r
2001
2002
2003
2004
2005
per capita crude
steel production
119.22
142.43
172.57
218.28
270.95
Y
e
a
r
2006
2007
2008
2009
2010
per capita crude
steel production
319.71
371.27
379.76
429.77
476.32
Table A-8. Pe
r capita
crude
steel of Cho
ngqin
g
(unit: kg )
Y
e
a
r
2001
2002
2003
2004
2005
per capita crude
steel production
59.8
63.9
75
92
93
Y
e
a
r
2006
2007
2008
2009
2010
per capita crude
steel production
120
136
138
146
212
Table A-9.Pe
r capita
Gene
ration Capa
ci
ty in China (u
nit: kWh
)
Y
e
a
r
2001
2002
2003
2004
2005
Per
capita
gener
ation
capacity 1164.29
1291.78
1482.91
1699.98
1917.79
Y
e
a
r
2006
2007
2008
2009
2010
Per
capita
gener
ation
capacity 2185.88
2490.01
2639.00
2790.08
3144.78
Table A-1
0
. Per ca
pita Gen
e
ration
Cap
a
c
ity in Chon
g
q
ing (u
nit: kWh)
Y
e
a
r
2001
2002
2003
2004
2005
Per capita gener
ation capacity
550.7
594.8
604
742
741
Y
e
a
r
2006
2007
2008
2009
2010
Per capita gener
ation capacity
865
1095
1194
1307
1383
Evaluation Warning : The document was created with Spire.PDF for Python.