Optimization Decision of the Location of Shared Car Par
k
A Study Base on Chongqing
Xiaofeng Yan and Jie Zhao
Foreign trade college, Chongqing Normal University,The No. 9 Xuefu Road of Hechuan Distict, Chongqing, China.
Keywords: Chongqing, Shared vehicle, Multi-objective function, Neural Network prediction.
Abstract: With the development of the shared economy, shared cars have sprung up like bamboo shoots after the rain,
providing great convenience for people to travel. How to choose a car park to optimize the use of resources
is one of the problems that the managers of all companies pay attention to. This paper through the rolling
forecast way to predict the future three years of Chongqing city share total car demand, and then estimate
the Chongqing city parking lot demand; then, from the point of view of considering the location factors, to
users, operators and government satisfaction as the objective function, the establishment of shop location
model; finally, using genetic algorithm to solve the model and in a certain the weight, select the optimal
solution. The research results show that the model can provide a more scientific decision-making method
for the location of car park in Chongqing.
1 INTRODUCTION
Although there are plenty of problems in the
development of shared cars, the market is getting
bigger and bigger. We should pay attention to the
market and focus on the choice of the way the local
people travel. In early studies, most scholars tend to
consider the social economic attribute '(age, gender
and income) and travel attributes (travel time and
cost), with the further study, they found that the
traveler's own values and way of life in the inner
factors also affect the choice of travel mode.
In addition, there are many research results based
on site location. Ben-Akiva (2002) by introducing
latent variables, discrete choice model, the
subjective psychological factors into the model
selection activities; Fatemeh and so on, put forward
the factory location model based on fuzzy number to
determine the evaluation index weight, and use the
simulated annealing algorithm to solve the weight.
Xia Jinghong (2005) uses the 0-1 mixed integer
programming method to establish the location
model. In the analytic hierarchy process, the grey
relational analysis is used, and the genetic algorithm
is used to solve the model. Gao Taiguang and Chen
Peiyou (2011) proposed rough set theory based on
Z.Pawlak. Qualitative and quantitative evaluation of
location problem was made by decision making
feature and method, and the location model that
could get the best location plan was established.
The development of shared cars in Chongqing
can not only check the flood of black cars, but also
create a diversified transportation system for
Chongqing citizens, and make Chongqing a
resource-saving and environmental protection city.
At the same time, it can also promote the
transformation and upgrading of the automobile
industry in Chongqing. Based on the above research,
this paper will use the neural network (BP) rolling
prediction in the future three years in Chongqing
city car sharing scale, consider three factors of
influence, sharing the car parking lot location model
of multiple objectives and planning, which is solved
by using genetic algorithm, to provide scientific
support for the Chongqing City car sharing the
popularity of.
Yan, X. and Zhao, J.
Optimization Decision of the Location of Shared Car Park - A Study Base on Chongqing.
In 3rd International Conference on Electromechanical Control Technology and Transportation (ICECTT 2018), pages 323-328
ISBN: 978-989-758-312-4
Copyright © 2018 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
323

2 THE CURRENT SITUATION OF
THE SHARED CAR MARKET
IN CHONGQING
2.1 The current situation of the shared
car market in Chongqing
At present, Chongqing has two major shared car
brand, one is car2go (car2go Dalem subsidiary), the
cars with blue and white smart sedan, without a
deposit, but to pay 99 ¥ to be registered, the
registration fee will not return, take lifelong
membership, 1.8 ¥ / km (+0.3 ¥ / minute flameout is
0.1 ¥ / minute), long fee cap price: 128 ¥ / day
(every 24 hour journey, the cost of another), if the
car is damaged or lost items, will have a special fee.
Because it is not electric car, so the endurance is
very strong, it is very convenient to use. The other is
the PandAuto, the 620EV and 330EV of the Li Fan.
Compared with the general car rental, we hope to
use the pure electric new energy car, which has no
exhaust, low noise, safety and environmental
protection. The body has a symbolic panda head.
The user needs to pay 1000 ¥ deposit before using
the car. Because there are two kinds of cars, so the
pricing way is also different: the 620EV car is 26 ¥ /
hour, the top price is 179 ¥ per day, and the 330EV2
is 19 ¥ / hour, the top price is ¥ per day, the price is
very low in the car rental industry. Lifan in
cooperation with Alipay, launched a user-friendly
Sesame Credit Car, as long as “sesame” in more
than 650, it can be removed 1000 ¥ deposit, if there
is illegal activities in the process of using, to deal
with illegal activities before deposit refund.
Although the cost is lower than impromptu, but
because it is an electric vehicle, so the continued
route is far less car2go.
Figure 1: PandAuto parking spots in Chongqing (about
400).
In addition to sharing brands above, There are
EVPOP, Chang An, EVCARD line trip , Chi road
travel and so on several operating platforms. In
addition to the free flow of car2go car rental model,
the rest of the platform take a fixed point to take
back, return to the parking lot themselves.
The Car sharing strong competitors is the local
rental car industry in Chongqing City, the car rental
industry entrenched for many years, the biggest
advantage is that there are rich models, automatic
and manual, and various brands, different cars are
more able to meet the needs of car rental. The
deposit of car rental generally fluctuates around
2000 ¥, and the charging mode of different vehicles
is different. The basic vehicle type is about 100 ¥,
and the entrust cost is at least 50 ¥. There is no need
to compensate vehicle damage after the accident.
Although Chongqing is a good car rental business, it
needs to sign a contract on the face, and the
formalities and materials are more. It's a lot of
trouble compared to the shared car. The deposit of
car rental generally fluctuates around 2000 ¥, and
the charging mode of different vehicles is different.
The basic vehicle type is about 100 ¥, and the entrust
cost is at least 50 ¥. There is no need to compensate
vehicle damage after the accident. Although
Chongqing is a good car rental business, it needs to
sign a contract on the face, and there are formalities
and materials. It's a lot of trouble compared to the
shared car.
2.2 The forecast of the development
scale of the shared automobile in
Chongqing
The development of most industries will experience
germination, growth and maturity. Sharing
automobile industry is no exception. Its development
is influenced by urban population and government
policies. Up to now, about 8000 vehicles have been
launched in Chongqing, covering most of the main
urban areas and a small number of remote counties.
According to the data of China's sixth (2010)
national population census, a neural network model
is used to predict the total population of Chongqing
in the next three years.
ICECTT 2018 - 3rd International Conference on Electromechanical Control Technology and Transportation
324

Table 1: The total population of the city of Chongqing for
2010-2016 years
year 2010 2011 2012 2013 2014 2015 2016
TPTPC 2885 2919 2945 2970 2991 3017 3048
TPC: Total population of Chongqing (10000)
This paper will use the prediction neural network
(BP) forecast model: a way to forecast the number of
population of fourth years in the first three years of
the total population, the number of the total
population of Chongqing as 2010, 2011, 2012 in
Chongqing City, the total population as input to
predict in 2013, so repeated until it meets the
accuracy requirement so far. First, the data in the
above table is normalized, that is, to keep all values
between (0,1) (using MatlabR2013a).
(/)
Z
RjLRjLR ωω =+ =
Combined with Chongqing
0 1389 0 1405
0 1430
0 1418 0 1430 0 1440
0 1430 0 1440 0 1452
0 144
0.
00
1418
0.1405 0.1
1452 0
48
14
1
67
..
.
=
...
...
...
P
⎡⎤
⎢⎥
⎢⎥
⎢⎥
⎢⎥
⎢⎥
⎢⎥
⎣⎦
[
]
0 1430 0 144 0 1452 0 1467 =. . . . T
The total number of the population every three
years after normalization as input, with a total
population of fourth years the number of normalized
as the target vector, create a BP neural network, each
input vector range, the hidden layer activation
function was Tansig, activation function of the
output layer is logsig, the training function as the
gradient descent function, namely the standard
learning algorithm, setting learning rate is 0.1, the
simulation and comparison results.
According to the population travel and the choice
of travel mode in Chongqing, the total volume of
automobile in 2020 is calculated in Chongqing. The
following formula is as follows:
M*C*
4
Q=10 *
**
N
p
tv
δ
（）*D
Table 2: The meaning of each symbol in the upper formula
Symbol Meaning
Q Estimated number of shared cars
M
Total population of urban residents (10000
people)
C
Daily average travel times per person (day /
time / person)
N
The ratio of citizens to choose to share car
trips
D
Average driving distance (km / times) of one
trip for a shared ca
r
δ
Percentage of the normal operation of a
shared ca
r
P
Passenger volume of a shared car trip (person
/ time)
t
Travel time (hours) of one trip for a shared ca
r
V
The speed of one trip for a shared car (km /
hour)
Through the above population forecast,
combined with the statistical yearbook of China and
the comprehensive transportation planning data of
Chongqing, we can get the results.
3200 2.65, 2% 11,MCND====，，
3.25, 4, 24.8, 90% δ PtV=== =
.
By bringing these data into the top, the total
number of shared cars in Chongqing in 2020 was
64296.
3 MODEL CONSTRUCTION
3.1 The basic principles and influencing
factors of the site selection
(1) Practicality: in accordance with the
characteristics of the population distribution in
Chongqing, it can provide convenient travel for the
citizens.
(2) Rationality: it can not only meet the needs of
urban residents but also guarantee the agreement
with the urban planning management of Chongqing.
(3) Sustainability: the principle is mainly for the
sharing of vehicle operation platform, to ensure that
the platform has certain benefits, and to support the
healthy operation of shared cars.
The Car sharing development mainly by users,
operators, government, three party effect, from the
user point of view, they want to take the car also
more locations and reasonable cost, so the car
sharing outlets generally focus on the larger flow of
people in commercial areas, residential areas, near
the school; while operators want to share vehicle
operation and management to maximize the
minimum cost and benefits, so they tend to be more
Optimization Decision of the Location of Shared Car Park - A Study Base on Chongqing
325

users and outlets located in places built on cheap
prices; compared to the two, the government pay
more attention to the city environment and traffic
management, sharing the selected automobile outlets
must not impede the operation of public transport,
and will not cause the public the complaint is.
3.2 Model hypothesis
Since the problem of shared vehicle location
involves many uncertain and unquantifiable
elements, the following assumptions are put forward
for the location model.
Hypothesis one: the service range of each shared
car node is fixed to 2 km, and the total demand for
shared vehicles within two thousand meters is
divided by 100, that is, network coverage.
Hypothesis two: Based on population factors, the
candidates for sharing cars are restricted only in
schools, commercial areas, residential areas, tourist
attractions, and public transport stations.
Hypothesis three: the distance between each
node does not consider various terrain and is
represented by only two points.
Hypothesis four: the area of all shared cars is 100
square meters, and the population density and the
land unit price of the net shop are positively related
to the pollution index of the land.
3.3 The establishment of a
mathematical model
Based on the above principles and assumptions,
under certain constraints, we select
m
preferred
demand nodes (
nm>
) from the
n
shared vehicle
candidate nodes, aiming at the largest user demand,
the lowest cost of operators and the highest level of
government satisfaction, and set up the following
objective functions:
The shortest distance between the supply and
demand of the shared vehicle is the maximum user
demand
max 1 nnq n
nNqQ
f
yx p•
∈∈
=•
∑∑
min 2 nnq
nNqQ
f
yd
∈∈
=•
∑∑
max 3 0.01 qnq
nNqQ
f
yx
∈∈
=•
∑∑
Shared car parked point is the least expensive
min 4 nn
nN
f
yh
∈
=•
∑
The greatest degree of government satisfaction
12 3
,
ma x 5 ( 0 . 0 1 )nn nqn
nN nNqQ
f
yp xc αα α
∈∈∈
=••+• +•
∑
∑
Constraints:
The total cost of the scheme should not be higher
than Z (million ¥)
nn
nN
y
hZ
⊂
•≤
∑
The scope of the total number of the car parking
points shared
max maxmin n
nN
MyM
⊂
≤≤
∑
No less than a certain distance between two
shared cars
min maxmnyyd d•• ≤
1When n is a parking spot
0When n isn't a parking spot
n
y
nN
⎧
=∀∈
⎨
⎩
0
0
1When
0When
nq
nq
nq
dd
x
dd
≤
⎧
=
⎨
>
⎩
Table 3: The meanings of symbolic
Symbol Meaning
N
Candidate set
Q
Demand point set
mind
The distance between any two nodes
np
Population density between demand
points
α
Factors that account for the weight
coefficient of government satisfaction
nc
Urban environment
nh
Share car park construction capital
ICECTT 2018 - 3rd International Conference on Electromechanical Control Technology and Transportation
326

4 MODEL SOLUTION
4.1 Genetic algorithm solving model
The model of this paper is a complex problem of
multivariate programming, which is usually solved
by particle swarm optimization, firefly algorithm,
genetic algorithm and simulated annealing
algorithm. Due to the large number of candidate
nodes and complex objective functions, In order to
simplify the solution, a genetic algorithm with the
non-dominated sorting is using to solve the location
model. At present, the algorithm is mostly used in
the actual optimization and scheduling engineering,
and is rarely applied to the location problem. At
present, the algorithm is mostly used in the actual
optimization and scheduling engineering, and is
rarely applied to the location problem.
According to the average resident population of
38 districts and counties in Chongqing, the average
number of 781 thousand and 900 people, In
combination with the concentration of tourist
attractions and the large population flow, the area of
15 people with concentrated population is selected.
Table 4: Partial candidate attribute.
Chongqing City
Permanent population
(10000)
Requirement
Population density
(man
/
p
er km)
Land price
2
10000 / m
Government satisfaction
Jian
g
Bei 84.98 1126.45 3845 1.89 5.44
Jiu Long Po 118.69 1530.43 2754 1.41 5.01
Yu Bei 155.09 1248.19 1064 1.32 5.07
Yu Zhon
g
64.95 893.57 28239 0.89 5.73
Ba Nan 100.58 792.70 552 0.75 5.08
Sha Pin
g
Ba 112.83 1462.29 2849 1.13 5.18
Nan An 85.81 841.50 3275 1.09 5.21
Fu Ling 114.08 927.01 388 0.80 5.24
Zuan Jian
g
107.84 889.65 393 0.61 3.46
Jiang Jin 133.19 1064.18 414 0.76 4.57
He Chuan 136.06 914.87 581 0.72 3.78
Yon
g
Chuan 109.61 873.04 694 0.64 4.61
Wan Zhou 160.74 1130.21 466 0.88 4.98
Yun Yan
g
89.66 438.46 247 0.69 3.02
Kai Xian 117.07 601.37 295 0.64 3.53
To realize the genetic algorithm of the non -
dominated sorting by MatlabR2013a programming.
The algorithm sets the population size
200N =
,
the genetic algebra
300Z =
, the cross
probability
1
0.8P =
, and the compilation
probability
2 0.1p =
. The 7 sets of solutions are
calculated through the software code, and the node
number of each group of optimal solutions is derived.
Table 5: The optimal solution of each group calculated by
Matlab
Scheme
number
Dot
number
Total
cost
(million
¥)
Total
distance
(km)
Government
satisfaction
4 88 121 66239 91
53 76 119 70105 85
9 60 110 71229 74
78 64 105 71062 79
94 83 108 64148 82
According to the above data, the comprehensive
score of Plan 9 is 85.7, and the highest score is in all
the programs. Therefore, the final location plan of
the shared car in Chongqing is 9, and the number of
outlets is 60.
4.2 Analysis of site selection of parking point
Please For the above results, we need to select the
final location plan from the above. The selection-
criteria is the largest user demand, the lowest
operator cost and the highest government
satisfaction.
Table 6: Comprehensive score of the scheme.
Scheme
number
Indicator1
Score
Indicator2
Score
Indicator3
Score
Comprehensive
score
4 89 72 82.5 81.2
53 83 76 80.9 80
9 86 80 81.1 85.7
78 82 81 79.2 80.7
94 80.9 79 74.9 78.3
32 81 84 82.3 82.4
66 70 99 86 85
Optimization Decision of the Location of Shared Car Park - A Study Base on Chongqing
327

5 RESEARCH CONCLUSIONS
Based on the actual survey data of Chongqing, this
paper analyse the problem of the location of shared
vehicles in Chongqing from three angles of users,
operators and government. According to the
topography and demographic characteristics of
Chongqing, the development prospect of shared cars
is more promising than the popularization of shared
bicycles. Moreover, people will travel more or more
depending on the long distance and convenient way
to travel. The research done in this article is for the
development of the shared car in Chongqing to be
more smoothly, and to contribute to the construction
of green traffic and civilized cities in Chongqing.
Moreover, a scientific and reasonable location
plan can not only reduce operators' management
costs, but also improve user satisfaction, which is
extremely important for the long-term development
of new shared automotive industry.
REFERENCES
Ju Peng, Zhou Jing, Xu Hongli, Zhang Junting, Research
on the behavior of car sharing selection based on
mixed selection model. Transportation system
engineering and information [J].2017 (04).
Cheng Yuan, A game study on the mode of urban traffic
travel under car sharing. Harbin Institute of
Technology (06) (06).
Lu Ting, Huang Jing, Jiang Wen, Construction of the
location model of shared car network shop -- Taking
Hongshan District of Wuhan as an example, [J].
"Transportation Research" 2017 (10)
Xia Jinghong, Logistics center and its location model
study [J]. Nanjing University of Information Science
and Technology.2005 (05).
Yearbook of the China Statistics Bureau -
Jiang Bujian, Research on the matching of automobile
supply and demand under the shared lease model [D].
HeFei University of Technology.2017.
ICECTT 2018 - 3rd International Conference on Electromechanical Control Technology and Transportation
328