D

Journal of Energy and Power Engineering 9 (2015) 731-740

doi: 10.17265/1934-8975/2015.08.007

DAVID

PUBLISHING

Impact Analysis of EV Charging with Mixed Control

Strategy
Di Wu1, Haibo Zeng2 and Benoit Boulet1
1. Department of Electric and Computer Engineering, McGill University, Montreal, QC H3A 0G4, Canada
2. Department of Electric and Computer Engineering, Virginia Tech, Blacksburg, VA 24061, USA
Received: May 20, 2015 / Accepted: June 15, 2015 / Published: August 31, 2015.
Abstract: EVs (electric vehicles) have been widely accepted as a promising solution for reducing oil consumption, air pollution and
greenhouse gas emission. The number of EVs is growing very fast over the years. However, the high adoption of EVs will impose a
burden on the power system, especially for neighborhood level network. In this paper, we propose a mixed control framework for EV
charging scheduling to mitigate its impact on the power network. A metric for modeling customers satisfaction is also proposed to
compare the user satisfaction for different algorithms. The impacts of the proposed algorithms on EV charging cost, EV penetration
and peak power reduction are evaluated with real data for a neighborhood level network. The simulation results demonstrate the
effectiveness of the proposed algorithms.
Key words: Neighborhood level network, electric vehicle, penetration level, mixed control, charging management, user satisfaction.

1. Introduction
With the growing concerns on environmental
protection and energy security, more and more
countries and regions are promoting the usages of
electric vehicles. EV (electric vehicle) is globally
treated as a promising alternative to conventional
vehicles to reduce greenhouse gas emission, and is
expected to see a fast development in the near future [1].
This growing trend of transportation electrification has
been reported in a few works [2, 3]. For example, it is
estimated that, 23% of all the vehicles will have an
alternative powertrain in 2020 [3].
On the other hand, the high adoption of EVs will
also impose great challenges on the power system. EV
charging power consumption is estimated to be
comparable to typical base load for neighborhood
level customers [4]. If the EV charging is left
uncontrolled, the peak power consumption will be
greatly increased, which accelerates the grid ageing.
Corresponding author: Di Wu, Ph.D. student, research
fields: stochastic optimization and optimal scheduling. E-mail:
[email protected].

This issue has drawn attention in the research

community [5-8]. It will be even worse for
neighborhood level networks, where EV adopters
often charge at similar times imposing clustering load
to the network [9, 10].
EV charging can be treated as a kind of flexible
load consumption [11], as we only need to get the EVs
charged with enough energy before the departure time.
Several control strategies have been proposed for EV
charging scheduling. There are mainly two groups of
control
strategies:
centralized
control
and
decentralized control strategies [12]. For the
centralized control strategies, usually there is a central
controller which can access all information and control
all the charge decisions for all the EVs. In Ref. [8], the
authors propose a centralized EV charging control
method which can help minimize the total charging
cost while considering the negative impacts on the
grid network. In Ref. [13], charging fairness for
customers is discussed in a centralized control
framework.
Centralized control could make full use of the

732

Impact Analysis of EV Charging with Mixed Control Strategy

charging flexibilities of all the EVs to reduce the

charging cost and mitigate the adverse effects on
power system. However, there will be a heavy
communication burden for the centralized control
strategies. Besides, the owner of the EV may not
necessarily be willing to give up the control to the
central controller, partly due to privacy issues.

be studied with simulation using real data.

The organization of this paper is organized as
follows: In Section 2, the system models are described;
in Section 3, the proposed mixed control framework is
presented; in Section 4, the simulation results are
presented with real data for a neighborhood level
network; finally, the paper is concluded in Section 5.

Because of these concerns, not all the people adopt the

centralized control strategy [14]. To overcome these
shortcomings, Refs. [15-20] discuss the decentralized
control strategies for EV charging which only require
no or little information exchange between electric
vehicles and the system controller. In Ref. [15], the

2. System Models

charging price is used as a signal to realize the

decentralized control for EVs. In Ref. [20], the authors
proposed a decentralized control strategy with genetic
algorithm.
Centralized control would be easier to get the global
optimum as it can control the charging for all the EVs.
However, decentralized control could provide the
customers with more flexibility. For some specific
neighborhood level networks, it could be possible that,
only some of the customers are willing to accept the
centralized control. To address this problem, we
propose a mixed EV charging control framework
which can recognize the need of decentralized control
while maximizing the benefit of centralized control
for both the grid companies and the customers.
Within the proposed control framework, the system
operator is responsible for the charging control and
price signal design. We consider two types of
customers in the neighborhood level network. For the
customers who are willing to give out the charging
control, the charging for these EVs will be controlled
by system controller. For those customers who are not
willing to give out the charging control, the charging
for these EVs will be influenced by time-of-use price
signal. To describe the user satisfaction, a new user
satisfaction model is proposed. The impacts of the
proposed mixed control algorithm on the charging
cost, EV penetration, and peak power reduction will

In this Section, we describe the proposed charging

control framework, including the models for EV
customers, base load power consumption, electric
vehicle charging power consumption and user
satisfaction.
2.1 EV Customers and Base Load Power Consumption
Fig. 1 shows the prototype for neighborhood level
network. In the neighborhood level network, there
could be one or more EVs in every home. We
consider that, there are two groups of customers for a
given neighborhood level network: contractual and
non-contractual customers. The contractual customers
would like to sign the contract with the utility
companies and give out the charging control of their
EVs. The other group of customers would prefer to
having the charging freedom and are unwilling to give
out the charging control.
For the neighborhood level network integrated with
EVs, we classify the power consumptions into two
categories: base load power consumption and EV
charging power consumption. The base load power
consumption includes all the other power consumptions
except the EV charging power consumption, for

Fig. 1

Neighborhood network integrated with EVs.

Impact Analysis of EV Charging with Mixed Control Strategy

example, heating, cooling, lighting and other facility

power consumptions. The base load power
consumptions of a neighborhood level network with
seven homes for two representative days, one in
summer and the other in winter, are demonstrated in
Fig. 2.
In Fig. 2, it can be seen that, the peak hours are
usually around noon in summer (due to the need for
cooling) and the evening in winter (for heating). For
uncontrolled charging, people would start charging
their EVs when they arrive home in the evening,
which may overlap with the base load consumption
peak hours. This will further increase the peak power
consumption and accelerate the grid ageing.
Fig. 3 shows the total power for the neighborhood
level network integrated with four electric vehicles. In
Fig. 3, the line total load 1 represents the total
power consumption of the network integrated with
four Honda Fit EVs, and the line total load 2 gives
the total power consumption integrated with four
Tesla Model S EVs. These EVs are assumed to be
charged continuously with their rated charging power.
From Fig. 3, it is clear that, even for the network
integrated with four EVs (a penetration level of 4/7),
the total power consumption will be very close to the
power limit for the neighborhood level network.

733

Fig. 2 Base load consumption for a neighborhood network

with seven homes in two representative days.

Fig. 3
Table 1

Total power consumption for the network.

EV specification.

Model
Tesla Model S EV
Honda Fit EV

Battery
capacity (kWh)
60
20

Rated charging
power (kW)
10
6.6

2.2 Electric Vehicle Charging Load

In this paper, we consider two kinds of EVs: Honda
Fit EV and Tesla Model S EV. The specifications for
these EVs are listed in Table 1 [21, 22]. Honda Fit is
relatively economic and can get fully charged around
3 h with its maximum charging power. Tesla Model S
has a much larger battery pack, and can get fully
charged in about 6 h. We consider continuous
charging, which means that the EVs can be charged
with any power from zero to maximal rated charging
power.
We use the survey results conducted in City of
Ottawa [23] to study the customers driving behaviors.
The results are described in Fig. 4, Tables 2 and 3.

Fig. 4

Driving range distributions.

Fig. 4 demonstrates the customers driving range

distributions. We can see that most people will
commute within 50 km. There will be a higher
probability for longer range driving in the weekend.

Impact Analysis of EV Charging with Mixed Control Strategy

734

Table 2
Hours

Customers departure time distribution.

< 7 a.m.

Percentage
18
(%)
Table 3
Hours

7 a.m.9 a.m.

9 a.m.12 p.m.

> 12 p.m.

70

Customers arrival time distribution.

< 12 p.m.

Percentage
3
(%)

12 p.m.- 2 p.m.- 5 p.m.- 7 p.m.> 9 p.m.

2 p.m.
5 p.m. 7 p.m. 9 p.m.
2

31

53

This is probably for shopping and pleasant during the

weekend. The departure and arrival time for the EV
owners are shown in Tables 2 and 3, respectively.
Seventy percent of all the drivers will leave home
between 7 a.m. and 9 a.m. More than half of the
people will arrive between 5 p.m. and 7 p.m.
For charging pricing, some utility companies adopt
the TOU (time-of-use) electricity pricing structure to
encourage customers to shift their power consumption
to valley hours. In this pricing structure, the electricity
price will be high for the peak hours of power
consumption and low for valley hours. We also utilize
the TOU pricing structure in our control framework.
2.3 Customer Satisfaction Model
Most of previous papers discuss the EV charging
control strategies to minimize the aggregated charging
cost or mitigate the negative impacts on power system,
with little work discussing the user satisfactions. In
order to compare user satisfaction for different
charging control algorithms, we propose a new user
satisfaction model. We observe three determining
factors: the user controllability of the charging, the
normalized final BSOC (battery state of charge) after
the charging and the relative total charging cost.

S SI (BSOC final / BSOCtarget )(Cun / C real )

(1)

Eq. (1) shows our proposed model for the

calculation of user satisfaction S, which is the product
of the three factors; SI denotes the satisfaction index.
For those people who can control the EV charging, SI
is set as 1; otherwise it is a constant positive fractional
number; BSOC final denotes the final BSOC after

the charging; BSOCtarget is the required BSOC

predefined by the customer. The ratio of these two
gives the indication of the battery charging status.
Finally, Cun is the charging cost for EV charging
without any control methods; and Creal shows the
charging cost with certain EV charging control
algorithm.
The user satisfaction for a specific network is defined
in Eq. (2), which is the average among all the
customers. In the equation, n means the number of EV
customers in the neighborhood level network. Si
shows the customer satisfaction value for i-th customer.
n

S average Si / n

(2)

i 1

3. Methodology
As mentioned in Section 2, we consider that, there
are two groups of customers in a neighborhood level
network: contractual customers and non-contractual
customers. It is assumed that, the EV charging control
and communication devices have been installed for all
the houses in the given neighborhood level network
for contractual customers. The system operator can
access full information of the contractual EVs: the real
time BSOC, arrival time, predefined departure time
and final BSOC requirement for their EVs. The
charging behaviors of all the contractual EVs will be
controlled by the system operator. For those
non-contractual EVs, there is less information
exchange between the EVs and the system operator.
But the system operator can access the charging
histories of these EVs. The non-contractual customers
are assumed to be rational and would follow the
electricity price signal to arrange the EV charging.
3.1 Mixed Control Algorithm
Fig. 5 shows the overview of the proposed energy
management system. It is assumed that, there is a
system operator who is responsible for controlling the
charging for all the contractual EVs and deciding the
time-of-use price structure for the neighborhood level
network.

Impact Analysis of EV Charging with Mixed Control Strategy

Fig. 6

735

Identification of zones for pricing.

3.2 Objective Function

Fig. 5 Energy management system with mixed control

strategy.

There are mainly three steps for the proposed mixed

control algorithm. The first step is the load forecasting
for base load power consumption. Different from Ref. [24],
in this paper, we focus on the forecasting for base load
power consumption. Load forecasting has been
discussed in several previous Refs. [25-28]. In this
paper, the time slot is chosen as 1 h. Weather
information, weekday/weekend classification and
hourly power consumptions for the previous day are
used for load forecasting. Neural network is used for
load forecasting for every hour in the next 24 h.
The second step is to determine the time-of-use
price structure based on the forecasted load
consumption. The pricing model proposed in Ref. [19]
is adopted to determine the TOU structure. Fig. 6
shows an example the TOU price structure. We first
find the maximal and minimal load consumption
based on the forecasted result. The obtained timing
region will be partitioned into three zones: valley,
medium and peak hours. We assign 0.112$/kWh,
0.135$/kWh and 0.072$/kWh for the medium hours,
peak hours and valley hours. The TOU price signal
will be informed to all the non-contractual customers
in the third step. TOU price acts as an incentive to
realize the decentralized control for those
non-contractual customers.

The objective for our proposed optimization

framework is to minimize the total charging cost for
the contractual customers. The objective function F is
shown in Eq. (3):
m

F min ct Pi , t

(3)

i 1 t

In total, we assume there are n customers and m of

all the customers are willing to give out the EV
charging control. Without loss of generality, we
assume that each customer owns one electric vehicle.
Then there are m electric vehicles owned by
contractual customers in the neighborhood level
network, and n m EVs owned by non-contractual
customers. Pi , t means the charging power for the
i-th EV in time interval t and ct is the electricity
price for that time interval.
3.3 Constraints
For the stability of the power system, all the
following constraints should be satisfied in all time
slots.
Total power constraint is shown in Eq. (4). It means
that, for every time interval, the total charging power
including the base load power consumption and the
EV charging power consumption should be less than a
certain value predefined by the transformer
specification. As mentioned in Eq. (4), PBi, t means
the base load consumption in i-th home for t-th time
interval, PCi, t describes the charging power

Impact Analysis of EV Charging with Mixed Control Strategy

736

consumption, and PTmax means the upper limit for the

total power consumption.
n

0 ( PBi , t PCi , t ) PTmax

(4)

i 1

EV charging power consumption is shown in Eq. (5).

Continuous EV charging is adopted in this paper,
which means that, the EV can be charged with power
from zero to PCmax, the maximal rated EV charging
power.

0 PCi , t PCmax

(5)

BSOC constraints are shown in Eqs. (6) and (7). Eq. (6)
shows that in all time slots the EV BSOC should be
smaller than a predefined value (BSOCmax). Eq. (7)
means that after the charging period, the final EV
BSOC (BSOCfinal) should satisfy the user predefined
BSOC (BSOCpredefined) requirement.

0 BSOCi , t BSOCmax

(6)

BSOC predefined BSOC final BSOCmax

(7)

Eq. (8) shows the BSOC updating constraint. Cap

in this equation means the battery capacity. C is
determined by the length of time interval, if time
interval is 1 h, C will be 1.
efficiency for EV charging.

shows the conversion

BSOCt 1 BSOCt C ( PCi / Cap) 100

(8)

4. Simulation Results
4.1 Simulation Setup
The proposed EV charging control framework is
tested with real data for a neighborhood level network
from a local utility company. There are in total seven
homes in the network as shown in Fig. 1.
Uncontrolled charging and centralized control EV
charging are used as comparisons for the proposed
mixed EV charging control framework. For
uncontrolled EV charging, we assume that, the
customers will start charging their EVs once they

arrive home with rated charging power. For

centralized charging control, we assume that, the
system controller can control the EV charging directly
for all the EVs in the network. Simulation results for
charging cost, user satisfaction, maximal EV
penetration and peak power reduction are analyzed.
As shown in Table. 4, we consider three cases for
the mixed charging control strategy according to the
number of contractual customers. For example, in
Case 2, four homes out of all the homes are
contractual customers. This is also defined as the basic
case for mixed control strategy.
Two kinds of electric vehicles: Honda Fit and Tesla
model S are considered. The specifications for these
two EVs are shown in Table. 1. The arrival BSOC,
arrival time and departure time for all these EVs will
be set according to the probability distribution of the
survey mentioned in Section 2. For the total power
constraints, PTmax is set as 60, the BSOCmax is set as
95, BSOCpre is set as 80. The average initial BSOC for
all EVs is 45%. For non-contractual customers, the
user satisfaction index SI is set to be 0.8. The
conversion efficiency for EV charging is set as 0.95.
The optimization problem is solved with CPLEX
(simplex method as implemented in the C
Programming Language) 12.0.
4.2 Charging Cost and User Satisfaction Result
As described in Section 3, we assume that, the total
operation cost and user satisfaction are two major
factors for neighborhood level customers. The charging
costs and user satisfactions for three representative
charging control strategies are shown in Figs. 7 and 8.
The charging cost for three control strategies are
shown in Fig. 7. We can see that, the charging cost for
the mixed charging control strategy is significantly
less than the uncontrolled strategy and close to the
centralized control strategies. With mixed control
strategy the charging cost will be reduced 16.2% and
19.5% for Honda Fit and Tesla Model S compared to
uncontrolled EV charging.

Impact Analysis of EV Charging with Mixed Control Strategy

Table 4

Mixed control strategy cases.

Cases

Case 1

Case 2

Case 3

#Contractual customers

Fig. 7

Charging cost for different control strategies.

737

mentioned in Ref. [30], a higher peak demand will

require a larger generation capacity and cause the
power system to be unstable. Fig. 9 shows the peak
power consumptions for different control strategies.
As the centralized control strategy could control the
charging of all the EVs, this control algorithm
provides the lowest peak power consumption. It can
also be seen that, with the proposed mixed control
strategy, the peak power consumption could be
reduced significantly. Compared with uncontrolled
EV charging, the peak power consumptions are
reduced 37.6% for Honda Fit and 33.22% for Tesla
Model S.
4.3 EV Penetration Result

Fig. 8

User satisfaction for different control strategies.

User satisfaction for three representative charging

control strategies is illustrated in Fig. 8. It can be seen
that, for the mixed strategy, customers can have the
highest overall user satisfactions. This is because the
non-contractual customers can get full control of the
EV charging and the overall charging cost is reduced.
We can see that with the proposed criteria, user
satisfaction can be well analyzed for different control
algorithms.
The charging costs for the three cases for mixed
control strategy are listed in Table. 5. We can see that,
with the increasing of the number of contractual
customers, we will have a lower charging cost.

The adoption of the EVs will increase the burden on

the power system which may require related
infrastructure upgrade. It would be an interesting
question for the utility companies to know the
potential EV penetration for a given neighborhood
level network. The possible EV penetration for a
given network is discussed in Refs. [5, 31]. However,
most of them focused on pure centralized or
decentralized control strategy. This may not be ideal
or realistic as discussed in previous sections.
The proposed mixed control framework can also be
adopted to find a more realistic EV penetration for a
Table 5

Contractual Customer number.

# Contractual customers

Charging cost Honda ($)

5.33

5.02

4.9

Charging cost Tesla ($)

15.41

14.38

13.7

4.3 Peak Power Consumption

Peak power consumption is another important
factor for utility companies, as the overloaded power
consumption will accelerate the grid ageing [29]. As

Fig. 9 Peak power consumption.

738

Impact Analysis of EV Charging with Mixed Control Strategy

given network. To find the EV penetration, we can set

the objective function in Section 4 as the number of
EVs in every home. With the proposed framework, the
simulation result will be more practical.
The EV penetration for three representative
charging control strategies is analyzed on the two
specific days. Simulation results are shown in Fig. 10.
As the base load consumption is higher in summer
than in winter, the EV penetration in winter is a little
higher than that in summer. In all three strategies, we
can get the highest EV penetration with the centralized
control strategy.
For typical summer day, EV penetration can only
be 0.92 for Honda Fit and 0.61 for Tesla Model S if
the EV charging is left uncontrolled. With mixed
charging control strategy, the EV penetration could be
3.46 for Honda Fit EV and 1.83 for Tesla Model S.
We can also see that as the Tesla Model S require high
EV charging rate which limits its penetration
potential.
EV penetrations for mixed control strategy with
different contractual customer numbers are shown in
Fig. 11. We can see that, with an increase of
contractual customers, the EV penetration will
increase accordingly.

5. Conclusions
The fast development of EVs will induce negative
impacts for power system and accelerate grid ageing if
the EV charging is left uncontrolled. In this paper, the
charging control for the EVs has been analyzed. A
mixed control strategy and its optimization framework
are proposed to mitigate the EV charging problem
while considering the willingness of the customers.
Compared with previous charging control strategies,
in the proposed framework we consider that in the
actual neighborhood level network, not all the
customers would like to sign related contract with the
utility companies and give out the charging control for
their EVs.
A new user satisfaction model is also described in

Fig. 10

EV penetration for different control strategies.

Fig. 11

EV penetration for mixed control strategies.

this paper, which will help study the user satisfactions

for different control strategies. The EV penetration,
charging cost and user satisfactions for the proposed
control framework, as well as the centralized control
strategy and uncontrolled charging are studied.
Simulation results show the efficiency of the proposed
smart charging control framework. With the proposed
method, we can get a higher EV penetration and lower
charging cost. We can see that, the proposed strategy
is beneficial for both the utility companies and the
customers.

References
[1]

[2]

[3]

DOE (US Department of Energy). 2011. One Million

Electric Vehicles by 2015. Virginia: US Department of
Energy.
Shireen, M. K., and Das, D. 2008. Biohydrogen as a
Renewable Energy ResourceProspects and Potentials.
International Journal of Hydrogen Energy 33 (1):
258-63.
Green, A. R. C., Wang, II, L., and Alam, M. 2011. The

Impact Analysis of EV Charging with Mixed Control Strategy

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

Impact of Plug-in Hybrid Electric Vehicles on

Neighborhood Networks: A Review and Outlook.
Renew. Sustain. Energy Rev. 15 (1): 544-53.
Azad, A. P., Olivier, B., and Pfeiffer, L. 2013. An
Optimal Control Approach for EV Charging with
Distribution Grid Ageing. In Proceedings of the 2013
First International Black Sea Conference on
Communications and Networking (BlackSeaCom), IEEE,
206-10.
Peter, R., Damian, F., and Andrew, K. 2012. Optimal
Charging of Electric Vehicles in Low-Voltage
Distribution Systems. IEEE Transactions on Power
Systems 27 (1): 268-79.
Rotering, N., and Ilic, M. 2011. Optimal Charge Control
of Plug-in Hybrid Electric Vehicles in Deregulated
Electricity Markets. IEEE Trans. Power Systems 26 (3):
1021-9.
Ahn, C., Li, C. T., and Peng, H. 2011. Decentralized
Charging Algorithm for Electrified Vehicles Connected
to Smart Grid. In Proceedings of the ACC (American
Control Conference), 3924-9.
Li, N., Chen, L., and Low, S. H. 2011. Optimal Demand
Response Based on Utility Maximization in Power
Networks. In Proceedings of the IEEE Power and
Energy Society General Meeting, 1-8.
Kahn, M. E. 2008. Green Market Geography: The
Spatial Clustering of Hybrid Vehicle and LEED
Registered Buildings. Working Paper 2008-19, UCLA
(University of California Los Angeles) Ziman Center for
Real Estate.
Aultman-Hall, L., Sears, J., Dowds, J., and Hines, P.
2012. Travel Demand and Charging Capacity for
Electric Vehicles in Rural States: A Vermont Case
Study. Presented at the Transport Research Board 92nd
Annual Meeting, Washington, USA.
Kempton, W., and Letendre, S. E. 1997. Electric
Vehicles as a New Power Source for Electric Utilities.
Transportation Research Part D: Transport and
Environment 2 (3): 157-75.
Chagngsun, C. H. 2011. Dencentralized Charging
Alogorithm for Electricfied Vehicles Connected to Smart
Grid. In Proceedings of the ACC (American Control
Conference), 3924-9.
Komang A. I., and Kab, S. 2013. A Centralized Scheme
Based on User Satisfaction Fairness and Cost. In
Proceedings of the IEEE ISGT Asia (Innovative Smart
Grid TechnologiesAsia), 1-4.
Papadaskalopoulos, D., and Strbac, G. 2012.
Decentralized Participation of Electric Vehicles in
Network-Constrained Market Operation. In Proceedings
of the 3rd IEEE PES International Conference and
Exhibition on ISGT (Innovative Smart Grid Technologies)

739

Europe, 1-8.
[15] Xiaomin, X., and Sioshansi, R. 2014. Using Price-Based
Signals to Control Plug-in Electric Vehicle Fleet
Charging. IEEE Transaction on Smart Grid 5 (3):
1451-64.
[16] Hong, L., and Shaoyun, G. 2013. Optimization of TOU
Price of Electricity Based on Electric Vehicle Orderly
Charge. In Proceedings of the IEEE PES (Power and
Energy Society) General Meeting, 1-5.
[17] Vaya, M. G., and Anderson, G. 2012. Centralized and
Decentralized Approaches to Smart Charging of Plug-in
Vehicles. In Proceedings of the IEEE Power and Energy
Society General Meetings, 1-8.
[18] Ruoyang, L., Qiuwei, W., and Oren, S. S. 2014.
Distributional Locational Marginal Pricing for Optimal
Electric Vehicle Charging Management. IEEE Trans.
Power Systems 29 (1): 203-11.
[19] Yan, Q. 2014. A Multi-tied Real-Time Pricing
Algorithm for Electric Vehicle Charging Station. In
Proceedings of the ITEC (IEEE Transportation
Electrification Conference) and Expo, 1-6.
[20] Grunewald, A., Hardt, S., and Mielke, M. 2012. A
Decentralized Charge Management for Electric Vehicles
Using a Genetic Algorithm: Case Study and
Proof-of-Concept in Java and FPGA. In Proceedings of
the IEEE CEC (Congress on Evolutionary Computation),
1-7.
[21] TESLA. 2014. 85 kWh Battery. Tesla Motors.
Accessed
May
20,
2015.
http://www.teslamotors.com/en_CA/charging#/basics.
[22] HONDA. 2014. 2014 Accord Plug-in. American Honda
Motor Co., Inc. Accessed May 20, 2015.
http://automobiles.honda.com/accord-plug-in/charging-an
-electric-vehicle.aspx.
[23] Oliver, B., Probe, P., and Flores, M. 2013. EMAP:
Informing the Development of an EV Deployment
Strategy in the City of Ottawa. Presented at Evve
Conference and Trade Show, Gatineau, Canada.
Accessed May 20, 2015. http://emc-mec.ca/evve2013/
pdfs/22OCT/TS8/3_Oliver_Flores.pdf.
[24] Wu, D., Haibo, Z., and Benoit, B. 2014. Neighborhood
Level Network Aware Electric Vehicle Charging
Management with Mixed Control Strategy. In
Proceedings of the 2014 IEEE IEVC (International
Electric Vehicle Conference), 1-7.
[25] Dongxiao, N., Yongli, W., and Desheng, D. W. 2010.
Power Load Forecasting Using Support Vector Machine
and ant Colony Optimization. Expert Systems with
Applications 37 (3): 2531-9.
[26] Bakirtzis, A. G. 1996. A Neural Network Short Term
Load Forecasting Model for the Greek Power System.
IEEE Transactions on Power Systems 11 (2): 858-63.

740

Impact Analysis of EV Charging with Mixed Control Strategy

[27] Senjyu, T. 2002. One-Hour-Ahead Load Forecasting

Using Neural Network. IEEE Transactions on Power
Systems 17 (1): 113-8.
[28] Lu, C. N., Wu, H. T., and Vemuri, S. 1993. Neural
Network Based Short Term Load Forecasting. IEEE
Transactions on Power Systems 8 (1): 336-42.
[29] Rutherford, M. J., and Yousefzadeh, V. 2011. The
Impact of Electric Vehicle Battery Charging on
Distribution Transformers. In Proceedings of the 26th

IEEE APEC (Applied Power Electronics Conference) and

Exposition, 396-400.
[30] Shizhen, Z., Xiaojun, L., and Minghua, C. 2013.
Peak-Minimizing Online EV Charging. Allerton.
[31] Rocha, A. P. M., Peas, L. J. A., and Soares, F. J. 2011.
Electric Vehicles Participating in Frequency Control:
Operating Islanded Systems with Large Penetration of
Renewable Power Sources. In Proceedings of the IEEE
Power Tech., 1-6.