1688 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO.

2, MARCH/APRIL 2021

Charging Scheduling of Electric Vehicle

Incorporating Grid-to-Vehicle and Vehicle-to-Grid
Technology Considering in Smart Grid
Sourav Das , Student Member, IEEE, Parimal Acharjee, Senior Member, IEEE,
and Aniruddha Bhattacharya, Member, IEEE

Abstract—In recent days, the deployment of electric vehicles G2V Grid to vehicle.
(EVs) in automobile sector is increasing the load demand in the HBC Higher battery capacity.
distribution system. To deal with this load demand, the charging HGSO Henry gas solubility optimization.
management needs to be improved. Nevertheless, an EV needs
several hours to complete charge. Reducing the charging time, ICSA Intelligent charging scheduling
energy consumption is a huge contest to deal with the promotion algorithm.
of EV over conventional vehicles. The condition of the itineraries ILPP Integer linear programming problem.
may affect the energy consumption of the EV, which needs to be IRV Input random variable.
considered before fulfilling the energy demand. In this article, MCS Monte Carlo simulation.
these are considered assigning suitable charging stations (CS) to
individual EVs and their scheduling is taken as an optimization PEM Point estimation method.
problem. The first part deals with the proper assignment of CS, ORV Output random variable.
which is a linear optimization problem and the second deals with PEV Plug-in electric vehicles.
the charging scheduling problem. An “intelligent charging schedul- PHEV Plug-in hybrid electric vehicles.
ing algorithm (ICSA)” is proposed with the integration of Henry QT Quade test.
gas solubility optimization to minimize total daily price incurred
by the CS operator. Later, ICSA is clubbed with other standard RTT Real-time tariff.
optimization techniques considering practical constraints. A 2 m SOC State of charge.
point estimation method has been utilized to tackle the uncertainty V2G Vehicle to grid.
and its performance has been compared with the Monte-Carlo WSRT Wilcoxon signed ranked test.
simulation technique. The robustness of ICSA is evaluated and
confirmed using the Wilcoxon signed rank test and Quade test. Parameters
Index Terms—Charging stations (CS), distribution system, AER All electric range.
electric vehicle (EV), Monte Carlo simulation (MCS), chrate,PEV_no min Minimum charging rate of EVs.
optimizations, scheduling, 2 m point estimation method (2m-PEM). chrate,PEV_no max Maximum charging rate of EVs.
De Demand of vehicles for CS.
NOMENCLATURE DOD Battery’s depth of discharge.
ETL Ensuing trip length.
Abbreviations
Lbatt Batteries lifecycle at fixed DOD.
CS Charging stations.
PWrated Charger’s power ratings in the CS.
CSO Charging station operator.
S Availability of charging slots.
EV Electric vehicles.
SOCmin Minimum SOC of the battery.
EVSE Electric vehicle supply equipment.
SOCmax Maximum SOC of the battery.
FFPV Fossil fuel powered vehicles.
η ch Charging efficiency of the vehicle.
η dis Discharging efficiency of the vehicle.
Manuscript received April 9, 2020; revised June 27, 2020 and September
10, 2020; accepted November 1, 2020. Date of publication December 1, 2020;
date of current version March 17, 2021. Paper 2020-SECSC-0520.R2, presented Indices
at the 2020 IEEE International Conference on Power Electronics, Smart Grid e Number of vehicles needs charge.
and Renewable Energy, Cochin, India, Jan. 2–4, and approved for publication f Total number of CS.
in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the Renewable
and Sustainable Energy Conversion Systems Committee of the IEEE Industry t Time needed to reach the CS.
Applications Society. (Corresponding author: Sourav Das.) k Jam coefficient constant.
The authors are with the Department of Electrical Engineer- m Number of uncertain IRV.
ing, National Institute of Technology, Durgapur, Durgapur 713209,
India (e-mail: [email protected]; [email protected]; no EV’s index arriving at CS in 30 min time
[email protected]). interval.
Color versions of one or more of the figures in this article are available online N Number of observations.
at https://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIA.2020.3041808 pos Location for IRV zl.

0093-9994 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1689

PEV_no Index of EVs at 30 min interval arriving SOCreq Required SOC by each EV in the park-
to the CS. ing duration.
slot Time index at 30 min time interval. SOCini Initial SOC of individual EV before
tot.PEV Number of EVs coming to the CS in 24 reaching to the CS.
h. slot_in Arrival slots of individual EV at CS.
y Location of ORV. slot.out Departure slots of individual EV at CS.
STPEV_no Strategy vector of PEV_noth number of
Variables
Battcap,PEV_no ; bcap Individual vehicles battery capacity. EV (PEV_no ∈ Z).
battcost,PEV_no Cost of the battery of EVs. STslot
PEV_no Charging strategy vector’s value of
CostTOT Total daily price incurred by CSO for PEV_no number of EV time at interval
tot.PEV EVs. index “slot” (PEV_no ∈ Z).
Costoch Overall cost for complete charging pro- STslot_in
PEV_no Value of charging strategy of PEV_noth
cess for tot.PEV EVs. number EV at arrival time interval
Costbatt.deg Degradation cost of batteries for “slot_in” (PEV_no ∈ Z).
slot.out
tot.PEV EVs. STPEV_no Value of charging strategy of PEV_noth
CostG2V Cost due to G2V mode for tot.PEV EVs. number EV at departure time interval
CostV2G Earn due to V2G mode tot.PEV EVs. “slot_out” (PEV_no ∈ Z).
chrate,PEV_no /ch Required charging rates for PEV_noth SOCslot
PEV_no State of charge of PEV_no th vehicle at
(PEV_no ∈ Z). time index “slot” (PEV_no ∈ Z).
CPEV_no Vector for charging strategy of PEV.noth T Complete time duration vector in 24 h.
EV (PEV_no ∈ Z). tin,PEV_no, / tarr Time of PEV_noth EV arriving at CS.
slot
CPEV_no The value of CPEV_no in the time inter- tout,PEVno /tdep Time of PEV_noth EV departing from
CS.
val  slot (PEV_no ∈ Z).
v(f, e, t) eth vehicle’s speed to f th CS at time t.
Costlab Labor cost of the battery of EVs.
vq (f, e, t) Vehicles flow between eth vehicle and
D Daily mileage of EVs.
f th CS at time t.
DCPEV_no Vector for discharging strategy of
Wl,pos Weighting factor, which specifies the
PEV_noth EV (PEV_no ∈ Z). dominance of the conforming location
slot
DCPEV_no The value of DPEV_no in the time inter- (pos = 1, 2).
val  slot (PEV_no ∈ Z). x Decision variable (1 or 0) to identify the
rate,PEV_no
dch /dch Required discharging rates for CS for individual EVs.
PEV_noth EVs (PEV_no ∈ 
Z Number of cars arriving in a CS.
Z).d(f, e, t)/dise_f Distance of f th Zl,pos Estimated location for IRV at pos = 1,
CS of eth vehicle at time t. 2.
df First trip distance of individual EVs. μzl Average value of IRV Zl.
E The value of expectations. σzl Standard deviation for IRV Zl.
Engdis Total energy discharge due to V2G from λZl,3 Skewness of IRV Zl.
individual EVs. λZl,4 Kurtosis of IRV Zl.
E con Energy consumption while reaching the ξZl,pos Standard location for IRV at pos = 1, 2.
CS by EVs. μc Average value of ORV.
Engreq Total energy drawn due to G2V for in- σC Standard value of ORV.
dividual EVs.

H Vector for time horizon.
I. INTRODUCTION
jjam (f, e, t) Traffic density between fth CS and eth
N THIS age of globalization, the usage of internal combus-
Lbatt
EV at time t.
Lifecycle of the EV’s battery at partic-
ular DOD.
I tion (IC) engine vehicles has a massive negative impact on
the environment. The problems with fossil fuel powered vehicles
P slots,PEV_no Parking duration for PEV_noth EV (FFPV) are as follows.
(PEV_no ∈ Z). 1) Fossil fuels are a scarce resource.
Prob ( Zl,j ) Probability of individual IRV Zl,j . 2) Spilling of oil may be hazardous.
Pwslots
PEV Total power requirement by all the ve- 3) It is very expensive nowadays.
hicles. Therefore, in recent days, due to the cheaper rate of electricity
RTT(slot) Value of RTT per slot. and zero pollution features, EVs are grabbing attention. EVs are
SOCarr Individual EV’s arrival SOC while ar- much more reliable and require less maintenance than FFPVs.
riving at CS. EV has quiet and smooth driving features. Among various types
SOCdep Individual EV’s departure SOC while of EVs, people are in specific tending toward PEV, as it is more
departing at CS. reliable and efficient than pure EVs.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1690 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

A huge deployment of EVs needs a good charging infras- But in this article, it has been shown that the proposed
tructure so that it may able to meet the charging demands of algorithm can handle both between G2V and V2G modes of
EVs smoothly and smartly. Intelligent charging management operation after a certain interval (atleast 30 min), in an efficient
and scheduling are the two key solutions to such a problem. manner, which is different from the earlier concepts adopted by
At a very initial stage, a simple scheduling has been performed the previous articles, which may result in a profitable scenario
by considering various driving cycle attributes, such as arrival for both CSO and the EV owner.
time, departure time, daily mileage, and first trip distance, as Again, in [12], fixed charging rates have been considered,
shown in [1], where the attributes are deterministic in nature. which are not economical for CSOs, as the electricity price
Researchers are, nowadays, focusing more to improve these may vary due to the grid’s generation and demands mismatch,
factors to make this better. But every new technology has its which makes the electricity tariff dynamic. Therefore, in this
own cons with which the researchers have to deal with. Every article, dynamic charging rates have been used so that CSO
time, EV drivers face the anxiety about “when” and “where” to can control and vary the charging rates as per the RTT at
charge their vehicles. Therefore, it is important to identify the different intervals, and consequentially, the operation can be
“appropriate” charging station (CS). To deal with this scenario, economical.
many researchers have proposed various methods. But most of From the critical literature survey [15]–[17], it has been ob-
them have considered the distance as the only feature to identify served that in these articles, fix data of driving cycles have been
the adequate CS for corresponding vehicles [2]. used to perform the charging scheduling. But these attributes
In [3]–[5], Wu et al. have proposed a quick charging solution are always uncertain in nature. Therefore, fluctuations of values
by increasing the charging current and voltage. Few of them used may be possible in these attributes, which are very critical and
ultracapacitors also to reduce the charging time and named those need to be considered to perform better charging scheduling.
as apt CS [6]. But, to identify an adequate CS, the availability Hence, in this article, the uncertainty related to driving cycles
of free charging slots in CS is the most important factor, which has been handled using 2m-PEM.
needs to be considered. Therefore, this factor is incorporated in Previous articles have not considered 2m-PEM in order to
this article along with other factors, such as less battery energy handle the probabilistic driving pattern. Different probabilistic
consumption and shortest distance, to identify an adequate CS. methods, such as MCS, approximate methods [18], truncated
After the allotment of relevant CS to the corresponding ve- Taylor series expansion method [19], and the discretization
hicles, charging scheduling is the next key challenge to deal method [20], are there, which have some major limitations while
with. Moreover, managing EV charging in an optimal way is handling a large number of uncertain variables. Scientist H. P.
always a great challenge. Many authors are focused on the Hong developed an efficient and modified 2m-PEM method for
methodology of EV charging using the G2V mode. How the probabilistic analysis, which is computationally moderate and
surplus energy of the EV can be utilized for the betterment of has the ability to deal with a large number of uncertainties in less
the grid is another major concern and this introduces a new era in time [21]. It uses deterministic routines to handle the stochastic
EV smart charging technology, termed as V2G technology. For frameworks and also has less computational complexity and is
the charging scheduling purpose, some authors have proposed efficient enough to handle a large number of uncertainties. One
a window optimization technique, which is suitable for online more advantage is that, at the end of the optimization problem,
application due to its incessant update of the information pattern. the best solution set, which is obtained, represents the mean
This determined the optimal scheduling with minimal charging value of the objective function and not the actual value of the
cost [7]. objective function. The mean value is used since the present
In [8], considering various driving patterns, it has been shown problem deals with a large number of uncertain variables. Hence,
that due to high pricing during the day time and offering less tariff it is always better to use the mean value than the best value while
at night, EV owners are provoked to charge the EVs during the dealing with the randomness of uncertain variables related to the
night. In the literature [9], an energy management scenario has driving cycle.
been established considering three types of traffic conditions, The proposed ICSA can be divided into two major parts. The
with the main objective to increase the fuel economy of PEV by first part of ICSA deals with the building up of mathematical
integrated traffic information. problem formulations for selecting the “appropriate” CS, deal-
In [10], an autonomous distributed V2G control scheme has ing with the uncertainties related to driving cycles by apply-
been used, where EV can be utilized as spinning reserves for the ing 2m-PEM, and combined charging scheduling of the EVs,
grid. In [11], the proposed charging strategy is allowing EVs to whereas the second part of the ICSA, tackles the minimization of
operate either in the V2G mode or in G2V mode while in the the total daily price incurred by CSO, by appropriately clubbing
CS. various optimization techniques.
The major limitation with these articles is, PEV has been The key contributions of this article are as follows.
operated either in the G2V mode or in the V2G mode, while in 1) Appropriate CS has been selected with respect to the
CS in a particular time. Few of them operated PEV in the G2V consumption of battery energy required to reach that very
mode continuously for a certain duration and also again in the CS and the availability of the slots.
V2G mode continuously for a certain duration, where frequent 2) Both G2V and V2G modes of operation after a certain
switching between the G2V mode and V2G mode is missing interval have been considered and handled by the proposed
[10]–[13]. ICSA.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1691

This contains two major parts, such as (Costoch ) and

(Costbatt.deg ) as shown
CostTOT = Costoch ∪ Costbatt.deg . (2)
och G2V V2G
Again, the Cost is the resultant of Cost and Cost ,
i.e.,
Costoch = (CostG2V + (−CostV2G )) or
Fig 1. Basic diagram for problem definition.
Costoch = CostG2V ∪ CostV2G . (3)
Now, in order to calculate these CostG2V , CostV2G , and
Costbatt.deg [20], the corresponding formulae have been shown
CostG2V
 tot_PEV
tot_slot     rate,PEV 
slot no
= ( CPEV no
ch RTT (slot) (4)
slot=1 PEV_no

CostV2G
Fig. 2. RTT from NEMS.
 tot_PEV
tot_slot    
= ( slot
DCPEV_no dchrate,PEV_no ∗RTT (slot)
slot=1 no=1
3) While performing the scheduling, variable charging rates (5)
have been considered since it is more economical than
fixed charging rates. 
tot_PEV
Costbatt.deg =
4) Various uncertain attributes of driving cycles used in ICSA PEV_no=1
are stochastic in nature and have been critically handled    
by 2m-PEM. The performance of 2m-PEM has been com- battcost,PEV_no ∗Battcap,PEV_no +Costlab ∗ Engdis 
× .
pared with the conventional MCS method. (Lbatt ∗Battcap,PEV.no ∗DOD)
(6)
II. PROBLEM FORMULATION FOR ICSA
C. Constraints in ICSA
A. Problem Positioning
While executing ICSA, few practical constraints need to be
The main anxiety works on EV drivers are “where” and considered, i.e., subjects to the charging rates, the battery SOC,
“when” they can charge their EV at the minimum queue in and the fulfillment of energy requirements. These need to be
a suitable CS. A CS can be said suitable when there will be considered while scheduling for the real-time feasibility.
free charging points available, which will provide the service 1) Charging Rates:
immediately after reaching. It must be located near some centers
of interest and one must be able to reach with the minimum traffic chrate,PEV_no < PWrated (7)
congestions with less energy consumptions. It is notable that by
ch rate,PEV_no
min ≤ ch rate,PEVno
≤ ch rate,PEVno
max . (8)
considering all the above-mentioned criteria, it is not necessary
that the nearest CS will be the most suitable one. Fig. 1 shows The charging or discharging rate should not surpass the charg-
the overall diagram for the problem definition. ing slot’s power ratings, as shown in (7). The charge controller
Now upon the arrival of the corresponding vehicle to their can vary its charging rate or discharging rate in between its rated
respective CS, SOCarr , tarr , tdep , ETL, and type of vehicles [bcap capacity, as shown in (8).
and all electric range (AER)] need to be provided to the CSO. 2) Energy Requirements:
Considering these factors, the required energy has been eval-
uated for every EVs. Now, as per tarr and tdep , the optimal 
slot.out

PEV_no · ch
STslot rate,PEV_no
= Engreq . (9)
scheduling has been executed by taking into consideration the
slot=slot_in
G2V mode and V2G mode to satisfy the required energy.
It must be satisfied within the parking time duration with
dynamic charging and discharging rates controlled by CSO.
B. Objective Function Formulation 3) Battery SOC:
The main objective of this scheduling is to minimize the total
SOCmin ≤ SOCslot
PEV_no ≤ SOC
max
. (10)
daily price incurred by CSO, which has been shown as follows:
The battery SOC should not surpass or go underneath its min-
min{CostTOT }. (1) imum and maximum level, which is 20% and 90%, respectively.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1692 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

4) Number of CS: The number of CS is very less than the Again, the parking duration for PEV_noth is given by
number of vehicles.
P slots,PEV = [tin,PEV_no, . . . , tPEVno , . . . , tout,PEV_no ] .
D. Charging and Discharging Strategy (20)
Since it is 24 h scheduling, therefore, it has been divided into
In (11) and (13), to determine the charging and discharging 48 slots of half hours each.
strategy, (12) and (14) have been developed The charging coordination of EVs (e.g., every 30 min) to
CPEV_no minimize the total daily price incurred by CSO by integrating
 the G2V mode and V2G mode needs to be executed, where the
1
= CPEV_no , . . . , C slot_in slot_out tot_slot
PEV_no , . . . , CPEV_no , . . . , CPEV_no time horizon vector is given by H 
(11)  = [1, . . . , slot, . . . , totslot ] .
H (21)
where
⎧ The number of cars arriving in a CS is denoted by the vector
1, if STslot PEVno = 1, ∀ slot : P
slots,PEV.no
⎪
⎪ ,
⎪
⎪   = [1, . . . , PEV_no, . . . , tot_PEV.
⎨ ∀P EV.no ∈ Z Z (22)
slot
CPEV.no = 0, or else
⎪
⎪ When PEVs are plugged in, three possible operations, i.e.,
⎪
⎪ 0, ∀ slot ∈ / P slots,PEVno ,
⎩ charging (+1), discharging (−1), and idle (0), will occur in order
∀P EV _no ∈ Z
to satisfy engreq .
(12)

DCPEVno 1
= DCPEV , . . . , DC slot in E. Energy Modeling
no PEV_no x , . . . ,
In order to satisfy the constraints given in (9), the energy
slotout totslot
DCPEV_ no
, . . . , DCPEV _no
(13) modeling is given by
where SOCreq · bcap
⎧ engreq = (23)
−1, η ch
⎪
⎪ PEVno = −1 ∀ slot :
if STslot
⎨
slot P slots,PEV_no
∀ P EV _no ∈ Z engdis req = SOCreq · bcap · η dis . (24)
DCPEV_no = .
⎪
⎪ 0, or else
⎩ Now, again to calculate SOCreq , the following equation is
0, ∀ slot ∈ / P slots,PEV_no , ∀ P EV _no ∈ Z
(14) given:
The charging strategy for PEV_noth vehicle can be defined ⎧
⎪
⎪ 1 − SOCarr , when SOC
dep
>1
by ⎪
⎪ SOC
⎪
⎪
dep
− SOC arr
,
 ⎨
STPEV_no = ST1PEVno , STslot in slot req when SOCarr < SOCdep < 1
PEVno , . . . , STPEVno , . . . , SOC = (25)
⎪
⎪ 0, when SOCarr = SOC 
dep
⎪
⎪
slot.out
, . . . , STtot_slot ⎪
⎪ − SOC − SOC
arr dep
,
STPEV_no PEV_no (15) ⎩
when 0.2 < SOCdep < SOCarr
where
⎧ where SOCarr [21] and SOCdep [21] can be calculated from the
⎨ 1, charging following:
STslot = −1, discharging , ∀PEV_no ∈ Z
 (16)  f 
PEV_no
⎩
0, idle SOCarr = 1 − d AER (26)
where the strategy should be such that  
SOCdep = ETL/AER + 20% . (27)

slot_out

PEV_no · ch
STslot rate
= Engreq ,
slot=slot_in Although SOCarr and SOCdep have been calculated, practi-
 cally, the SOC status after the arrival and the ensuing trip distance
∀ PEV_no ∈ Z. (17)
can be obtained using EVs telematics system. Moreover, since
Energy discharged and power requirements by all the vehicles the full charging of the battery may reduce the battery life cycle
can be formulated as [25], [26], it is avoided.

tot_slot From the above-mentioned discussion, it is clear that in order
Engdis_no = slot
DCPEV_no ∗chrate,PEVno ∀ PEV_no ∈ Z to perform the scheduling, df , tarr , tdep , and D are basically
slot=1 needed. But all these attributes are uncertain. Due to the un-
(18) availability of historical data, the statistical analysis (2m-PEM)

tot_PEV
  rate,PEV_no has been performed to handle the uncertainty and the required
Pwslots
PEV =
slot
CPEV no
− DCPEV_no
slot
ch estimated data have been generated. The details regarding the
PEV_no=1 estimation have been discussed in the following section and
−
→ the details regarding the data generation have been clarified in
∀ slots ∈ H. (19) Section V-C.
Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1693

III. UNCERTAINTY MODELING TECHNIQUE TABLE I

STRUCTURE OF ASSIGNMENT MATRIX
A. Hong’s 2m-PEM
To deal with the uncertainties, two deterministic points have
been estimated from one probabilistic point. For one probabilis-
tic input random variable (IRV) set, two deterministic sets of
data can be achieved using 2m-PEM.
Mathematically, the deterministic agenda of the charging cost
minimization problem can be represented as
C = fn (ri ) . (28) The expectations of the ORV can be calculated as
m 
 2
The function fn transfers the uncertainty from IRVs to output
variables E (C y ) = (Wl,pos × ((Cl,pos ))y ) , y = 1, 2. (38)
l=1 pos=1
C = fn (Cr , Z1 , Z2 , . . . , Zm ) (29)
Finally, the μc and σC of the ORV can be determined as
where Zl (l = 1, 2, . . . , m) are the input variables under vague- follows:
ness with the probability function fnZm .  1 
The statistical moment’s output of IRVs has been determined μc = E C and σC = (E (C 2 )) − (E (C 1 ))2 . (39)
using μzl , σzl , λZl,3 , and λZl,4 . The 2m-PEM generates two
deterministic concentrations for each IRV Zl , as (Zl,1 Wl,1 ) and After explaining the problem formulation in Section II and
(Zl,2 Wl,2 ), which implies the dominance of the conforming generating the data using 2m-PEM in Section III, the solution
location in the calculation of the output random variable (ORV). strategy to find out the appropriate CS and combined scheduling
Here Zl,pos can be determined as follows: of the EVs have been performed applying two stages of ICSA
in the following section.
Zl,pos = μzl + ξZl,pos ∗ σZl , pos = 1, 2. (30)
In (30), the posth standard location, ξZl,pos , can be assessed IV. SOLUTION STRATEGY OF ICSA
as follows: A. First Level of ICSA
 √
ξZl,pos = λZl,3 /2 + (−1)(3−pos) λZl,4 − ( 3λZl,3 /2)2 .
2 The main aim is to allocate each PEV to “appropriate” CS, by
(31) keeping the arrival SOC at its maximum by consuming very less
In (31), λZl,3 [19] and λZl,4 [19] of Zl have been calculated amount of energy while reaching its recommended CS, where
as follows: the vacant slots are available. This has been done using integer
 linear programming (ILP).
λZl,3 = m3 (λZl,3 )3 (32) To formulate ILP problem (ILPP), few sets of parameters need
to be defined. The assignment matrix C can be defined as C (f,
where
e) (where e >> f), where each EVs e can be assigned to CSf .
N
 The assignment matrix is based on the well-known assignment
m3 = E(Zl − μzl )3 = (Zl − μzl )3 · Prob (Zl,j ) (33) problem [23] with a mixture of the transportation problem [24],
j=1
as given in Table I. Since there is a dynamic behavior in the

λZl,4 = m4 (λZl,4 )4 (34) assignment matrix coefficient, therefore, instead of C (f, e), it
can be defined as C (f, e, t), where t ∈ time.
where The demand of the vehicle is to get only one charging point
N
 at only one CS; therefore, the demand is unity.
m4 = E(Zl − μzl )4 = (Zl − μzl )4 · Prob (Zl,j ) . (35) In this ILPP, the main objective is to keep the energy con-
j=1 sumption at its lowest value, i.e.,
⎧ ⎫
One deterministic set is kept fixed and other IRVs are consid- ⎨ f  e ⎬
ered equal to its average value in each simulation of 2m-PEM, min (E con (f, e, t) · x (f, e, t)) (40)
as follows: ⎩ ⎭
f =1 e=1

C(l,pos) = fn (Cr , μ1 , μ2 , . . . , zl,pos , . . . , μZm ) (36) where


where pos = 1 and 2 and l = 1, 2, …, m. 1, if EVe is assigned to CSf at timet
x (f, e, t) = (41)
In (36), Zl,1 and Zl,2 are the specified locations of the IRV 0, otherwise
Zl , and μZm is the average value of the rest of the IRVs.
subject to
Two weighting factors of Zl can be calculated as
(−1)pos ξZl,3−pos 
cs

Wl,pos = × . (37) x (f, e, t) = 1 ∀ EVe (42)

m ξZl,1 −ξZl,2 f =1

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1694 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

where 1 ≤ e ≤ EV and each vehicle can be assigned to only Stage 3: Updating Henry’s coefficient and solubility
one CS at a time   
l 1 1

ev (H l (k + 1)) = (H l (k)) × exp −C −
x (f, e, t) ≤ CSf ∀CSf (43) T (k) T̃
e=1 (52)
where 1 ≤ f ≤ CS the capacity of each CS should not exceed soli,l (k) = α ∗ H l (k + 1) ∗ pi,l (k) . (53)
its limit.
The df should not be lower than the EV’s dise_f Stage 4: Position update
The positions can be updated using the following equations:
dise_f ≤ df . (44)  
Y i,l (k+1) = Y i,l (k)+F ∗ rand ∗ σ ∗ Y i,best (k)−Y i,l (k)
1) Energy Modeling in Linear Optimization Problem: In or-
der to model the energy consumption in the assignment matrix, + F ∗ rand ∗ μ
various factors, such as (v(f, e, t)), (d(f, e, t)), AER(e), its  
∗ soli,l (k) ∗ Y best (k) − Y i,l (k) (54)
(bcap (e)), (vq (f, e, t)), and (jjam (f, e, t)), have been considered
[25]. σ = β ∗ exp(−F best (k)+/F i,l (k)+),  = 0.05.
The required time for covering the distance (d(f, e, t)) by the (55)
EV can be expressed as
Stage 5: Escape of local optimum and update of the worst
T = d/v , where T, d, v : fn (f, e, t) . (45) agent’s position
Again, from the equation of Greenshields model [27], it has    
N worst = N ∗ rand r2 − r1 + r1 (56)
been proved that vq is directly proportional to the jjam  
vq ∝ jjam (46) Gi,l = Gmin(i,l) + rand Gmax(i,l) − Gmin(i,l) ). (57)

or vq = k · jjam (47) For further details regarding these above-mentioned equations

and nomenclatures, refer [28]. The algorithm of HGSO has
where k = v1 and vq , jjam ∈ fn(f, e, t). been shown in the following part. In Fig. 3, the whole working
The consumed energy Econ [2] under normal conditions is procedure has been described.
bcap · d
Econ = (48)
AER V. PERFORMANCE EVALUATION, RESULTS, AND DISCUSSIONS
where AER ∈ fn(e) and d ∈ fn(f, e, t). A. Test Case-1
So, by considering the vehicles flow vq and jjam , it can be
modified as [2] This test case has been divided into six subtest cases named
TC1a, TC1b, TC1c, TC1d, TC1e, and TC1f with 10, 20, 30, 40,
bcap · T · vq 50, and 60 EVs, respectively, of lower battery capacity (LBC).
Econ = . (49)
AER · jjam Each test case consists of three kinds of EVs, such as PHEV-30,
PHEV-40, and BEV [29] in the ratio of 70%, 20%, and 10%,
B. Second Level of ICSA whose specifications are given in [27].
HGSO has been used to solve the second level of optimization
problem. Using the outcome of ILP along with the parameter B. Test Case-2
shown in Section II-E, all the required attributes have been
In the same manner, test case 2 also has been divided into six
assessed, and the corresponding population sets have been gen-
subtest cases named TC2a, TC2b, TC2c, TC2d, TC2e, and TC2f
erated. The HGSO is a very recent optimization proposed by
with 10, 20, 30, 40, 50, and 60 EVs of higher battery capacity
Hashim et al. [30] and inspired by the nature science-based
(HBC), and each of them consists of Tesla Roadster [30], BMW
algorithm.
Mini E [31], and BEV Version of BMW i3 [31] model EV’s in
The stages of HGSO and its corresponding equations have
the ratio of 20%, 55%, and 25%, respectively.
been shown as follows.
Stage 1: The process of Initialization
Within a constraint’s boundary or range, a random population C. Input Data
can be generated and initialized by The attributes related to driving cycles, such as tarr , tdep , D, df ,
 k  ETL, and SOCini are uncertain and to generate these, 2m-PEM
Y i (k + 1) = Ymin
k
+ rand Ymax k
− Ymin . (50)
has been adopted, as shown in Section III-A, which follows a
Stage 2: Clustering and evaluation of Henry’s gas constant normal distribution [8], [26]. The data regarding these attributes
⎫ have been given in Tables III and IV. In Table II, the charging and
H l (k) = C 1 ∗ rand (0, 1) ⎬
discharging rates for each charging category have been decided
pi,l = C 2 ∗ rand (0, 1) . (51)
 i,l = C 3 ∗ rand (0, 1) ⎭
as per the battery capacity of the vehicles, and the charging rates
C are always greater than the discharging rates [14].

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1695

D. Real-Time Tariff
The information about RTT is highly volatile. The data are
taken from NEMS of any particular day [1], where RTT of every
half an hour (30 min) interval is provided in Fig. 2.

E. Assumptions
Few assumptions have been considered in ICSA. It is assumed
that all these CS are homogeneous regarding the EVSE ratings
and each CS have fewer charging points than the total number
of vehicles. For example, if total ten EVs are there, then at each
CS, five, three, and two charging slots are there, respectively.
Likewise, for other test cases also, the number of charging slots
at each CS is less than the total number of vehicles. CSs are
situated near the public areas, such as schools, markets, and
universities, where EVs can stay for at least a couple of hours. In
Table II, three types of charging categories have been set, as per
the energy demand by the EVs and accordingly, two input data,
i.e., min. charging rate (or discharging rate) and max. charging
rate are there for each type of charging category [14], [15].

F. Results and Discussions

As per the problem formulation, in the first level of ICSA,
appropriate CS has been allotted to each PEV with the minimum
energy consumption of the battery. Thereafter, considering all
the constraints, as mentioned in (42)–(44) and as per the FTD
and ETL, the values of SOCarr [see Fig. 5(a1) and (a2)] and
SOCdep [see Fig. 5(b1) and (b2)], the SOCreq [see Fig. 5(c1)
Fig. 3. Basic workflow of ICSA with the HGSO integration. and (c2)] and its corresponding engreq [see Fig. 5(d1) and (d2)]
has been determined for TC1a and TC2a, where ten EVs of LBC
and HBC are considered. Likewise, for other test cases also, a
TABLE II similar kind of operation has been performed.
VARIABLE CHARGING RATES Using these calculated attributes, in the second level of the
ICSA, the HGSO optimization has been applied to evaluate the
objective function, i.e., total daily price incurred by CSO and
its corresponding charging strategies, as shown in Fig. 6, where
black blocks are representing the V2G mode of operations and
white blocks are representing the G2V mode of operations, and
the rest of the Grey blocks are idle. From TC1a–TC1f, it can
TABLE III be observed that, due to the majority of LBC EVs, most of
PARAMETERS OF UNCERTAIN ATTRIBUTES the time, each and every EV is participating more in the G2V
mode of operations and for very few times it is changing its
switching sequence from the G2V mode to the V2G mode of
operations. But as EVs are increasing from 10 to 60, it can also be
noticed that gradually the V2G mode of operations is increasing,
but then also in several cases, despite higher RTT, EVs are
participating in the G2V mode, rather than the V2G mode of
operation.
TABLE IV Unlike the former scenario, if the charging strategies of TC2a–
PARAMETERS OF VEHICLES LOCATION FROM CS
TC2f are observed, then it can be realized that, due to the majority
of HBC EVs, when the RTT is higher, EVs are participating in the
V2G mode of operation and when RTT is lower, it is participating
in the G2V mode of operations, which is actually required for
the minimization of the daily price incurred by CSO. When the
number of EVs is increasing, it can be seen that the V2G mode

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1696 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

Fig. 4. For TC1a and TC2a. (a1) and (a2) Estimated daily mileage of ten LCB and ten HCB. (b1) and (b2) Estimated first trip distance of ten LCV and ten HCB.
(c1) and (c2) Estimated arrival time and departure time of both ten LCB and HCB. (d1) and (d2) Ensuing trip length of ten LCB and ten HCB.

Fig. 5. For TC1a and TC2b. (a1) and (a2) Calculated SOCarr of ten LCB and ten HCB. (b1) and (b2) Calculated SOCdep of ten LCV and ten HCB. (c1) and
(c2) Calculated SOCreq of both ten LCB and ten HCB. (d1) and (d2) Calculated engreq of ten LCB and ten HCB of operation can be possible.

of operations is increasing than the G2V mode of operations for SOC departure, SOC requirement, and its corresponding energy
each and every EV. Therefore, comparing these two scenarios, requirement have been calculated, which are 58.884%, 82.018%,
it can be said that, if the number of EVs is more having HBC, and 4.48047 kWh, respectively. This needs to be kept in mind
then better coordination between the G2V mode of operation that the departure slot has been chosen by the EV user only.
and V2G mode of operation can be possible. Within this interval, the CSO needs to meet the energy demand of
Nevertheless, irrespective of LBC EVs or HBC EVs, the that EV. Consequently, optimization has been performed using
coordination of this dual mode of operation is always beneficial (11), to keep the total daily price incurred by CSO at optimum.
for both CSO and EV owner. The charging strategy of the fifth EV has been shown in the first
In support of this claim, a critical analysis has been performed row of Fig. 9, where both G2V and V2G modes of operation
by taking a subtest case TC1a, where ten LBC EVs are there, and have been performed. On the other hand, in the second row, the
among them, the fifth number EV has been taken to demonstrate charging strategy using only the G2V mode of operation has
the benefit. From Fig. 6(a1), it can be noticed that the fifth EV been shown. As per the RTT, if the cost of these two operations
enters the CS at slot 22 and the departure slot is 43. Now, as per can be compared, then from Table V, it can be realized that
the vehicle’s daily mileage [see Fig. 4 (a1)], FTD [see Fig.4(b1)], the combination of both G2V and V2G modes of operation is
ETL [see Fig. 4(d1)], and that very vehicle’s AER, SOC arrival, always beneficial for CSO, as it is getting a profit of 0.205385

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1697

Fig. 6. For test cases 1 and 2: (a1) and (a2) charging strategy of TC1a and TC2a; (b1) and (b2) charging strategy of TC1b and TC2b; (c1) and (c2) charging
strategy of TC1c and TC2c; (d1) and (d2) charging strategy of TC1d and TC2d; (e1) and (e2) charging strategy of TC1e and TC2e; (f1) and (f2) charging strategy
of TC1f and TC2f.

TABLE V
VARIOUS COST OF THE FIFTH EV IN TC1a

Fig. 7. Cost comparison of TC1a.

Fig. 9. Charging strategy for the fifth EV of TC1a.

these two test cases, from Figs. 10 and 12, it can be analyzed
that, due to more LBC EVs, the margin of profit for the dual mode
of operation is less. On the other hand, for TC2e, if Figs. 11 and
Fig. 8. Cost comparison of TC2a. 13 can be realized, then it can be said that, due to the presence of
more HBC EVs, more amount of V2G operations is occurring.
$. Moreover, after comparing the costing between only the G2V Therefore, the margin of profit for CSO is also getting higher.
mode of operation and dual mode of operation of TC1a and Although for both the scenarios, CSO is gaining profit as it is
TC2a, from Figs. 7 and 8, it can be observed that, for every earning more from the grid by performing more V2G mode of
individual case, the CSO is getting more profit by conducting operation. After performing the coordinate charging, it can be
the dual mode of operation rather performing only the G2V observed from Fig. 14 that, for both test cases with the increase
mode of operation and in case of HBC EVs, the margin of profit of set of EVs from 10 to 60, the cost of charging is increasing.
is more. But, in comparison with HBC EVs in test case 2’s scenarios,
Again, for TC1e and TC2e, having 50 LBC EVs and 50 HBC the cost of the G2V mode of operation is more for LBC EVs
EVs, a similar kind of analysis has been performed. Between under test case 1. Again, if only the cost of V2G mode can be

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1698 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

Fig. 15. Comparison of earning from the V2G mode of operation between
LBC and HBC EV.
Fig. 10. Comparison of individual costs of TC1e.

Fig. 16. Comparison of earning from the battery degradation for the V2G
mode of operation between LBC and HBC EV.

Fig. 11. Comparison of individual costs of TC2e.

Fig. 17. Comparison of the total daily price incurred by CSO for both G2V
and V2G modes of operation between LBC and HBC EV.

Fig. 12. Overall cost comparison between only the G2V mode and combina-
tion of both G2V and V2G modes for TC1e.

Fig. 18. Comparison of cost for only the G2V mode of operation between
LBC and HBC EV.

Fig. 19. Comparison of profit incurred by CSO for both G2V and V2G modes
Fig. 13. Overall cost comparison between only the G2V mode and combina- of operation between LBC and HBC EV.
tion of both G2V and V2G modes for TC2e.

considered, then from Fig. 15, it can be observed that the earning
of revenue from the V2G mode is increasing with the increment
of EVs for both the test cases. But, compared with LBC EVs,
the earning of revenue is more for HBC EVs. Since the earning
of revenue from battery degradation is correlated with the V2G
mode of operation, thus, from Fig. 16 also, the same kind of
scenario can be observed. On the contrary, when only the G2V
Fig. 14. Comparison of cost for the G2V mode of operation between LBC mode of operation has been considered, for all the test systems,
and HBC EV. then it can be assessed that, with the increase of EVs, the cost

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1699

TABLE VI
COMPARISON OF TOTAL COSTS AND PROFITS CONSIDERING BOTH G2V AND V2G MODES OF OPERATION FOR ALL THE TEST CASES UNDER TEST CASES 1 AND 2

TABLE VII
COMPARISON OF COSTS BY APPLYING ICSA WITH 2M-PEM AND MCS TECHNIQUE BY CLUBBING VARIOUS OPTIMIZATION

of charging is increasing, and there is no question of earning operation, then for all the test cases, both CSO and EV owner
revenues. will gain profit and with the increment in the number of EVs,
Hence, after combining the above-mentioned three scenarios, this profit will arise more. At the same time, the margin of profit
it can be said that, if the combination of G2V and V2G mode will be more for HBC EVs, as shown in Fig. 19. The overall
of operation can be performed instead of only the G2V mode of comparison of total costs and profits, with the increment of EVs

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1700 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

TABLE VII
CONTINUED

DE: Differential evolution; BBO: Biogeography-based optimization; GWO: Grey-wolf optimization; SOS: Sum of square optimization; BA: Bat algorithm; PSO: Particle swarm
optimization; GA: Genetic algorithm.

for both the test cases have been shown in Table VI.Hence, from of square optimization, bat algorithm, particle swarm optimiza-
Table VI and Figs. 14–19, it can be observed that, with the tion, and genetic algorithm, have been applied to verify its per-
increase of LBC EVs, the profit will increase. But, compared formance. From Table VII, by observing the mean and standard
with the EVs having HBC, the margin of profit is lower. deviation of the costs, it can be said that ICSA is robust enough to
Now, in order to establish the robustness of the ICSA, var- give consistent outputs for all the test cases by clubbing HGSO.
ious optimization techniques, such as differential evolution, Moreover, to analyze the performance of 2m-PEM, it has been
biogeography-based optimization, grey-wolf optimization, sum compared with the MCS method [18] after 50 trial runs, and from

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
DAS et al.: CHARGING SCHEDULING OF EV INCORPORATING G2V AND V2G TECHNOLOGY CONSIDERING IN SMART GRID 1701

TABLE VIII capacity, the margin of profit from charging scheduling will
WSRT RESULT FOR TEST CASES 1 AND 2
be greater. Likewise, for better coordination of the G2V and
V2G mode of operation, a greater number of vehicles with HBC
are desirable. Thereafter, from the results it has been observed
that 2m-PEM is superior regarding the handling of uncertain
variables. Furthermore, the results after applying HGSO have
been compared with the other optimization techniques and it
has been observed that HGSO is superior to others. Later, by
the statistical analysis, it can be said that the proposed ICSA in
coordination with the optimization techniques is robust enough
to produce consistent output for both lower and higher number
of vehicles.

REFERENCES
[1] S. Das, P. Acharjee, and A. Bhattacharya, “Charging scheduling of electric
TABLE IX vehicle incorporating grid-to-vehicle (G2V) and vehicle-to-grid (V2G)
QT RESULT FOR TEST CASES 1 AND 2 technology in smart-grid,” in Proc. IEEE Int. Conf. Power Electron., Smart
Grid Renewable Energy, Kerala, India, Jan. 2–4, 2020, pp. 1–6.
[2] R. Mkahl, “Contribution to the modeling, dimensioning and management
of the energy flows of an electric vehicle charging system: Study of
the interconnection with the electric network,” Ph.D. dissertation, Dept.
Eng. Sci. Microengineering, Univ. Technol. Belfort-Montbéliard, Belfort,
France, 2015.
[3] D. Wu, D. C. Aliprantis, and L. Ying, “Load scheduling and dispatch for
aggregators of plug-in electric vehicles,” IEEE Trans. Smart Grid, vol. 3,
Table VII, by observing the best value and standard deviation, no. 1, pp. 368–376, Mar. 2012.
it can be said that, 2m-PEM is superior in comparison with the [4] F. Zhang, X. Hu, R Langari, and D. Cao, “Energy management strategies of
connected HEVs and PHEVs: Recent progress and outlook,” Prog. Energy
MCS method. Moreover, the time of simulation is much higher Combustion Sci., vol. 73, pp. 235–256, Jul. 1, 2019.
for the MCS method, compared with the 2m-PEM. This is how [5] X. Liu and Z. Bie, “Optimal allocation planning for public EV charging
the superiority of the 2m-PEM for handling the uncertainty in station considering AC and DC integrated chargers,” Energy Procedia,
vol. 159, pp. 382–387, Feb. 2019.
the proposed ICSA algorithm can be established. [6] K. Saichand and V. John, “A time-varying virtual resistance control for
1) WSRT and Quade Test: For both test case 1 (TC1a–TC1f) ultracapacitor based dc–dc converters,” IEEE Trans. Veh. Technol., vol. 68,
and 2 (TC2a–TC2f), WSRT [34] and QT [34] have been per- no. 6, pp. 5548–5556, Apr. 2019.
[7] C. Ma, J. Rautiainen, D. Dahlhaus, A. Lakshman, J.-C. Toebermann, and
formed (at a 95% confidence interval) by simulating each test M. Braun, “Online optimal charging strategy for electric vehicles,” Energy
case for 30×. The absolute value for 30 samples is 137 (from Procedia, vol. 73, pp. 173–181, Jun. 2015.
α-distribution table). From Table VIII, it is hereby observed [8] A. Ehsan and Q. Yang, “Active distribution system reinforcement planning
with EV charging stations—Part I: Uncertainty modeling and problem
that for every test case the test statistic value is higher than the formulation,” IEEE Trans. Sustain. Energy, vol. 11, no. 2, pp. 970–978,
absolute value. This implies the acceptance of null hypothesis Apr. 2020.
H0 , which further shows the consistency and sturdiness of [9] A. Abdalrahman and W. Zhuang, “PEV charging infrastructure siting
based on spatial–temporal traffic flow distribution,” IEEE Trans. Smart
optimization techniques in ICSA. Likewise, in the QT, from Grid, vol. 10, no. 6, pp. 6115–6125, Nov. 2019.
Table IX, it can be observed that for both T1 and T2, the test [10] Y. Ota, H. Taniguchi, T. Nakajima, K. M. Liyanage, J. Baba, and
statistic value is greater than its absolute value (i.e., 2.04 for 30 A. Yokoyama, “Autonomous distributed V2G (vehicle-to-grid) satisfying
scheduled charging,” IEEE Trans. Smart Grid, vol. 3, no. 1, pp. 559–564,
samples) as obtained from the f-distribution table. This infers Mar. 2012.
rejection of null hypothesis H0 , i.e., applying HGSO results [11] E. Sortomme, M. M. Hindi, S. D. J. MacPherson, and S. S. Venkata, “Co-
in a significant change as compared with another optimization ordinated charging of plug-in hybrid electric vehicles to minimize distri-
bution system losses,” IEEE Trans. Smart Grid, vol. 2, no. 1, pp. 198–205,
techniques in ICSA. Mar. 2011.
[12] R. Zgheib, K. Al-Haddad, and I. Kamwa, “V2G, G2V and active
VI. CONCLUSION filter operation of a bidirectional battery charger for electric vehi-
cles,” in Proc. IEEE Int. Conf. Ind. Technol., Taipei, Taiwan, 2016,
In this article, by applying the proposed ICSA, EVs have pp. 1260–1265.
[13] M. C. B. P. Rodrigues, I. D. N. Souza, A. A. Ferreira, P. G. Barbosa,
been allocated at appropriate CS. Thereafter, the minimization and H. A. C. Braga, “Simultaneous active power filter and G2V (or V2G)
of the total daily price acquired by CSO is determined and its operation of EV on-board power electronics,” in Proc. 39th Annu. Conf.
conforming charging scheduling has been set on. As per the IEEE Ind. Elect. Soc., Vienna, Austria, 2013, pp. 4684–4689.
[14] A. Mohamed, V. Salehi, T. Ma, and O. Mohammed, “Real-time energy
energy requirement, the charging scheduling is done by keeping management algorithm for plug-in hybrid electric vehicle charging parks
the vehicles in the G2V, V2G, and idle mode. The significance involving sustainable energy,” IEEE Trans. Sustain. Energy, vol. 5, no. 2,
of coordination of the G2V mode and V2G mode of operation pp. 577–586, Apr. 2014.
[15] L. Gan, U. Topcu, and S. H. Low, “Stochastic distributed protocol
has been analyzed and it has been showed that both CSO and for electric vehicle charging with discrete charging rate,” in Proc.
EV owner can be in a beneficial position. Moreover, it has been IEEE Power Energy Soc. Gen. Meeting, San Diego, CA, USA, 2012,
detected that, with the increase of EVs along with its battery pp. 1–8.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.
1702 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 57, NO. 2, MARCH/APRIL 2021

[16] J. Rios-Torres, J. Liu, and A. Khattak, “Fuel consumption for various Sourav Das (Student Member, IEEE) received the
driving styles in conventional and hybrid electric vehicles: Integrating B.Tech. degree in electrical engineering from the
driving cycle predictions with fuel consumption optimization,” Int. J. Maulana Abul Kalam University of Technology, West
Sustain. Transp., vol. 13, no. 2, pp. 123–137, Feb. 2019. Bengal (formerly known as West Bengal University of
[17] P. Ping, W. Qin, Y. Xu, C. Miyajima, and K. Takeda, “Impact of driver Technology), Kolkata, India, in 2016, and the M.Tech.
behavior on fuel consumption: Classification, evaluation and prediction degree in electrical power system from the National
using machine learning,” IEEE Access, vol. 7, pp. 78515–78532, 2019. Institute of Technology Durgapur, Durgapur, India, in
[18] D.-M. Kim, P. Benoliel, D.-K. Kim, T. H. Lee, J. W. Park, and J.-P. Hong, 2018, where he is currently working toward the Ph.D.
“Framework development of series hybrid powertrain design for heavy- degree in electrical engineering.
duty vehicle considering driving conditions,” IEEE Trans. Veh. Technol., His current research interest include power system
vol. 68, no. 7, pp. 6468–6480, Jul. 2019. optimization, and renewables and electric Vehicle.
[19] C. Barrows et al., “The IEEE reliability test system: A proposed 2019
update,” IEEE Trans. Power Syst., vol. 35, no. 1, pp. 119–127, Jan. 2020.
[20] Z. Qin et al., “A universal approximation method and optimized hardware
architectures for arithmetic functions based on stochastic computing,”
IEEE Access, vol. 8, pp. 46229–46241, 2020.
[21] C. Li, Y. Chen, T. Ding, Z. Du, and F. Li, “A sparse and low-order
implementation for discretization-based eigen-analysis of power systems
with time-delays,” IEEE Trans. Power Syst., vol. 34, no. 6, pp. 5091–5094,
Nov. 2019. Parimal Acharjee (Senior Member, IEEE) received
[22] A. Saha, A. Bhattacharya, P. Das, and A. K. Chakraborty, “A novel the B.E.E. degree from the University of North Ben-
approach towards uncertainty modeling in multiobjective optimal power gal, Siliguri, India, in 1996, and the M.E.E. and
flow with renewable integration,” Int. Trans. Elect. Energy Syst., vol. 29, Ph.D. degrees in electrical power system from Ja-
no. 12, Dec. 2019, Art. no. e12136. davpur University, Kolkata, India, in 2001 and 2007,
[23] W. Kempton and J. Tomic, “Vehicle-to-grid power fundamentals: Calcu- respectively.
lating capacity and net revenue,” J. Power Sources, vol. 144, pp. 268–279, He has three years of industrial experience and 15
Jun. 2005. years of teaching experience. He is currently with
[24] S. Rezaee, E. Farjah, and B. Khorramdel, “Probabilistic analysis of plug-in the National Institute of Technology Durgapur, Dur-
electric vehicles impact on electrical grid through homes and parking lots,” gapur, India, as an Associate Professor. His current
IEEE Trans. Sustain. Energy, vol. 4, no. 4, pp. 1024–1033, Oct. 2013. research interest is the application of soft computing
[25] E. Wikner and T. Thiringer, “Extending battery lifetime by avoiding high techniques in various power system problems, flexible ac transmission system
SOC,” Appl. Sci., vol. 8, no. 10, pp. 1–16, 2018. devices, smart grid, distributed generation, and phasor measurement unit.
[26] S. Rezaee, E. Farjah, and B. Khorramdel, “Probabilistic analysis of plug-in Dr. Acharjee was a recipient of the Indo-U.S. Research Fellowship.
electric vehicles impact on electrical grid through homes and parking lots,”
IEEE Trans. Sustain. Energy, vol. 4, no. 4, pp. 1024–1033, Oct. 2013.
[27] A. Aktel, B. Yagmahan, T. Özcan, M. M. Yenisey, and E. Sansarcı,
“The comparison of the metaheuristic algorithms performances on airport
gate assignment problem,” Transp. Res. Procedia, vol. 22, pp. 469–478,
Jan. 2017.
[28] J. Schrieber, D. Schuhmacher, and C. Gottschlich, “DOTmark—A bench-
mark for discrete optimal transport,” IEEE Access, vol. 5, pp. 271–282,
2016.
[29] N. Shlayan, K. Challapali, D. Cavalcanti, T. Oliveira, and Y. Yang, “A novel Aniruddha Bhattacharya (Member, IEEE) received
illuminance control strategy for roadway lighting based on greenshields the B.Sc. Eng. degree in electrical engineering from
macroscopic traffic model,” IEEE Photon. J., vol. 10, no. 1, Feb. 2018, the National Institute of Technology, Jamshedpur,
Art. no. 8200211. Jamshedpur, India, in 2000, and the M.E.E. and Ph.D.
[30] F. A. Hashim, E. H. Houssein, M. S. Mabrouk, W. Al-Atabany, and degrees in electrical power system from Jadavpur
S. Mirjalili, “Henry gas solubility optimization: A novel physics-based al- University, Kolkata, India, in 2008 and 2011, respec-
gorithm,” Future Gener. Comput. Syst., vol. 101, pp. 646–667, Dec. 2019. tively.
[31] N. G. Omran and S. Filizadeh, “Location-based forecasting of vehicular His employment experience include Siemens Me-
charging load on the distribution system,” IEEE Trans. Smart Grid, vol. 5, tering Limited, India; Jindal Steel and Power Limited,
no. 2, pp. 632–641, Mar. 2014. Raigarh, India; Bankura Unnyani Institute of Engi-
[32] Specifications of Tesla Roadster, 2020. [Online]. Available: neering, Bankura, India; Dr. B. C. Roy Engineering
https://www.tesla.com/roadster College, Durgapur, India; National Institute of Technology Agartala, India. He
[33] Specifications of BMW EVs, 2020. [Online]. Available: is currently an Assistant Professor with Electrical Engineering Department,
https://www.bmwusa.com/vehicles/bmwi.html National Institute of Technology Durgapur, Durgapur, India. His areas of interest
[34] Z. Liu, E. Blasch, and V. John, “Statistical comparison of image fusion include power system optimal power flow, hydrothermal scheduling, power
algorithms: Recommendations,” Inf. Fusion, vol. 36, pp. 251–260, 2017. system stability, power system optimization, and electric vehicle.

Authorized licensed use limited to: University of Delhi. Downloaded on April 04,2025 at 09:39:12 UTC from IEEE Xplore. Restrictions apply.