Received 18 August 2023, accepted 18 September 2023, date of publication 22 September 2023,

date of current version 27 September 2023.

Digital Object Identifier 10.1109/ACCESS.2023.3317894

Multi-Speed Transmission Optimization of

Electric Vehicles Based on Shifting Pattern
Considering Dynamic Inertia Efficiency
KIJONG PARK1 , SANG-KIL LIM 2, AND KIHAN KWON 2, (Member, IEEE)
1 Researchand Development Division, Hyundai Motor Company, Hwaseong 18280, South Korea
2 Department of Automotive Engineering, Honam University, Gwangju 62399, South Korea

Corresponding author: Kihan Kwon ([email protected])

This work was supported by the Regional Innovation Strategy (RIS) through the National Research Foundation of Korea (NRF) funded by
the Ministry of Education (MOE) under Grant 2021RIS-002.

ABSTRACT Multi-speed transmissions for electric vehicles (EVs) can achieve superior economic and
dynamic performances than single-speed transmissions. Since gear shifting causes an equivalent inertia
variation in multi-speed transmissions, the optimal shifting pattern should be determined by considering the
inertia variation effect to maximize EV performances. To consider the dynamic inertia variation effect owing
to gear shifting, the equivalent inertia for each speed gear and dynamic inertia efficiency are mathematically
derived. An EV analysis model is constructed to evaluate the EV performances, and energy efficiency
and acceleration ability are adopted as quantification measures for economic and dynamic performances,
respectively. The result comparison of the optimal shifting patterns when considering and not considering
the dynamic inertia efficiency exhibits the importance of the optimal shifting pattern considering the
dynamic inertia efficiency for the superior transmission design of EVs. A multi-objective optimization
problem is formulated that includes the design variables as gear ratios and shifting patterns and the
objective functions as energy efficiency and acceleration ability. As an alternative to the excessive calculation
burden of conducting multi-objective optimization, an artificial neural network (ANN)-based multi-objective
optimization process is utilized. To verify the importance of the dynamic inertia efficiency on economic
and dynamic performances, the gear ratios and shifting patterns are optimized by considering the dynamic
inertia efficiency and none. The different optimum solutions and objective function values demonstrate the
necessity of considering the dynamic inertia efficiency owing to gear shifting; the economic and dynamic
performances are improved from 2.7% to 7.8% and 2.8% to 3.0%, respectively.

INDEX TERMS Electric vehicle, two-speed transmission, dynamic inertia efficiency, shifting pattern, ANN-
based optimization.

ABBREVIATION WLTP World harmonized light-duty vehicles test

procedure.
EV Electric vehicle.
EFF Energy efficiency of EV.
ANN Artificial neural network.
SOC State of charge of battery.
DCT Dual-clutch transmission.
WOT Wide-open throttle.
UDDS Urban dynamometer driving schedule.
APS Accelerator pedal sensor.
HWFET Highway fuel economy test.
NSGA Non-dominated sorting genetic algorithm.
NEDC New European driving cycle.
NRMSE Normalized root mean square error.

I. INTRODUCTION
The associate editor coordinating the review of this manuscript and The electric vehicle (EV) market has been rapidly growing
approving it for publication was Ton Duc Do . in the automotive industry as an alternative to vehicles that
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
VOLUME 11, 2023 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 103765
 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

consume fossil fuels [1]. In an EV powertrain, a transmission components [23]. Therefore, it is necessary to consider
that converts the torque and speed of an electric motor the transmission efficiency that varies with the gear shift
into wheels generally employs a single-speed gear because to further accurately optimize the shifting pattern as well
it has a simple structure with no gear shifting, enabling as the motor efficiency [19]. In addition, optimizing both
excellent high transmission efficiency and drivability [2]. gear ratios and shifting patterns can potentially improve the
However, compared to multi-speed transmissions, single- economic and dynamic performances of EVs. In summary,
speed transmissions have several drawbacks in terms of eco- the gear-shifting pattern for EVs should be elaborately
nomic and dynamic performance [3], [4], [5]. For economic designed to improve their economic and dynamic perfor-
performance, even if a motor drives in the low powertrain mances, considering variations in the motor and transmission
efficiency region under a specific vehicle driving condition, efficiencies with gear ratio optimization.
a single-speed transmission cannot control the motor oper- When a vehicle is driven, translational and rotational
ation because the motor torque and speed are determined motions exist in various vehicle components. Based on the
by a fixed transmission gear ratio and the requested vehicle law of conservation of kinetic energy, these motions can be
speed and force [6], [7]. However, a multi-speed transmission represented by the motion of equivalent inertia, which is
allows the motor to operate in a high-efficiency region with virtual inertia, and its kinetic energy is equal to the sum of the
appropriate gear shifting depending on the vehicle’s driving kinetic energies of each vehicle component [24]. Therefore,
conditions [8]. For dynamic performance, the vehicle traction the translational motion of body mass and the rotational
force is represented by the product of the motor torque and motion of powertrain component inertias can be integrated
gear ratio, and the vehicle speed is expressed as the motor into one virtual inertia for vehicle analysis. The equivalent
speed divided by the gear ratio. Therefore, it is appropriate to inertia of the input side (motor and part of the transmission)
use high and low gear ratios to satisfy high traction force and differs according to the driving gear ratio. Assuming that the
vehicle speed, respectively. vehicle travels at a constant speed, the operating speed of the
Multi-speed transmissions are more advantageous than motor using a high or low gear ratio of the transmission is
single-speed transmissions for EV performance because high or low, respectively. As the kinetic energy is proportional
selecting the desired speed gear with an appropriate gear to the square of the rotational speed, using a high gear ratio
ratio among various speed gears is possible. The gear increases the equivalent inertia of the vehicle. This effect is
shifting allows the motor to operate in areas with high negligible as vehicle acceleration approaches zero. However,
powertrain efficiency and increases the traction force and it significantly affects the requested motor power as vehicle
available vehicle speed. Since multi-speed transmissions acceleration increases. For economic performance, up- or
are more complex than single-speed transmissions, design down-shifting can improve the motor and transmission effi-
optimization of the transmission is essential to find an ciencies by changing the motor torque and speed. However,
optimal solution that maximizes the economic and dynamic the gear shift entails an additional inertia effect on the input
performances of EVs [9]. Gear ratio optimization is essential side under acceleration conditions. Although the motor and
because the critical design parameter of a multi-speed transmission efficiencies after the gear shift are higher or
transmission is the gear ratio of each speed [10]. However, lower than the previous gear, the total energy consumption
the gear-shifting pattern controls the driving gear according rate may be reversed by considering the variation in inertial
to the vehicle driving conditions; therefore, optimizing the energy due to vehicle acceleration. For dynamic performance,
shifting pattern is also significant. Several previous studies using a high gear ratio can multiply the input torque more
have proposed shifting patterns for multi-speed EVs. than that when using a low gear ratio. Although a significant
These studies revealed the following: First, the shifting increase in the input torque ensures a high acceleration of
pattern generally focuses on economic performance, and the vehicle, a high acceleration also involves a large inertia
the optimal pattern is determined primarily to maximize resistance. Therefore, using a high gear ratio increases the
only motor efficiency [11], [12], [13], [14], [15], [16]. equivalent inertia of the input side, which may result in
Second, some studies have conducted gear ratio optimization negative effects on vehicle acceleration. In summary, the
simultaneously because the optimal shifting pattern depends equivalent inertia variation in the speed gear selection process
on the combination of gear ratios [10], [17], [18], [19], [20]. should be considered to maximize the economic and dynamic
Thus, to effectively optimize the shifting pattern, economic performances of EVs.
and dynamic performances should be considered in the Previous studies have determined the gear-shifting pattern
shifting pattern optimization since both are equally important. based on static characteristics, such as constant input
Some studies have proposed a shifting pattern that focuses torque and speed. However, to derive a further optimal
on both performances [10], [17], [18], [21], [22]. However, shifting pattern, it should be expressed by including the
these studies have practical limitations because the shifting dynamic characteristic of the equivalent inertia variation.
patterns differ for each performance rather than the integrated Therefore, the shifting pattern should be optimized accord-
pattern. Moreover, increased speed gears should reduce ing to the degree of vehicle acceleration to reflect the
the transmission efficiency because of additional shifting inertia variation effect for each speed gear. In addition,

103766 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

the optimal shifting pattern and gear ratios can ensure

excellent economic and dynamic performances. However,
their enhancement is challenging because of the trade-off
relationship meaning that optimizing the shifting pattern
and gear ratios for economic performance can diminish the
dynamic performance [10], [19]. Therefore, a multi-objective
optimization method can be an alternative to this problem by
providing a Pareto front consisting of diverse optimal design
solutions [25].
To overcome the aforementioned research limitations on
the transmission optimization of EVs, this study proposes
an effective optimization method of multi-speed transmission
for EVs considering the effect of equivalent inertia variation
on the gear-shifting pattern. A reference powertrain con-
figuration of a two-speed transmission EV was introduced
to analyze the equivalent inertia for each speed. Based
on this, the equivalent inertia was mathematically derived
and an EV analysis model was developed to evaluate the
economic and dynamic performances of EVs. To verify the
importance of the inertia variation effect, a dynamic inertia
efficiency, which quantifies this inertia variation effect,
was mathematically derived. The optimal shifting patterns
with dynamic inertia efficiency and nothing were obtained,
and the performance results were compared. To address
the trade-off relationship between economic and dynamic
FIGURE 1. Power flows of DCT for each driving gear.
performance, a multi-objective optimization problem was
formulated that included the design variables as gear ratios
and shifting patterns and the objective functions as energy
efficiency and acceleration ability. As an alternative to the 3) The different optimization results of the gear ratios
excessive calculation burden needed for performing the and shifting patterns according to the equivalent
multi-objective optimization, this study utilized an artificial inertia variation showed better objective values of the
neural network (ANN) to predict the relationship between optimal design based on the inertia variation effect.
the design variables and objective functions and implemented Therefore, it verifies the performance superiority of the
an adaptive sampling method to minimize the samples to transmission design considering the equivalent inertia
construct the ANN model. The optimization results presented variation due to the gear ratios and shifting.
different Pareto fronts as the optimal solutions based on the
shifting pattern from the dynamic inertia efficiency. These II. ANALYSIS OF MULTI-SPEED EV
optimization results demonstrate the necessity of considering An EV model employing multi-speed transmission was
the inertia variation effect in the design of gear-shifting developed to analyze the economic and dynamic perfor-
patterns. The contributions of this study can be summarized mances. Previous studies on multi-speed EVs have proposed
as follows: various powertrain configurations. In the gear-shifting pro-
1) By referring to the limitations of previous studies on cess, gear-shifting events generally occur more when vehicle
the multi-speed transmission optimization of EVs, this acceleration is considered than when not considered. Since
study conducted an integrated optimization of gear frequent gear-shifting causes poor drivability, the multi-speed
ratios and shifting patterns considering both economic transmission structure should address this shortcoming
and dynamic performances and variable efficiencies of appropriately. Therefore, this study adopted a dual-clutch
the motor and transmission. transmission (DCT) that exhibits high transmission efficiency
2) Unlike the previous studies on shifting pattern opti- and stable drivability [26].
mization, this paper proposed an optimization process The equivalent inertia for each driving speed gear can be
of the shifting pattern, including inertia variation for derived from the powertrain configuration equipped with the
each speed, to confirm the importance of the equiv- DCT. To analyze the equivalent inertia of EV, the reference
alent inertia effect on energy efficiency and vehicle inertia component is determined. Several studies have calcu-
acceleration. The comparison results of economic and lated the equivalent inertia by referring to the wheel, which
dynamic performances in consideration of dynamic converts the rotational motion to the translational motion of a
inertia efficiency or none quantitatively exhibited the vehicle. Figure 1 shows the power flow of each driving-speed
effect of inertia variation on EV performances. gear. Although the transmission consists of various rotational

VOLUME 11, 2023 103767

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

TABLE 1. Inertia values of each component. where ρ is the air density, cd is the drag coefficient, Af is
the frontal area, g is the acceleration owing to gravity, µr is
the rolling resistance coefficient, θ is the road slope, and Tbrk
is the braking torque. Tbrk can be classified into mechanical
(Tmec ) and regenerative torques, as follows:

Tbrk = Tmec − Tm ri rf · ηt (6)

Here, Tmec is generated by the mechanical brake, and Tm is the

negative value determined by the regenerative braking torque
distribution, as mentioned in [28].
Energy efficiency and acceleration ability were adopted as
quantitative measures to evaluate the economic and dynamic
performances of the EVs. For energy efficiency, many
components, such as clutches, gears, bearings, and shafts, this vehicle manufacturers have presented official efficiency
study considers clutches and gears to calculate the equivalent values based on various standard driving cycles, such as
inertia for the transmission because the inertia values of other the urban dynamometer driving schedule (UDDS), highway
components are negligible compared to those of clutches and fuel economy test (HWFET), new European driving cycle
gears. The inertia values of each component are summarized (NEDC), and world harmonized light-duty vehicles test
in Table 1 and determined from the component specifications procedure (WLTP). The energy efficiency of an EV (EFF)
introduced in [27]. Considering motor, transmission, and means the distance that can be driven using 1 kWh of energy
vehicle inertias, the equivalent inertia at the wheel driven by and it is calculated as follows:
the first-speed gear (Jeq.1 ) can be expressed as follows: dcyc
EFF = (7)

2
Jeq.1 = Jm + Jc1 + Jp1 · r1 rf + Jg1 + Jg2 + Jpf · rf2


Cbat · dSOC

r1 rf 2 where dcyc is the driving cycle distance, Cbat is the battery

 
+ Jp2 · + Jgf + MR2t (1) capacity, and dSOC is the state of charge (SOC) consumption.
r2
The dSOC is calculated from the analysis of the battery
where M is the mass of the vehicle, and Rt is the effective tire equivalent circuit. The battery voltage (Vbat ) was derived
radius. Likewise, the equivalent inertia at wheel driven by the using the circuit voltage equation, as follows:
second-speed gear (Jeq.2 ) can be expressed as follows:

2 Vbat = VOCV − Ri Ibat (8)
Jeq.2 = Jm + Jc2 + Jp2 · r2 rf + Jg1 + Jg2 + Jpf · rf2



where VOCV is the open-circuit voltage, Ri is the internal
r2 rf 2
 
+ Jp1 · + Jgf + MR2t (2) battery resistance, and Ibat is the battery current. Here, Ibat
r1 is determined by the battery charging and discharging con-
Using the equivalent inertia (Jeq ), vehicle acceleration (a) ditions using the mechanical-electrical power relationship,
and speed (v) are determined as follows: as follows:
Tm ωm

Twhl Rt
a=
 if Tm ≥ 0 (discharging)
 ηm ηi Vbat


J 
Z eq Ibat = (9)
v = a dt (3) η η T ω

m i m m
if Tm < 0 (charging)




Vbat
where Twhl is the wheel torque, represented by the driving
(Tdrv ) and resistance (Tres ) torques (Twhl = Tdrv − Tres ). Tdrv where ωm is the motor speed, and ηm and ηi are the motor
can be expressed as follows: and inverter efficiencies, respectively. dSOC is calculated by
integrating Ibat as follows:
Tdrv = Tm ri rf · ηt (4) R
VOCV · Ibat dt
where Tm is the motor output torque, ri is the speed gear dSOC = (10)
Cbat
ratio (r1 or r2 ), and ηt is the transmission efficiency. Here,
ηt is determined by considering the losses of the clutch, gear, The EV analysis model was built by MATLAB/Simulink,
bearing, and concentric shaft in a DCT [19]. Tres can be as shown in Figure 2.
expressed as follows: For dynamic performance, the 0−100 km/h acceleration
  time, maximum speed, and ascendable gradient have been
1
Tres = Rt ρcd Af v + Mg (µr cos θ + sin θ) + Tbrk (5)
2 generally employed as evaluation measures in several studies.
2 These measures can be determined under different vehicle

103768 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

TABLE 2. Specifications of reference EV.

FIGURE 2. 2-speed EV analysis model.

speed conditions. Under a wide-open throttle (WOT) con-

dition, the 0−100 km/h acceleration time (ta ) satisfies the
following equation:
Z ta
v= a dt = 100 km/h (11)
0
Next, the maximum speed (vmax ) can be theoretically
determined as follows:
ωmax
vmax = Rt (12)
r2 rf
where ωmax denotes the maximum motor speed. However,
the driving torque (Tdrv ) must be greater than the resistance
torque (Tres ) at this speed. Otherwise, vmax is determined
at a speed that satisfies the condition Tdrv > Tres . Lastly,
because the ascendable gradient (θgrad ) considers the drivable
condition (v > 0), assuming that the vehicle speed is zero,
it can be determined as a gradient value equalizing Tdrv and
Tres . Therefore, θgrad satisfies the following equation:
Tmax (0) · r1 rf = Rt Mg µr cos θgrad + sin θgrad


(13)
where Tmax is the maximum motor torque, which varies
with the motor speed. Here, if θgrad is large, µr cos θgrad is
negligible. Therefore, θgrad can be expressed as:
 
Tmax (0) · r1 rf
θgrad = sin−1 (14)
Rt Mg
The significant parameter values of a reference EV are
summarized in Table 2.

III. EFFECT OF DYNAMIC INERTIA EFFICIENCY ON

OPTIMAL SHIFTING PATTERN
The economic and dynamic performances vary considerably
FIGURE 3. Powertrain efficiency (ηm ηt ) maps for each driving gear.
depending on the gear ratio used under certain driving
conditions. Therefore, the design of the gear-shifting pattern
is as important as that of the gear ratios. The gear-shifting
pattern determines the speed gear according to the accelerator acceleration, the shifting pattern is generally determined
pedal sensor (APS) and vehicle speed values, which represent by considering the optimal up-shift conditions, and the
the driver’s request and vehicle status, respectively. In the down-shift is determined by offsetting the vehicle speed on
two-speed transmission, the shifting pattern represents an up- the up-shift to avoid frequent gear shifts [21].
or down-shift, meaning the first to the second gear shift, With respect to economic performance, a gear shift
or vice versa. Since the up-shift occurs mainly under vehicle should be performed to minimize the required motor power

VOLUME 11, 2023 103769

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

FIGURE 4. Optimal shifting pattern for economic performance.

FIGURE 5. Optimal shifting pattern for dynamic performance.

under the same wheel power, represented as the product 0% to 100%, and the gear-shifting speeds were determined
of wheel torque and speed, which maximizes the motor from the cross-torque points, as shown in Figure 5. From
and transmission efficiencies. Although the motor efficiency the mentioned optimal shift pattern determination methods
(ηm ) is determined only by the input torque and speed, for economic and dynamic performances, they utilize only
the transmission efficiency (ηt ) additionally varies with the the static characteristics of efficiency and output torque to
driving gear. Therefore, the powertrain efficiency expressed determine the gear-shifting pattern. Although many previous
as the product of the motor and transmission efficiencies studies have used these methods, they are not the best under
(ηm ηt ) is different for each driving gear, as shown in dynamic conditions such as vehicle acceleration. To obtain
Figure 3. These efficiency maps can be converted equally into a more optimized shifting pattern for EVs, the static and
efficiency values according to the APS and vehicle speed, dynamic characteristics should be considered together in the
and an optimal shifting line can be determined by selecting optimal shifting pattern process as follows.
a gear in a more efficient area when driven with each gear, When the vehicle is accelerating, the acceleration resis-
as shown in Figure 4. Based on this shifting line, the optimal tance torque on the wheel can be calculated as follows:
shifting pattern can be determined to maximize the economic
Tacc = Jeq · αwhl (15)
performance of the EV.
With respect to dynamic performance, a gear shift should where αwhl denotes the angular acceleration of the wheel.
be performed to maximize the output torque according to From the equivalent inertia of first- and second-gear sides
the vehicle speed. Therefore, the driving torques (Tdrv ) of mentioned in Eqs. (1) and (2), respectively, the equivalent
each speed gear were calculated for the APS values from inertia variation (Jvar ) caused by the gear shifting between

103770 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

the first and second gear drives can be derived by subtracting where ηin is the dynamic inertia efficiency, which is the
Jeq.2 from Jeq.1 as follows: difference in efficiency when shifting from the first-speed
   gear to the second. Since ηin is the relative value of the first
Jvar = Jm r12 − r22 + Jc1 r12 − Jc2 r22 + Jp1 r12 − Jp2 r22 gear when driving in the second gear, as expressed in Eq. (17),
# it is subtracted from fE (P) when only the first-speed gear is
Jp1 r12 Jp2 r22 2 driven. Therefore, the objective function fE (P) can be revised
+ 2 − 2 rf (16)
r2 r1 by considering the inertia variation effect, as follows:
When the gear ratios (r1 , r2 ) are fixed, Jvar is determined

 ηm ηt (APS, v) − ηin (αwhl ) if ri = r1

as a constant value. In addition, a large difference in
fE (P) = (21)
gear ratios also increases the inertia variation, indicating
ηm ηt (APS, v)

if ri = r2

that the equivalent inertia depending on the gear shift
significantly affects vehicle performance. Therefore, the For the value of fE , considering the dynamic inertia efficiency
method proposed in this paper considers the motor and under acceleration conditions (fE = ηm ηt − ηin ) is
transmission efficiencies for the optimal gear-shifting pattern, necessarily lower than the value of considering only the static
as well as the inertia efficiency, which indicates the requested characteristics (fE = ηm ηt ) when driving with the first-speed
torque variation from the inertia effect by the gear shift as gear, which means that the up-shifting is performed at a lower
follows. speed than considering only the static characteristics.
For economic performance, the shifting pattern optimiza- For dynamic performance, the shifting pattern optimiza-
tion problem considering only the static characteristic can tion problem considering only the static characteristic can be
be formulated to maximize the powertrain efficiency (ηm ηt ) formulated to maximize the driving torque (Tdrv ) according
according to the APS and vehicle speed (v) as follows: to the APS and vehicle speed (v) as follows:
maximize fE (P) = ηm ηt (APS, v)
P maximize fP (P) = Tdrv (APS, v)
h i P
where P = viopt , · · · , vN
opt , i = 1, · · · , N (17) h i
where P = viopt , · · · , vnopt , i = 1, · · · , N (22)
where P is the shifting speed set, vopt is the optimal shifting
speed, and superscript i is the total number of APS cases From this optimization problem, the optimal P can be easily
(0−100%). Here, P is for the up-shifting process, and the obtained by combining the values of Tdrv for each speed gear.
shifting speeds for the down-shifting process are determined Using the first gear during low-speed driving is generally
by offsetting from the speed values of P [21]. The shifting advantageous in vehicle traction because a high motor torque
pattern optimization that considers dynamic characteristics is multiplied by a large gear ratio. However, an evaluation
can be defined by including the acceleration resistance torque of dynamic performance should be conducted by considering
variation effect to Eq. (17). the inertia variation effect because the acceleration that
The motor torque (Tm ) by APS is expressed as follows: occurs under dynamic performance is greater than that in
the economic performance evaluation. Therefore, when the
Tm = Tmax (ωm ) · APS (18) acceleration is large, even at a low speed, the gear up-
Tmax is the variable value according to the motor speed (ωm ) shift (first to second) may be advantageous because it can
and can be determined by the vehicle speed and driving gear significantly reduce the inertia resistance torque.
as follows: Unlike maximizing the efficiency for economic perfor-
v mance, the analysis of dynamic performance, such as accel-
ωm = · ri rf (19)
Rt eration ability, should consider the time-variant behavior of
Therefore, Tm can be determined if APS and v are provided. the vehicle. Therefore, it is advisable to directly evaluate the
Because of r1 > r2 in the gear ratio design, from Eqs. (1) acceleration ability through simulations using the developed
and (2), Jeq.1 is always greater than Jeq.2 . Hence, although EV model. To consider Tdrv with the inertia effect for the
driving with the first gear accelerated is more advantageous optimal gear-shifting pattern, the optimization problem can
than driving with the second gear because a large gear ratio be expressed using Eq. (11) as follows:
can significantly multiply the motor torque transmitted to the minimize fP (P) = ta
wheel, in contrast, it is disadvantageous in terms of inertia P
Z ta
resistance. To compare the inertia effect with the motor and
subject to a dt = vtar (23)
transmission efficiency, the inertia resistance torque can be 0
converted into an expression in a unit of efficiency expression
where vtar is the speed of the target vehicle. In several
using the ratio of the resistance torque to the motor torque
previous studies, vtar was adopted at 100 km/h, and the
(Tm ) as follows:
acceleration time (ta ) was evaluated under WOT conditions.
Jvar αwhl
ηin = (20) Therefore, Eq. (23) is acceptable for obtaining an optimal
r2 rf Tm gear-shifting speed at an APS of 100%.

VOLUME 11, 2023 103771

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

FIGURE 6. Example of APS values under the WLTP driving cycle.

Most APS values did not exceed 60% in the standard

driving cycle, as shown in Figure 6. Therefore, to obtain an
optimal shifting pattern that considers economic and dynamic
performance, the optimal shifting speeds can be determined
by focusing on the economic and dynamic performance at low
and high APS values, respectively. Optimization problem that
integrates Eqs. (21) and (23) can be formulated as follows:

 maximize fE (P) if APS ≤ 60%
P= P (24)
 minimize fP (P) if APS = 100%
P

Here, optimal shifting speeds between 60% and 100% APS

hwere obtained iusing the interpolation method. Since P =
viopt , · · · , vnopt must consist of ascending speeds, along
with an increase in APS because of the drivability problem,
it should satisfy the following constraint:
FIGURE 7. Example of optimal shifting patterns according to wheel
acceleration (r1 = 3.321, r2 = 1.681).
vi−1
opt ≤ viopt , i = N, · · · , 2 (25)

An example of optimal shifting patterns based on Eqs. (24) TABLE 3. Comparison of results for economic and dynamic performances.
and (25) considering dynamic inertia efficiency, is shown
in Figure 7. As the wheel acceleration (αwhl ) increases, the
optimal shifting lines gradually move toward the low-speed
side. Since a high αwhl involves a large inertia resistance when
driven by the first gear, the up-shifting is performed rapidly
with increasing vehicle speed to reduce the inertia resistance
even in areas where the powertrain efficiency (ηm ηt ) of the
second gear is lower than that of the first gear. In addition, the
shifting speed of the WOT condition (100% APS) is constant and acceleration time (ta ) for the economic and dynamic
regardless of αwhl because this speed already considers the performances of EVs, respectively.
inertia variation effect at full acceleration based on Eq. (23).
Table 3 compares the results of applying the optimal shifting IV. OPTIMIZATION RESULTS OF TWO-SPEED
pattern considering dynamic inertia efficiency with those TRANSMISSION
not considering it. These results demonstrate the importance To maximize the economic and dynamic performances of
of the inertia variation effect of gear-shifting in obtaining EVs, it is appropriate to utilize multi-objective optimization
optimal shifting patterns for the energy efficiency (EFF) methods that can address the trade-offs between performance

103772 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

FIGURE 8. ANN-based multi-objective optimization process.

objectives. The critical design parameters for optimizing a In addition, to solve the multi-objective optimization problem
two-speed transmission of an EV are the gear ratios of each in Eq. (26), this study employs a non-dominated sorting
speed and shifting pattern. In particular, since the optimal genetic algorithm (NSGA) [31], which has been used for
shifting patterns for each combination of gear ratios are various multi-objective optimization problems in the engi-
different, they should depend on the gear ratios. Therefore, neering field. Here, a genetic algorithm in the NSGA requires
the multi-objective optimization problem is formulated by numerous calculations of objective functions compared to
employing the objectives of EFF and ta as follows: gradient-based algorithms. Therefore, an ANN-based multi-
objective optimization method is used as an alternative to
minimize f¯ (−EFF, ta ) the excessive calculation problem [32]. The ANN algorithm
r,P
subject to r ∈ c (26) trains the given sample results to derive the relationship
between the input (design variable) and output (target) [33]
where r = [r1 , r2 ] is the vector of gear ratios and c and builds a predictive model which estimates the objective
is the feasible region of gear ratios from the dynamic function values from the given design variable values of
constraints. Here, the dynamic constraints are adopted with transmission without analyzing the EV model. Therefore,
a maximum speed (150 km/h) and an ascendable gradient it can significantly decrease the calculation effort owing to
(40%), as determined by Eqs. (12) and (14), respectively. the excessively iterative calculations in the multi-objective
In addition, because a large step ratio (r1 /r2 ) results in optimization process using the NSGA.
shifting difficulties [29], the step ratio was limited to three The overall optimization process is illustrated in Figure 8.
or less. In the objective function, EFF is expressed as a First, the sample points are selected in a feasible region of the
negative value (−EFF) because the optimization problem gear ratios, and the optimal shifting pattern of each sample
is formulated to minimize the objective function values, is determined by Eq. (24). Next, the EV performances are
although EFF should be maximized for better economic analyzed using the gear ratio samples and optimal shifting
performance. patterns, and the ANN model is constructed using the analysis
The results of multi-objective optimization are displayed as results of the samples. Here, the cross-validation method
the Pareto front, which is a set of optimal solutions. Because is utilized to validate the ANN model accuracy, and if
the multi-objective optimization algorithm should update the constructed ANN model does not satisfy an acceptable
many solutions simultaneously, performing this process normalized root mean square error (NRMSE), the new
requires significant computational effort [30]. Therefore, sample points are added to the previous sample points and
solving multi-objective optimization problems is more chal- the ANN model is reconstructed. After the construction of the
lenging compared to single-objective optimization problems. ANN model satisfying the criteria of NRMSE is complete,

VOLUME 11, 2023 103773

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

FIGURE 9. Comparison of Pareto fronts for each cycle.

FIGURE 10. Comparison of the optimum gear ratios for each cycle.

a multi-objective optimization is finally performed using the multi-objective optimizations were conducted in two cases:
NSGA and the ANN model to obtain a Pareto front, which i) considering the dynamic inertia efficiency and ii) not con-
means a set of optimal solutions. sidering it. In addition, three cases (WLTP, UDDS+HWFET,
To confirm the inertia variation effect on the optimal and NEDC) were used for the driving cycles for economic
gear-shifting pattern for economic and dynamic performance, performance evaluation to validate the effectiveness of

103774 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

shifting patterns considering dynamic inertia efficiency. The

Pareto fronts of the two cases were compared from the
optimization results for each driving cycle, as shown in
Figure 9. These Pareto fronts depict the trade-off between
economic and dynamic performances. For all driving cycles,
the values of the optimum solutions show significant
differences when the inertia variation effect is considered.
Applying the optimal shifting pattern considering inertia
variation substantially outperforms the energy efficiency
(EFF) and acceleration time (ta ) compared to the case in
which it does not. Therefore, these results demonstrate that
the dynamic inertia efficiency should be reflected in the
optimization of gear-shifting patterns.
Figure 10 illustrates the optimum gear ratios for each
Pareto front, as shown in Figure 9. These results indicate
the following. First, combining low and high gear ratios
provides economic and dynamic performance advantages.
A low gear ratio is generally beneficial for economic
performance owing to its small equivalent inertia. However,
the optimum gear ratios in Figures 10 (b) and (c) are not
in the lowest gear ratio area. Therefore, while a low gear
ratio ensures low equivalent inertia, it is more economically
dominant in UDDS+HWFET and NEDC to improve the
overall efficiency by changing the motor operating points,
depending on the gear ratio and shifting pattern. For dynamic
performance, a high gear ratio is generally helpful because
it can significantly multiply the motor torque transferred to
the wheels. However, the optimum gear ratios in Figure 10
are not in the highest gear ratio area. Since a high gear ratio
exhibits high wheel torque and high acceleration resistance
owing to the large equivalent inertia, considering only the
wheel torque is not appropriate for the vehicle’s acceleration
ability.
Next, the distributions of the optimum gear ratios vary
significantly depending on the optimal shifting pattern, FIGURE 11. Optimal shifting patterns for economic performance (A:
inertia effect, B: none).
considering or without considering the dynamic inertia
efficiency. In particular, in Figures 10 (a) and (b), the
optimum gear ratios derived by the optimal shifting pattern
without considering the inertia effect are more distributed in the Case 2 pattern, the EFF and ta values indicated the
the low gear ratio area than in the pattern considering the opposite tendency.
inertia effect, whereas the optimum gear ratios considering Figure 11 shows the optimal shifting patterns for the best
the effect are more distributed in the high gear ratio area solutions for EFF in Table 4. Solutions A and B have the
than not considering the effect. This means that which same second gear ratio (0.863), whereas the first gear ratio
shifting pattern is applied affects the EV performance and of solution A (2.139) is greater than that of solution B
optimum gear ratios. For the WLTP case in Figure 10 (a), (1.649). Due to the large step ratio of solution A, it is
Table 4 compares the results of the best solutions for the possible to change the motor operating points more than
economic and dynamic performances between the cases that of solution B by shifting the speed gear to improve
with and without the inertia effect. To confirm the effect the powertrain efficiency (ηm ηt ). However, because the
of the optimal shifting patterns on the performance values equivalent inertia of the first gear side of solution A is larger
when using different optimal shifting patterns that consider than that of solution B, it negatively affects the acceleration
(Case 1) and do not consider (Case 2) the inertia effect, the resistance of solution A, as expressed in Eq. (15). When
performance values of each combination of gear ratios are comparing the optimal shifting patterns in Figure 11, the
evaluated. For economic and dynamic performances, when shifting speeds of solution A are more distributed on the
applying the Case 1 pattern, the EFF and ta values of the left side (low speed) at 60 km/h than those of solution B.
gear ratios considering the inertia effect were superior to the This means that the second-gear driving in solution A is
values not considering the effect. However, when applying more frequent than that in solution B. Since it reduces the

VOLUME 11, 2023 103775

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

TABLE 4. Comparison of the best gear ratio solution results based on optimal shifting pattern in WLTP (Case 1: considering inertia effect, Case 2: not
considering.)

acceleration with a positive effect is greater than the negative

effect of Tdrv because ta of solution C is faster than that of
solution D. This demonstrates that the inertia effect on the
acceleration ability is greater than that on the static driving
torque. These results are particularly remarkable in designs
with large gear and step ratios. In summary, the dynamic
inertia efficiency, quantifying the equivalent inertia variation,
should be considered in the design of the gear ratios and
shifting patterns for the performance of EVs.

V. CONCLUSION
This paper proposes an effective optimization method
for the multi-speed transmission of EVs considering the
dynamic inertia efficiency for gear-shifting patterns. The
equivalent inertia for each speed was analyzed from the
power flow of each driving speed gear to confirm the
FIGURE 12. Comparison of driving torque for dynamic performance equivalent inertia variation. An EV analysis model with
(C: inertia effect, D: none).
variable motor and transmission efficiencies was constructed,
and the quantitative performance criteria were determined
to evaluate the economic and dynamic performances. From
acceleration resistance by shifting the speed gear at lower the EV analysis, the performance results based on the
speeds, solution A is superior to solution B in terms of EFF, optimal shifting pattern, considering the inertia effect, were
despite the larger equivalent inertia of the first gear side. compared. These results demonstrate the importance of
Figure 12 shows the driving torque at WOT condition in dynamic inertia efficiency, quantifying the inertia variation
the best solutions for ta in Table 4. Solutions C and D have effect, by gear-shifting in obtaining optimal shifting patterns
the same first-gear ratio (3.894), whereas the second-gear for the energy efficiency and acceleration time for economic
ratio of solution C (1.636) is greater than that of solution D and dynamic performances of EVs, respectively.
(1.298). Since the equivalent inertia of the second gear side Due to the trade-off relationship between economic
of solution C is larger than that of solution D, it has a and dynamic performances, a multi-objective optimization
negative effect on the acceleration ability of solution C. problem, including the design variables such as gear ratios
When comparing the optimal shifting speeds, solution C and shifting patterns and the objective functions of energy
performs gear-shifting at a speed lower than solution D. efficiency and acceleration time, was formulated. As an
In particular, the up-shifting of solution C is conducted before alternative to the excessive calculation burden needed for
the driving torque of the first gear reaches the maximum performing the multi-objective optimization using NSGA,
driving torque of the second gear. Therefore, the total driving the ANN-based multi-objective optimization method was
torque in the red-colored area is insufficient compared to that utilized. In particular, to confirm the inertia variation effect
of solution D. However, although the Tdrv of solution C is on the optimal gear-shifting pattern, optimizations were
noticeably lower than that of solution D in the red-colored performed by dividing the optimal shifting patterns consid-
area, ta of solution C is superior to that of solution D. ering the dynamic inertia efficiency and not considering it.
For the acceleration ability, Tdrv of solution C is a negative In addition, various driving cycles (WLTP, UDDS+HWFET,
effect. In contrast, since the shifting speed of solution is and NEDC) were employed for economic performance
significantly faster than that of solution D, the equivalent evaluation to determine the effectiveness of shifting patterns
inertia of solution C is lower than that of solution D after considering equivalent inertia variation.
gear shifting speed. It is advantageous for the acceleration The optimization results demonstrate the importance of
ability and indicates that the equivalent inertia effect on the considering the dynamic inertia efficiency for the optimal

103776 VOLUME 11, 2023

K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

shifting pattern. In each driving cycle, compared to the [10] P. Wu, P. Qiang, T. Pan, and H. Zang, ‘‘Multi-objective optimization of
shifting pattern without the inertia variation effect, economic gear ratios of a seamless three-speed automated manual transmission for
electric vehicles considering shift performance,’’ Energies, vol. 15, no. 11,
and dynamic performances were improved from 2.7% to p. 4149, Jun. 2022.
7.8% and 2.8% to 3.0%, respectively. In addition, the gear [11] P. D. Walker, S. Abdul Rahman, B. Zhu, and N. Zhang, ‘‘Modelling, sim-
ratio values of the optimum solutions were obtained differ- ulations, and optimisation of electric vehicles for analysis of transmission
ratio selection,’’ Adv. Mech. Eng., vol. 5, Jan. 2013, Art. no. 340435.
ently when the dynamic inertia efficiency was considered
[12] T. Liu, Y. Zou, and D. Liu, ‘‘Energy management for battery electric
and not. Applying the optimal shifting pattern considering vehicle with automated mechanical transmission,’’ Int. J. Vehicle Des.,
dynamic inertia efficiency substantially outperforms the vol. 70, no. 1, pp. 98–112, Apr. 2016.
energy efficiency (EFF) and acceleration time (ta ) compared [13] S. Tan, J. Yang, X. Zhao, T. Hai, and W. Zhang, ‘‘Gear ratio
optimization of a multi-speed transmission for electric dump truck
with the case in which it does not. Therefore, these operating on the structure route,’’ Energies, vol. 11, no. 6, p. 1324,
results demonstrate that the equivalent inertia variation effect May 2018.
should be reflected in the optimization of gear-shifting [14] J. Ruan, P. D. Walker, J. Wu, N. Zhang, and B. Zhang, ‘‘Development of
continuously variable transmission and multi-speed dual-clutch transmis-
patterns. In conclusion, considering the dynamic inertia sion for pure electric vehicle,’’ Adv. Mech. Eng., vol. 10, no. 2, Feb. 2018,
efficiency from gear ratios and shifting can achieve superior Art. no. 168781401875822.
performance in the transmission design of EVs. In future [15] C. Lin, M. Zhao, H. Pan, and J. Yi, ‘‘Blending gear shift strategy design
work, we will demonstrate the effectiveness of this study by and comparison study for a battery electric city bus with AMT,’’ Energy,
vol. 185, pp. 1–14, Oct. 2019.
validating before and after optimization results through the [16] K. Zhao, Z. Li, Y. Chen, and Y. Liu, ‘‘Multi-objective optimization
experimental results. In addition, because the optimization of gearshift trajectory planning for multi-speed electric vehicles,’’ Eng.
results for each driving cycle are different, presenting the best Optim., vol. 54, no. 6, pp. 949–972, Jun. 2022.
[17] Y. Li, B. Zhu, N. Zhang, H. Peng, and Y. Chen, ‘‘Parameters optimization
solution considering the real-driving conditions is limited in
of two-speed powertrain of electric vehicle based on genetic algorithm,’’
this study. Therefore, we will study the method to represent Adv. Mech. Eng., vol. 12, no. 1, Jan. 2020, Art. no. 168781402090165.
the best optimal solution under various driving conditions [18] G.-B. Sun, Y.-J. Chiu, W.-Y. Zuo, S. Zhou, J.-C. Gan, and Y. Li,
and conduct the experiment on the presented solutions on a ‘‘Transmission ratio optimization of two-speed gearbox in battery electric
passenger vehicles,’’ Adv. Mech. Eng., vol. 13, no. 6, Jun. 2021,
real-world system with variable parameters. Art. no. 168781402110228.
[19] K. Kwon, J. Jo, and S. Min, ‘‘Multi-objective gear ratio and shifting
ACKNOWLEDGMENT pattern optimization of multi-speed transmissions for electric vehicles
considering variable transmission efficiency,’’ Energy, vol. 236, Dec. 2021,
(Kijong Park and Sang-Kil Lim contributed equally to this Art. no. 121419.
work.) [20] K. Kwon, J.-H. Lee, and S.-K. Lim, ‘‘Optimization of multi-speed
transmission for electric vehicles based on electrical and mechanical
efficiency analysis,’’ Appl. Energy, vol. 342, Jul. 2023, Art. no. 121203.
REFERENCES
[21] B. Zhu, N. Zhang, P. Walker, X. Zhou, W. Zhan, Y. Wei, and N. Ke,
[1] S. R. Khasim, C. Dhanamjayulu, and S. M. Muyeen, ‘‘A single inductor ‘‘Gear shift schedule design for multi-speed pure electric vehicles,’’ Proc.
multi-port power converter for electric vehicle applications,’’ IEEE Access, Inst. Mech. Eng., D, J. Automobile Eng., vol. 229, no. 1, pp. 70–82,
vol. 11, pp. 3367–3385, 2023. Jan. 2015.
[2] B. Gao, D. Meng, W. Shi, W. Cai, S. Dong, Y. Zhang, and [22] X. Lin, Y. Li, and B. Xia, ‘‘An online driver behavior adaptive shift
H. Chen, ‘‘Topology optimization and the evolution trends of two-speed strategy for two-speed AMT electric vehicle based on dynamic corrected
transmission of EVs,’’ Renew. Sustain. Energy Rev., vol. 161, Jun. 2022, factor,’’ Sustain. Energy Technol. Assessments, vol. 48, Dec. 2021,
Art. no. 112390. Art. no. 101598.
[3] J. Wu, X. Hong, G. Feng, and Y. Zhang, ‘‘Seamless mode shift control
[23] J. Ruan, P. D. Walker, N. Zhang, and J. Wu, ‘‘An investigation of hybrid
for a new Simpson planetary gearset based dual motor powertrain
energy storage system in multi-speed electric vehicle,’’ Energy, vol. 140,
in electric vehicles,’’ Mechanism Mach. Theory, vol. 178, Dec. 2022,
pp. 291–306, Dec. 2017.
Art. no. 105056.
[24] K. Kwon, J. Lee, and S. Min, ‘‘Motor and transmission multiobjective
[4] Y. Liu, D. Gao, K. Zhai, Q. Huang, Z. Chen, and Y. Zhang, ‘‘Coordinated
optimum design for tracked hybrid electric vehicles considering equivalent
control strategy for braking and shifting for electric vehicle with two-
inertia of track system,’’ IEEE Trans. Transport. Electrific., vol. 7, no. 4,
speed automatic transmission,’’ eTransportation, vol. 13, Aug. 2022,
pp. 3110–3123, Dec. 2021.
Art. no. 100188.
[5] Y. Liu, Y. Chen, Z. Li, K. Zhao, and Z. Lin, ‘‘Multi-objective optimal [25] L. F. Amorim, A. P. de Paiva, P. P. Balestrassi, and J. R. Ferreira, ‘‘Multi-
gearshift control for multispeed transmission electric vehicles,’’ IEEE objective optimization algorithm for analysis of hardened steel turning
Access, vol. 8, pp. 129785–129798, 2020. manufacturing process,’’ Appl. Math. Model., vol. 106, pp. 822–843,
[6] F. A. Machado, P. J. Kollmeyer, D. G. Barroso, and A. Emadi, ‘‘Multi- Jun. 2022.
speed gearboxes for battery electric vehicles: Current status and future [26] Z. Lei, D. Sun, Y. Liu, D. Qin, Y. Zhang, Y. Yang, and L. Chen, ‘‘Analysis
trends,’’ IEEE Open J. Veh. Technol., vol. 2, pp. 419–435, 2021. and coordinated control of mode transition and shifting for a full hybrid
[7] Y. Liu, J. Xie, D. Qin, Y. Zhang, Z. Chen, G. Li, and Y. Zhang, ‘‘Design, electric vehicle based on dual clutch transmissions,’’ Mechanism Mach.
control, and validation of two-speed clutchless automatic transmission Theory, vol. 114, pp. 125–140, Aug. 2017.
for electric vehicle,’’ IEEE/ASME Trans. Mechatronics, vol. 27, no. 3, [27] X. Zhou, P. Walker, N. Zhang, B. Zhu, and J. Ruan, ‘‘Numerical
pp. 1299–1310, Jun. 2022. and experimental investigation of drag torque in a two-speed dual
[8] P. Dong, J. Zhao, X. Xu, Y. Liu, S. Wang, H. Huang, R. Wang, clutch transmission,’’ Mechanism Mach. Theory, vol. 79, pp. 46–63,
L. Zheng, and Z. Zhou, ‘‘Performance comparison of series–parallel Sep. 2014.
hybrid transmissions with multiple gears and modes based on effi- [28] L. Li, X. Wang, R. Xiong, K. He, and X. Li, ‘‘AMT down-
ciency model,’’ Energy Convers. Manage., vol. 274, Dec. 2022, shifting strategy design of HEV during regenerative braking pro-
Art. no. 116442. cess for energy conservation,’’ Appl. Energy, vol. 183, pp. 914–925,
[9] M. R. Ahssan, M. M. Ektesabi, and S. A. Gorji, ‘‘Electric vehicle Dec. 2016.
with multi-speed transmission: A review on performances and com- [29] B. Gao, Q. Liang, Y. Xiang, L. Guo, and H. Chen, ‘‘Gear ratio optimization
plexities,’’ SAE Int. J. Altern. Powertrains, vol. 7, no. 2, pp. 169–182, and shift control of 2-speed I-AMT in electric vehicle,’’ Mech. Syst. Signal
Dec. 2018. Process., vols. 50–51, pp. 615–631, Jan. 2015.

VOLUME 11, 2023 103777

 K. Park et al.: Multi-Speed Transmission Optimization of EVs Based on Shifting Pattern

[30] K. Kwon, M. Seo, H. Kim, T. H. Lee, J. Lee, and S. Min, ‘‘Multi-objective SANG-KIL LIM received the B.S. degree in
optimisation of hydro-pneumatic suspension with gas–oil emulsion for mechanical engineering from Jeonbuk National
heavy-duty vehicles,’’ Vehicle Syst. Dyn., vol. 58, no. 7, pp. 1146–1165, University, in 2008, and the M.S. and Ph.D.
Jul. 2020. degrees in electrical engineering from Chonnam
[31] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, ‘‘A fast and elitist National University, Gwangju, South Korea, in
multiobjective genetic algorithm: NSGA-II,’’ IEEE Trans. Evol. Comput., 2010 and 2017, respectively. Since 2019, he has
vol. 6, no. 2, pp. 182–197, Apr. 2002. been holding a postdoctoral research position with
[32] K. Kwon, M. Seo, and S. Min, ‘‘Efficient multi-objective optimization
the EV Components and Materials Research and
of gear ratios and motor torque distribution for electric vehicles with
Development Group, Korea Institute of Industrial
two-motor and two-speed powertrain system,’’ Appl. Energy, vol. 259,
Feb. 2020, Art. no. 114190.
Technology. Since 2020, he has been a Professor
[33] M. Khajehzadeh, M. R. Taha, S. Keawsawasvong, H. Mirzaei, and with the Department of Automotive Engineering, Honam University. His
M. Jebeli, ‘‘An effective artificial intelligence approach for slope stability research interests include power electronics, power conditioning system
evaluation,’’ IEEE Access, vol. 10, pp. 5660–5671, 2022. DC–DC converters for renewable energy, and power conversion system
for EVs.

KIHAN KWON (Member, IEEE) received the

KIJONG PARK received the Ph.D. degree in Ph.D. degree in automotive engineering from
mechanical design and production engineering Hanyang University, Seoul, South Korea, in 2020.
from Hanyang University, Seoul, South Korea, From 2009 to 2015, he was a Research Engi-
in 2004. He is currently a Senior Research Engi- neer with Hyundai Motor Company, Hwaseong,
neer with Hyundai Motor Company, Hwaseong, South Korea. He is currently a Professor with the
South Korea. His research interests include Department of Automotive Engineering, Honam
HEV/EV powertrain system design, CAE, and University, Gwangju, South Korea. His research
optimization. interests include HEV/EV powertrain system
design, control, and optimization.

103778 VOLUME 11, 2023