electronics
Article
Estimation of the Hot Swap Circulation Current of a Multiple
Parallel Lithium Battery System with an Artificial Neural
Network Model
Nam-Gyu Lim 1 , Jae-Yeol Kim 2 and Seongjun Lee 2, *






Citation: Lim, N.-G.; Kim, J.-Y.; Lee,
S. Estimation of the Hot Swap
Circulation Current of a Multiple
Parallel Lithium Battery System with
an Artificial Neural Network Model.
Electronics 2021, 10, 1448. https://
doi.org/10.3390/electronics10121448
Academic Editors: Mauri Marco and
Vincenzo Musolino
Received: 17 May 2021
Accepted: 14 June 2021
Published: 17 June 2021
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2021 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Electronics 2021, 10, 1448. https://doi.org/10.3390/electronics10121448
1 Department of Mechanical System and Automotive Engineering, Chosun University, Gwangju 61452, Korea;
lnk9100@chosun.kr
2 Department of Mechanical Engineering, Chosun University, Gwangju 61452, Korea; jykim@chosun.ac.kr
* Correspondence: lsj@chosun.ac.kr; Tel.: +82-62-230-7173
Abstract: Battery applications, such as electric vehicles, electric propulsion ships, and energy storage
systems, are developing rapidly, and battery management issues are gaining attention. In this
application field, a battery system with a high capacity and high power in which numerous battery
cells are connected in series and parallel is used. Therefore, research on a battery management system
(BMS) to which various algorithms are applied for efficient use and safe operation of batteries is
being conducted. In general, maintenance/replacement of multi-series/multiple parallel battery
systems is only possible when there is no load current, or the entire system is shut down. However,
if the circulating current generated by the voltage difference between the newly added battery
and the existing battery pack is less than the allowable current of the system, the new battery can
be connected while the system is running, which is called hot swapping. The circulating current
generated during the hot-swap operation is determined by the battery’s state of charge (SOC), the
parallel configuration of the battery system, temperature, aging, operating point, and differences in
the load current. Therefore, since there is a limit to formulating a circulating current that changes
in size according to these various conditions, this paper presents a circulating current estimation
method, using an artificial neural network (ANN). The ANN model for estimating the hot-swap
circulating current is designed for a 1S4P lithium battery pack system, consisting of one series and
four parallel cells. The circulating current of the ANN model proposed in this paper is experimentally
verified to be able to estimate the actual value within a 6% error range.
Keywords: energy storage system (ESS); hot swap; battery; artificial neural network (ANN); Fitnet;
multiple parallel; multiple series; equivalent circuit model (ECM); circulating current
1. Introduction
As environmental and resource depletion problems become more serious, research on
eco-friendly systems that use batteries, such as electric vehicles, electric-powered ships,
and large-capacity energy storage system (ESS) power generation systems that do not use
fossil fuels, is accelerating. Since large capacity and high power are required in these fields,
battery packs in which numerous battery cells are connected in series/parallel are used.
Various studies are being conducted because the role of the battery management system
(BMS) is very important to maintain the optimal state of these battery packs and perform
protection operations [1–6].
Figure 1 is an example of a large-capacity battery system configuration applied to
an energy storage system and an electric propulsion ship. A total of 200 to 300 lithium
battery cells are connected in series to form one high-voltage rack, and several racks
are connected in parallel to expand the capacity [7]. Hot swap refers to the function of
detaching/attaching a module from the DC bus by applying a high voltage relay on the
top of the rack for maintenance/replacement of cell unit batteries in the above multi-
series/multi-parallel system. At this time, the hot swap action should be performed only
https://www.mdpi.com/journal/electronics
 Electronics 2021, 10, x FOR PEER REVIEW 2 of 20

Electronics 2021, 10, 1448 2 of 19

maintenance/replacement of cell unit batteries in the above multi-series/multi-parallel

system. At this time, the hot swap action should be performed only under certain condi-
under
tions. certain conditions.
The reason The
is that the reason isinthat
difference thethe difference
state of chargeinofthe
thestate of charge
newly of the
added battery
newly
and theadded battery
working and causes
battery the working battery circulating
an excessive causes an excessive
current. circulating current.

Figure1.1.Large-capacity
Figure Large-capacitybattery
batterysystem
systemconsisting
consistingofofhigh
highvoltage
voltagemodules
modulesconnected
connectedininparallel.
parallel.

In
Inorder
orderto tominimize
minimizeor oravoid
avoidthethecirculating
circulatingcurrent
currentduring
duringthe
thehot-swap
hot-swapoperation,
operation,
the
thehot-swap
hot-swapoperation
operationshould
shouldbe beperformed
performedafteraftermatching
matchingthe thestate
stateofofcharge
chargebetween
between
the
thenewly
newlyadded
addedbattery
batteryandandthethe
existing working
existing workingbattery. To solve
battery. this problem,
To solve there there
this problem, was
awas
study on applying
a study a DC/DC
on applying converter
a DC/DC to match
converter the output
to match voltage
the output of the battery.
voltage Jiang
of the battery.
etJiang
al. used
et al.aused
buckaregulator and Moo
buck regulator andetMoo
al., applied a buck/boost
et al., applied converter
a buck/boost to minimize
converter to mini-
the
mizecirculating currentcurrent
the circulating by controlling the battery
by controlling voltagevoltage
the battery [8,9]. These
[8,9]. studies deviseddevised
These studies ways
to basically suppress the circulating current caused by the voltage difference
ways to basically suppress the circulating current caused by the voltage difference of the of the battery,
but they are
battery, butproblematic in that the system
they are problematic sizesystem
in that the increases and
size the efficiency
increases and the decreases when
efficiency de-
acreases
converterwhenis used to connect
a converter the device
is used to a high
to connect the voltage/power
device to a highbattery system. battery
voltage/power
On the other hand, studies have been conducted to connect batteries in parallel
system.
without Onusing
the othera DC/DC converter.
hand, studies haveHsubeenetconducted
al. presented an algorithm
to connect batteriesforinopening and
parallel with-
closing switches according to priority through differential classification
out using a DC/DC converter. Hsu et al. presented an algorithm for opening and closing of SOCs of batteries
for both individual
switches accordingand parallel charging
to priority of batteries [10].
through differential In this study,
classification an efficient
of SOCs battery
of batteries for
parallel connection method without using a DC/DC converter was presented,
both individual and parallel charging of batteries [10]. In this study, an efficient battery but the
parallel
parallelconnection
connection condition
method withoutwas simply
using set as SOC,
a DC/DC and only
converter thepresented,
was case where butthetheload
par-
current was the charging condition was considered. In addition, studies
allel connection condition was simply set as SOC, and only the case where the load current to estimate the
circulating currentcondition
was the charging have not been conducted. Song
was considered. et al. controlled
In addition, studies tothe switchthe
estimate connected to
circulating
the battery by using a control algorithm based on fuzzy control logic [11].
current have not been conducted. Song et al. controlled the switch connected to the battery They succeeded
in lowering the circulating current due to the parallel connection and were able to achieve
by using a control algorithm based on fuzzy control logic [11]. They succeeded in lowering
efficient energy management. In Song’s study, similar to that of Hsu, the circulating current
the circulating current due to the parallel connection and were able to achieve efficient
was reduced by classifying the battery, according to the SOC, and the circulating current
energy management. In Song’s study, similar to that of Hsu, the circulating current was
estimation study was not performed. Two previous studies [10,11] used the state of charge
reduced by classifying the battery, according to the SOC, and the circulating current esti-
(SOC) as a condition for the parallel connection of batteries. Since the state of charge
mation study was not performed. Two previous studies [10,11] used the state of charge
of the battery is an estimated value, there may be an error compared to the actual state
(SOC) as a condition for the parallel connection of batteries. Since the state of charge of
of charge. Since the SOC estimation error of parallel-connected battery racks represents
the battery is an estimated value, there may be an error compared to the actual state of
a large voltage difference between the battery racks, using only the SOC as a hot swap
charge. Since the SOC estimation error of parallel-connected battery racks represents a
condition may be problematic. Therefore, in this paper, the terminal voltage of the battery
is determined to be a hot-swap condition.
Electronics 2021, 10, 1448 3 of 19

Lee et al. predicted the circulating current of the battery based on an equivalent
model of the battery. To calculate the inrush current that occurs during the hot-swap
operation, the equivalent model of the battery is based on the open-circuit voltage and Ri
model. Cell unit tests and the simulation were used to verify the model by applying an EV
driving cycle. However, these authors did not verify the situation in which the direction of
the charging/discharging load current is changed, nor did they study the characteristics
of the battery according to the state change, such as the temperature and aging of the
battery [12,13]. Chou et al. conducted a study on the design of a Smart Redundant Battery
(SRB) that reflects the direct current internal resistance (DCIR) and pack impedance of the
battery (cable, etc.) and the hot-swap conditions that can operate within a set current [14].
Since this study did not consider the hot-swap circulating current, according to the parallel
configuration of batteries, it is possible to exceed the current limit of the battery system
under the parallel condition of three or more batteries.
In previous studies, various studies on circulating current generation in multi-parallel
battery systems were performed, but most studies determined that the circulating current
generated during hot-swap operation was due to differences in the SOC of the battery.
However, no study was conducted to analyze and estimate the influence of major factors
affecting the magnitude of the circulating current generated during hot swapping.
In this paper, the effects of voltage deviation, temperature, parallel configuration, and
load current were analyzed as major factors that determine the circulating current generated
during a hot-swap operation. After designing a 1S4P battery pack to perform the hot-swap
analysis, a 2 RC-based battery equivalent simulation model design was performed. Based
on the simulation model, the case study analysis led to hot-swap conditions that could
operate within the established current limits. The nD-LUT (Look-Up Table) model can be
used to perform a hot-swap operation that reflects various conditions but can only be used
as a condition to turn the action on and off. However, the estimated current can be utilized
for power management before and after the hot-swap operation as well as determining the
hot-swap operation by estimating the circulating current generated during the hot-swap
operation.
In the previous study of Lee et al., a formula for calculating the magnitude of circulat-
ing current generated during the hot swap was derived from a simplified circuit model of
the entire system [13]. The formula for estimating the circulating current can be applied
well under the stabilized condition in which the state of charge of the existing connected
batteries is the same, but it cannot be used in a situation where the current between the
battery cells is unbalanced, due to the continuous hot-swap operation. Therefore, in this
paper, an artificial neural network (ANN) model was used to estimate the circulating
current generated during the hot-swap operation, considering various conditions in real
time.
The ANN model was suggested as a method for estimating the state of the battery,
such as the state of charge (SOC) and state of health (SOH) in battery systems but has not
yet been applied to studies to estimate the hot-swap circulating current [15–18]. Among
the ANN models, the Fitnet model can be fitted to any degree of accuracy if sufficient input
values are available. Thus, the results of the simulation-based hot-swap case study analysis
were used as the input data, and the circulating current generated under the corresponding
condition is displayed as the result [19–21]. Therefore, it is possible to determine the
conditions for the hot-swap operation from the results of the circulating current estimation
with the ANN model, which can take various conditions into consideration. In this study, an
ANN model with two hidden layers was constructed to estimate the hot-swap circulating
current of a 1S4P lithium battery pack. The performance of the model in terms of the
circulating current estimation was confirmed to be approximately 94% accurate.

2. Design of Lithium Battery Model

The battery system of an ESS consists of numerous battery cells. Specifically, units
containing 200–300 lithium battery cells that are connected in series are connected in
 Electronics 2021, 10, x FOR PEER REVIEW 4 of 20
Electronics 2021, 10, x FOR PEER REVIEW 4 of 20

2. Design 2. of
Design
Lithium of Lithium
Battery Model Battery Model
Electronics 2021, 10, 1448 4 of 19
The battery The system
batteryof system
an ESS ofconsists
an ESS consists
of numerous of numerous battery
battery cells. cells. Specifically,
Specifically, units units
containing containing
200–300 200–300 lithiumcells
lithium battery battery thatcells that are connected
are connected in series in areseries are connected
connected in par- in par-
allel. The battery model for the analysis of the ESS of a cell causes the simulation model model
allel. The battery model for the analysis of the ESS of a cell causes the simulation
to become parallel. Thecomputationally
to computationally
become battery complex,
model for the analysis
complex,
and and of
the analysis thethe ESSnot
analysis
may of may
abecell causes
not the simulation
be possible.
possible. Since Since amodel
a high- high-
to become
voltage computationally
battery in a complex,
multi-series and
configuration the analysis
requires
voltage battery in a multi-series configuration requires a passive or active cell-balancing may
a not
passive be possible.
or active Since a high-
cell-balancing
voltage battery can inbeasimplified
multi-series to aconfiguration requires amodel passive or active cell-balancing
function,function,
it can be itsimplified to a single-cell single-cell
equivalent equivalent
model under the under the assumption
assumption that that
function, it
the cell-to-cell can be simplified to a single-cell equivalent model under the assumption that
the cell-to-cell voltage isvoltage
consistent is consistent
[22]. [22].
the cell-to-cell
In this voltage
study, the ishot-swap
consistent [22]. was performed, using a scaled down 1S4P lith-
analysis
In this study, the hot-swap analysis was performed, using a scaled down 1S4P lith-
iumpackIn this study,
battery pack theto forhot-swap
account foranalysis
multiple was performed,
parallel systems. using a scaled down 1S4P lithium
ium battery to account multiple parallel systems.
battery pack to account for multiple parallel systems.
2.1. Design2.1. Design of LithiumEquivalent
Battery Equivalent Circuit
2.1.ofDesign
Lithium of Battery
Lithium Battery Equivalent Circuit Circuit
The lithium The battery
lithium battery equivalent circuit(ECM)
model (ECM) can tobedescribe
used to describe the non-
linear The lithium equivalent
characteristics battery of
circuit model
equivalent
batteries circuit
and to modelcan
estimate the
be used
(ECM) can of
state be the
used the non-the
to describe
batteries [23–25]. non-
The
linear characteristics
linear characteristicsof batteries and to estimate
of batteries the statethe
and to estimate of state
the batteries [23–25].[23–25].
of the batteries The The
model can model can be based
be based on an ECM consisting of sources,
voltage sources, resistors, and capacitors as
model can beonbased an ECM on an consisting
ECM consisting of voltage of voltage resistors,resistors,
sources, and capacitors as
and capacitors as
shown in shown in Figure
Figurein2.Figure 2.
Figure2.2a,b Figure
shows2a,b shows the first- and second-order RC circuits, respectively.
shown Figure 2a,bthe first-the
shows andfirst-
second-order RC circuits,
and second-order RC respectively.
circuits, respectively.
As the number
As the number of RCs of RCs increases,
increases, the accuracy the accuracy
of the of themay
model model may increase,
increase, but the but the param-
As the number of RCs increases, the accuracy of the model may increase, butparam-
the parameters
eters
eters constituting constituting
thethemodel the model increase, which increases the computational
intensityintensity and
constituting modelincrease,
increase,which whichincreases
increases the the computational
computational intensity andanalysis
and
analysis
analysistime.
time.In time.
Inaddition, In addition,
addition,becausebecausethe because the
thenonlinearity
nonlinearity nonlinearity of the battery can be sufficiently de-
of of
thethe battery
battery cancan be sufficiently
be sufficiently de- with
described
scribed with
scribed with the2RC, the equivalent circuitwas model was designed using the 2RC[26]. model [26]. In
2RC, 2RC, equivalent
the equivalent circuitcircuit
modelmodel was designed designed
using using
the 2RC themodel
2RC model
[26]. In this In study, the
this study,this study,
the lithium the lithium battery equivalent circuit was designed using MATLAB/Simulink
lithium batterybattery
equivalent equivalent
circuit was circuit was designed
designed using MATLAB/Simulink
using MATLAB/Simulink software. Two
software. software.
Two approachesTwo approachescould becould
followed be followed for the design.
for the design. First, asinshown in3a,
Figure 3a,
approaches could be followed for the design. First, asFirst,
shown asinshown
Figure 3a, Figure
the Simscape or
the
the Simscape Simscape
or Power or Power
Sim librarySim library
provided provided
by Simulink by Simulink
includes includes
a method a method
for for
configur- configur-
Power Sim library provided by Simulink includes a method for configuring electrical circuit
ing electrical
ing electrical circuitSecond,circuit elements.
elements. Second, the
canECM can be represented by Equation (1), and
elements. the Second,
ECM can thebeECMrepresented be represented
by Equationby(1), Equation
and it can (1), be
and modeled
itmodeled
it can bebasedcan onbe modeled
based onbased
equations usingon
equations equations
basic using
blocks using
basic
as basic
blocks
shown asblocks
in shownas
Figure inshown
3b. Figurein3b. Figure 3b.

(a) (a) (b) (b)

Figure
Figure 2.Figure 2.equivalent
Battery2. Battery equivalent
Battery equivalent circuit
circuit
circuit model model(a)
model
(ECM): (ECM):
(ECM): (a)1-RC
(a)
1-RC ECM, 1-RC ECM,
(b)ECM, (b)2-RC
2-RC (b)
ECM. 2-RCECM.
ECM.

(a) (a) (b) (b)

Figure 3.3.Modeling the battery ECM

ECMby
byMATLAB/Simulink: (a)
(a) modeling using the
thesimscape library; (b)
(b)equation-based
Figure 3. Figure
ModelingModeling theECM
the battery battery MATLAB/Simulink:
by MATLAB/Simulink: (a) modeling modeling
using the using
simscape simscape library;
library; (b) equation-based
equation-based
modeling
modeling using
usingbasic
basicblocks.
blocks.
modeling using basic blocks.
Each
Each parameter
parameter constitutingthe
constituting thelithium
lithiumbattery
batteryequivalent
equivalent circuit
circuit is derived
is from
derived fromfrom
the
Each parameter
the voltage constituting
characteristic the lithium
behavior battery
of the equivalent
current flowingcircuit is
through derived
the lithium the
battery. In
voltage characteristic
voltage characteristic behavior
behavior of the of the
current current
flowing flowing
through through the
thetemperature, lithium
lithium battery. battery.
In In ad-
ad- are not
addition,
dition, the characteristics
the characteristics of the battery,
of the such
battery, such
such as the
as the SOC,
SOC, temperature, and
andare aging,
aging, are not
dition, the characteristics
constant and varyofwith
the time.
battery, as the
Therefore, SOC,
in this temperature,
study, a discharge and aging,
experiment not carried
was
out using battery charging/discharging equipment (Neware, Hong Kong, China, BTS-4000)
and a temperature chamber (JEIO Tech, Daejeon, Korea, LCH-11) to reflect the extent to
which the characteristics of the battery change, depending on the SOC and temperature in
onics 2021, 10, x FOR PEER
Electronics 2021, REVIEW
10, x FOR PEER REVIEW 5 of 20 5 of 20

Electronics 2021, 10, 1448 5 of 19

constant and vary withand
constant time. Therefore,
vary with time.in this study, in
Therefore, a discharge
this study,experiment
a dischargewas carried was carried
experiment
out using battery charging/discharging
out using equipment (Neware,
battery charging/discharging equipment Hong Kong, Hong
(Neware, China,Kong,
BTS- China, BTS-
4000) and a temperature
4000) and a chamber (JEIO
temperature Tech, Daejeon,
chamber (JEIO Tech,Korea, LCH-11)
Daejeon, to reflect
Korea, the ex-
LCH-11) to reflect the ex-
the model. The temperature levels were 0, 10, 23, 35, and 45 ◦ C and the SOC was tested at
tent to which the
tentcharacteristics of the battery of
to which the characteristics change, depending
the battery change, ondepending
the SOC and ontemper-
the SOC and temper-
10%
ature in the model.
atureintervals.
The
in The experimental
thetemperature
model. Thelevels were
temperature profile10,of23,
0,levels the35,charging/discharging
were and 4523,
0, 10, °C 35,
andandthe45
SOC equipment
°C was usedwas
and the SOC in
the experiment
tested at 10% tested
intervals.
at 10% is shown
Theintervals. in
experimental Figure 4, and the
profile of the profile
The experimental results of the
charging/discharging voltage/current measurements
equipment equipment
of the charging/discharging
in the in
used discharge
used in the experiment period
theisexperiment
shown are
in Figureshown
is shown4, andin Figure
in Figure
the 5 [27].
results
4, andof the results
voltage/current meas-
of the voltage/current meas-
urements in the discharge
urements period
in the are shown
discharge periodin Figure
are shown5 [27].in Figure 5 [27]. !
1 1 V
Vterminal = OCV − 1 × Ri − 1s 𝑉C
1 Iload 1di f f + Rdi f f𝑉× Cdi f f (1)
V𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑙 = OCV − 𝐼𝑙𝑜𝑎𝑑=×OCV
V𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑙 𝑅𝑖 − 𝐼𝑙𝑜𝑎𝑑( × 𝑅+𝑖 − ( + ) ) (1) (1)
𝑠 𝐶𝑑𝑖𝑓𝑓 𝑅𝑑𝑖𝑓𝑓 × 𝐶𝑑𝑖𝑓𝑓 𝑅𝑑𝑖𝑓𝑓 × 𝐶𝑑𝑖𝑓𝑓
𝑠 𝐶𝑑𝑖𝑓𝑓

Figure 4. PulseFigure
current
Figure4.4.discharge test discharge
Pulsecurrent
Pulse current profile.
dischargetest
testprofile.
profile.

Figure 5. Voltage/current
Figure
Figure results from
5. Voltage/current
5. Voltage/current pulse
results current
results
from from discharge
pulsepulse experiment.
current
current discharge
discharge experiment.
experiment.

The parameters The

The parameters
ofparameters ofofthe
the equivalent equivalent
model
the weremodel
equivalent derived
model werefrom
were derived from
the dynamic
derived from thecharacteris-
dynamic
the dynamic characteristics
characteris-
of theofvoltage
tics of the voltage
tics during during
the
the voltage the discharge
discharge
during period,
the period,
and the
discharge and theand
battery
period, battery
terminal terminal
the battery voltage voltage is expressed
is ex-voltage
terminal is ex-
by
pressed by EquationEquation
pressed (2). (2). OCV
The
by Equation The (2).
OCV parameter
parameter
The OCV was was derived
derived
parameter from from
the
was derived the
voltagefromvoltage
inthe in the steady
thevoltage
steady in the steadystate
before
state before the
state the for
current
before current
theeach for each
discharge
current discharge
section
for each section
was
discharge applied.wasThe
section applied.
was series
applied. The series
resistance
The seriesRresistance
resistanceRR
i is i iisis
calculated by dividing the voltage difference
as ①ΔV denoted as ∆V , as shown in Figure 6,
as ①ΔV
1 1
calculated by calculated
dividing the byvoltage
dividing difference
the voltage denoted
difference 1, as shown
denoted in1,Figure
as shown 6, byin Figure 6, by
by the applied current. The Rdiff and Cdiff parameters were derived by exponentially
the applied current.
the appliedThe R diff and The
current. Cdiff parameters
Rdiff and Cdiffwere derivedwere
parameters by exponentially fitting
derived by exponentially fitting
fitting the voltage values, excluding the OCV and IR-Drop in ∆V 2 2 in Figure 6. The
the voltage values, excluding
the voltage
MATLAB the OCV
values, excluding
curve-fitting and
ToolBox the IR-Drop
wasOCV in
usedand ②ΔV in in ②ΔV
Figure
IR-Dropfitting.
2
for function 6. The MATLAB
2 in Figure 6. The MATLAB
The fitting was performed by
curve-fitting applying
ToolBox
curve-fittingwas used
Equation for
ToolBox(3)function
was
to the used fitting. The fitting
for function
Custom Equation was
fitting.
item performed
The
of fitting was
curve-fitting, byperformed
applying
and the Rby applying
diff1 , Rdiff2 ,
Equation (3) C to the
Equation Custom toEquation
the Custom item of curve-fitting,
Equation item of and the R
curve-fitting,diff1 , R
diff1 , and Cdiff2 parameters were derived from Equations (5) and (6) as a result of, the
(3) and
diff2 ,
theC R
diff1, and
diff1 , R diff2 , C diff1 and
fitting [28,29]. Figure 7 shows the results that were obtained using the second-order RC
ECM and MATLAB/Simulink/Simscape software, and the parameters derived from the
 Cdiff2 parameters were derived from Equations (5) and (6) as a result of the fitting [28,29].
Figure 7 shows the results that were obtained using the second-order RC ECM and
MATLAB/Simulink/Simscape software, and the parameters derived from the above re-
sults reflect the characteristics according to the values of SOC and Temp, using the nD-
Electronics 2021, 10, 1448 lookup table. To verify the accuracy of the battery ECM model in Figure 7, we performed 6 of 19
a terminal voltage comparison between the voltage measurement data of the pulse dis-
charge test and the simulation model. The comparison result is shown in Figure 8, and the
ECM model confirmed that the voltage error was less than about 30 mv in most regions
abovefor
except results reflect
the low SOCthe(<3
characteristics
V) region, whereaccording to the values
the nonlinear of SOC and
characteristics ofTemp, using
the battery
are severe, and the MSE (Mean Square Error) was found to be 0.0011; through this,7,the
the nD-lookup table. To verify the accuracy of the battery ECM model in Figure we
performed a terminal voltage comparison
accuracy of the single cell ECM model was verified. between the voltage measurement data of the
pulse discharge test and the simulation model. The comparison result is shown in Figure 8,
and the ECM model confirmed that the voltage error 𝑡 𝑡
was less than about 30 mv in most
𝑉𝑡 = OCV − 𝐼𝑙𝑜𝑎𝑑 × 𝑅𝑖 − 𝐼𝑙𝑜𝑎𝑑 × 𝑅𝑑𝑖𝑓𝑓1 𝑒 𝜏1 − 𝐼
𝑙𝑜𝑎𝑑 × 𝑅𝑑𝑖𝑓𝑓2 𝑒 𝜏2 (2)
regions except for the low SOC (<3 V) region, where the nonlinear characteristics of the
battery are severe, and the MSE (Mean Square Error) was found to be 0.0011; through this,
the accuracy of the single cell ECM model−𝑡was verified. −𝑡
Y_fit = A × exp 𝐵 + C × 𝑒𝑥𝑝 𝐷 (3)
t t
Vt = OCV − Iload × Ri − Iload × Rdi f f 1 e τ1 − Iload × Rdi f f 2 e τ2 (2)
𝑅𝑖 = ∆V1/𝐼𝑙𝑜𝑎𝑑
−t −t (4)
Y_fit = A × exp B +C × exp D (3)
A Ri = ∆V1/Iload C (4)
𝑅𝑑𝑖𝑓𝑓1 = , 𝑅𝑑𝑖𝑓𝑓2 = (5)
𝐼𝑙𝑜𝑎𝑑 A 𝐼𝑙𝑜𝑎𝑑C
Rdi f f 1 = , Rdi f f 2 = (5)
Iload Iload
B D
𝐶𝑑𝑖𝑓𝑓1 = B, 𝐶𝑑𝑖𝑓𝑓2 = D (6)
Cdi f f 1𝑅=
𝑑𝑖𝑓𝑓1 , Cdi f f 2 𝑅
=𝑑𝑖𝑓𝑓2 (6)
Rdi f f 1 Rdi f f 2

Figure
Figure6.6.Voltage,
Voltage,OCV,
OCV,and
andcurrent
currentwaveform
waveformwhen
whenSOC
SOCdecreased
decreased from
from 70%
70% to
to SOC
SOC 60%.
60%.

2.2. 1S4P Lithium Battery Pack Design for Experimental Verification

To verify the lithium battery software model and the hot-swap experiment, a 1S4P
battery pack was designed. Current sensors and relays were allocated for each cell unit
to implement the hot-swap function for the parallel connection of batteries in series units
of the ESS system. Table 1 lists the components of the lithium battery pack. The INR
18650-30Q was used as the lithium battery, the ACS 712 product with <5 A performance
was used as the current sensor, and measurements were performed in units of 100 ms,
using a myRio-1900 controller. In the case of the voltage measurement, a direct connection
was made from the lithium battery to satisfy the analog input specification (< 5 V) of the
controller. For the voltage/current measurement and relay control of the lithium battery
pack, an experimental set was constructed, using the myRio-1900 controller and LabVIEW
software; the composition is shown in Figure 9.
 Electronics 2021,
Electronics 2021, 10, x1448 7 of 19
Electronics 2021, 10, x10,
FORFOR PEER
PEER REVIEW
REVIEW 7 of 720of 20

Figure
Figure
Figure 7.7.Second-order
Second-order
7. Second-order RC
RCECM
RC ECM ECM modeling
modeling
modeling (MATLAB/Simulink/Simscape
SW).SW).
(MATLAB/Simulink/Simscape
(MATLAB/Simulink/Simscape SW).

Figure
Figure 8. Derived
8. Derived parameter
parameter validation
validation simulation
simulation results.
results.
Figure 8. Derived parameter validation simulation results.

2.2.2.2.
1S4P1S4P Lithium
Lithium Battery
Battery PackPack Design
Design for Experimental
for Experimental Verification
Verification
To To verify
verify the the lithium
lithium battery
battery software
software model
model andand
the the hot-swap
hot-swap experiment,
experiment, a 1S4P
a 1S4P
battery pack was designed. Current sensors and relays were allocated for each cell unitunit
battery pack was designed. Current sensors and relays were allocated for each cell to to
implement
implement the the hot-swap
hot-swap function
function for for
the the parallel
parallel connection
connection of batteries
of batteries in series
in series units
units of of

Electronics 2021, 10, 1448
used as the current sensor, and measurements were performed in units of 100 ms, using a
myRio-1900 controller. In the case of the voltage measurement, a direct connection was
made from the lithium battery to satisfy the analog input specification (< 5 V) of the con-
troller. For the voltage/current measurement and relay control of the lithium battery pack,
an experimental set was constructed, using the myRio-1900 controller and LabVIEW
8 of 19 soft-
ware; the composition is shown in Figure 9.

Table 1. Components of 1S4P lithium battery pack.

Table 1. Components of 1S4P lithium battery pack.
Item Model Name
Item
Battery Cell ModelINR-18650
Name 30Q
Battery(18650
Cell Type) INR-18650 (3040
30QmAh)
(18650 Type) (3040
5V mAh)
Relay Module
Relay (< 10 A) 5 V Relay Module
Relay (<10 A) (SZH-RLBG-012)
(SZH-RLBG-012)
Current
Current Sensor Sensor
(<5 A) (< 5 A) ACS 712 ACS 712
Cable Cable 1.5
2 mm
1.5 mm (cooper) 2 (cooper)

Nickel Nickel
Plate Plate 5 mm/0.15 T
5 mm/0.15 T
NI myRio-1900,
NI myRio-1900,
Controller
Controller Lab view
Lab view
3D Printing
Structure 3D Printing
Structure (PLA)
(PLA)

(a) (b)
Figure 9. Configuration
Figure of the 1S4P
9. Configuration battery
of the pack forpack
1S4P battery the experiment: (a) block
for the experiment: (a)diagram of the experimental
block diagram construction;
of the experimental construction;
(b) hot-swap experiment and configuration of the measurement system.
(b) hot-swap experiment and configuration of the measurement system.

2.3. Designing Software

2.3. Designing Model for
Software 1S4P
Model forLithium BatteryBattery
1S4P Lithium Pack Pack
The lithium batterybattery
The lithium cell model was composed
cell model of a second-order
was composed RC ECM,
of a second-order RC as shown
ECM, as shown
in Figure 7, and a7,relay
in Figure and amodel was arranged
relay model for each
was arranged forcell
eachto implement the relay
cell to implement thefunction,
relay function,
as shown in Figure
as shown 10. The10.
in Figure relay
Theon-resistor was confirmed
relay on-resistor experimentally
was confirmed with 30with
experimentally mΩ.30 mΩ.
The cable and contact resistance used in the pack configuration was confirmed by
The cable and contact resistance used in the pack configuration was confirmed by carryingcarrying
out an out
experiment with 5 with
an experiment mΩ, 5using
mΩ,one resistance
using model model
one resistance for each forcell.
eachThe ECM
cell. Themodel
ECM model
of the battery pack consisted
of the battery of a cell
pack consisted of amodel, a relay
cell model, and and
a relay other cable
other resistance
cable models.
resistance models. To
To verify the the
verify accuracy of the
accuracy modeling,
of the modeling,the the
modeling
modelingaccuracy waswas
accuracy judged by comparing
judged by comparing the
the circulating current generated by performing a hot swap experiment on four batteries
with different states of charge. Cell voltage accuracy was performed in Section 2.1. First,
as shown in Figure 11, when the 3.932 V cell and the 3.822 V cell performed the hot-swap
operation, the measured current was 1.02 A and the ECM simulation result was 1.055 A.
Second, when the 3.827 V and 3.766 V cells were hot swapped, the measured current was
0.6 A and the ECM simulation result was 0.585 A. Finally, it was confirmed as shown in
Table 2 that the measured current of the 3.766 V battery and the 3.402 V battery during
the hot-swap operation ws 3.41 A and the ECM simulation result was 3.54 A. From the
above results, it was confirmed that the ECM model of the 1S4P lithium battery pack had
an accuracy of 96% or more.
 1 3.932 V 1.055 A
1 1.02 A
2 3.822 V (3.4%)
2 3.827 V 0.585 A
2 0.6 A
3 3.766 V (2.5%)
Electronics 2021, 10, 1448 3 3.766 V 3.54 A 9 of 19
3 3.41 A
4 3.402 V (3.8%)

Electronics 2021, 10, x FOR PEER REVIEW 10 of 20

Figure
Figure10.
10.Software
Softwaremodel
modelofof1S4P
1S4Pbattery
batterypack.
pack.

(a) (b)
Figure 11. Comparison of the experimental and simulated results for the hot swap (3.932 V vs. 3.822 V): experimental and
Figure 11. Comparison of the experimental and simulated results for the hot swap (3.932 V vs. 3.822 V): experimental and
simulated
simulated(a)
(a)voltage
voltageand
and(b)
(b)current.
current.

Table 2. Components
3. Hot-Swap of the 1S4P lithium battery pack.
Analysis
The purpose of hot swap analysis is to establish the conditions Measured under which excessive
Simulated
Cell.N Voltage
circulating current is not generated when batteries with different
(Max. A/Cell) charge states are con-
(Max. A/Cell)
nected in parallel. The magnitude
1 of the3.932
circulating
V current generated during 1.055 the
A hot-
1
swap operation is determined2 not only by theVdifference in1.02
3.822 theAstate of charge of the bat-
(3.4%)
tery, but also by various factors
2 that determine
3.827 Vthe characteristics of the battery. Therefore,
0.585 A
2 0.6 A
when deriving the conditions 3 for hot swap 3.766 V
operation, (2.5%)
the optimal hot swap algorithm can
be designed by considering 3 the factors that3.766affect
V the circulating current. In3.54 A study,
this
3 3.41 A
4 3.402 V (3.8%)
based on the simulation model of the 1S4P configuration, we derived the optimal hot-
swap operating conditions by conducting a case study under various conditions.
3. Hot-Swap Analysis
3.1. Derivation
The purposeof Key Factors
of hot swapfor Hot Swap
analysis (Case
is to Study)the conditions under which excessive
establish
circulating current is not generated when batteries
The key factors that determine the circulating current with different
generated charge
duringstates are con-
the hot-swap
nected in parallel.
operation The magnitude
were derived. This wasofaccomplished
the circulatingbycurrent generated
analyzing during configuration
the parallel the hot-swap
operation is determined
of the lithium battery packs not and
onlyhot-swap
by the difference in the
case studies state of the
by varying charge of the battery,
temperature, volt-
but also by various factors that determine the characteristics of the battery.
age deviation, and load current based on a simulation model of the 1S4P configuration. Therefore,

3.1.1. Impact of Parallel Configuration of Battery

In the 1S4P configuration, the hot-swap parallel configuration in which one battery
is inserted is divided into three cases: 1 (On) vs. 1 (Off), 2 (On) vs. 1 (Off), and 3 (On) vs. 1
Electronics 2021, 10, 1448 10 of 19

when deriving the conditions for hot swap operation, the optimal hot swap algorithm can
be designed by considering the factors that affect the circulating current. In this study,
based on the simulation model of the 1S4P configuration, we derived the optimal hot-swap
operating conditions by conducting a case study under various conditions.

3.1. Derivation of Key Factors for Hot Swap (Case Study)

The key factors that determine the circulating current generated during the hot-swap
operation were derived. This was accomplished by analyzing the parallel configuration of
the lithium battery packs and hot-swap case studies by varying the temperature, voltage
deviation, and load current based on a simulation model of the 1S4P configuration.

3.1.1. Impact of Parallel Configuration of Battery

In the 1S4P configuration, the hot-swap parallel configuration in which one battery
is inserted is divided into three cases: 1 (On) vs. 1 (Off), 2 (On) vs. 1 (Off), and 3 (On) vs.
1 (Off). Table 3 provides the simulation results with one 4.1 V (+0.4 V) or 3.3 V (−0.4 V)
battery inserted depending on the number of parallel batteries operating under the same
conditions (23 ◦ C, load 0 A, voltage 3.7 V). In the case in which N batteries are in operation,
it was confirmed that the circulating current is exchanged between N batteries and one
battery when one battery with a different charge state is inserted. Therefore, the larger the
number of batteries connected in parallel, the greater the circulating current flow in the
inserted battery. These results confirmed that it is necessary to reflect the configuration
information of the existing connected battery during the hot-swap operation.

Table 3. Effect of battery parallel configuration on hot-swap circulating current.

4.1 V (+0.4 V) 3.3 V (−0.4 V)

Battery
Configuration X_Cell Y_Cell X_Cell Y_Cell
Current (A) Current (A) Current (A) Current (A)

1 vs. 1 −3.836 A 3.836 A 3.732 A −3.732 A X vs. Y

X: Existing_cell
2 vs. 2 −2.555 A 5.11 A 2.52 A −5.041 A Y: Hot Swap_cell
Load: 0 A
3 vs. 3 −1.915 A 5.746 A 1.886 A 5.657 A Temp: 23 ◦ C

3.1.2. Effect of Battery Temperature

Figure 12 shows the results that were obtained by deriving the values of the Figure 12a
OCV and Figure 12b Ri parameters for each state of charge at each temperature (0, 10, 23,
35, and 45 ◦ C). These results indicated that, as the temperature decreased, the slope of the
OCV decreased and the internal resistance increased. In the absence of a load current and
when the 3.7 V battery is connected in parallel, a circulating current flows through the 3.3 V
inserted battery. The variation in this current with temperature is shown in Figure 13a. The
higher the temperature, the larger the magnitude of the circulating current generated under
the same conditions. These results confirmed that it is necessary to reflect the temperature
information during the hot-swap operation.
 23, 35, and 45 °C). These results indicated that, as the temperature decreased, the slope of
the OCV decreased and the internal resistance increased. In the absence of a load current
and when the 3.7 V battery is connected in parallel, a circulating current flows through
the 3.3 V inserted battery. The variation in this current with temperature is shown in Fig-
Electronics 2021, 10, 1448
ure 13a. The higher the temperature, the larger the magnitude of the circulating 11 current
of 19
generated under the same conditions. These results confirmed that it is necessary to reflect
the temperature information during the hot-swap operation.

Electronics 2021, 10, x FOR PEER REVIEW 12 of 20

(a) (b)
Figure
Figure 12. ECM12. ECM parameter
parameter derivation
derivation result byresult by varying
varying SOC temperatures:
SOC temperatures: (a) SOC-OCV;
(a) SOC-OCV; (b)i. SOC-Ri .
(b) SOC-R

3.1.3. Influence of Deviation in Battery Voltage

The circulating current generated during the hot-swap operation is generated in the
process of maintaining the energy balance from the difference in voltage (SOC) of the bat-
tery. Figure 13b shows the circulating current that is generated when hot swapping one
battery with a voltage deviation of −0.3~+0.3 V when three batteries at 23 °C, a load current
of 0 A, and 3.7 V are connected in parallel. The results of the above analysis confirmed
that, in the absence of a load current, the circulating current generated during the hot-
swap operation under the corresponding condition maintains a linear relationship with
the voltage deviation. Therefore, in the absence of a load current, the voltage deviation
and circulating current were confirmed to increase and decrease linearly during the hot-
swap operation.

(a) (b)

Figure Influence
13.13.
Figure of battery
Influence statements
of battery on on
statements hothot
swap circulating
swap current
circulating (a) (a)
current at various temperatures
at various andand
temperatures (b)(b)
as aasfunction
a function
of the voltage deviation.
of the voltage deviation.

3.1.4. Influence of Load Current

3.1.3. Influence of Deviation in Battery Voltage
TheFigure 14 shows
circulating the results
current of the
generated simulation
during analysisoperation
the hot-swap when a battery with aincharge
is generated the
state of ±0.3 V, compared to the terminal voltage, is input when
process of maintaining the energy balance from the difference in voltage (SOC) of the a 3 A/Cell discharge load
current is applied to three batteries connected in parallel with a 3.7
battery. Figure 13b shows the circulating current that is generated when hot swapping one V charge state. Figure
14a shows
battery with athe case deviation
voltage where a battery of ② +0.3
of −0.3~+0.3 V isthree
V when inputbatteries 23 ◦ C, a load
when aatdischarge current is ap-
current
of plied
0 A, and
to the3.7existing
V are connected
connectedin parallel.
battery. WhenThetheresults
resultofisthe aboveto
changed analysis
charging current ③,
confirmed
that, in the absence of a load current, the circulating current generated during
a battery with a low existing charging state receives a charging current from the external the hot-swap
operation under the corresponding condition maintains a linear relationship with the
charging current and the inserted cell. Figure 14b shows the case in which a battery of ②
voltage deviation. Therefore, in the absence of a load current, the voltage deviation and
−0.4 V is current
circulating insertedwere
when a discharge
confirmed load current
to increase is applied
and decrease to an existing
linearly during the connected
hot-swap bat-
tery.
operation.To charge the inserted battery, the existing battery supplies an additional discharge
current in addition to the discharge load current. When the load current changes to the
3.1.4. Influence
charging current at ③,
of Load Current
the latter current can confirm the result of the required current
concentration
Figure 14 shows needed thetoresults
chargeofthe
theinserted
simulationbattery. This when
analysis result aconfirmed
battery withthatathe allow-
charge
state ±0.3 voltage
ableof(+)/(−) V, compared to the
deviation terminal
range basedvoltage, is input of
on the direction when a 3 A/Cell
the load current discharge
(charge/dis-
load current
charge) is applied
is not symmetric. to three batteries
Therefore, it isconnected
necessaryin to parallel
reflect thewith a 3.7 V of
condition charge state.
the load cur-
Figure 14a shows the case where a battery of
rent when establishing the hot-swap operation condition. 2 +0.3 V is input when a discharge current is
applied to the existing connected battery. When the result is changed to charging current
,
3 a battery with a low existing charging state receives a charging current from the external
charging current and the inserted cell. Figure 14b shows the case in which a battery of 2
Electronics 2021, 10, 1448 12 of 19

−0.4 V is inserted when a discharge load current is applied to an existing connected battery.
To charge the inserted battery, the existing battery supplies an additional discharge current
in addition to the discharge load current. When the load current changes to the charging
current at ,3 the latter current can confirm the result of the required current concentration
needed to charge the inserted battery. This result confirmed that the allowable (+)/(−)
voltage deviation range based on the direction of the load current (charge/discharge) is
Electronics 2021, 10, x FOR PEER REVIEW 13 of 20
not symmetric. Therefore, it is necessary to reflect the condition of the load current when
establishing the hot-swap operation condition.

(a) (b)
Figure 14. Simulation results of the hot-swap operation when inserting a battery with ±0.3 V difference: (a) +0.3V battery
Figure 14. Simulation results of the hot-swap operation when inserting a battery with ±0.3 V difference: (a) +0.3V battery
hot
hotswap; (b)−
swap;(b) 0.3 V
−0.3 V battery
battery hot
hot swap.
swap.

3.2.
3.2. Derivation
Derivationofof Hot Swap
Hot Swap Operating
OperatingConditions
Conditions
The
The case
case study
study analysis
analysis in in Section
Section 3.1
3.1 enabled
enabled aa hot-swap
hot-swap condition
condition to to be
be derived
derived in
in
which
whichthe themagnitude
magnitudeofofthe thecirculating
circulatingcurrent
current generated
generated during
duringthethe
hot-swap
hot-swap operation in
operation
the 3 vs.
in the 1 parallel
3 vs. configuration,
1 parallel configuration, which
whichis is
the worst
the worstcondition,
condition,does
doesnot
notexceed
exceed3 3A/Cell
A/Cell
(1C). The factors considered in the hot-swap operation
(1C). The factors considered in the hot-swap operation were were the load current, temperature,
temperature,
and
and voltage deviation; the
voltage deviation; theresults
resultsappear
appearininTable
Table4.4.The
The hot-swap
hot-swap operation
operation cancan be
be ap-
applied
plied byby either
either using
using a lookup
a lookup table
table as as a table
a table or by
or by dividing
dividing a specific
a specific section
section by
by the
the temperature
temperature andandloadload current.
current. However,
However, in case
in the the case
of a of a hot-swap
hot-swap operation
operation basedbased
on a
on a table,
table, the dimension
the dimension of theof the increases
table table increases
rapidlyrapidly
with the with the addition
addition of conditions,
of conditions, and the
and
disadvantage is that a look-up table can only be used to turn on/off a simple relay.simple
the disadvantage is that a look-up table can only be used to turn on/off a There-
relay. Therefore,
fore, research is research
needed to is estimate
needed tothe estimate the circulating
circulating current generated
current generated during a
during a hot-swap
hot-swap
operation.operation.
Table 4. Hot swap operating voltage conditions for the 1S4P system.

Load
Mode 0 ◦C 10 ◦ C 23 ◦ C 35 ◦ C 45 ◦ C
(A/Cell)
3A −0.04–0.23 V −0.01–0.14 V 0.01–0.11 V 0.00–0.10 V 0.00–0.10 V
Discharge 2A −0.12–0.25 V −0.08–0.17 V −0.06–0.14 V −0.06–0.13 V −0.06–0.13 V
1A −0.23–0.28 V −0.18–0.20 V −0.14–0.17 V −0.14–0.16 V −0.13–0.16 V
0 −0.31–0.30 V −0.23–0.23 V −0.20–0.20 V −0.19–0.19 V −0.19–0.19 V
1A −0.28–0.23 V −0.20–0.15 V −0.17–0.13 V −0.16–0.12 V −0.16–0.12 V
Charge 2A −0.24–0.05 V −0.17–0.06 V −0.14–0.06 V −0.13–0.06 V −0.13–0.05 V
3A −0.21 to −0.07 V −0.14 to −0.03 V −0.11 to −0.01 V −0.10 to −0.01 V −0.10 to −0.01 V
Electronics 2021, 10, 1448 13 of 19

4. Artificial Neural Network (ANN) Models

As mentioned in the discussion of the hot-swap analysis in Section 3, various key
factors must be considered to set the hot-swap operating conditions, and a look-up table
model can be used to consider various conditions. However, as conditions are added, the
dimensions of the table increase and can only be used as simple relay on/off controls. If the
circulating current generated during the hot-swap operation can be estimated, hot-swap
execution and power management before and after the execution become possible.

4.1. Design of ANN Model (Fitnet)

The Fitnet model was used among several ANN models of the deep learning toolbox
provided by MathWorks. The Fitnet model can be fitted with any accuracy if there is
sufficient input value, as mentioned in the introduction, so it is possible to estimate the
circulating current generated during the hot-swap operation with simulation-based input
data.

4.1.1. Inputs and Target Data

For the input and target data of the Fitnet model, the results of the hot swap case study
were used as input; the data type is listed in Table 5. Four conditions (parallel configuration,
temperature, load current, and voltage deviation) derived through the case study were set
as inputs, and the circulating current generated under the corresponding condition and
the current when the load current was reversed were set as target data. For the input and
target data of the Fitnet model, the results of the hot-swap case study were used as input;
the data type is provided in Table 5. Four conditions (parallel configuration, temperature,
load current, and voltage deviation) derived through the case study were set as inputs, and
the circulating current generated under the corresponding condition and the current when
the load current was reversed were set as target data. There was a total of 500 input data
values; 70% of the input data were randomly classified as train_set, 15% as validation_set,
and the remaining 15% as test_set, after which the training proceeded.

Table 5. Inputs, targets, and data type of the ANN.

Input Parameters Unit of Measure Target Parameters Unit of Measure

Parallel State
N (1,2,3) Existing_BAT A/cell
(Ex, 2 = 2 vs. 1)
Temperature ◦C Hot_Swap_BAT A/cell
Load Current A/cell Existing_BAT A/cell
Deviation Voltage V Hot_Swap_BAT A/cell

4.1.2. Structure of ANN Model: Hidden Layers and Neurons

As shown in Figure 15, the performance of the ANN model is determined by the
hidden layers, neurons, and training functions that exist between the inputs and outputs.
As the number of hidden layers and neurons increases, the accuracy of the model may
increase, but this increases the learning time, and the overfitting problem occurs. Therefore,
the number of hidden layers and neurons was varied, and the results are provided in Table
6. The results indicate that when two hidden layers with 10 neurons in each layer are used,
the model design is optimized.
 As the number of hidden layers and neurons increases, the accuracy of the model may
increase, but this increases the learning time, and the overfitting problem occurs. There-
Electronics 2021, 10, 1448 fore, the number of hidden layers and neurons was varied, and the results are provided
14 of 19
in Table 6. The results indicate that when two hidden layers with 10 neurons in each layer
are used, the model design is optimized.

Figure 15.
Figure 15. Composition
Composition of
of the
the artificial
artificial neural
neural network
network model
model [20].
[20].

Table 6. Accuracy of the ANN 1 depending on the number of hidden layers and neurons.
Table 6. Accuracy of the ANN 1 depending on the number of hidden layers and neurons.
Hidden Layer Neurons MSE22 MAE33
Hidden Layer Neurons MSE MAE
1 5 0.02520 0.1246
11 5
10 0.02520
0.00070 0.1246
0.0177
1 10 0.00070 0.0177
11 15
15 0.00040
0.00040 0.0138
0.0138
22 55 55 0.00170
0.00170 0.0274
0.0274
2 5 10
10 0.00010
0.00010 0.0096
0.0096
22 10
10 5 5 0.00008
0.00008 0.0071
0.0071
2 10 10 0.00005 0.0055
22 10
15 15
10 0.00005
0.00002
0.0055
0.0027
23 15
5 515 5 0.00002
0.00097 0.0027
0.0196
3 555510 5 0.00097
0.00067 0.0196
0.0167
33 55 10
5 1010 0.00006
0.00067 0.0058
0.0167
3 5 10 5 0.00110 0.0238
3 5 10 10 0.00006 0.0058
3 10 10 5 0.00200 0.0329
33 105 10
10 105 0.00110
0.00002 0.0238
0.0035
3
1 Training function: 10 10train_set
Levenberg–Marquardt, 5 0.00200
70%, validation_set 0.0329
15%, test_set 15%, 2 mean square error,
3 error.
3 mean absolute 10 10 10 0.00002 0.0035
1Training function: Levenberg–Marquardt, train_set 70%, validation_set 15%, test_set 15%, 2 mean
4.1.3. Training
square Function
error, 3 mean absolute error.
The process that was followed to determine the number of hidden layers and neurons
4.1.3.
for theTraining
ANN model Function
is described in Section 4.1.2. In this section, the training functions are
The process
compared that wasthe
to determine followed to determine
most suitable the number
function. The deep oflearning
hidden layers
toolboxandprovided
neurons
by MathWorks
for the ANN model provides approximately
is described in Section12 training functions.
4.1.2. In this section,The
themodel was
training configured
functions are
by using each function under the same conditions, which enabled
compared to determine the most suitable function. The deep learning toolbox providedthe performance of the
model
by for each provides
MathWorks function approximately
to be confirmed 12(Table
training7). functions.
In this model, whichwas
The model is designed to
configured
estimate
by usingtheeachcirculating
function current
under thefrom the conditions,
same results in Table 7, the
which trainlmthe
enabled function appliedofwith
performance the
the Levenberg–Marquardt
model for each function toalgorithm
be confirmedand the trainbr
(Table function
7). In applied
this model, withisthe
which Bayesian
designed to
regularization
estimate algorithmcurrent
the circulating deliveredfrom superior estimation
the results in Table performance,
7, the trainlmcompared
functionwith the
applied
other the
with training functions. Of thesealgorithm
Levenberg–Marquardt two functions,
and thethe trainbr
trainlm function
function applied
was significantly
with the
faster than the trainbr function; therefore, the trainlm function
Bayesian regularization algorithm delivered superior estimation performance, was used in compared
the ANN
model
with the[30].
other training functions. Of these two functions, the trainlm function was signif-
icantly faster than the trainbr function; therefore, the trainlm function was used in the
4.2. Validation of ANN Model
ANN model [30].
The learned ANN model was validated on the basis of the hot-swap simulation results
and hot-swap
4.2. Validation of experimental
ANN Model results obtained for the 1S4P battery packs.

4.2.1.The learned ANN model

Simulation-Based was validated
ANN Model on the basis of the hot-swap simulation re-
Validation
sults and hot-swap experimental results obtained for the 1S4P battery packs.
In the 3 vs. 1 configuration, the charging state of the three cells that are already
connected is 3.7 V, and the circulating current is generated when a battery with a load
current of −1A/Cell and a battery with +0.13 V, compared to the terminal voltage, are
input. The direction of the load current changes after 3 s. The flowing load current was
compared with the simulation model. The results obtained using the following input
values for the ANN model are shown in Figure 16: module configuration 3, temperature
(23 ◦ C), load current −1 A/Cell, and voltage deviation 0.13 V. As a result of comparing
 Electronics 2021, 10, 1448 15 of 19

the simulation results in Table 8 with the ANN model, it can be seen that the circulating
current estimation performance of the ANN model operated around 3%.

Table 7. Results of performance analysis of model according to training function.

Training Function Algorithm MSE MAE

trainlm Levenberg–Marquardt 0.00006 0.0055
trainbr Bayesian Regularization 0.00003 0.0038
trainbfg BFGS Quasi-Newton 0.0023 0.0358
trainrp Resilient Backpropagation 0.0139 0.0898
trainscg Scaled Conjugate Gradient 0.0088 0.0702
traincgb Conjugate Gradient with Powell/Beale Restarts 0.0094 0.0738
traincgf Fletcher–Powell Conjugate Gradient 0.0314 0.1376
traincgp Polak–Ribiére Conjugate Gradient 0.0272 0.1262
trainoss One-Step Secant 0.0322 0.1416
traingdx Variable Learning Rate Gradient Descent 0.2575 0.4019
traingdm Gradient Descent with Momentum 35.2494 4.9497
ronics 2021, 10, x FOR PEER REVIEW traingd Gradient Descent 17 of 20
0.2088 0.3584

Figure 16. Verification

Figure 16.ofVerification
the results obtained for the
of the results hot-swap
obtained forcirculating current
the hot-swap with the
circulating ANN with
current model.
the(a)
ANNANN input
data, (b) ANNmodel.
Estimation Circulation
(a) ANN Current.
input data, (b) ANN Estimation Circulation Current.

Table 8. Comparison of ANN model with simulation results.

Cell_1 Cell_4 (New)

Load Reverse Load Load Reverse Load
Simul ANN Simul ANN Simul ANN Simul ANN
1.435 A 1.425 A −1.47 A −1.47 A 0.065 A 0.063 A 2.801 A 2.815 A
 Electronics 2021, 10, 1448 16 of 19

Table 8. Comparison of ANN model with simulation results.

Cell_1 Cell_4 (New)

s 2021, 10, x FOR PEER REVIEW Load Reverse Load Load Reverse Load
18 of 20
Simul ANN Simul ANN Simul ANN Simul ANN
1.435 A 1.425 A −1.47 A −1.47 A 0.065 A 0.063 A 2.801 A 2.815 A
and 3 currents, estimated using the ANN model, are 1.495 A, −0.15 A, and −1.32 A, which
are indicated by4.2.2.
circles in Figure 17. Table
Experiment-Based ANN9 summarizes the magnitude of the peak cur-
Model Validation
rent generated duringFigure
the hot17swap
shows the experimental results and
from the above experiment simulation
in which cell 2 results.
and cellAs
3 are sequentially
shown in Table hot-swapped
9, it can be confirmed that the peak value of the circulating current that
at 10 s intervals while battery cell 1 is connected to the DC bus in a 1S4P
occurs when thebattery
cell to be hot-swapped
pack. is connected
At this time, to theofDC
the magnitude thebus can be estimated
circulating with
current estimated in the ANN
the ANN modelmodel
proposed in this paper.
is shown together with a circle.

Figure 17.Figure 17. Experimental

Experimental result
result of hot of hot
swap swap operation
operation of three batteries
of three batteries with deviation
with voltage voltage deviation and
and simulation result of
simulation result of estimated circulating current of ANN model.
estimated circulating current of ANN model.

Table 9. Comparison ofAs

peak currentofestimation
a result the 1 vs.results
1 hot and experimental
swap resultsdifference
with a voltage during hot of
swap.
0.101 V at the time
of 1010s,sit was experimentally measured that the 20 discharge
s current of 0.94 A in cell 1 and
Cell_1 the charging current
Cell_2 (New) of − 0.95 A in cell
Cell_12 occurred. At Cell_3time,
this (New) it can be seen that the
ANN model estimates that the circulating current is 0.99 A for cell 1 and −0.96 A for cell 2,
Measured ANN Measured ANN Measured ANN Measured ANN
and the current generated during the hot swap is estimated to be within ±0.05 A. At the
0.94 A 0.99 A point
time −0.95of A20 s, −0.96
cell 3 A
is hot1.49 A
swapped 1.495 Acell 1−1.32
while A 2 −1.38
and cell A
are connected. From the
experimental results, it can be seen that the peak values of the circulating currents of cells
5. Conclusions 1, 2, and 3 are 1.49 A, −0.17 A, and −1.32 A. Under the same experimental conditions, cell
In this study,
1, 2,a and
method for estimating
3 currents, estimated theusing
circulating
the ANN current generated
model, are 1.495during the A, and −1.32 A,
A, −0.15
hot-swap operation which applied to multiple
are indicated series and
by circles multiple
in Figure parallel
17. Table battery systems
9 summarizes using
the magnitude of the peak
the ANN modelcurrentwas proposed. The circulating current generated during the hot swap
generated during the hot swap from the above experiment and simulation results. is
determined not As only by simple
shown differences
in Table in the
9, it can be state ofthat
confirmed charge, but also
the peak valuebyofcomplex re-
the circulating current that
lationships, suchoccurs
as thewhen the configuration
parallel cell to be hot-swapped
of multiple is connected to the DC
parallel systems, bus can be estimated with
temperature
and load current.the ANN model
Therefore, proposedtoincalculate
it is difficult this paper.and estimate the circulating current
considering various conditions. In this study, the multi-simulation results used in the hot-
swap analysis to build an artificial neural network (ANN) model can be used to estimate
circulation currents for training battery configuration, temperature, load current, and
power deviation conditions. For performance validation of the ANN model reflecting the
above four conditions (parallel configuration, temperature, current, i.e., deviation), when
compared with the hot-swap simulation results, the circulation current estimate was
found to be around 97%. We also performed periodic current estimation for approxi-
Electronics 2021, 10, 1448 17 of 19

Table 9. Comparison of peak current estimation results and experimental results during hot swap.

10 s 20 s
Cell_1 Cell_2 (New) Cell_1 Cell_3 (New)
Measured ANN Measured ANN Measured ANN Measured ANN
0.94 A 0.99 A −0.95 A −0.96 A 1.49 A 1.495 A −1.32 A −1.38 A

5. Conclusions
In this study, a method for estimating the circulating current generated during the
hot-swap operation applied to multiple series and multiple parallel battery systems using
the ANN model was proposed. The circulating current generated during the hot swap
is determined not only by simple differences in the state of charge, but also by complex
relationships, such as the parallel configuration of multiple parallel systems, temperature
and load current. Therefore, it is difficult to calculate and estimate the circulating current
considering various conditions. In this study, the multi-simulation results used in the hot-
swap analysis to build an artificial neural network (ANN) model can be used to estimate
circulation currents for training battery configuration, temperature, load current, and
power deviation conditions. For performance validation of the ANN model reflecting the
above four conditions (parallel configuration, temperature, current, i.e., deviation), when
compared with the hot-swap simulation results, the circulation current estimate was found
to be around 97%. We also performed periodic current estimation for approximately 94%
of the ANN model performance for the 1S4P battery pack.
The ANN model proposed in this study was validated in a 1s4p system with a small
number of cells connected in parallel. Therefore, in addition to this study, it is necessary
to design an ANN estimation model for a battery system with more series and parallel
connections to verify its estimation performance. In addition, it is necessary to study
the algorithm to increase the safety and efficiency of the multi-parallel battery system by
applying the estimated circulating current estimation result to the power management
algorithm.

Author Contributions: N.-G.L. contributed to the main ideas of this study and wrote the paper. J.-Y.K.
and S.L. comprehensively revised the paper. All authors approved the final version of the manuscript
for publication. All authors have read and agreed to the published version of the manuscript.
Funding: This study was supported by the Korea Institute for Advancement of Technology (KIAT)
grant funded by the Korean Government (MOTIE) (P0002092, The Competency Development Pro-
gram for Industry Specialist). This work was supported by the National Research Foundation of
Korea (NRF) grant funded by the Korea government (MSIT) (No. NRF-2021R1F1A1063150).
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations

SOC State of Charge

SOH State of Health
ECM Equivalent Circuit Model
BMS Battery Management System
EV Electric Vehicles
ESS Energy Storage System
ANN Artificial Neural Network
Fitnet Fitting Network
MSE Mean Square error
MAE Mean Absolute error
LUT Look up Table
OCV Open Circuit Voltage
Electronics 2021, 10, 1448 18 of 19

References
1. Gabar, H.A.; Othman, A.M.; Abdussami, M.R. Review of Battery Management Systems(BMS) Development and Industrial
Standards. Technologies 2021, 9, 28. [CrossRef]
2. Lelie, M.; Braun, T.; Knips, M.; Nordmann, H.; Ringbeck, F.; Zappen, H.; Sauer, D.U. Battery Management System Hardware
Concepts: An Overview. Appl. Sci. 2018, 8, 534. [CrossRef]
3. Luo, X.; Kang, L.; Lu, C.; Linghu, J.; Lin, H.; Hu, B. An Ehanced Multicell-to-Multicell Battery Equalizer Based on Bipolar-
Reseonant LC Converter. Electronics 2021, 10, 293. [CrossRef]
4. Aiello, O. Electromagnetic Susceptibility of Battery Management Systems’ ICs for Electric Vehicles: Experimental Study. Electronics
2020, 9, 510. [CrossRef]
5. Doridant, A.; Abouda, K.; Givelin, P.; Thibaud, B. Battery Management System Demonstrator Board design using EMC System
simulation. In Proceedings of the 2019 International Symposium on Electromagnetic Compatibility—EMC EUROPE, Barcelona,
Spain, 2–6 September 2019; pp. 427–432.
6. Wei, Z.; Zhao, J.; He, H.; Ding, G.; Cui, H.; Liu, L. Future smart battery and management: Advanced sensing from external to
embedded multi-dimensional measurement. J. Power Sources 2021. [CrossRef]
7. Baronti, F. Investigation of series-parallel connections of multi-module batteries for electrified vehicles. In Proceedings of the
IEEE International Electric Vehicle (IEVC) (2014), Florence, Italy, 17–19 December 2014; pp. 1–7.
8. Jiang, Y.; Habib, A.H.; Zhao, X.; Shrinkle, L.J.; Callafon, R. Centralized recursive optimal scheduling of parallel buck regulated
battery modules. Proceedings of IEEE Conference on Decision and Control (CDC) (2017), Melbourne, VIC, Australia, 12–15
December 2017; pp. 2029–2034.
9. Moo, C.-S.; Ng, K.S.; Hsieh, Y.-C. Parallel Operation of Battery Power Modules. IEEE Trans. Energy Convers. 2008, 23, 701–707.
10. Hsu, C.T.; Sheng, S.; Sun, Q.; Li, P.; Lehman, B. Increased energy delivery for parallel battery packs with no regulated bus. In
Proceedings of the Intelec 2012, Scottsdale, AZ, USA, 30 September–4 October 2012; pp. 1–8.
11. Song, C.; Shao, Y.; Song, S.; Chang, C.; Zhou, F.; Peng, S.; Xiao, F. Energy management of parallel-connected cells in electric
vehicles based on fuzzy logic control. Energies 2017, 10, 404. [CrossRef]
12. Lee, S.; Kim, J. Inrush current estimation and hot-swapping for safe parallel battery pack. Int. J. Electron. 2020, 107, 1609–1624.
[CrossRef]
13. Lee, S.; Kim, J.; Ha, M.; Song, H. Inrush current estimation for hot swap of the parallel connected large capacity battery pack.
Proceedings of IEEE Energy Conversion Congress and Exposition (ECCE) (2018), Portland, OR, USA, 23–27 September 2018; pp.
2489–2492.
14. Chou, C.J.; Jiang, S.B.; Yeh, T.L.; Sun, C.C. Fault-Tolerant Battery Power Network Architecture of Networked Swappable Battery
Pack in Parallel. Energy 2012, 14, 2841.
15. Lin, H.-T.; Liang, T.-J.; Chen, S.-M. Estimation of Battery State of Health Using Probabilistic Neural Network. IEEE Trans. Ind.
Inform. 2013, 9, 679–685. [CrossRef]
16. Mahmoud, I.; Rioch, D.; Ahmed, E.; Ryan, A.; Saeid, H. Battery state of charge estimation using an Artificial Neural Network. In
Proceedings of the 2017 IEEE Transportation Electrification Conference and Expo (ITEC) (2017), Chicago, IL, USA, 22–24 June
2017; pp. 342–349.
17. Farmann, A.; Waag, W.; Marongiu, A.; Sauer, D.U. Critical review of on-board capacity estimation techniques for lithium-ion
batteries in electric and hybrid electric vehicles. J. Power Sources 2015, 281, 114–130. [CrossRef]
18. Zhu, Q.; Xu, M.; Liu, W.; Zheng, M. A state of charge estimation method for lithium-ion batteries based on fractional order
adaptive extended kalman filter. Energy 2019, 187, 115880. [CrossRef]
19. Ismail, M. Estimation Strategies for Training of Deep Learning Neural Networks. Ph.D. Thesis, McMaster University, Hamilton,
ON, USA, 2018.
20. Math Works Deep Learning Tool Box. Neural Network Patternnet. Available online: https://mathworks.com/products/deep-
learning.html#net (accessed on 16 June 2021).
21. Romero, A.; Ballas, N.; Kahou, S.E.; Chassang, A.; Gatta, C.; Bengio, Y. Fitnets: Hints for thin deep nets. arXiv 2014,
arXiv:1412.6550.
22. Hemavathi, S. Overview of Cell Balancing Method for Li-ion Battery Technology. Energy Storage 2020, 203, 1–12.
23. Hu, J.; He, H.; Wei, Z.; Li, Y. Disturbance-Immune and Aging-Robust Internal Short Circuit Diagnostic for Lithium-Ion Battery,
IEEE Trans. Ind. Electron. 2021. [CrossRef]
24. Wei, Z.; He, H.; Pou, J.; Tsui, K.L.; Quan, Z.; Li, Y. Signal-Disturbance Interfacing Elimination for Unbiased Model Parameter
Identification of Lithium-Ion Battery. IEEE Trans. Ind. Inform. 2020. [CrossRef]
25. Wei, Z.G.; Hu, J.; He, H.; Li, Y.; Xiong, B. Load Current and State of Charge Co-Estimation for Current Sensor-Free Lithium-ion
Battery. IEEE Trans. Power Electron. 2021. [CrossRef]
26. Westerhoff, U.; Kurbach, K.; Lienesch, F.; Kurrat, M. Analysis of Lithium-Ion Battery Models Based on Electrochemical Impedance
Spectroscopy. Energy Technol. 2016, 4, 1620–1630. [CrossRef]
27. Robyn, J.; Michael, S.; Pravesh, S.; Javier, G.; Tarun, H.; Massimo, C. SAE Technical Paper Series—Battery Model Parameter
Estimation Using a Layered Technique: An Example Using a Lithium Iron Phosphate Cell. In Proceedings of the SAE International
SAE 2013 World Congress & Exhibition, Detroit, MI, USA, 16–18 April 2013; pp. 1–14.
Electronics 2021, 10, 1448 19 of 19

28. Zhang, L.; Peng, H.; Ning, Z.; Mu, Z.; Sun, C. Comparative Research on RC Equivalent Circuit Models for Lithium-Ion Batteries
of Electric Vehicles. Appl. Sci. 2017, 7, 1002. [CrossRef]
29. Renault Nissan Technology & Business Centre India Private Limited, “Battery Parameter Estimation”. Available on-
line: https://www.mathworks.com/content/dam/mathworks/mathworks-dot-com/images/events/matlabexpo/in/2018
/lithium-ion-battery-parameter-estimation-for-hil-sil-and-mil-validation.pdf/ (accessed on 19 April 2018).
30. Hagan, M.T.; Menhaj, M.B. Training feedforward networks with the Marquardt algorithm. IEEE Trans. Neural Netw. 1994, 5,
989–993. [CrossRef] [PubMed]