energies

Review
Review of the Li-Ion Battery, Thermal Management, and
AI-Based Battery Management System for EV Application
Maryam Ghalkhani * and Saeid Habibi *

Department of Mechanical Engineering, McMaster University, Hamilton, ON L8S 4L8, Canada

* Correspondence: ghalkham@mcmaster.ca (M.G.); habibi@mcmaster.ca (S.H.)

Abstract: With the large-scale commercialization and growing market share of electric vehicles (EVs),
many studies have been dedicated to battery systems design and development. Their focus has been
on higher energy efficiency, improved thermal performance and optimized multi-material battery
enclosure designs. The integration of simulation-based design optimization of the battery pack
and Battery Management System (BMS) is evolving and has expanded to include novelties such as
artificial intelligence/machine learning (AI/ML) to improve efficiencies in design, manufacturing,
and operations for their application in electric vehicles and energy storage systems. Specific to BMS,
these advanced concepts enable a more accurate prediction of battery performance such as its State of
Health (SOH), State of Charge (SOC), and State of Power (SOP). This study presents a comprehensive
review of the latest developments and technologies in battery design, thermal management, and the
application of AI in Battery Management Systems (BMS) for Electric Vehicles (EV).

Keywords: lithium-ion batteries; battery management systems; AI-based monitoring systems; electric
vehicle

1. Introduction
1.1. A Brief History of EVs
The first EVs were built around the mid to late 19th century but conceded their
Citation: Ghalkhani, M.; Habibi, S. commercial footprint to cars powered by Internal Combustion Engines (ICE) [1]. Today’s
Review of the Li-Ion Battery, Thermal transportation sector primarily uses ICE, contributing to almost a quarter of all energy-
Management, and AI-Based Battery related greenhouse gas emissions. This issue initiated the demand for replacing ICE
Management System for EV vehicles with advanced technology vehicles such as EVs. Although EVs can reduce fuel
Application. Energies 2023, 16, 185. costs significantly because of the high efficiency of electric-drive systems compared to
https://doi.org/10.3390/en16010185 internal combustion engines, EVs suffer much greater constraints in terms of their limited
driving range, scarcity of charging stations, charging times, and higher initial costs as
Academic Editor: Carlos
Miguel Costa
compared to ICE vehicles [2–4]. As such, an integrative review would be suitable to
understand the development of this emerging topic by providing a clear understanding of
Received: 30 September 2022 what are the key barriers and motivators of EV adoption on the sustainability dimensions.
Revised: 6 December 2022
Accepted: 9 December 2022 1.2. A Brief History of LIBs
Published: 24 December 2022
For the past 3 decades major commercial and academic progress has been made in the
development of Li-based battery technologies. This has been driven by the market demand
for high-performance rechargeable batteries to reduce the cost and weight of EVs while
Copyright: © 2022 by the authors.
increasing their range and longevity.
Licensee MDPI, Basel, Switzerland. Around 30 years ago Sony Co. commercialized the world’s first lithium-ion battery
This article is an open access article (LIB) which led to a large increase in research in battery technologies. The research was fu-
distributed under the terms and eled by environmental concerns and the impact of fossil fuels on greenhouse gas emissions
conditions of the Creative Commons Governments around the world subsequently have invested considerably in support of
Attribution (CC BY) license (https:// green technologies (solar, wind, etc.) and electric vehicles [5].
creativecommons.org/licenses/by/ Lithium-ion batteries (LIBs) store energy through the storage of charge through the
4.0/). motion of lithium ions between positive and negative electrodes via a liquid electrolyte.

Energies 2023, 16, 185. https://doi.org/10.3390/en16010185 https://www.mdpi.com/journal/energies

Energies 2022, 15, x FOR PEER REVIEW 2 of 16

Energies 2023, 16, 185 2 of 16

Lithium-ion batteries (LIBs) store energy through the storage of charge through the
motion of lithium ions between positive and negative electrodes via a liquid electrolyte.
The cathode is usually made of graphite while the anode can be made of various types of
The cathode is usually made of graphite while the anode can be made of various types of
lithium oxides. Many studies provide a comprehensive review of the properties of differ-
lithium oxides. Many studies provide a comprehensive review of the properties of different
ent LIB chemistries such as the lithium Iron Phosphate cell (LiFePO4), lithium manganese
LIB chemistries such as the lithium Iron Phosphate cell (LiFePO4 ), lithium manganese
oxide (LiMn2O4), lithium manganese oxide, or Lithium Cobalt Oxide (LiCoO2) [6–10].
oxide (LiMn2 O4 ), lithium manganese oxide, or Lithium Cobalt Oxide (LiCoO2 ) [6–10].
While LIBs provide an increased energy density and cycle life from the previous gen-
While LIBs provide an increased energy density and cycle life from the previous
eration of batteries, with continuous technological advancements, they are operating in-
generation of batteries, with continuous technological advancements, they are operating
creasingly closer to their theoretical limit [11]. Therefore, in addition to further improving
increasingly closer to their theoretical limit [11]. Therefore, in addition to further improving
the design and manufacturing process of current LIBs, research efforts have focused on
the design and manufacturing process of current LIBs, research efforts have focused on
developingnext-generation
developing next-generationlithium
lithiumbatteries,
batteries,such
suchasassolid-state
solid-stateandandmetal-air
metal-airbatteries.
batteries.
LIBs are typically made of four major parts: cathode, anode, separator,
LIBs are typically made of four major parts: cathode, anode, separator, and electrolyte. and electro-
lyte. Figure 1 provides a schematic of a battery cell during the charge/discharge
Figure 1 provides a schematic of a battery cell during the charge/discharge process. The process.
The battery
battery is filled
is filled withwith an electrolyte
an electrolyte to help
to help thethe movement
movement ofoflithium
lithiumions
ionsbetween
betweenthe the
electrodes. The arrows indicate the direction and magnitude of the current
electrodes. The arrows indicate the direction and magnitude of the current flowing into flowing into
thecurrent
the currentcollectors.
collectors.TheThecurrent
currentfrom
fromthetheexternal
externalcircuit
circuitflows
flowsinto
intothe
thecell,
cell,through
throughthethe
tabs and then flows into all the local regions containing active materials distributed
tabs and then flows into all the local regions containing active materials distributed on the on the
entirecurrent
entire currentcollectors
collectors[7].
[7].

Figure1.1.Schematic
Figure Schematicofofaaunit
unitcell
cellof
ofthe
thebattery
batteryincluding
includingpositive
positiveand
andnegative
negativeporous
porouselectrodes,
electrodes,aa
porous separator, and current collectors during the charge and discharge process
porous separator, and current collectors during the charge and discharge process [7].[7].

Severalreview
Several reviewpapers
paperson onbattery
batterysafety
safetyhave
havebeen
been recently
recently published,
published, covering
covering top-
topics
ics such
such as cathode
as cathode andandanodeanode materials,
materials, electrolytes,
electrolytes, advanced
advanced safety
safety batteries,
batteries, andand bat-
battery
tery thermal
thermal runaway runaway
issues issues
[12,13].[12,13].
Amongst Amongst all the known
all the known batterymodes,
battery failure failurethe
modes, the
internal
internal
short short
is one ofisthe
onemajor
of thesafety
majorconcerns
safety concerns
for the for the lithium-ion
lithium-ion battery battery industry.
industry. AnotherAn-
important consideration
other important is temperature
consideration variation
is temperature leading
variation to thermal
leading runaways.
to thermal The cell
runaways. The
temperature
cell temperatureis affected by the
is affected bynumber and thickness
the number of electrode
and thickness layers
of electrode and and
layers generally the
generally
battery size. size.
the battery Therefore, controlling
Therefore, heat generation
controlling is extremely
heat generation important
is extremely for high-power
important for high-
devices, such assuch
power devices, electric vehiclevehicle
as electric batteries, wherewhere
batteries, excessive heat can
excessive heatcause damage
can cause and
damage
reduce the battery’s longevity lifetime
and reduce the battery’s longevity lifetime [7]. [7].
Thermal
Thermal runaway
runaway isisaacontinuous,
continuous,temperature
temperatureincrease
increaseinside
insideaaLi-ion
Li-ionbattery
batterythat
that
can
canbebeprecipitated
precipitatedby bymanufacturing
manufacturing defects
defects within
within the
the battery
battery oror by
by overheating,
overheating, over-
over-
charging,
charging, or orshort
shortcircuit.
circuit. When
When internal
internal heating
heating extends
extends to toaamaximum
maximumtemperature,
temperature, itit
causes the generation of gases, and the increase in internal pressure that leads to battery
rupture, fire, and explosion [14,15]. The maximum surface temperature can rise to 943◦ for
a fully charged 18,650 battery [16].
Energies 2023, 16, 185 3 of 16

1.3. Batteries for EVs

There are currently three main types of anode materials used in modern electric vehicles
(EVs) batteries, namely Lithium Iron Phosphate (LFP), Lithium Nickel Cobalt Aluminum
Oxide (NCA), and lithium nickel manganese cobalt oxide (NMC) [6,7,17]. LIBs have had a
considerable impact on the EV industry in terms of high energy density, lifespan, nominal
voltage, power density, and cost. While they provide an increased energy density and cycle life
from the previous generation of batteries, with continuous technological advancements, they
are operating increasingly closer to their theoretical limit [11]. The different chemistries vary in
energy density, power capability, cost, and cycle life among other areas. A hybrid battery pack
consists of two or more chemistries so it can combine the advantages of different chemistries to
improve the overall pack performance. Therefore, in addition to further improving the design
and manufacturing process of current LIBs, research efforts have focused on developing
next-generation batteries, such as solid-state batteries, metal-air, and lithium-sulphur cell
chemistries. Solid-state batteries (SSBs) replace the liquid electrolyte used in LIBs with a
solid compound that allows the movement of lithium ions. Compared to LIBs, they have
the potential to provide much higher energy densities and therefore longer driving ranges
for EVs. The solid electrolyte is also less flammable than the liquid electrolytes used in
LIBs, which helps prevent thermal runaway in battery cells. The current challenges in SSB
developments include the interfacing of various layers and costs. The current consensus is
that SSBs will likely become the leading direction of the next-generation EV batteries [18].
Indeed, many leading automotive manufacturers, such as Toyota, Volkswagen, BMW, Ford,
General Motors, Stellates, Mercedes-Benz and Hyundai, have invested in solid-state battery
technologies [19,20].
Lithium-sulfur (Li-S) battery is another type of battery with the potential to become
the next generation of cells to be used in energy storage systems because of their extremely
high theoretical energy density. Li-S cells use lithium metal as the anode and sulphur
as the cathode. During the charge/discharge process, a reversible redox reaction occurs
between the lithium and sulphur instead of the intercalation that occurs in LIBs, offering
a much higher theoretical energy density of 2500 Wh/kg. A breakthrough in 2009 that
achieved stable cycling of over 20 cycles in Li-S cells was followed by extensive research
to improve their specific capacity and cycle life [21]. In March 2022, American battery
manufacturer Sion Power announced that they have achieved more than 2500 cycles with
their 17 Ah Licerion Electric Vehicle cells which target the automotive market [22]. The
German battery start-up Theion featured in a recent Forbes article [23] also has plans to
deliver Li-S batteries suitable for automotive applications by 2024. While the technology
is still in the prototyping phase, with the extremely high theoretical energy density and
low dependence on rare earth metals, Li-S batteries have the potential to become the future
generation cells after or combined with SSBs.
In addition to various types and generations of Li-ion batteries, metal-air batteries are
also considered by some to be a viable future alternative for their high capacity and energy
density [24,25]. The anode is made of a metal, such as lithium, aluminum-magnesium,
etc., and the external air acts as the cathode. Compared to Li-ion batteries, metal-air
batteries are still in a relatively early stage of development and may not see large-scale
commercialization before the next iteration of Li-ion batteries.
To develop high energy density battery chemistries, combing different cell chemistries
is another way to improve the overall performance or reduce the cost of the battery pack
without significantly increasing its mass and volume. The different chemistries vary in
energy density, power capability, cost, and cycle life. A hybrid battery pack consists of two
or more chemistries so it can combine the advantages of different chemistries to improve
the overall pack performance.
Several small-to-medium-scale companies have been trying to commercialize hybrid
battery packs for use in EVs. UK start-up Chimera Energy [26] or Upgrade Technology
Engineering [27], the same team under different company names aims to develop a multi-
chemistry battery system to leverage the advantages of different chemicals and improve
Energies 2023, 16, 185 4 of 16

overall performance. They also own a patent on the control strategy of such battery
packs [28].
In 2019, the UK company Williams Advanced Engineering developed an “Adaptive
Multi-Chem” technology that combines high-energy-density Nickel Manganese Cobalt
(NMC) pouch cells and high-power-density NMC cylindrical cells in one pack to improve
the overall performance [29,30]. Finally, Italian sports car start-up Automobile Estrema
announced their Fulminea project, which uses a hybrid battery pack combining superca-
pacitors and solid-state Li-ion cells to provide the best sports performance. The vehicles
are expected to be delivered to customers in 2023 [31].
Apart from their use in EVs, hybrid battery packs can also be useful in microgrid
applications. They not only allow the combination of different battery chemistries as in
EVs but their combination with recycled second-life automotive batteries. In 2020, Japanese
battery manufacturer GS Yuasa built an energy storage system combining lead-acid and
Li-ion batteries to reduce costs [32].
The Li-ion batteries are used to provide high energy output for EV charging and
the lead-acid batteries are used to capture the renewable energy generated from nearby
photovoltaic arrays [33]. Retired EV batteries can still be used in other applications where
the power capability is not as critical as in the automotive sector. A group from the
University of Oxford is exploring the possibility of combining different chemistry batteries
to provide a low-cost energy storage solution in sub-Saharan countries [34]. This topic was
also studied by a group from Oak Ridge National Laboratory, where a system architecture
for a multi-chemistry second-life battery system was proposed to integrate BMS and power
electronic converters from multiple manufacturers [35].
While hybrid battery packs have the potential to provide cost reduction and moderate
improvement in the overall energy and power performance, they cannot provide the signif-
icant improvement offered by SSBs and Li-S batteries. Therefore, major EV manufacturers
are more interested in using novel cell chemistries than SSbs [36].

2. Battery Thermal Conditioning

The battery thermal management system is one of the main aspects of an EV. Therefore,
a major focus area for research is the optimal working temperature range of batteries during
charging/discharging mode and the associated thermal management.
Since thermal runaway leads to fire in battery packs [37], regulating the battery
temperature within a safe range of 25 ◦ C to 40 ◦ C during charging and discharging cycles
in EVs is essential for the battery’s longevity and safety [38]. A 10 Ah pouch Li-ion cell
has been tested at various C-rates to measure the time-dependent temperature behaviour
constant current charge/discharge. The battery was placed in a small chamber with a fixed
temperature of 14 ◦ C. Figure 2a–c demonstrates the temperature variation of the battery
surface versus time at 0.1, 0.5 and 0.8 C rates, respectively. Figure 2d displays the positions
of thermocouples (T1–T4) on the battery surface. These results show the growth of surface
temperature close to the positive tab for T1 about 0.6 ◦ C, 6.4 ◦ C and 10.9 ◦ C above the initial
temperature after one charge/discharge cycle at 0.1 C, 0.5 C, and 0.8 C, respectively [7,17].
The study was expanded at a higher C rate to validate the battery thermal behaviour,
and the simulation results are compared with the experimental measurements at 0.5 C,
1 C, 2 C and 4 C discharge rates as reported in Figure 3. The results suggest that an
electrochemical-thermal model can reliably estimate the thermal behaviour of an NCA
Li-ion cell batteries, enabling a rapid strategy to investigate the temperature distribution
of the cell under different C-rates [7,17]. The temperature increase at the end of 0.5 C
discharge is about 3.8 ◦ C, and the cell temperature significantly rises above the ambient
temperature as the C-rate grows.
Energies 2022, 16,
Energies 2023, 15, 185
x FOR PEER REVIEW 55 of
of 16

temperature profiles
Figure 2. Surface temperature profiles at
at (a)
(a) 0.1
0.1 C-rate,
C-rate,(b)
(b) 0.5
0.5 C-rate
C-rateand
and(c)(c)0.8
0.8C-rate.
C-rate.(d)
(d)Tempera-
Temper-
atureisismeasured
ture measuredclose
closetotothe
thepositive
positivetab
tab(T1,
(T1,blue
bluelines),
lines), close
close to
to the negative tab (T2, orange lines),
center (T3,
center (T3, red
red lines)
lines) and
and at at the
the bottom
bottom ofof the
the battery
battery (T4,
(T4, yellow
yellow lines)
lines) [7].
[7].

The study was

Numerical expanded atinathe
investigations higher C rate showed
literature to validate thethe
that battery thermal of
conductivity behaviour,
the elec-
and theincreases
trolyte simulation results
with are compared
temperature, causingwith thecurrent
more experimental measurements
to be directed to hotteratsections
0.5 C, 1
of
C, the
2 C battery
and 4 C[17,39,40].
discharge A range
rates of thermal
as reported in management
Figure 3. The strategies are commonly
results suggest used
that an electro-
in battery packs tomodel
chemical-thermal ensurecan
temperature uniformity
reliably estimate amongst
the thermal cells andoffor
behaviour anmaintaining the
NCA Li-ion cell
temperature within the range [41]:
batteries, enabling a rapid strategy to investigate the temperature distribution of the cell
•underAirdifferent
cooling C-rates [7,17]. The temperature increase at the end of 0.5 C discharge is
•aboutLiquid
3.8 °C,cooling
and the cell temperature significantly rises above the ambient temperature
•as the C-rate grows.
Phase change materials (PCM)
• Numerical
Heat pipes investigations in the literature showed that the conductivity of the elec-
•trolyte
andincreases with temperature,
a combination of the abovecausing more current to be directed to hotter sections
strategies.
of theThe
battery [17,39,40]. A range of thermal management
liquid and air cooling systems are considered strategies
active coolingare commonly
systems since used
they
in battery packs to ensure temperature uniformity amongst cells and for
include external pumps, fans, and other auxiliary systems [42]. Active cooling systems maintaining the
temperature within the range [41]:
demand more space and power, while PCMs are known as passive thermal management
•
systems
Air and do not generally require additional components for their operation [43]. Key
cooling
•
considerations in the selection of thermal management systems include energy efficiency,
Liquid cooling
•
cell temperature
Phase change uniformity,
materials and overall weight and volume.
(PCM)
• Mengliang
Heat pipes et al. proposed a heat pipe and refrigerant-based BTMS coupled with
• air-conditioning
an and a combinationsystem forabove
of the a battery module to investigate the battery temperature
strategies.
distribution, and energy efficiency of the BTMS [44]. The results reveal that once the
initial temperature is increased from 25 ◦ C to 30 ◦ C and 35 ◦ C, the average coefficient
of performance increased by 16.95% and 38.41%, respectively; and the average exergy
efficiency of the BTMS improved by 2.63% and 5.07%, respectively [38].
Energies 2022,16,
Energies2023, 15,185
x FOR PEER REVIEW 66 of
of16
16

Figure3.3.Numerical
Figure Numericaland
andexperimental
experimentalvalidation
validationofofaverage
averagesurface
surfacetemperature
temperatureduring
duringvarious
various
discharge rates [17].
discharge rates [17].

The liquid
Kokkula andinvestigated
et al. air cooling the systems
thermalare management
considered active cooling systems
of pouch-type since bat-
lithium-ion they
include
teries withexternal pumps,
a straight fans, and othercold
mini-channel-based auxiliary systems [42].
plate sandwiched Activetwo
between cooling systems
consecutive
demand more
pouch-type space
LiFePO 4 and
cells power,
to form while
a PCMs
module are
coolantknownpass as passive
through thermal
the plate. management
The results
systems and
revealed that do not generally
a cold requirean
plate including additional
even number components
of channels for their
has operation [43]. Key
a higher pressure
considerations
drop than an odd innumber
the selection of thermal
of channels due tomanagement systems Additionally,
the flow resistance. include energy theefficiency,
average
cell temperature uniformity, and overall weight and volume.
temperature of the battery module decreased with an increase in coolant flow rate at the
expense Mengliang
of a moreetconsiderable
al. proposedpressure
a heat pipedrop and
andrefrigerant-based
increased powerBTMS coupled[45].
consumption with an
air-conditioning system for
A three-dimensional a battery
numerical module
model to investigate
is developed the battery
by Peng et al. totemperature
study the PCM dis-
tribution,
process andand energy
its effect onefficiency of the BTMS
battery thermal [44].at
behaviour The resultsC-rate
different revealdischarge
that onceprocesses
the initial
temperature
(0.5 C, 1 C andis2increased from 25PCM
C), and different °C toproperties
30 °C and (different
35 °C, themass average coefficient
fractions of perfor-
of expandable
mance increased
graphite) by 16.95%
of the battery and 38.41%,
module. The resultsrespectively;
suggest aand the average
non-uniform exergy
PCM efficiency
liquid fractionof
distribution during the discharge process
the BTMS improved by 2.63% and 5.07%, respectively [38]. since the outer layer and top portion of PCM
meltedKokkula
first. Alternatively, by adding
et al. investigated the high
thermalthermal conductivity
management nanoparticles
of pouch-type to a compos-
lithium-ion bat-
ite PCM
teries of 12
with wt% EGmini-channel-based
a straight (expanded graphite), coldthe results
plate showed that
sandwiched the heat
between two dissipation
consecutive
ofpouch-type
the batteryLiFePO4
pack responded well to
cells to form the increased
a module coolant flowpassrate than that
through theofplate.
pure The
PCM. Ad-
results
ditionally,
revealed that theyaexperimentally
cold plate including observed the good
an even number effects of composite
of channels has aPCMhigherforpressure
battery
thermal
drop than management
an odd number systems [46].
of channels due to the flow resistance. Additionally, the average
Based on the above analysis
temperature of the battery module and decreased
perception with
few important
an increase future research
in coolant perspectives
flow rate at the
are highlighted. From the literature, thermal fins are mostly created
expense of a more considerable pressure drop and increased power consumption [45]. of copper, bronze, steel,
nickel,Astainless steel, and aluminum
three-dimensional numerical alloy.
modelAlthough
is developed the integration
by Peng et of al.metal plates
to study the with
PCM
the fins can
process andeffectively
its effect onimprove
batterythe cooling
thermal performance,
behaviour the weight
at different C-rate of discharge
the systemprocesses
is still a
concern
(0.5 C, 1and,
C andexploring
2 C), and novel designs
different PCMof fins using advanced
properties (different manufacturing
mass fractionstechniques
of expandablestill
requires more attention. Moreover, to efficiently dissipate heat
graphite) of the battery module. The results suggest a non-uniform PCM liquid fraction generated during battery
operation
distribution at aduring
higher temperature
the dischargeand duringsince
process a faster-charging
the outer layer rate,
andthetopuseportion
of evaporative
of PCM
fluid
melted first. Alternatively, by adding high thermal conductivity nanoparticlesofto
as a potential added technique can significantly improve the performance existing
a com-
air-cooled
posite PCM battery
of 12management
wt% EG (expanded systems. graphite), the results showed that the heat dissipa-
tionIn ofaddition,
the battery thepack
prediction
respondedof lithium-ion
well to the battery temperature
increased flow rateperformance
than that ofin different
pure PCM.
operating conditions, for different values of PCM material, thicknesses,
Additionally, they experimentally observed the good effects of composite PCM for battery and discharge rates
would be beneficial for accelerating
thermal management systems [46]. the industrialization of this innovative cooling strategy.
Various types of PCMs can absorb and release a large quantity of thermal energy
Based on the above analysis and perception few important future research perspec-
through the phase change procedure from solid–liquid, solid–solid and then liquid–gas.
tives are highlighted. From the literature, thermal fins are mostly created of copper,
 the performance of existing air-cooled battery management systems.
In addition, the prediction of lithium-ion battery temperature performance in differ-
ent operating conditions, for different values of PCM material, thicknesses, and discharge
rates would be beneficial for accelerating the industrialization of this innovative cooling
Energies 2023, 16, 185
strategy. 7 of 16
Various types of PCMs can absorb and release a large quantity of thermal energy
through the phase change procedure from solid–liquid, solid–solid and then liquid–gas.
Although paraffin wax (PW) is the most favoured PCM due to its high energy density,
Although
nontoxicityparaffin
and lowwax (PW)pressures,
vapour is the most
thefavoured PCMofdue
disadvantage PWtois its
lowhigh energy
thermal density,
conductiv-
nontoxicity and low vapour pressures, the disadvantage of PW
ity and the risk of liquid leakage during the phase change procedure. is low thermal conductivity
and the risk of liquid leakage during the phase change procedure.
From the literature to increase the thermal conductivity of PW, different thermally
From the literature to increase the thermal conductivity of PW, different thermally
conductive fillers, such as ceramic fillers Al2O3 AlN and metal nanoparticles Cu, Al and
conductive fillers, such as ceramic fillers Al2O3 AlN and metal nanoparticles Cu, Al and Ag
Ag can be added to PW [47–49]. Additionally, the liquid leakage issue of PW can be
can be added to PW [47–49]. Additionally, the liquid leakage issue of PW can be avoided by
avoided by mixing polymer and PW since the polymeric matrix can fix PW by a strong
mixing polymer and PW since the polymeric matrix can fix PW by a strong intermolecular
intermolecular force [50,51].
force [50,51].
Y. Zhang et al.l, used graphite/paraffin/silicone rubber composite PCMs to control
Y. Zhang et al., used graphite/paraffin/silicone rubber composite PCMs to control the
the temperature and improve the safety performance and service life in heat energy stor-
temperature and improve the safety performance and service life in heat energy storage,
age, battery management and thermal interface materials for electronic devices. In this
battery management and thermal interface materials for electronic devices. In this study,
study, the expanded graphite (EG), paraffin wax (PW) and silicone rubber (SR) matrix are
the expanded graphite (EG), paraffin wax (PW) and silicone rubber (SR) matrix are blended
blended with the mixture EG/PW/SR composite shown in Figure 4. Further analysis
with the mixture EG/PW/SR composite shown in Figure 4. Further analysis proved that
proved
the that
fusion the fusion
latent latent
heat and heat and the crystallization
the crystallization latent heat of latent heat of the
the composite PCMcomposite
were 43.6PCMJ/g
were 43.6 J/g and 41.8 J/g, respectively. Moreover, The shape stable test
and 41.8 J/g, respectively. Moreover, The shape stable test indicates that the EG/PW/SRindicates that the
EG/PW/SRPCM
composite composite PCM
may well maybaking
resist well resist baking
at 150 ◦ C forat24
150 °C for
h with no24shift
h with
[52].no shift [52].

graphite (EG)/paraffin
Figure 4. The preparation process of expanded graphite (EG)/paraffin wax
wax [52].
[52].

3. Battery Management
3. Battery Management System
System
A
A battery-management system
battery-management system (BMS)
(BMS) detects
detects unusual
unusual circumstances
circumstances and
and validates
validates
the
the proper
proper method
method for
for controlling
controlling the
the temperature
temperature behaviour
behaviour of
of the
the battery
battery to
to avoid
avoid any
any
negative impact on the power-intake profile. BMS plays a critical role in the safe and
efficient operation of batteries. It uses state estimation which is a broad field of research.
BMS should be designed to mitigate the effects of operations at different states, power
demands, temperatures, and states of health. It generally employs model base estimation
that requires an accurate battery model and a robust estimation strategy to work efficiently.
Since LIBs charge faster than conventional battery technologies, a well-designed BMS is
essential to help with the safety, dependability, and overall performance of lithium-ion
battery systems.
Since LIBs charge faster than conventional battery technologies, a well-designed BMS
is essential to help with the safety, dependability, and overall performance of lithium-ion
battery systems. The accurate estimation of the SOC of a Li-ion battery is challenging
because the Li-ion battery is a highly time-variant, non-linear, and complex electrochemi-
cal system. The SOC estimation methods have been classified into four main categories,
namely the direct measurement method, bookkeeping estimation method, model-based
method, and computer intelligence method. A critical explanation, including their merits,
 Since LIBs charge faster than conventional battery technologies, a well-designed BMS
is essential to help with the safety, dependability, and overall performance of lithium-ion
battery systems. The accurate estimation of the SOC of a Li-ion battery is challenging be-
cause the Li-ion battery is a highly time-variant, non-linear, and complex electrochemical
Energies 2023, 16, 185 system. The SOC estimation methods have been classified into four main categories, 8 of 16
namely the direct measurement method, bookkeeping estimation method, model-based
method, and computer intelligence method. A critical explanation, including their merits, lim-
itations, andand
limitations, estimation errors
estimation from
errors other
from studies,
other is provided.
studies, Some
is provided. recommendations
Some recommendations de-
pending on the development of technology are suggested to improve online
depending on the development of technology are suggested to improve online estima- estimation [53].
The BMS is responsible for monitoring the SOC, SOH, SOP, and the remaining useful
tion [53].
life (RUL) of the
The BMS is battery packfor
responsible as well as for cell
monitoring thebalancing,
SOC, SOH, thermal
SOP, management, and safety.
and the remaining useful
Figure
life (RUL) of 5the
demonstrates
battery packthe global
as well as market for battery-management
for cell balancing, systemsand
thermal management, for differ-
safety.
ent applications [54].
Figure 5 demonstrates the global market for battery-management systems for different
applications [54].

Figure5.5.Growth
Figure Growthof
ofBattery-management
Battery-managementsystems
systemsfrom
from2020
2020to
to2030
2030[54].
[54].

Battery
Battery packs
packs are
are the
the most
most expensive
expensive components
components in in EVs
EVs and
and the
the largest
largest factor
factor con-
tributing
tributingto tothe
the price
price differential
differential between
between EVs EVs and conventional
conventional ICE-powered
ICE-powered vehicles
vehicles [7].
In
In the
the small andandmidsize
midsizecar carsegments,
segments, thethe average
average EVEV costscosts $12,000
$12,000 more more to produce
to produce than
than comparable ICE-powered vehicles. The reason behind this cost is
comparable ICE-powered vehicles. The reason behind this cost is that most original equip- that most original
equipment manufacturers
ment manufacturers over-engineer
over-engineer battery
battery packspacks by 10–14%
by 10–14% in terms
in terms of capacity
of capacity to
to slow
slow
downdown the battery
the battery degradation
degradation rateto
rate due due to SBMS
SBMS limitations.
limitations. This over-engineering
This over-engineering could
could be mitigated
be mitigated by implementing
by implementing accurateaccurate and robust
and robust SOC, SOC,SOH,SOH,
and SOPand estimation
SOP estimation
strat-
strategies onboard
egies onboard the BMS.
the BMS.
Multiple
Multiple models
models cancan be
be considered
considered or orlearned
learnedforforSOCSOCand andSOHSOHestimation.
estimation. These
These
models
models are used in combination with an AI-empowered filter [55]. However, aa combina-
are used in combination with an AI-empowered filter [55]. However, combina-
tion
tion of
of linear
linear and
andnonlinear
nonlinear filters
filters can
canbebedeployed
deployedfor forestimating
estimating thethestates
stateswhose
whose time
time
evolutions
evolutions are governed by linear and nonlinear dynamics, respectively. In this way,
are governed by linear and nonlinear dynamics, respectively. In this way,the
the
computational
computationalburdenburdencan canbebereduced
reduced[56].
[56].
Next-generation management will serve as the vital link between EVs and the energy
society, which consists of numerous EVs, charging stations, and power plants. It is essential
to have an accurate estimation of battery voltage, heat generation rate, and state of health
under different conditions to maintain the safe and efficient operations of the BMS for
EV application.
With the increasing number of onboard batteries, advanced management is needed
for battery modules. Advanced management systems take different forms, including cen-
tralized systems and distributed. The advanced management system focuses on improving
the battery performance and the user’s driving experience and enables the monitoring of
battery dynamics. Battery modelling and state estimation, thermal management, battery
equalization, charging control, and fault diagnosis are the required functions [57].
While AI technology would improve and transform the implementation of LIBs
for EV applications, the deployment of AI/ML algorithms into real-world scenarios for
predicting and discovering battery materials and estimating the state of the battery system
is challenging [58].
ML techniques can be used to link data, by creating a new dataset construction
and/or existing dataset development where a critical correlation in material science is the
 tery equalization, charging control, and fault diagnosis are the required functions [57].
While AI technology would improve and transform the implementation of LIBs for
EV applications, the deployment of AI/ML algorithms into real-world scenarios for pre-
dicting and discovering battery materials and estimating the state of the battery system is
challenging [58].
Energies 2023, 16, 185 9 of 16
ML techniques can be used to link data, by creating a new dataset construction and/or
existing dataset development where a critical correlation in material science is the struc-
ture-property relation. A predictive AI/ML approach helps to extract complicated and
structure-property
nonlinear relation.
patterns from A predictive
training AI/ML
datasets and approach
translate helps to into
the meta-data extract complicated
statistical mod-
and nonlinear patterns from training datasets and translate the meta-data
els. Therefore, fault prognosis as an additional function to the local fault detection into statistical
func-
models.
tion of theTherefore,
BMS would fault
useprognosis
historicalas an and
data additional
machine function toto
learning the local fault
predict detection
or prevent the
function of of
occurrence thea BMS
fault would use historical
in the battery systemdata and machine learning to predict or prevent
[59–61].
the occurrence of a fault inbased
The SOH estimation the battery
on ML system [59–61]. is shown in Figure 6. The overall
technologies
The SOH estimation based on ML technologies
framework is based on two parts, the offline training process, is shown inthe
and Figure 6. Theprocess
estimation overall
framework is based on two parts, the offline training process, and the estimation process
which could be either offline or online. Sui et al. analyzed the impact of entropy as a fea-
which could be either offline or online. Sui et al. analyzed the impact of entropy as a feature
ture for capacity estimation of the battery by monitoring the variation of voltage, current,
for capacity estimation of the battery by monitoring the variation of voltage, current, and
and temperature during the aging process. The performance of these sample entropy (SE)-
temperature during the aging process. The performance of these sample entropy (SE)-based
based estimators, revealed that the entropy-based SOH estimation method will be im-
estimators, revealed that the entropy-based SOH estimation method will be improved once
proved once the battery SOC gets to the polarization zone [62].
the battery SOC gets to the polarization zone [62].

Figure 6.
Figure The battery
6. The battery SOH
SOH estimation
estimation structure
structure using
using ML
ML algorithms
algorithms [61].
[61].

Currently, with the rapid development of AI in every industry, AI is finding its way
Currently, with the rapid development of AI in every industry, AI is finding its way
into our everyday lives. Generative AI has been determined as a strategic AI technology
into our everyday lives. Generative AI has been determined as a strategic AI technology
by Gartner [63]. Additionally, in an earlier report, it was predicted that AI would lead to
by Gartner [63]. Additionally, in an earlier report, it was predicted that AI would lead to
breakthroughs in key areas such as [64]:
breakthroughs in key areas such as [64]:
• Augmented workforce
• Cybersecurity
• Metaverse
• Autonomous vehicles
Energy management, climate, healthcare, drug discovery, and robotics are among
other fields where AI can make a difference.
The learning architectures and transformer-based models that benefit from an attention
mechanism have been playing a critical role in different AI applications, especially in natural
language processing. Putting language models aside, regarding the previously mentioned
trends, the following state-of-the-art algorithms represent the recent top AI achievements
for different application domains:
• Computer vision:
– Time-Space Transformer has been developed by Meta AI (Facebook AI) for video
understanding through action recognition. Compared to a 3D convolutional
neural network (CNN), TimeSformer is faster to train and requires less computing
power. Hence, it is more suitable for real-time or on-demand video-processing
applications [63].
Energies 2023, 16, 185 10 of 16

– Perceiver has been developed by Google DeepMind based on transformers to be

compatible across various modalities including image, video, and point cloud. Us-
ing an asymmetric attention mechanism, it can handle large inputs by iteratively
distilling inputs into a latent information bottleneck [65].
• Generative AI:
– DALL-E has been developed by OpenAI to create images from text descriptions,
which applies to a wide range of concepts that can be expressed using natural
language. Building on transformers, DALL-E models the text and image tokens
as one data stream in an autoregressive manner [66].
– Google DeepMind has proposed an AI framework to play the role of an assistant
for mathematicians helping them in discovering new conjectures and theorems.
The algorithm relies on seeking potential patterns and relations between mathe-
matical objects [67].
• Deep reinforcement learning:
– AlphaFold has been developed by Google DeepMind for predicting the protein
structures in the human proteome [68].
– A deep reinforcement learning algorithm has been proposed by Google DeepMind
for high-dimensional and high-frequency control of tokamak, which is a nuclear
fusion reactor. The proposed controller autonomously learns to command the
magnetic actuator coils in a way to satisfy a diverse set of requirements over a
wide range of plasma configurations [65].
• Multimodal learning:
– To cope with the insufficiency of multimodal training data, a transformer-based
multimodal learning architecture has been proposed [69]. In this architecture,
a set of unimodal pre-trained transformer models are used to encode different
modal inputs. Then, representations, which are provided by these encoders, are
fused by another set of transformer layers.
– In multimodal learning, some modalities likely face issues such as missing data,
noisy data, lack of annotated data, unreliable labels, and scarcity of training data.
To cope with such issues, the notion of multimodal co-learning has been proposed
for performing knowledge transfer from resource-rich modalities to resource-poor
ones [70].
– Since data plays an important role in the learning process, recently, in the AI
community, a campaign has been started by Andrew Ng to promote data-centric
AI as opposed to model-centric AI [71]. According to Andrew Ng, “data-centric
AI is the discipline of systematically engineering the data needed to successfully
build an AI system”. The main idea is to have smart-size learning models instead
of overly complex ones. However, the model quality should not be compensated
for simplicity. To obtain a high-quality AI system, such simpler models must
be trained on datasets, which are carefully collected smartly. In this regard,
observability deserves special attention.
Observability plays a key role in estimation, control, and decision-making. Regarding
condition monitoring as well as fault diagnosis and prognosis, the notion of observability
has been recently extended from the system level to the data level. Aiming for high-quality
data, the following five pillars have been proposed for data observability [72]:
• Freshness refers to the availability of up-to-date data.
• Distribution refers to whether data is in the acceptable range.
• Volume refers to the completeness of data.
• Schema is concerned with structure change for the data.
• Lineage is concerned with both upstream and downstream negative impacts of
data downtime.
Energies 2023, 16, 185 11 of 16

To build an automated end-to-end data pipeline, AI must be part and parcel of each
one of these pillars. AI can handle partial observability and compensate for missing data
through data imputation [73]. Furthermore, deploying a cognitive controller alongside
the physical controller in the system will pave the way for improving observability [74].
While the physical controller controls the flow of energy in the system, the cognitive
controller is responsible for controlling the flow of information toward minimizing risk in
the decision-making process [75].
The sequential data collected as time series reflects the dynamic state of the system un-
der study, and the fusion of features associated with different modalities into an industrial
knowledge graph will facilitate condition monitoring and prognostics [76]. Graph neural
networks [77] and tools from graph signal processing [78] can be deployed to develop
such a framework. Furthermore, graph neural networks allow for handling unstructured
data [79]. Using such a knowledge graph allows the AI-empowered decision-making
process to move beyond robustness and aim at achieving antifragility through learning and
system reconfiguration. While a robust or resilient system is supposed to resist shocks and
stay the same, an antifragile system must improve. Antifragile systems should be immune
to prediction errors. Moreover, in case of an adverse event, the antifragile system must be
able to quickly restore its normal status and recover its normal performance [80].
For condition monitoring and prognostics, state/parameter estimation would be a key
element. Through shifting from a model-centric to a data-centric approach, a large spec-
trum of AI-empowered filters can be developed and used for state/parameter estimation
regarding the deployed models and learning methods [56]. In this regard, multiple-linear,
adaptive, kernel-based, and deep-learning models can be tailored to and learned for the
application at hand. For training such models, supervised, semi-supervised, unsuper-
vised, weakly supervised, self-supervised, noise-robust, and reinforcement learning can be
used [81]. Such models can then be used in combination with different filtering algorithms
as the required state-space models. For instance, to handle partially known dynamics,
recurrent neural networks (RNNs), as well as a combination of RNNs and time-varying
state-space models, were used to aid the Kalman filter [82]. Alternatively, the filtering
procedure can also be viewed as a learning process as well. In this regard, for time-series
analysis, probabilistic transformers combine a state-space model with the transformer
architecture to benefit from an attention mechanism [83].
Habibi presented a new hybrid filter-based method for state estimation as the smooth
variable structure filter (SVSF). The SVSF method is model-based and applies to smooth
nonlinear dynamic systems. It allows for the explicit definition of the source of uncertainty
and can guarantee stability given an upper bound for uncertainties and noise levels. The
performance of the SVSF improves with the more refined definition of upper bounds
on parameter variations or uncertainties. By combining SVSF) with different models
and learning methods, a large spectrum of novel AI-empowered filtering algorithms
will be derived and used for condition monitoring and prognostics, especially in energy
management systems (EMS). To be more specific, such filters will be used as part and parcel
of a BMS for battery state estimation. Then, these estimates will be used for diagnosing
and prognosing electrical faults or cell failures as well as improving battery efficiency [84].
The combination of the SVSF with deep learning, reinforcement learning, and graph
neural networks to derive novel filtering algorithms is uncharted territory to improve the
overall accuracy and stability of the estimates. Therefore, there is great potential from both
theoretical and practical perspectives to make a difference in the literature. Moreover, using
the main idea behind the SVSF may contribute to AI research in the sense of deriving novel
learning algorithms [85].
Regarding the advantages of SVSF over the Kalman filter such as robustness against
model uncertainty and the existence of a secondary set of indicators in addition to the
innovation vector, it is expected that the AI-empowered SVSF algorithms will come out
as winners in competitions against the available deep learning-based filters that are built
Energies 2023, 16, 185 12 of 16

on Kalman filter. Moreover, it is expected that the AI-empowered SVSF algorithms find
applications beyond the domain of applicability of current deep learning-based filters.
Data collection is an essential aspect of developing and training AI models. This
proposal will generate extensive and high-quality data for various battery chemistry and
cell types. These data will become an asset for researchers to conduct further analysis,
use for training and modelling, and benchmark against. The datasets will enable more
researchers to start innovating without building up facilities first, accelerating the pace of
battery research.

4. Conclusions
The climate change concern and other environmental issues due to the immense
exploitation of fossil fuels and the emission of greenhouse gases result in increased con-
sumption of rechargeable batteries. Although there are various types of primary batteries
and rechargeable batteries available in the market, lithium-ion LIBs are the most common
energy storage systems due to their high specific capacity, high energy density and good
cycling stability especially for EV applications [7].
In the current study, different cooling methods were investigated to improve the tem-
perature performance of LIBs have been summarized including air cooling, liquid cooling,
PCM cooling, and heat pipes. It is noticed that the air-cooling system has advantageous
features such as safe, consistent, and simple design, but the lower heat capacity and thermal
efficiency of the air as a cooling method. Liquid-cooled is a very effective cooling technique
with greater thermal conductivity and greater heat capacities compared to air cooling in
which a liquid is used as a coolant to eliminate the heat generated by a battery. To increase
thermal conductivity, PCM cooling allows simple cooling designs to wrap batteries, with
graphite sheets between batteries, increasing the heat loss and improving the temperature
uniformity of the battery pack. To achieve better cooling performance PCM cooling can
also be combined with liquid cooling or heat pipes.
Moreover, a BMS is an essential device for charging and discharging the batteries, over-
coming many challenges, and improving the operating performance of battery modules. On
the other hand, using AI-based predictive algorithms in BMS can improve the availability
of testing datasets and robust processing of data in real-time for EV applications.
According to the above analysis, a future investigation using a Kalman Filter Algo-
rithm is necessary to improve the existing algorithms by including both SOH and SOC
estimators to calculate the ageing of the battery in terms of power management during
long periods of use. This response to the demand side of safety challenges in packing
lithium-ion battery energy management.
Moreover, from the perspective of theory and application, ML technologies possess
play a major role in battery SOH estimation. Therefore, the outlook of the research on
future research should focus on implementing advanced battery models and algorithms
onto the cloud-based BMS and the difficulty of onboard implementation. In particular,
an EMS system that runs optimally under harsh weather can significantly improve the
mileage and usability of electric vehicles in Canada and similar regions with cold climates.
This will encourage customer acceptance of electric vehicles and accelerate the pace toward
carbon peak and carbon neutrality goals.

Author Contributions: Conceptualization, M.G.; Formal analysis, M.G. and S.H.; Investigation,
validation, M.G., resources, M.G.; data curation, M.G.; writing—original draft preparation, M.G.;
writing—review and editing, S.H.; visualization, M.G.; supervision, S.H.; project administration, S.H.;
funding acquisition, S.H. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: Not applicable.
Acknowledgments: The authors also wish to thank all the support provided by the Centre for
Mechatronics and Hybrid Technologies (CMHT) at McMaster university.
Energies 2023, 16, 185 13 of 16

Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript:
EVs Electric Vehicles
BMS Battery Management System
AI Artificial Intelligence
ML Machine Learning
SOC State of Health
SOH State of Charge
EV Electric Vehicle
LIBs Lithium-ion batteries
LiFePO4 Lithium Iron Phosphate
LiCoO2 Lithium Cobalt Oxide
NCA Lithium Nickel Cobalt Aluminum Oxide
LFP Lithium Iron Phosphate
SSB Solid-state batteries
Li-S Lithium-sulfur
NRC National Research Council Canada
NMC Nickel Manganese Cobalt
PCM Phase change materials
PW Paraffin Wax
EG Expanded Graphite
SR Silicone Rubber
RUL Remaining Useful Life
CNN Convolutional Neural Network
RNNs recurrent neural networks
SVSF smooth variable structure filter
EMS energy management systems

References
1. Zhang, X.; Li, Z.; Luo, L.; Fan, Y.; Du, Z. A review on thermal management of lithium-ion batteries for electric vehicles. Energy
2022, 238, 121652. [CrossRef]
2. Kumar, R.R.; Alok, K. Adoption of electric vehicle: A literature review and prospects for sustainability. J. Clean. Prod. 2020, 253, 119911.
[CrossRef]
3. Chen, Z.; Liu, W.; Yin, Y. Deployment of stationary and dynamic charging infrastructure for electric vehicles along traffic corridors.
Transp. Res. Part C Emerg. Technol. 2017, 77, 185–206. [CrossRef]
4. Nian, V.; Hari, M.; Yuan, J. A new business model for encouraging the adoption of electric vehicles in the absence of policy
support. Appl. Energy 2019, 235, 1106–1117. [CrossRef]
5. Li, M.; Lu, J.; Chen, Z.; Amine, K. 30 years of lithium-ion batteries. Adv. Mater. 2018, 30, 1800561. [CrossRef]
6. Zubi, G.; Dufo-López, R.; Carvalho, M.; Pasaoglu, G. The lithium-ion battery: State of the art and future perspectives. Renew.
Sustain. Energy Rev. 2018, 89, 292–308. [CrossRef]
7. Ghalkhani, M.; Bahiraei, F.; Nazri, G.-A.; Saif, M. Electrochemical–thermal model of pouch-type lithium-ion batteries. Electrochim.
Acta 2017, 247, 569–587. [CrossRef]
8. Saw, L.; Ye, Y.; Tay, A. Electrochemical–thermal analysis of 18650 Lithium Iron Phosphate cell. Energy Convers. Manag. 2013, 75,
162–174. [CrossRef]
9. Du, S.; Jia, M.; Cheng, Y.; Tang, Y.; Zhang, H.; Ai, L.; Zhang, K.; Lai, Y. Study on the thermal behaviors of power lithium iron
phosphate (LFP) aluminum-laminated battery with different tab configurations. Int. J. Therm. Sci. 2015, 89, 327–336. [CrossRef]
10. Chen, S.; Wan, C.; Wang, Y. Thermal analysis of lithium-ion batteries. J. Power Sources 2005, 140, 111–124. [CrossRef]
11. Grey, C.P.; Hall, D.S. Prospects for lithium-ion batteries and beyond-a 2030 vision. Nat. Commun. 2020, 11, 6279. [CrossRef]
[PubMed]
12. Saxena, S.; Kong, L.; Pecht, M.G. Exploding e-cigarettes: A battery safety issue. IEEE Access 2018, 6, 21442–21466. [CrossRef]
13. Yao, X.-Y.; Kong, L.; Pecht, M.G. Reliability of cylindrical li-ion battery safety vents. IEEE Access 2020, 8, 101859–101866. [CrossRef]
14. Loveridge, M.J.; Remy, G.; Kourra, N.; Genieser, R.; Barai, A.; Lain, M.J.; Guo, Y.; Amor-Segan, M.; Williams, M.A.; Amietszajew,
T. Looking deeper into the Galaxy (Note 7). Batteries 2018, 4, 3. [CrossRef]
15. Kim, J.; Oh, J.; Lee, H. Review on battery thermal management system for electric vehicles. Appl. Therm. Eng. 2019, 149, 192–212.
[CrossRef]
Energies 2023, 16, 185 14 of 16

16. Fu, Y.; Lu, S.; Li, K.; Liu, C.; Cheng, X.; Zhang, H. An experimental study on burning behaviors of 18650 lithium ion batteries
using a cone calorimeter. J. Power Sources 2015, 273, 216–222. [CrossRef]
17. Bahiraei, F.; Ghalkhani, M.; Fartaj, A.; Nazri, G.-A. A pseudo 3D electrochemical-thermal modeling and analysis of a lithium-ion
battery for electric vehicle thermal management applications. Appl. Therm. Eng. 2017, 125, 904–918. [CrossRef]
18. Li, C.; Wang, Z.-Y.; He, Z.-J.; Li, Y.-J.; Mao, J.; Dai, K.-H.; Yan, C.; Zheng, J.-C. An advance review of solid-state battery: Challenges,
progress and prospects. Sustain. Mater. Technol. 2021, 29, e00297. [CrossRef]
19. Available online: https://www.nytimes.com/business/energy-environment/next-generation-auto-battery.html (accessed on 7
June 2022).
20. Available online: www.forbes.com/sites/arielcohen/2021/02/11/what-batteries-will-power-the-future (accessed on 7 June 2022).
21. Ji, X.; Lee, K.T.; Nazar, L.F. A highly ordered nanostructured carbon–sulphur cathode for lithium–sulphur batteries. Nat. Mater.
2009, 8, 500–506. [CrossRef]
22. Zhuang, W.; Liu, Z.; Su, H.; Chen, G. An intelligent thermal management system for optimized lithium-ion battery pack. Appl.
Therm. Eng. 2021, 189, 116767. [CrossRef]
23. Available online: www.forbes.com/sites/jamesmorris/2022/04/02/sulfur-battery-technology-could-make-electric-cars-go-
three-times-further-by-2024 (accessed on 7 June 2022).
24. Chawla, N. Recent advances in air-battery chemistries. Mater. Today Chem. 2019, 12, 324–331. [CrossRef]
25. Manuel Salado, E.L. Advances, challenges and environmental impacts in metal-air battery electrolytes. Mater. Today Energy 2022,
28, 101064. [CrossRef]
26. Available online: http://www.chimera-energy.com/index.html (accessed on 7 June 2022).
27. Available online: https://www.upgradetech.co.uk/ (accessed on 7 June 2022).
28. Hale, C. Battery management. US Patent 11316352, 26 April 2022.
29. Available online: https://wae.com/project/multi-chem/ (accessed on 7 June 2022).
30. Available online: https://www.emobility-engineering.com/diversity-gives-pack-balance/ (accessed on 7 June 2022).
31. Available online: https://fulminea.com/twitter-tech-news/ (accessed on 7 June 2022).
32. Available online: https://www.yuasa.co.uk/2020/08/gs-yuasa-portsmouth-port-battery-system-container-delivery-begins-
next-phase-of-innovative-project/ (accessed on 7 June 2022).
33. Available online: https://chargethefuture.org/blog/groundbreaking-dual-chemistry-energy-storage-system-launches-at-uk-
port-to-deliver-net-zero-carbon-operation/ (accessed on 7 June 2022).
34. Available online: https://globalresearch.admin.ox.ac.uk/article/multi-chemistry-battery-pack-using-second-life-batteries-grid-
systems-developing-countries (accessed on 7 June 2022).
35. Smith, M.T.; Starke, M.R.; Chinthavali, M.; Tolbert, L.M. Architecture for Utility-Scale Multi-Chemistry Battery Energy Storage.
In Proceedings of the 2019 IEEE Energy Conversion Congress and Exposition (ECCE), Baltimore, MD, USA, 29 September–3
October 2019; pp. 5386–5392.
36. Chung, S.; Trescases, O. Hybrid Energy Storage System With Active Power-Mix Control in a Dual-Chemistry Battery Pack for
Light Electric Vehicles. IEEE Trans. Transp. Electrif. 2017, 3, 600–617. [CrossRef]
37. Feng, X.; Ouyang, M.; Liu, X.; Lu, L.; Xia, Y.; He, X. Thermal runaway mechanism of lithium ion battery for electric vehicles: A
review. Energy Storage Mater. 2018, 10, 246–267. [CrossRef]
38. Shim, J.; Kostecki, R.; Richardson, T.; Song, X.; Striebel, K.A. Electrochemical analysis for cycle performance and capacity fading
of a lithium-ion battery cycled at elevated temperature. J. Power Sources 2002, 112, 222–230. [CrossRef]
39. Qu, J.; Ke, Z.; Zuo, A.; Rao, Z. Experimental investigation on thermal performance of phase change material coupled with
three-dimensional oscillating heat pipe (PCM/3D-OHP) for thermal management application. Int. J. Heat Mass Transf. 2019, 129,
773–782. [CrossRef]
40. Javani, N.; Dincer, I.; Naterer, G.; Yilbas, B. Heat transfer and thermal management with PCMs in a Li-ion battery cell for electric
vehicles. Int. J. Heat Mass Transf. 2014, 72, 690–703. [CrossRef]
41. Wang, G.; Gao, Q.; Yan, Y.; Wang, Y. Thermal Management Optimization of a Lithium-Ion Battery Module with Graphite Sheet
Fins and Liquid Cold Plates. Automot. Innov. 2020, 3, 336–346. [CrossRef]
42. Shahid, S.; Agelin-Chaab, M. Experimental and numerical studies on air cooling and temperature uniformity in a battery pack.
Int. J. Energy Res. 2018, 42, 2246–2262. [CrossRef]
43. Huang, R.; Li, Z.; Hong, W.; Wu, Q.; Yu, X. Experimental and numerical study of PCM thermophysical parameters on lithium-ion
battery thermal management. Energy Rep. 2020, 6, 8–19. [CrossRef]
44. Yao, M.; Gan, Y.; Liang, J.; Dong, D.; Ma, L.; Liu, J.; Luo, Q.; Li, Y. Performance simulation of a heat pipe and refrigerant-based
lithium-ion battery thermal management system coupled with electric vehicle air-conditioning. Appl. Therm. Eng. 2021, 191, 116878.
[CrossRef]
45. Monika, K.; Chakraborty, C.; Roy, S.; Dinda, S.; Singh, S.A.; Datta, S.P. Parametric investigation to optimize the thermal
management of pouch type lithium-ion batteries with mini-channel cold plates. Int. J. Heat Mass Transf. 2021, 164, 120568.
[CrossRef]
46. Peng, P.; Wang, Y.; Jiang, F. Numerical study of PCM thermal behavior of a novel PCM-heat pipe combined system for Li-ion
battery thermal management. Appl. Therm. Eng. 2022, 209, 118293. [CrossRef]
Energies 2023, 16, 185 15 of 16

47. Jiang, X.; Luo, R.; Peng, F.; Fang, Y.; Akiyama, T.; Wang, S. Synthesis, characterization and thermal properties of paraffin
microcapsules modified with nano-Al2O3. Appl. Energy 2015, 137, 731–737. [CrossRef]
48. Li, Y.; Li, J.; Feng, W.; Wang, X.; Nian, H. Design and preparation of the phase change materials paraffin/porous Al2O3@ graphite
foams with enhanced heat storage capacity and thermal conductivity. ACS Sustain. Chem. Eng. 2017, 5, 7594–7603. [CrossRef]
49. Wang, W.; Yang, X.; Fang, Y.; Ding, J.; Yan, J. Enhanced thermal conductivity and thermal performance of form-stable composite
phase change materials by using β-Aluminum nitride. Appl. Energy 2009, 86, 1196–1200. [CrossRef]
50. Sarı, A. Form-stable paraffin/high density polyethylene composites as solid–liquid phase change material for thermal energy
storage: Preparation and thermal properties. Energy Convers. Manag. 2004, 45, 2033–2042. [CrossRef]
51. Molefi, J.; Luyt, A.; Krupa, I. Comparison of LDPE, LLDPE and HDPE as matrices for phase change materials based on a soft
Fischer–Tropsch paraffin wax. Thermochim. Acta 2010, 500, 88–92. [CrossRef]
52. Zhang, Y.; Li, W.; Huang, J.; Cao, M.; Du, G. Expanded graphite/paraffin/silicone rubber as high temperature form-stabilized
phase change materials for thermal energy storage and thermal interface materials. Materials 2020, 13, 894. [CrossRef]
53. Ali, M.U.; Zafar, A.; Nengroo, S.H.; Hussain, S.; Junaid Alvi, M.; Kim, H.-J. Towards a smarter battery management system for
electric vehicle applications: A critical review of lithium-ion battery state of charge estimation. Energies 2019, 12, 446. [CrossRef]
54. Krishna, G.; Singh, R.; Gehlot, A.; Akram, S.V.; Priyadarshi, N.; Twala, B. Digital Technology Implementation in Battery-
Management Systems for Sustainable Energy Storage: Review, Challenges, and Recommendations. Electronics 2022, 11, 2695.
[CrossRef]
55. Kurle, R.; Rangapuram, S.S.; de Bézenac, E.; Günnemann, S.; Gasthaus, J. Deep rao-blackwellised particle filters for time series
forecasting. Adv. Neural Inf. Process. Syst. 2020, 33, 15371–15382.
56. Setoodeh, P.; Habibi, S.; Haykin, S. Nonlinear Filters: Theory and Applications; John Wiley & Sons: Hoboken, NJ, USA, 2022.
57. Liu, K.; Li, K.; Peng, Q.; Zhang, C. A brief review on key technologies in the battery management system of electric vehicles.
Front. Mech. Eng. 2019, 14, 47–64. [CrossRef]
58. Lv, C.; Zhou, X.; Zhong, L.; Yan, C.; Srinivasan, M.; Seh, Z.W.; Liu, C.; Pan, H.; Li, S.; Wen, Y. Machine learning: An advanced
platform for materials development and state prediction in lithium-ion batteries. Adv. Mater. 2022, 34, 2101474. [CrossRef]
[PubMed]
59. Curtarolo, S.; Morgan, D.; Persson, K.; Rodgers, J.; Ceder, G. Predicting crystal structures with data mining of quantum
calculations. Phys. Rev. Lett. 2003, 91, 135503. [CrossRef]
60. Tran, M.-K.; Panchal, S.; Khang, T.D.; Panchal, K.; Fraser, R.; Fowler, M. Concept review of a cloud-based smart battery
management system for lithium-ion batteries: Feasibility, logistics, and functionality. Batteries 2022, 8, 19. [CrossRef] [PubMed]
61. Sui, X.; He, S.; Vilsen, S.B.; Meng, J.; Teodorescu, R.; Stroe, D.-I. A review of non-probabilistic machine learning-based state of
health estimation techniques for Lithium-ion battery. Appl. Energy 2021, 300, 117346. [CrossRef]
62. Sui, X.; Stroe, D.-I.; He, S.; Huang, X.; Meng, J.; Teodorescu, R. The effect of voltage dataset selection on the accuracy of
entropy-based capacity estimation methods for lithium-ion batteries. Appl. Sci. 2019, 9, 4170. [CrossRef]
63. Bertasius, G.; Wang, H.; Torresani, L. Is space-time attention all you need for video understanding? In Proceedings of the
International Conference on Machine Learning, Online, 14–24 July 2021; p. 4.
64. Marr, B. The 7 biggest artificial intelligence (AI) trends in 2022. Forbes, 24 September 2021.
65. Degrave, J.; Felici, F.; Buchli, J.; Neunert, M.; Tracey, B.; Carpanese, F.; Ewalds, T.; Hafner, R.; Abdolmaleki, A.; de Las Casas, D.
Magnetic control of tokamak plasmas through deep reinforcement learning. Nature 2022, 602, 414–419. [CrossRef]
66. Ramesh, A.; Pavlov, M.; Goh, G.; Gray, S.; Voss, C.; Radford, A.; Chen, M.; Sutskever, I. Zero-shot text-to-image generation. In
Proceedings of the International Conference on Machine Learning, Online, 14–24 July 2021; pp. 8821–8831.
67. Davies, A.; Veličković, P.; Buesing, L.; Blackwell, S.; Zheng, D.; Tomašev, N.; Tanburn, R.; Battaglia, P.; Blundell, C.; Juhász, A.
Advancing mathematics by guiding human intuition with AI. Nature 2021, 600, 70–74. [CrossRef]
68. Tunyasuvunakool, K.; Adler, J.; Wu, Z.; Green, T.; Zielinski, M.; Žídek, A.; Bridgland, A.; Cowie, A.; Meyer, C.; Laydon, A. Highly
accurate protein structure prediction for the human proteome. Nature 2021, 596, 590–596. [CrossRef] [PubMed]
69. Miyazawa, K.; Kyuragi, Y.; Nagai, T. Simple and Effective Multimodal Learning Based on Pre-Trained Transformer Models. IEEE
Access 2022, 10, 29821–29833. [CrossRef]
70. Rahate, A.; Walambe, R.; Ramanna, S.; Kotecha, K. Multimodal co-learning: Challenges, applications with datasets, recent
advances and future directions. Inf. Fusion 2022, 81, 203–239. [CrossRef]
71. Press, G. Andrew Ng Launches A Campaign For Data-Centric AI. Forbes 16 June 2021.
72. Carlo, M. The Big Book of Data Observability. Available online: https://www.montecarlodata.com/wp-content/uploads/2022
/03/The-Big-Book-of-Data-Observability.pdf (accessed on 7 June 2022).
73. Pournabi, M.; Mohammadi, M.; Afrasiabi, S.; Setoodeh, P. Power system transient security assessment based on deep learning
considering partial observability. Electr. Power Syst. Res. 2022, 205, 107736. [CrossRef]
74. Fatemi, M.; Setoodeh, P.; Haykin, S. Observability of stochastic complex networks under the supervision of cognitive dynamic
systems. J. Complex Netw. 2017, 5, 433–460. [CrossRef]
75. Haykin, S.; Fatemi, M.; Setoodeh, P.; Xue, Y. Cognitive control. Proc. IEEE 2012, 100, 3156–3169. [CrossRef]
76. Zhou, B.; Shen, X.; Lu, Y.; Li, X.; Hua, B.; Liu, T.; Bao, J. Semantic-aware event link reasoning over industrial knowledge graph
embedding time series data. Int. J. Prod. Res. 2022. [CrossRef]
Energies 2023, 16, 185 16 of 16

77. Wu, L.; Cui, P.; Pei, J.; Zhao, L. Graph Neural Networks: Foundations, Frontiers, and Applications; Springer: Berlin/Heidelberg,
Germany, 2022.
78. Ortega, A. Introduction to Graph Signal Processing; Cambridge University Press: Cambridge, UK, 2022.
79. Després, N.; Cyr, E.; Setoodeh, P.; Mohammadi, M. Deep learning and design for additive manufacturing: A framework for
microlattice architecture. Jom 2020, 72, 2408–2418. [CrossRef]
80. Taleb, N.N. Antifragile: Things That Gain from Disorder; Random House: New York, NY, USA, 2012; Volume 3.
81. Matsuo, Y.; LeCun, Y.; Sahani, M.; Precup, D.; Silver, D.; Sugiyama, M.; Uchibe, E.; Morimoto, J. Deep learning, reinforcement
learning, and world models. Neural Netw. 2022, 152, 267–275. [CrossRef]
82. Revach, G.; Shlezinger, N.; Ni, X.; Escoriza, A.L.; Van Sloun, R.J.; Eldar, Y.C. KalmanNet: Neural network aided Kalman filtering
for partially known dynamics. IEEE Trans. Signal Process. 2022, 70, 1532–1547. [CrossRef]
83. Tang, B.; Matteson, D.S. Probabilistic transformer for time series analysis. Adv. Neural Inf. Process. Syst. 2021, 34, 23592–23608.
84. Habibi, S. The smooth variable structure filter. Proc. IEEE 2007, 95, 1026–1059. [CrossRef]
85. Avzayesh, M.; Abdel-Hafez, M.; AlShabi, M.; Gadsden, S.A. The smooth variable structure filter: A comprehensive review. Digit.
Signal Process. 2021, 110, 102912. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.