Review of State Estimation and Remaining Useful Li

sustainability
Review
Review of State Estimation and Remaining Useful Life
Prediction Methods for Lithium–Ion Batteries
Jiahui Zhao , Yong Zhu *, Bin Zhang, Mingyi Liu, Jianxing Wang, Chenghao Liu and Xiaowei Hao
China Huaneng Group, Clean Energy Research Institute (CERI), Beijing 102209, China;
[email protected] (J.Z.)
* Correspondence: [email protected]
Abstract: The accurate estimation of the state of charge, the state of health and the prediction of
remaining useful life of lithium–ion batteries is an important component of battery management. It
is of great significance to prolong battery life and ensure the reliability of the battery system. Many
researchers have completed a large amount of work on battery state evaluation and RUL prediction
methods and proposed a variety of methods. This paper first introduces the definition of the SOC, the
SOH and the existing estimation methods. Then, the definition of RUL is introduced, and the main
methods are classified and compared. Finally, the challenges of lithium–ion battery state estimation
and RUL prediction are summarized, and the direction for future development is proposed.
Keywords: lithium–ion battery; SOC estimation; SOH estimation; RUL prediction
1. Introduction
Lithium–ion batteries have the advantages of high energy density, low self-discharge
and long life, and have been used in many fields [1,2]. Given the background of an
expanding lithium–ion battery global market, related technical research such as state-of-
charge (SOC) estimation needs to be gradually improved to effectively ensure the promotion
and application of lithium–ion batteries in multiple scenarios. Accurate estimation of the
Citation: Zhao, J.; Zhu, Y.; Zhang, B.; SOC plays an important role in preventing battery overcharge, overdischarge and extending
Liu, M.; Wang, J.; Liu, C.; Hao, X. the battery cycle life. However, lithium–ion batteries are highly nonlinear, and their SOC is
Review of State Estimation and affected by many factors, making it difficult to accurately estimate the SOC. Therefore, the
Remaining Useful Life Prediction SOC estimation method of lithium–ion batteries is of widespread concern.
Methods for Lithium–Ion Batteries. In addition, in the continuous charging and discharging process of lithium–ion bat-
Sustainability 2023, 15, 5014. https:// teries, battery performance will deteriorate with the decrease in capacity and increase in
doi.org/10.3390/su15065014
impedance; this may induce battery faults, such as internal short circuits and thermal run-
Academic Editor: Julia Kowal away, resulting in equipment and system failure and catastrophic accidents [3]. Therefore,
it is also critical to accurately estimate the state of health (SOH) and predict the remaining
Received: 9 February 2023
useful life (RUL) to improve the reliability of the battery [4].
Revised: 28 February 2023
At present, some scholars have summarized and solved the estimation and prediction
Accepted: 7 March 2023
methods of the SOC, the SOH and the RUL. Reference [1] summarized the methods and
Published: 11 March 2023
models for SOH estimation and compared their advantages and disadvantages, but the
methods are not well classified or comprehensively covered. Reference [2] provided a
coherent review of battery state estimation and RUL prediction techniques and provided
Copyright: © 2023 by the authors. an available commercial dataset for analysis; however, the relationship between these
Licensee MDPI, Basel, Switzerland. methodologies is not clearly explained. Reference [5] reviewed model-based methods and
This article is an open access article data-driven methods for SOC estimation, but the models of battery are not compared in
distributed under the terms and detail. With the development of the battery industry, there have been a lot of innovations
conditions of the Creative Commons in battery state estimation and life prediction methods. However, previous reviews cannot
Attribution (CC BY) license (https:// fully cover innovative methods. Thus, this paper explains and summarizes recent extensive
creativecommons.org/licenses/by/ scholarly research.
4.0/).
Sustainability 2023, 15, 5014. https://doi.org/10.3390/su15065014 https://www.mdpi.com/journal/sustainability

cannot fully cover innovative methods. Thus, this paper explains and summarizes rece
extensive scholarly research.
Sustainability 2023, 15, 5014
The structure of this paper is as follows: Section 1 introduces the SOC estimati
2 of 22
methodology. Section 2 introduces the SOH estimation methodology. Section 3 summ
rizes the estimation method of RUL. Section 4 summarizes and suggests possible futu
research
Thedirections
structure ofbased on the
this paper is existing problems
as follows: Section 1in current research.
introduces the SOC estimation
methodology. Section 2 introduces the SOH estimation methodology. Section 3 summarizes
2. the
SOCestimation method
Estimation of RUL. Section 4 summarizes and suggests possible future research
Method
directions based on the existing problems in current research.
2.1. Classification of SOC Estimation Methods
2. SOC Estimation
Battery Method
SOC refers to the available state of the remaining charge in the battery. It
2.1. Classification of SOC Estimation Methods
the ratio of the remaining charge margin Qremain of the battery to the nominal (rate
Battery SOC refers to the available state of the remaining charge in the battery. It is the
charge
ratio ofcapacity Qratedcharge
the remaining as shown
margininQFigureof1.the
Generally, it isnominal
battery to the expressed as charge
(rated) follows:
remain
capacity Qrated as shown in Figure 1. Generally,Qit is expressed as follows:
SOC = remain
× 100% .
Q Qrated
SOC = remain × 100% (1)
Qrated
Figure 1. The relationship between SOC, SOH and RUL.

Figure 1. The relationship between SOC, SOH and RUL.
In this paper, the SOC estimation methods for a lithium–ion battery are summarized
intoIn this
the paper,four
following the categories
SOC estimation
[5]: methods for a lithium–ion battery are summariz
into the following four categories [5]:
(1) The experiment-based SOC estimation method [6] involves the accurate estimation
of the
(1)SOC
Theby experimentally measuring
experiment-based the value ofmethod
SOC estimation cell features;
[6] involves the accurate estim
(2) The model-based SOC estimation method [7,8] includes the equivalent circuit
tion of the SOC by experimentally measuring the value of cell features;
model, the electrochemical model and other relevant battery models which are usually
(2) The model-based SOC estimation method [7,8] includes the equivalent circu
employed to estimate the state parameters of the battery’s SOC;
model,(3)the
Theelectrochemical model and
data-driven SOC estimation other[9]
method relevant
fits and battery
combinesmodels which of
a large number are usua
employed to estimate
data to estimate the state
the battery’s SOCparameters of the battery’s SOC;
through machine-learning;
(3) The data-driven SOC estimation method
(4) The fusion of multiple methods [10] aims [9] fits
to balance and combines
the accuracy a large
of estimation andnumber
reduce
data the calculation
to estimate time. These
the battery’s SOCmulti-class
throughmethods learn from each other.
machine-learning;
Figure 2 shows the selection of the estimation methods for the SOC.
(4) The fusion of multiple methods [10] aims to balance the accuracy of estimati
and reduce the calculation time. These multi-class methods learn from each other.
Figure 2 shows the selection of the estimation methods for the SOC.
Sustainability 2023, 15, x FOR PEER REVIEW 3 of 23
Sustainability 2023, 15, 5014 3 of 22
Ampere integration
Open-circuit voltage method

Experimental State estimation model
method Electrochemical impedance
spectroscopy
Rint model
……
nth Thevenin
Equivalent circuit model
Electrical model
PNGV model
model
OCV - SOC model
GNL model
Model-based
method P2D model
SOC Electrochemical
model SP model State estimation method
estimation
methods for Simplifed P2D model Kalman filtering and
LIB Filter
its improvement
Neural network and its improvement
Particle filter and its
Regression method and its improvement
Data-driven
improvement
method H infinite filter and its
Fuzzy logic method improvement
……
Observer ……
Integration of model method and data
driven method
Joint method
Integration of different data-driven
methods
Figure 2. Classification of lithium–ion battery SOC estimation methods.
2.2. The2. Experimental

Figure Classification SOC Estimation
of lithium–ion Method
battery SOC estimation methods.
2.2.1. Ampere Integration
2.2. The Experimentalintegration
Ampere–hour SOC Estimation
(alsoMethod
known as the current integration method or the
Coulomb counting method)
2.2.1. Ampere Integration estimates the SOC of the battery through the accumulated
charge and discharge. Compared with other SOC estimation methods, the current integra-
Ampere–hour integration (also known as the current integration method or the Cou-
tion method is relatively reliable and can dynamically estimate the SOC of the battery, thus,
lomb counting method) estimates the SOC of the battery through the accumulated charge
it is widely used in real applications. It is expressed as:
and discharge. Compared with other SOC estimation methods, the current integration
method is relatively reliable and can dynamically 1 estimate
Z t the SOC of the battery, thus, it
S = S −
is widely used in real applications. It is expressed η Idt (2)
C as:
OC OC0
t N 0
1 t
= SOC
SOCtime
where SOC is the state of charge at OC0 ist the
t; 0S− η IdSOC
t , at the beginning time of(2)
t0 ;
C N
0
C is the nominal capacity of the battery under the current standard conditions; η is the
N
coulomb
where is the stateI is
SOCefficiency; ofthe current,
charge which
at time is negative
t; SOC0 is the SOCwhen charging
at the andtime
beginning positive
of t0;when
CN is
discharging.
the The factors
nominal capacity affecting
of the theunder
battery accuracythe of the ampere
current integral
standard methodη can
conditions; be cou-
is the seen
from Equation (2):
lomb efficiency; I is the current, which is negative when charging and positive when dis-
(1) The
charging. Theinitial SOC0
factors is mainly
affecting therelated
accuracyto the sampling
of the ampereaccuracy
integral and frequency
method can beofseen
the
current
from sensor; (2):
Equation
(2) The
(1) The initial
CoulombSOC0 efficiency η is deeply
is mainly related to theaffected
samplingbyaccuracy
the working state of battery
and frequency of the
(e.g., charged
current sensor;state, temperature, current, etc.), and is difficult to measure accurately;
(3) The
(2) The Coulomb capacity CηNisisdeeply
available efficiency affected by battery
affected aging
by the and operating
working conditions
state of battery (e.g.,
(charge/discharge ratio, temperature, etc.).
charged state, temperature, current, etc.), and is difficult to measure accurately;

(3) The available capacity CN is affected by battery aging and operating conditions
(charge/discharge ratio, temperature, etc.).
Due
Duetotoerror
erroraccumulation,
accumulation,the theampere–hour
ampere–hour integration
integration method
method doesdoesnotnotmeet re- re-
meet
quirements for accuracy of SOC estimation. Some scholars have
quirements for accuracy of SOC estimation. Some scholars have carried out research carried out research on on
improving the accuracy of the ampere–hour integration method.
improving the accuracy of the ampere–hour integration method. Reference [11] tested theReference [11] tested the
lithium
lithium iron
iron phosphate
phosphate powerpower battery
battery and
and compared
compared the the importance
importance of of various
various parame-
parameters
ters for improving the accuracy of SOC estimation. The
for improving the accuracy of SOC estimation. The results showed that results showed that thetheinitial SOC
initial SOC
correction
correctionwas wasthe themost
mostimportant
importantfor forimproving
improving thethe
accuracy
accuracy of of
thetheamp–hour
amp–hour integra-
integra-
tion
tion method.
method.Reference
Reference [12] points
[12] points outoutthatthat
in ainsimulation
a simulation process, if theiftest
process, the current is
test current
not corrected in time, it will lead to large cumulative measurement
is not corrected in time, it will lead to large cumulative measurement errors due to the errors due to the open-
loop working
open-loop mode of
working ampere–hour
mode of ampere–hourintegration. At the same
integration. At thetime,
same reference [12] pro-[12]
time, reference
posed the ampere–hour integral method with capacity correction
proposed the ampere–hour integral method with capacity correction for the error caused for the error caused by
the traditional method which treats the available capacity of the
by the traditional method which treats the available capacity of the battery as a fixedbattery as a fixed value.
Reference [13], focusing
value. Reference on the error
[13], focusing on the caused by the inability
error caused to update
by the inability the Coulomb
to update effi-
the Coulomb
ciency online in the traditional amp–hour integration method, proposed
efficiency online in the traditional amp–hour integration method, proposed an amp–hour an amp–hour in-
tegration method to modify the electricity after piecewise integration
integration method to modify the electricity after piecewise integration by adopting various by adopting various
Coulomb
Coulombefficiencies.
efficiencies.
2.2.2.
2.2.2.The
TheOpen-Circuit
Open-CircuitVoltage-Based
Voltage-BasedMethodMethod
Since the terminal voltage of the
Since the terminal voltage of the battery battery has
hasa fixed functional
a fixed functional relationship
relationship with thethe
with
SOC under the condition of long-term standing, the battery’s
SOC under the condition of long-term standing, the battery’s SOC can be obtained SOC can be obtained by by
measuring
measuringthe thecurve
curvecorresponding
corresponding toto
the open
the open circuit voltage
circuit voltage (OCV)
(OCV) andandthethe
SOC SOC[14].[14].
When the ampere–hour integration method is used, it often relies
When the ampere–hour integration method is used, it often relies on the OCV–SOC curve toon the OCV–SOC curve
to calibrate
calibrate at at the
the beginning/endofofcharge
beginning/end chargeand anddischarge.
discharge.However,
However,there thereare
arethree
threeproblems
prob-
lems with the open-circuit voltage
with the open-circuit voltage method: method:
(1)
(1)When
Whenmeasuring
measuringopen-circuit
open-circuitvoltage,
voltage, thethebattery
battery needs
needsto to
be be
rested forfor
rested a long
a long
time to achieve voltage stability, which is difficult for real-time
time to achieve voltage stability, which is difficult for real-time applications; applications;
(2)
(2)ItItisisdifficult
difficulttotodetermine
determinehow howlong
longthe therest
resttime
timeis;is;
(3)
(3) As shown in Figure 3, the OCV–SOC curve has hysteresis
As shown in Figure 3, the OCV–SOC curve has a a hysteresis effect [15];
effect thethe
[15]; samesame
voltage corresponds to different SOCs when charging and discharging,
voltage corresponds to different SOCs when charging and discharging, which will also which will also
leads
leadstotolarge
largeerrors
errorsininSOC
SOCcalculation.
calculation.
OCV–SOCcurve
Figure3.3.OCV–SOC
Figure curveofoflithium–ion
lithium–ionbattery.
battery.
2.2.3. Other Methods

2.2.3. Other Methods
Electrochemical impedance spectroscopy uses a small amplitude sinusoidal AC signal
Electrochemical impedance spectroscopy uses a small amplitude sinusoidal AC sig-
to perturb the electrochemical power supply system. By measuring the battery impedance
nal to perturb the electrochemical power supply system. By measuring the battery imped-
at different frequencies, the internal reaction mechanism of the battery that is most relevant
ance at different frequencies, the internal reaction mechanism of the battery that is most
to the change in the battery’s SOC is found, so as to estimate the battery’s SOC value [15–17].
relevant to the change in the battery’s SOC is found, so as to estimate the battery’s SOC
Electrochemical impedance spectroscopy can quickly and directly reflect the dynamic
value [15–17]. Electrochemical impedance spectroscopy can quickly and directly reflect
characteristics of batteries, and can decouple the factors affecting the SOC. However, it is
the dynamic characteristics of batteries, and can decouple the factors affecting the SOC.
difficult to measure electrochemical impedance spectroscopy in practical applications.
The load voltage-based method is also commonly used to estimate the SOC. Since the
variation law of load voltage on the SOC is similar to that of an open-circuit voltage with
SOC, the SOC can be roughly calculated based on voltage variation [18]. This method has
However, it is difficult to measure electrochemical impedance spectroscopy in practical
However,
However,ititisisdifficult
applications. difficultto tomeasure
measureelectrochemical
electrochemicalimpedance
impedancespectroscopy
spectroscopyin inpractical
practical
applications.
The load
applications. voltage-based method is also commonly used to estimate the SOC. Since the
Sustainability 2023, 15, 5014 The
Theload
variation law
loadofvoltage-based
load voltagemethod
voltage-based method isisalso
on the SOC is commonly
also similar
commonly used
to thatusedofto anestimate
to the
theSOC.
open-circuit
estimate Since
voltage
SOC. the
5with
Sinceof the
22
variation
SOC, the law
SOC of
can load
be voltage
roughly on the SOC
calculated is similar
based on to that
voltage of an
variation
variation law of load voltage on the SOC is similar to that of an open-circuit voltage with open-circuit
[18]. This voltage
method with
has
SOC,
the
SOC, the
theSOC
SOCcan
advantages be
beroughly
of simple
can calculated
calculation
roughly andbased
calculated high on
based onvoltage
voltagevariation
estimation variation[18].
efficiency. This
Thismethod
However,
[18]. there is
method hasa
has
the advantages
“platform area”
the advantages
advantages of of
[5] simple
in the
of simple calculation
relationship
simple calculation and
calculation andcurve high
and high estimation
between the
high estimation efficiency.
terminal
estimation efficiency. However,
voltage
efficiency. However, and SOC
However, therethere
there isisaaa
of is
the
the
“platform
lithium area”
battery.
“platformarea” [5]
As
area”[5] in
shownthe
[5]ininthe relationship
in Figure
therelationship 4, curve
relationshipcurvea between
certain
curvebetween the
open-circuit
betweenthe terminal
theterminalvoltagevoltage
terminalvoltage in the
voltageandand SOC
SOCof
platform
andSOC the
area
of the
“platform of the
lithium
corresponds
lithium battery.to
battery. As
a As shown
wideshown in
inFigure
capacityFigure 4,
range,4,a acertain
and an
certain open-circuit
inaccurate
open-circuit voltage
terminal
voltage in the
voltage
in the platform area
estimation
platform area
lithium battery. As shown in Figure 4, a certain open-circuit voltage in the platform area
corresponds
within to aawill
this range
corresponds widelead capacity range,
to a large and an
SOC estimation an inaccurate terminal
error. In addition, voltage
due to theestimation
existence
corresponds totoa widewide capacity
capacity range,
range, and and inaccurate
an inaccurate terminal terminal
voltagevoltage estimation
estimation within
within
of this
polarization
within range
thiswill will
voltage,
range will lead
the to a large
variation SOC
in estimation
battery error.
terminal In
voltageaddition,
is very due to the
different existence
when the
this range lead tolead to a SOC
a large largeestimation
SOC estimation error.error. In addition,
In addition, due due to the
to the existence
existence of
of polarization
current changes
of polarization voltage,
rapidly,
voltage, the variation
which will in
also battery terminal
introduce errors voltage
in the is
SOCvery different
calculation. when
whenthe
polarization voltage, thethe variation
variation in in battery
battery terminalvoltage
terminal voltageisisvery verydifferent
differentwhen the
the
current
currentchanges
changesrapidly, which
rapidly,which will
whichwill also
willalso introduce
alsointroduce errors
introduceerrors
errorsinin the
inthe SOC
theSOC calculation.
SOCcalculation.
calculation.
current changes rapidly,
Figure 4. Curve of terminal voltage SOC for LFP battery. Reprinted with permission from Ref. [19]
(© 2020 byCurve
Figure Zhihu) .
Figure4.
Figure 4.4. Curve
of
Curveof terminal
ofterminal
terminalvoltage
voltage
voltage
SOC
SOCSOC for LFP
for for
LFPLFP
battery.
battery.
battery.
Reprinted
Reprinted
Reprinted
with
withpermission
permission
with permission
from
from
from Ref.
Ref. [19]
[19]Ref. [19]
(© 2020
(©(©2020
2020byby Zhihu)
Zhihu)
.
.
by Zhihu).
2.3. The Model Method
2.3.
2.3.The
2.3. TheModel
The
The Model
Model Method
lithium–ion
Methodbattery model includes an external characteristic model and an elec-
Method
The
trochemical lithium–ion
model;
Thelithium–ion
The lithium–ion battery
both can be
battery
battery model
usedincludes
model
model for SOCan
includes
includes external
externalcharacteristic
calculation.
an
an external characteristicmodel
characteristic modeland
model andan
and anelec-
elec-
trochemical
trochemical model;
model; both
both can
can be
be used
used for
for
trochemical model; both can be used for SOC calculation.SOC
SOC calculation.
calculation.
2.3.1. SOC Estimation Method Based on the Electrical Model
2.3.1.
2.3.1.SOC
TheSOC
SOC Estimation
Estimation
method of SOC
Estimation Method
Method Based
estimation
Method Based on
onthe
based onElectrical
the the external
Electrical Model
characteristic model is relatively
Model
The
simple,The method of SOC estimation based on the externalcharacteristic
method
involves
method of
less
of SOC
SOC estimation
computation
estimation based
and
based ison
on the
easier
the external
to apply
external in engineering.
characteristic
characteristic model
modelThe
model isisrelatively
is electrical
relatively
relatively
simple,
simple, involves less computation and is easier to apply in engineering. The electrical
model
simple, involves
includes
involves theless
less computation
equivalent
computation circuitand is
model and
easier the
to OCV–SOC
apply in
in engineering.
models.
engineering. The
Commonly used
electrical
model
model includes the equivalent circuit model and the OCV–SOC models. Commonlyused
includes
equivalent the
circuit equivalent
models circuit
include the
circuit model
Rint and
model, the OCV–SOC
the PNGV
OCV–SOC models.
model Commonly
[20], the nth-order
used
equivalent
equivalent
Thevenin
equivalent circuit
circuit
model models
models
[20,21]
circuit models include
includethe
include
(nth-order the
RC Rint
Rintmodel,
Rint
themodel), model,
the GNL
model, the
thePNGV
the model model
PNGV
PNGV model
[20],
model [20],
etc. Thethe
[20], nth-order
comparison
the nth-order
Thevenin
of different
Thevenin model
model
models
model [20,21]
[20,21] (nth-order
(nth-order
is listed
[20,21] in Table
(nth-order RC
RCRC model),
1.model),
model), the
thethe GNL
GNLGNL model
model
model [20],
[20], etc.
etc.
[20], The
The
etc. comparison
comparison
The comparison of
different
of different models
modelsis listed
is listedin Table
in
of different models is listed in Table 1. 1.
Table 1.
Table 1. Classification and comparison of models for a lithium–ion battery.
Table 1.
Table Classification and comparison of models for a lithium–ion battery.
Classification Table1.
1.Classification
Classificationand
Configuration andcomparison
comparison ofofmodels
Description modelsfor foraalithium–ion
lithium–ion
Advantage
battery.
battery. Disadvantage
Classification
Classification
Classification Configuration
Configuration
Configuration Description
Description
Description Advantage
Advantage
Advantage Disadvantage
It cannot reflect the dy-
Disadvantage
Disadvantage
The model is sim-
An ideal voltage source namic
ItItcannot
cannotcharacteristics
reflect
reflect of
thethedy-
ple,model
The and the isispa-
sim- It cannot reflectthe dy-
Rint model isAnconnected
ideal voltageto a resistor
source The model sim- thenamic battery and
characteristics
dynamic has low
characteristics of
An ideal
An idealvoltage
voltage source
source rameter
ple,Theandmodel is simple, namic characteristics of
measure-
the
RintRint
modelmodel is connectedin series.
is connected to a
to aresistor
resistor ple,
andand thepa-
the parameterpa- the of the battery and has
accuracy
battery with
and a small
has low
Rint model is connected to a resistor rameter
in series. ment is easy.
measure-
measurement is easy.
the battery
low accuracyand withhas alow
in rameter measure- range ofsmall
applications.
inseries.
series. ment
accuracy
accuracy with
with aasmall
range
small
The n-order Thevenin mentisiseasy. easy. The range
change
of applications.
of applications.
in open-cir-
Electrical The RC loop is range of applications.
equivalent
The
Then-order
The circuit
n-order
n-order Theveninmodel
Thevenin
Thevenin cuit
Thevoltage
change and
in thein self-
inopen-cir-
The change
model
Electrical
Electrical equivalent circuit model
used
The
The
The
to
RC
RC
RC
simulate
loop
looploop
is
isis to Theopen-circuit
used
change open-cir-
voltage and
is based
equivalent
equivalent on the
circuit
circuit Rint
model
modelthe discharge
cuit voltage and caused
andthe theby
self-
model
model
Electrical is based on the Rint used dynamic
to
simulate
used to char- cuit voltage
simulate
the dynamic
simulate the self-discharge self-
Sustainability Thevenin
2023, 15, x FOR PEER REVIEW model; n
isisbased
model;RC
basedon non circuits
RCthe Rint
circuits
the are
Rintare characteristics of load current
discharge
caused bycaused
discharge accumu-
load current
caused by
6 ofby23
model Thevenin model
connected in seriesto to
acteristics
the
the dynamic
dynamic
batteries,
of batter-
and the
char-
char-
higherlated accumulated over
model
Thevenin
Thevenin connected
model;
model; nnRC RC incircuits
series
circuits are load over
current time aretime
accumu- not
represent the areacteristics
ies, and the
the n, of
the higher
batter-
higher load
are current accumu-
not considered. The
model represent
connected thein polariza-
series to acteristics of batter- considered.
lated The bigger
model polarization
connected in series to ies, theand n, the
the precision
the higher
higher lated over time arenot
over
bigger time
the n,are
the not
tion phenomenon
phenomenon
represent the of the
of
polariza- cell.
the ies, and the higher the n, more
the more
considered. computation.
The compu-
bigger
represent
Based oncell. the polariza-
the 1st order the then, precision
the higher considered. The bigger
tion The
the n, calculation
the higher the n, thetation.
tionphenomenon
Thevenin phenomenon
Basedequivalent
of
on the 1st order
the
ofcir-
the the precision the n, themore morecompu- compu-
cell. burden
the is low;
precision tation.
Thevenin
cell. equivalent The calculation burden tation.
cuit model, capacitor
circuit model, capacitorCp Compared with the
is low; Compared It does not solve the bat-
with
It does not solve the
is added Cp is to
added to describe
describe the 1stthe 1st order Thevenin
PNGVPNGV
modelmodel
the change in
order Thevenin
equivalent circuit
tery self-discharge
battery prob-
self-discharge problem.
change in open-voltage
open-circuit circuit equivalent model, thecircuit
accuracy lem.
caused by load current
voltage caused
over time.
by load model,isthe higher.
accu-
current over time. racy is higher.
Compared with
PNGV, the open
The two RC loops repre-
circuit voltage
sent concentration po-
change caused by Compared with the
larization and electro-
load current over PNGV model, the calcu-
chemical polarization,
Based on
Based on thethe 1st
1st order
order The calculation
calculation
The
Thevenin equivalent
Based onequivalent
Thevenin the 1st order cir-
cir- Theburden is low;
low;
calculation
burden is
cuit model,
Thevenin
cuit model, capacitor
equivalent
capacitorcir- Cp Compared
Cp Compared
burden iswith with It does
the It
low;the does not
not solve
solve thethe bat-
bat-
Sustainability 2023,PNGV
15, 5014model is added to describe the 1st order Thevenin tery self - discharge 6 of 22
prob-
PNGV model cuit model,
is added to capacitor
describe the Cp Compared with the It
1st order Thevenin terydoes
selfnot solve theprob-
-discharge bat-
PNGV model change
ischange
addedin in open
toopen --circuit
describe circuit
the 1stequivalent
order Thevenin
equivalent circuit tery self-discharge
circuit lem.
lem. prob-
voltage
change in caused
open-by by model, the
load equivalent
circuit model, thecircuit
accu-
accu- lem.
voltage caused load
Table 1. Cont. current
voltage causedover time. load model, racy is
racy isthe higher.
accu-
higher.
current over by time.
current over time. Compared
racy is higher.
Compared with
with
Classification Configuration Description Advantage Disadvantage
PNGV,
Compared the open
the with
The two
The two RC RC loops repre- PNGV,
loops repre- open
circuit the voltage
Thesent
sent concentration
two RC loops repre-
concentration po- PNGV,
po-
circuit open
voltage
change caused
circuitcaused
change voltageby by Compared with
Compared with thethe
larization
sent Theand
and
concentration
larization two RC
electro-
electro-po- load
loops Compared with PNGV,
change
load current
caused over
by PNGV model,
Compared
model,with the calcu-
the
chemical polarization,
represent thecurrent over PNGV the calcu-
GNL model
model larization
chemical andconcentration
electro- time
polarization,
polarization and is
open circuit
considered,
change caused
voltage
by loadlation model,
is morethe
Compared complex
with the
GNL respectively, and the load
time current
is considered,over PNGV
lation is more calcu-
complex
chemical
respectively,polarization,
and the
electrochemical and current
the over time is with larger
battery PNGVcalculation
model, the
GNL
GNL model
model structure ispolarization,
closer to the time isthe
considered,
andconsidered,
batteryand the lation iscalculation
with larger more complex
calculation
is more
respectively,
structure is
respectively,
and
closer and
the
to the
the self-discharge
battery is with burden.
internal characteristics and theself-discharge
batteryis
self-discharge is complex
larger
burden. with larger
calculation
Electrical structure
internal is closer
characteristics
structure is closertotothethe considered;
considered; higher
higher calculation burden.
ofcharacteristics
the characteristics
internal cell. self-discharge
considered;precision is
higher
and burden.
model internal of the cell.
of the cell. precision and
wider applicability.
considered;
precision higher
and
of the cell. wider applicability.
precision
wider and
applicability.
wider applicability. Some parameters
Some parameters of of the
the
The battery
The battery terminal
terminal
Open-circuit
Open-circuit model
Some
model Somedo not
parameters
do not havehave
parameters ac-
of the
of ac-
the
Open-circuit voltage
The
voltage
Theisbattery
battery calculated
terminal
terminal
is calculated model do not have
voltage SOCSOC
Open-circuit
voltage voltage is calculated SimpleSimple calculation
calculation model tual physical
do signifi-
notphysical
have ac-
voltage with thewith
relationship be- Simple calculation tual physical actual signifi-
SOC model
model withvoltage
the isthe
calculated
relationship
relationship be- cance, and the the accuracy
significance, accuracy
and the
voltage
model SOC tween OCV
between and SOC.
OCV SOC.
and SOC. Simple calculation tual
cance, physical
and signifi-
with
tweentheOCVrelationship
and be- accuracy
is low.
low. is low.
model cance, and is the accuracy
tween OCV and SOC.
is low.
The lithium–ion
The lithium–ion battery battery
The is equivalent
The lithium–ion
is lithium–ion
equivalent to to the
battery
battery
the
is equivalent to the
It is
It is too
too complex
complex and and
structure
is equivalent of electrode It is too complex and
structure structure ofto
of electrode the
electrode computationally inten-
(positive and
(positive negative
and HighHigh
negative High accuracy and
accuracy
It is toocomputationally
complex and
and computationally inten-
P2D
P2D model
model structure
(positive of electrode
and negative accuracy and sive, and the analytical
intensive,analytical
and the
P2D model electrode);
electrode); diaphragm
diaphragm wide applicability
wideaccuracy
applicability computationally
sive, and the
analytical
inten-
solution can
(positive
electrode); andand negative
diaphragm
electrolyte are High
wide and solution
applicability can be is diffi-
P2D model and electrolyte
composed are
ofare
numerous sive, and is the
com- wide applicability solution can be is obtain.
be analytical
difficult to diffi-
electrode);
and electrolyte diaphragm com- cult to obtain.
posed spherical solid
of numerous
numerous particles. solution can be
cult to obtain. is diffi-
and electrolyte
posed of are com-
spherical solid particles. cult to obtain.
posed of
spherical numerous
solid particles.
Electro- spherical solid particles.
chemical
model
Electrochemical
Electrochemical Under the the condition
condition of of
Under
model
Electrochemical
model Two spherical particles high rate
Under charge
the conditionand of
Two spherical particles Under therate
high high
rate condition
charge
chargeand and
of
model are
Two
are used
used
Twoto represent
spherical
spherical
to represent the
particles
particles Simple structure discharge,
high rate the assump-
chargethe and
to representthe the Simple structure
structure anddischarge, the assump-
discharge,
are used Simple
SP model
SP model arepositive and
usedpositive
positive toand negative
represent
negative and
the and small
Simple
small amount
structure tion of
amount discharge,
amount tion of the
the model
assumption
the
ofisthe
assump- not
model is not valid,not
model is
SP model and negative small
the
SP model terminals
terminals of
terminals
positive and lithium
of lithiumion
negative
of lithium
ion
ion and of calculation
of calculation valid, the calculation er-
ofsmall amount valid,
calculation tion of thecalculation
the model
calculation erroris not
er-
is large
batteries, respectively.
batteries,of
terminals
batteries, respectively.
lithium ion
respectively. of calculation ror is
valid,
ror isthe
large
and theand
large the ap-
application
calculation
and the ap-
er-
range is small.
batteries, respectively. plication
ror is large
plication range
and is
range is
thesmall.
ap-
small.
The
The calculation plication range is small.
Thecalculation
calculation burden
burden
burden theofP2D
of themodel
Theofcalculation
the P2Dis
P2D Unable to solve the
Simplify the The PDE of the P2D Unable
greatly reduced. It is Unable to solve
to
inherentsolve the in-
the
problems in-
of
P2D model model isthe
simplified.
model
burden
model is
more
greatly
isofgreatly
the P2D
accurate
re-
re-
and herent theproblems
P2D, it is difficult to
Simplifythe The PDE of P2D of the
Simplify the The PDE of the P2D duced.
model is It is more
applicable
greatlymore the Unable
thanre- toapply
herent problemssolve the
online. in-
of the
P2D model model is simplified.
simplified. duced. It is P2D, it is
is difficult
difficultof totheap-
Simplify the
P2D model The PDE
model is of the P2D accurate SP and model. herent
appli- P2D, it problems to ap-
duced. Itand
accurate is more
appli- ply online.
P2D model model is simplified. cable than
than the SP P2D, itply is difficult
online. to ap-
accurate
cable andthe appli-
SP
The Rint model is the simplest equivalent circuit model,
model. which ply online.
consists of an ideal
cable model.
than the SP
voltage source and an internal resistance in series. The model is simple, and the model
model.
parameters
The Rintare
The Rint easyis
model
model istothe
measure.
the simplestHowever,
simplest equivalent
equivalent it cannot reflect the
circuit model,
circuit model, which
whichtransient process
consists
consists of an
of of the
an ideal
ideal
battery
voltage and
source
Thesource
voltage has
and
Rint modellow
an
and an accuracy
internal
is internal and
the simplest a small
resistance in
equivalent
resistance application
series. The
circuit
in series. range.
model
Themodel,
modelwhichisCompared
simple, and
consists
is simple, with
andof the the Rint
model
theanmodel
ideal
model,
parameters
voltage the Thevenin
are
source easy model
to adds
measure. an RC
However, circuit
it to reflect
cannot the
reflect polarization
the transientprocess
processinside
of the
the
parameters are and
easyantointernal resistance
measure. However, in itseries.
cannot The model
reflect theistransient
simple, and the model
process of the
cell.
batteryThe
parametersandmore
has RC
lowcircuits added,
accuracy and the
a more
small accurate
application the circuit
range. model
Compared and the
with higher
the the
Rint
battery andare haseasy
low to measure.
accuracy andHowever,
a small it cannot reflect
application range. theCompared
transient process
with theofRintthe
computational
model,
battery the
and has complexity
Thevenin
low model
accuracy and
addsamount.
and ana RC The
circuit
small second-order
to
application reflect Thevenin
the
range. modelprocess
polarization
Compared is also
with theknown
inside
Rint
model, the Thevenin model adds an RC circuit to reflect the polarization process inside
as the
the cell.dual
model, the polarization
TheThevenin
more RCRCmodel (DP)adds
circuits model,
an in
added, which
the
RC moreone
circuit to RC
accuratecircuit
reflectthe represents
circuit
thecircuit theand
model
polarization electrochemical
the higher
process higher
inside
the cell. The more circuits added, the more accurate the model and the
polarization of the battery, and the other represents the concentration
the cell. The more RC circuits added, the more accurate the circuit model and the higher polarization of the
battery. However, the Thevenin model does not consider the change in open-circuit voltage
and self-discharge caused by load current over time. Compared with the 1st order Thevenin
model, the PNGV model adds a capacitor in the loop to describe the change in the open-
circuit voltage with the current integral. The GNL model integrates the advantages of
the above three models. It not only reflects the ohmic polarization, the electrochemical
the computational
polarization and thecomplexity
concentration andpolarization
amount. The second-order
processes of the Thevenin
battery, but model is also
also solves
known
the as the
problem of dual polarizationvoltage
the open-circuit (DP) model, affected in which
by the one RC circuit
integral represents
load current elec-
and the self-
trochemical
discharge of polarization
the battery. of the battery,
It has a wide range and the ofother represents
applications, but the
theconcentration
complexity of polari-
the
zationand
model of thethebattery.
amountHowever, the Thevenin
of calculation increase model
at the same does not time.consider the change in open-
circuitThevoltage
estimationandmethod
self-discharge
based on caused by load current
the OCV–SOC curve can over
alsotime. Compared
estimate with the
the lithium–ion
1 orderSOC.
battery’s
st Thevenin model,
The main the models
fitting PNGV [22] model adds the
include a capacitor
Shepherdinmodel the loop to describe
(Equation the
(3)), the
change inuniversal
Unnewehr the open-circuit voltage (4)),
model (Equation withthe theNernst
current modelintegral. The GNL
(Equation model
(5)) and integrates
the compound
model (Equation (6)):
the advantages of the above three models. It not only reflects the ohmic polarization, the
electrochemical polarization and = Econcentration
Uk the 0 − Rik − K0 /S polarization
OCk processes of the battery, (3)
but also solves the problem of the open-circuit voltage affected by the integral load current
Uk = E0 − Rik − K1 SOCk (4)
and the self-discharge of the battery. It has a wide range of applications, but the complex-
ity of the model and the amount Uk = 0 − Rik + K
of Ecalculation 2 ln( SOCk
increase at) the same time. (5)
The estimation method Uk based
= E0 − onRithe−OCV–SOC
K0 /SOCk −curve K1 SOCkcan also estimate the lithium–
k
ion battery’s SOC. The main (6)
+K2 ln(SOCk ) + K3 ln(1 − SOCk ) Shepherd model (Equation
fitting models [22] include the
(3)), the Unnewehr universal model (Equation (4)), the Nernst model (Equation (5)) and
where U k is the terminal
the compound voltage of(6):
model (Equation the battery at time step k; ik is the current at time step k;
E0 is the open circuit voltage after the battery is fully charged; R is the internal resistance of
the battery; SOCk is the SOC of kthUtime k = Estep;
0 − RiK k 0−,K K10, /KS2OCandk , K3 are the parameters to be (3)
fitted. A detailed description and comparison of the above models are shown in Table 1.
When estimating SOC based on = Eelectrical
U k the 0 − Rik − Kmodel, 1 SOC k , the battery system is usually
(4)
discretized, the SOC is taken as the system state and the SOC is estimated using a filter
U k = E0 − Rik + K 2 ln( SOC k ) , (5)
based on the ampere integration method.
Taking the 1st order Thevenin U k = equivalent
E0 − Rik − Kcircuit0 / S OC k − K1 S OC k
as an example, as shown in Figure 5,
the discretized equation of state and the output equation are shown , in Equations (7) and (6)
+ K 2 ln( SOC k ) + K 3 ln(1 − SOC k )
(8), respectively:
where U is the terminal voltage of the battery at time# step k; ik is the current at time step
" Ts
U
k; E0 is the open circuit voltage 1,k after exp
the (−
battery
R C )is 0 charged;
fully U 1,k − 1 R is the internal resistance
= 1 1 +
of the battery; SOCk is " SOCk
the SOC of kth time0 step; K 01, K 1, KS and
2 OCk −K 1 3 are the parameters to be
# are shown in Table(7)
fitted. A detailed description R1 (1 − andexp (− RT1 Cs 1 )) of the abovewmodels
comparison 1.

1,k −1
When estimating SOC based Ton the [ I
electrical ] +
k −1 model, the battery system is usually
−Q s w2,k−1
discretized, the SOC is taken as theI system state and the SOC is estimated using a filter
based on the ampere integration [Uk ] = U method.
oc ( SOCk ) − U1,k − R0 Ik + vk (8)
Taking the 1st order Thevenin equivalent circuit as an example, as shown in Figure
5, theUdiscretized
where 1,k is the voltage of the
equation ofbattery
state and polarization
the output capacitor;
equation subscript
are shown k is the current time;
in Equations (7)
Tand
s is sampling time;
(8), respectively: and w k and v k are system noise and observation noise, respectively.
They are Gaussian white noise with mean is zero, covariance Q and covariance R.
Figure5.5.First
Figure Firstorder
orderofofequivalent
equivalentcircuit
circuitfor
forbattery.
battery.
Uoc (SOCk ) is the function of open circuit voltage Uoc with respect to SOC, which
can be obtained by fitting the curve of OCV–SOC. R1 , C1 are the internal resistance and
capacitance of polarization, respectively. R0 is the ohmic internal resistance of the battery;
QI is the actual total capacity under the current I. The battery’s SOC can be estimated by
combining the filter and its derived algorithm.
The commonly used filters include the Kalman filter (KF), the particle filter (PF) and
the H infinite filter (HIF).
The KF can use the output data to continuously revise the system state variables and
provide the optimal estimation of the state quantity at the next moment [23]. The KF is
widely used in the research of battery SOC because of its good resistance to noise and low
dependence on an initial value.
Since lithium–ion batteries are nonlinear systems, the extended Kalman filter (EKF) is
used to estimate the SOC [24], which solves problems caused by system nonlinearity by
performing a Taylor expansion at the operating point.
According to Equations (7) and (8), the error of the EKF comes from the following
three points:
(1) Only the first-order term is taken when the Taylor expansion is carried out on the
nonlinear system at the operating point, and the higher-order term is ignored;
(2) Noise wk and vk ,and their covariances Q and R, are constants and do not change
dynamically;
(3) Inaccurate battery model parameters of R0 , R1 and C1 will lead to large errors.
For problem (1), the unscented Kalman filter (UKF) is employed [25]. The UKF elim-
inates the larger error caused by the partial derivative of the first order using the Taylor
expansion of the nonlinear function in the EKF. However, this method needs accurate
statistical characteristics for the system process noise and the observation noise. In addition,
because the calculation process is based on a matrix, when there are drastic fluctuations of
voltage and current in the system, the matrix cannot guarantee strict positive quality, result-
ing in divergence of state estimation and unstable estimation [26]; the amount of calculation
is also large. To solve the problem of the UKF, some scholars proposed the square root
unscented Kalman filter (SR–UKF) and ensured the semi-positive nature of the covariance
matrix by Cholesky decomposition factor updating and matrix QR decomposition, which
increases the stability of calculation [27]. However, problem (2) still remains.
The adaptive extended Kalman filter (AEKF) method was proposed by reference [28]
to solve problem (2). Based on the EKF, the AEKF focused on the problem of filtering
divergence caused by the unknowability of the covariance matrices Q and R. A forgetting
factor is added to the EKF, and the covariance of error is corrected by the observed value.
The convergence rate is improved by dynamic changes in Q and R. However, the AEKF
does not consider the change in model and equation parameters, and the fixed forgetting
factor will also introduce errors.
The idea of the PF is based on the Monte Carlo method, which uses sets of particles to
represent probabilities, and can be applied to any form of state–space model. Compared
with the KF, the PF is a nonlinear and non-Gaussian filtering method, which does not make
a linear assumption or a Gaussian hypothesis of posterior probability. In reference [29], the
PF was used to estimate the SOC of Fe–phosphate lithium–ion batteries, and the results
showed that it could well describe the external characteristics of nonlinear systems and
significantly improve the prediction accuracy of the SOC. However, particle degradation
was prone to occur. To solve problems such as particle scarcity and noise disturbance,
reference [30] proposed an improved estimation algorithm, the unscented particle filter
(UPF), to achieve accurate SOC estimation. However, the UPF requires high computing
ability for hardware devices.
The HIF method is an improved algorithm based on the EKF. In the HIF, Q and R are
parameters designed according to the prior knowledge of wk and vk . Compared with the
EKF and the PF, model inaccuracy and noise uncertainty can be better tolerated [31,32]. In
reference [33], the HIF algorithm was used to obtain the SOC estimation results with high
accuracy and robustness. Reference [34] uses a dual HIF algorithm to estimate the SOC,
which can stabilize the estimation error within 2%. Most of the HIF algorithms can achieve
high accuracy; however, due to their robustness, there is still a problem of insensitivity to
system state and model uncertainty to noise. In reference [35], a nonlinear observer based
on HIF was proposed to estimate the SOC for lithium–ion batteries. Experimental results
show that this algorithm can suppress model errors under different working conditions,
but the algorithm is slow to respond to changes in model parameters.
Similar to filters, observer-based methods can also observe model states in real time
by calculating systematic errors [36]. Among them, the sliding mode observer (SMO) is a
closed-loop state observer, and the output variable provides correction for the observer sys-
tem. Due to its closed-loop characteristics, the SMO has strong robustness, low sensitivity
to battery model structure and parameter disturbance [37], no prior information to predict
noise [38] and its estimation accuracy is close to that of the PF. Reference [39] proposes
an SOC estimation method for lithium–ion batteries based on an improved SMO, which
inherits the good robustness of the SMO.
2.3.2. SOC Estimation Method Based on the Electrochemical Model

Although the electrical model is simple, it cannot fully reflect the state change caused
by the change in the internal mechanism of the battery. The electrochemical model of
the lithium–ion battery describes the chemical reaction inside the battery and has a more
definite physical meaning. The battery electrochemical model mainly includes a pseudo
two-dimensional (P2D) model [40], a single particle (SP) model and a simplified P2D model.
The comparison of the three models is also shown in Table 1.
The P2D model regards the lithium battery as a structure composed of electrodes
(positive and negative), a diaphragm and an electrolyte full of numerous spherical solid
particles. A series of partial differential equations (PDE) is used to describe the dynamic
mechanism inside the battery. It can carry out accurate battery state estimation and has
universality and scalability. It is suitable for batteries of different material systems and can
be developed and extended to more complex multi-field coupling models. However, for
complex models, in addition to excessive parameters, it is also difficult to find analytical
solutions for PDE [8].
The SP model is the simplest electrochemical model of the lithium battery, which is
simplified from the P2D model. In the SP model, two spherical particles are regarded
as the positive and negative electrodes of the battery, respectively. It is assumed that the
embedding and ejection process of lithium ions occurs on the spherical particles, and the
concentration of the electrolyte and its internal potential are considered constant. The
single particle model is simple in structure and small in calculation. However, the model
assumption is not valid under the conditions of high rates of charge and discharge, so it
will introduce large errors.
Since the equations of the P2D model are too complex and the accuracy of the SP
model is poor, many scholars have proposed different simplified methods for the P2D
model in different application scenarios to ensure calculation accuracy and reduce the
amount of calculation. The existing simplification methods mainly include geometric
structure simplification, solid–liquid diffusion process simplification and transformation
simplification by mathematical algorithm. In reference [41], mathematical equations were
established to describe the internal physical and chemical behaviors of power batteries
during charging and discharging based on the porous electrode theory and the concentrated
solution theory; the finite analysis method was used to grid and discretize the PDE, reducing
its order to an ordinary differential equation. However, the simplified model cannot solve
the inherent problem of excessive P2D parameters and will reduce the model accuracy to
varying degrees. It is still worth thinking about how to balance the amount of calculation
and the model’s accuracy [42–45].
When calculating the SOC based on the electrochemical model, different from Equation (1),
the SOC of the electrochemical mechanism model (such as the P2D model and the SP
model) is divided into surface SOC and average SOC [46], which can be expressed as
θs,i − θs0%
SOCsurf = (9)
θs100% − θs0%
θs,i − θs0%
SOCbulk = 100%
(10)
θs − θs0%
where SOCsurf and SOCbulk represent surface SOC and average SOC, respectively. The
parameters θ s,i and θs,i represent electrode utilization and the average utilization rate of the
electrode, respectively, while θ s,i is the ratio of solid lithium ion concentration to the largest
solid lithium ion concentration ratio. The electrode utilization ratio of θ s 0% and θ s 100% is
for 0% SOC and 100% SOC, respectively.
Similar to the calculation of the SOC based on an external characteristic model, the
SOC of an electrochemical model can be estimated by a filter or observer.
2.4. Data-Driven Methods

The relationship between battery parameters and the SOC is complex and non-linear,
making it difficult to establish a model with traditional mathematical methods [47]. Data-
driven methods do not rely on a specific model of the battery, and can directly rely on
the mapping relationship between the input and output of the system to estimate the
SOC. At present, widely used data-driven methods include neural networks, regression
analysis, etc.
2.4.1. Neural Networks and their Improvement Method

The SOC estimation method based on neural networks takes the battery operating
data, such as voltage, current and other parameters as input and SOC as output and
trains the system through sample data to find the mapping relationship between various
parameters [48].
The basic structure of a neural network (NN) consists of a three-layer formation as
shown in Figure 6. The input layer takes the vector of instantaneous current, voltage and
temperature values. The output layer is the instantaneous SOC value. By training the NN,
it is able to form a non-linear map that accurately models the input–output relationship
without any prior knowledge of the internal structure of the battery [49]. The relationship
between the input layer and the output layer is developed using a suitable number of
hidden layers, hidden neurons and an activation function. The SOC in the output layer can
be expressed by n o
SOCi = f i ∑k w j,k o j + θ j,k (11)
Figure6.6.The
Figure Thegeneral
generalarchitecture
architectureofof the
the 3-layer
3-layer neural
neural network
network forfor SOC
SOC estimation.
estimation. Reprinted
Reprinted with
with permission
permission from
from Ref. Ref.
[49] (© [49]
2018(©
by2018 by Elsevier).
Elsevier).
AAback
backpropagation
propagation(BP)(BP)neural
neuralnetwork
networkisisa atypical
typicalneural
neuralnetwork
networkalgorithm.
algorithm.ItsIts
network
networkstructure
structureisissimple,
simple,and
anditithas
hasa astrong
strongnonlinear
nonlinearmapping
mappingability.
ability.Reference
Reference[50]
[50]
summarized
summarizedthe theinfluencing
influencingfactors
factorsofofSOC
SOCestimation
estimationand
andproposed
proposeda amethod
methodofofSOCSOC
estimation based on a BP neural network. However, BP neural networks tend to easily
converge to local optimum. In addition, the network structure is different, making it easy
for non-convergence to appear. Currently, there exists a number of common NN architec-
tures. Recurrent neural network (RNN) is a neural network that takes sequence data as
estimation based on a BP neural network. However, BP neural networks tend to easily

converge to local optimum. In addition, the network structure is different, making it
easy for non-convergence to appear. Currently, there exists a number of common NN
architectures. Recurrent neural network (RNN) is a neural network that takes sequence
data as input and makes recursion in the evolution direction of the sequence; all loop
nodes are connected by a chain, which is suitable to estimate the SOC and other temporal
data [51]. However, during its training, gradient vanishing and gradient explosion will
appear. To solve these two problems, some scholars propose a long short-term memory
(LSTM) neural network [52–54], a gated recurrent unit (GRU) neural network [55,56] and
a Bi LSTM neural network [57]. LSTMs solve the problem of gradient vanishing and
gradient explosion during long sequence training. Compared to an ordinary RNN, an
LSTM can perform better in longer sequences. A GRU is an improvement on an LSTM,
with the same accuracy as LSTM, but with fewer parameters and faster computation. In
reference [58], an RNN with the GRU is established to evaluate the SOC for a LIB. The
ensemble optimization method based on a Nadam and AdaMAx optimizer is used to
improve the training operation and determine the optimal parameters.
However, all neural networks and their derivative methods have three disadvantages:
they need to collect a large amount of data, which is not applicable for online estimation;
training results are too dependent on sample quality; and the overfitting problem appears
easily since the battery’s model is not involved.
2.4.2. The Regression Analysis Method and Its Improvement

The regression analysis method involves processing a large number of statistical data
to determine the correlation between dependent variables and some independent variables.
It establishes a regression equation with good correlation and can be extrapolated to predict
the future change in dependent variables.
A support vector machine (SVM) is an algorithm that uses classification and regression
analysis techniques to process data [59]. This method is effective in12high-dimensional
Sustainability 2023, 15, x FOR PEER REVIEW of 23
pattern recognition and nonlinear regression problems. An SVM attempts to construct
hyperplanes in high-dimensional space in order to separate data of one class from another
as shownininFigure
as shown Figure 7 [60].
7 [60]. An optimal
An optimal separation
separation boundary boundary
is achievediswhen
achieved when the distance
the distance
from thehyperplane
from the hyperplane to the
to the nearest
nearest data of
data point point of any
any class class is maximized.
is maximized. Reference [61]Reference [61]
employed
employed SVM SVM to predict
to predict theofSOC
the SOC a LIB of
as aafunction
LIB as ofa cell
function
current,of cell current,
voltage and tem- voltage and
perature. However,
temperature. when the
However, sample
when the size is expanded
sample to a certainto
size is expanded point, the complexity
a certain point, the complexity
brought by
brought bythe SVM
the SVMoptimization will bewill
optimization significantly increased,increased,
be significantly and the accuracy
and theof the
accuracy of the
model will decrease [62,63].
model will decrease [62,63].
Figure 7. Example of hyperplane to separate distinct classes for the SVM. Reprinted with permis-
Figure 7. Example of hyperplane to separate distinct classes for the SVM. Reprinted with permission
sion from Ref. [60] (© 2005 by Springer).
from Ref. [60] (© 2005 by Springer).
To solve the above problems, reference [62] proposed a least squares SVM, which
To solve
converts the above
the learning problems,
problem reference
of the standard SVM[62]
into proposed a solving
a problem of least squares SVM, which con-
linear equa-
verts the learning
tions with problem
fewer variables and of
lessthe standard SVM
computation, into solving
with faster a problem
speedofand
solving
better linear
ro- equations
with fewer variables and less computation, with faster solving speed and better robustness.
bustness.
2.5. The Joint Method

Due to the advantages and disadvantages of each method, some scholars use the in-
tegration of multiple methods to estimate the SOC, with complementary advantages. The
fusion types include the fusion of a model-based method and a data-driven method, the

Due to the advantages and disadvantages of each method, some scholars use the
integration of multiple methods to estimate the SOC, with complementary advantages. The
fusion types include the fusion of a model-based method and a data-driven method, the
fusion of different data-driven methods, etc. Reference [64] proposed a convolutional neural
network (CNN) and an LSTM integration model, which established a mapping relationship
between the SOC at the current moment and the discharge data of lithium batteries at
several historical moments, improving the accuracy of SOC estimation. Reference [65]
analyzed the nonlinear relationship between the OCV and the SOC based on the Thevenin
equivalent circuit model and combined the models, employing improved UKF with neural
networks to estimate the SOC; this was more accurate than the ordinary UKF.
After a comprehensive analysis of the above methods for estimating the lithium–ion
battery’s SOC, the advantages and disadvantages of common methods for estimating the
lithium–ion battery’s SOC are compared, as shown in Table 2.
Table 2. Comparison of advantages and disadvantages in common SOC estimation methods for
lithium–ion batteries.
Classification Method Advantage Disadvantage

Experimental Non-relative internal mechanism of the Easy to generate cumulative error, which
Ampere integration
method battery; simple requires high initial value and sensor precision
The SOC cannot be calculated online in real
OCV–SOC method Non-relative to battery types; simple
time [5]
Electrical model Simple and practical Poor accuracy
Basic model Electro- It reflects the internal characteristics of
Complex and computationally intensive [5]
chemical model the battery
Model-
The convergence speed is fast, and the noise
based method The system noise is uncertain, which requires
KF method suppression ability is strong.
high accuracy of the model
State estima- Low sensitivity to initial value
tion method
Strong robustness and low requirement for Prone to particle degradation
PF method
model accuracy Computationally intensive
Easy to disappear gradient and fall into
local optimization
Neural network No reliance on high-precision battery models
Computationally intensive
Data- Easy to overfit, poor generalization capability
driven method
It can achieve good results in high-dimensional
Regression analysis pattern recognition, nonlinear regression and Only applicable to small data samples
other problems
Model-based method and
data-driven method
Joint method The accuracy and reliability of the estimation High complexity and computationally
Data-driven method and results are high intensive [2]
data-driven method
3. RUL Prediction Method

3.1. Classification of RUL Prediction Methods
Due to the electrochemical side reaction in the anode and cathode electrolyte of the
battery, as well as the influence of mechanical stress, manufacturing process defects, envi-
ronmental temperature and other external factors, the degradation of batteries has become
a complex problem [66]. Determining the RUL of a battery under different operating
conditions is of great significance to accurately estimate the battery replacement cycle
or maintain the battery [67]. It is generally believed that when the real-time capacity of
lithium–ion battery drops to 80% of the rated capacity, the critical value of battery failure is
reached [68].
The RUL refers to the number of charge/discharge cycles before the battery health
state has deteriorated to a point where it reaches EOL [69]. As Figure 1 shows, the RUL is
the remaining cycle numbers of the battery from the current SOH to EOL. Thus, the SOH
and the RUL can be estimated with the same methodology. The SOH estimation methods
are not further reviewed in this paper. Similar to the SOC estimation, the RUL estimation
and prediction methods for the lithium–ion battery comprise a model-based method, a
data-driven method and a joint method since the RUL can also be regarded as a battery
state. Different from the SOC, the RUL is not an instantaneous characteristic so it can be
predicted through an empirical model, a semi-empirical model and a time series prediction
data-driven method.
3.2. The Model-Based Method

As Figure 8 shows, according to different modeling mechanisms, that the model-based
method can be categorized into an empirical model, a semi-empirical model, an electrical
model and an electrochemical model.
Single exponential
model
Double exponential
model
Basic empirical
Empirical model Linear model
model
Polynomial model
Verhulst model
Basic empirical model

Model-based Arrhenius model
Semi-empirical model &
method
Stress factor model Inverse power model
RUL Electrochemical model Double exponential
prediction model
methods for Electrical model
LIB
Neural network and its
improvement
Time series prediction Regression method and Basic time series
method its improvement prediction method
Entropy-based method
Data-driven
method
Basic time series
prediction method
Feature-based method &
Aging feature
Model-based method &
data-driven method
Joint method
Data-driven method &
data-driven method
Figure 8. Classification of RUL prediction methods for a lithium–ion battery.

Figure 8. Classification of RUL prediction methods for a lithium–ion battery.
3.2.1.Empirical
3.2.1. EmpiricalModel
Model
InInempirical
empirical models,
models, battery
battery capacity
capacityattenuation
attenuationisisrepresented
represented asas
a mathematical
a mathematical
functionofofnumber
function number ofof cycles.
cycles. Different
Differentmathematical
mathematicalfunctions functions areare
used
usedto to
describe
describe thethe
battery capacity deterioration trend. Most empirical models for RUL estimation
battery capacity deterioration trend. Most empirical models for RUL estimation are devel- are devel-
opedbased
oped basedononaasingle
single exponential
exponential model
model[70],
[70],aadouble
doubleexponential
exponential model
model [71], a linear
[71], a linear
model [72], a polynomial model [73], a Verhulst model
model [72], a polynomial model [73], a Verhulst model [74], etc. [74], etc.
single exponential model : Cmax = a1 exp(a2 n) + a3 , (12
single exponential model : Cmax = a1 exp( a2 n) + a3 (12)
double exponential model : Cmax = b1 exp(b2 n) + b3 exp(b4 n) , (13)
double exponential model : Cmax = b1 exp(b2 n) + b3 exp(b4 n) (13)
double exponential model : Cmax = b1 exp(b2 n) + b3 exp(b4 n) , (14)
double exponential model : Cmax = b1 exp(b2 n) + b3 exp(b4 n) (14)
linear
linear model: C
model = cc1nn++
: Cmax = c2 c, (15)
(15)
max 1 2
2 (16)
polynomial model : C max = d1 n + d 2 n + d 3 ,
e1 / e2
Verhulst model : Cmax = , (17)
1 + [e1 / (e2C0 ) − 1] exp(e1n)
polynomial model : Cmax = d1 n2 + d2 n + d3 (16)

e1 /e2
Verhulst model : Cmax = (17)
1 + [e1 /(e2 C0 ) − 1] exp(e1 n)
where n is the equivalent cycle number; Cmax is the remain capacity at nth cycle; and a, b, c,
d, e are constant coefficients to be fitted.
The empirical model only needs one function to evaluate the RUL, which is easy to
operate and small in calculation. However, due to its lack of real physical meaning, the
estimation accuracy is low.
3.2.2. The Semi-Empirical Model

The semi-empirical model considers the influence of various physical factors on battery
aging by adopting the stress factors related to battery aging and modeling the change in
stress factors. Stress factors related to battery aging are usually modeled based on the
inverse power law [75], the Arrhenius model [76] and the double exponential model. The
stress factors can be classified according to different criteria:
(1) Battery parameters including the SOC [77], solid electrolyte interphase (SEI) film
formation and capacity plummeting [78], etc.;
(2) Operating conditions including charge or discharge rate [74], charge or discharge cut-off
voltage [79], depth of discharge (DOD) [80,81], Ampere hour (Ah) throughput [82], etc.;
(3) Operating environment including temperature [81], etc.
Different stress factors can be coupled to obtain the comprehensive capacity decline
rate to redevelop a basic empirical model. Reference [83] applied a semi-empirical model
related to normal temperature, abnormal temperature and discharge rate to evaluate
the RUL. The result showed high accuracy with an estimation error of 2.22%. Besides
discharge rate and temperature, reference [84] coupled the stress factor of cut-off voltage,
the discharge rate and temperature to simulate the battery aging process. Reference [78]
established a capacity decline model based on the SEI films formation mechanism. However,
the semi-empirical model is no longer applicable when the operating conditions of the
battery change significantly since the variation in stress factors cannot be modeled [85].
3.2.3. The Electrical Model

Similar to the SOC estimation, the RUL can also be regarded as a battery state. Thus,
model-based methods for SOC estimation can also be employed to predict the RUL. Ref-
erence [86] proposed an improved UPF to predict the RUL with an error of less than 5%.
Reference [87] proposed a new PF framework based on a conditional variational self-coder
for RUL prediction of lithium batteries. In the process of particle resampling, a reweight-
ing strategy is introduced to prevent the loss of particle diversity. Compared with the
traditional method, this method improves the prediction accuracy.
3.2.4. The Electrochemical Model

The electrochemical model describes the influence of internal chemical and physical
reactions on battery aging, such as electrical conductivity loss, activated substances loss
and lithium–ion loss. However, the complexity of the electrochemical model and poor
dynamic accuracy limit its application in RUL prediction with a large number of aging
related parameters to be identified as compared to the electrical model [88]. Reference [89]
employed the mixtures of the inverse Burr distribution with inverse Gaussian and inverse
Weibull distributions to analyze a battery dataset and predict the RUL through the Monte
Carlo method.
3.3. The Data-Driven Method

The data-driven method selects some historical data as training samples, and then uses
the training algorithm to predict the RUL after training. This method avoids the complex
mathematical model establishment and expertise required for traditional methods. It can
be roughly classified as a time series prediction method and feature-based method.
3.3.1. The Time Series Prediction Method

The time series prediction method predicts future capacity directly based on the
3.3.1. The Time Series Prediction Method
change trend of historical capacity time series data.
The time series prediction method predicts future capacity directly based on the
The regression model, NN and entropy-based methods are usually employed to pre-
change trend of historical capacity time series data.
dict using time series. Reference [90] enhanced a PF-based empirical method by introducing
The regression model, NN and entropy-based methods are usually employed to pre-
a NN to model battery degradation trends under various operation conditions to improve
dict using time series. Reference [90] enhanced a PF-based empirical method by introduc-
their
ing ageneralization and nonlinear
NN to model battery degradation representing
trends underability,
variousand recursively
operation update
conditions the NN
to im-
model’s parameters with a particle filter to reduce the degeneracy and
prove their generalization and nonlinear representing ability, and recursively update the impoverishment
ofNN
themodel’s
PF. Reference [91] with
parameters proposed an RUL
a particle filter prediction
to reduce themethod basedand
degeneracy on an autoregressive
impoverish-
integrated
ment of the PF. Reference [91] proposed an RUL prediction method based on an auto-the sta-
moving average (ARIMA) model. However, ARIMA model requires
bility of theintegrated
regressive time series dataaverage
moving and has(ARIMA)
high requirements
model. However,for battery
ARIMAoperating conditions.
model requires
Reference [92]
the stability of constructed a CNN
the time series model
data and has for
highRUL predictionfor
requirements with an accuracy
battery operatingofcon-
over 90%.
However, gradient[92]
ditions. Reference disappearance
constructed aisCNNa typical
modelproblem
for RUL for all NN with
prediction methods. The following
an accuracy of
over 90%.of
problems However,
the timegradient disappearance
series prediction methodis a typical problem
still require for all NN methods. The
solution:
following problems of the time series prediction method still require solution:
(1) Low accuracy for local capacity variation;
(1) Not
(2) Low applicable
accuracy fortolocal capacity
predict the variation;
RUL in the battery’s entire life cycle due to different
(2) degradation
Not applicable to predict the
characteristics inRUL
eachin aging
the battery’s
period;entire life cycle due to different
degradation characteristics in each aging period;
(3) High dependence on the training set and data quality;
(3) High dependence on the training set and data quality;
(4) Problem of overfitting.
(4) Problem of overfitting.
3.3.2. The Feature-Based Method
3.3.2. The Feature-Based Method
The
The feature-based methodimproves
feature-based method improves thethe robustness
robustness of basic
of the the basic
time time
seriesseries prediction
prediction
method by considering health factors (HFs) which indicate battery aging.
method by considering health factors (HFs) which indicate battery aging. In comparison In comparison
totodirect timeseries
direct time series prediction,
prediction, the feature-based
the feature-based method method has performance
has a better a better performance
in pre- in
predicting
dicting thetheRULRUL curve’s
curve’s locallocal fluctuations
fluctuations (e.g.,
(e.g., the the capacity
capacity regeneration
regeneration effect). effect).
The
The procedure for the
procedure for the feature-based
feature-baseddata-driven
data-driven method
method is summarized
is summarized as data
as data ac- acqui-
sition, datadata
quisition, preprocessing,
preprocessing, feature
featureextraction,
extraction, feature electionand
feature election anddata-driven
data-driven prediction,
predic-
as shown
tion, in Figure
as shown 9. 9.
in Figure
Figure 9. Procedure
Figure9. forfeature-based
Procedure for feature-based data-driven
data-driven method.
method.
The batteryHFs
The battery HFsextracted
extracted
areare mainly
mainly derived
derived from thefrom the voltage,
voltage, current
current and and tem-
temper-
perature curves. The HFs that have appeared in other researches are
ature curves. The HFs that have appeared in other researches are summarized in Table 3 summarized in
Table 3 [93–98]. Then, feature selection is conducted if necessary. Since there
[93–98]. Then, feature selection is conducted if necessary. Since there are various HFs re- are various
latedrelated
HFs to battery
to aging andaging
battery the HFs
and arethe
coupled
HFs aretogether,
coupledin order to reduce
together, the calculation
in order to reduce the
burden, some researchers analyze the correlation of the HFs
calculation burden, some researchers analyze the correlation of the HFs with the RUL and select
with the
the RUL
most
and relevant
select the HFs
mostthrough
relevantranking. The Pearson
HFs through coefficient,
ranking. the Spearman
The Pearson coefficient
coefficient, and
the Spearman
gray relation
coefficient and analysis are usually
gray relation used are
analysis to analyze
usuallythe correlation
used between
to analyze health factors
the correlation between
and capacity attenuation. In reference [95], HFs that are highly correlated
health factors and capacity attenuation. In reference [95], HFs that are highly with the capac-
correlated
ity degradation
with the capacitycurve are chosen
degradation by means
curve of gray
are chosen by relation
means of analysis and used
gray relation to predict
analysis and used
the RUL with GPR. Dimension reduction means are also employed to reduce the size of
to predict the RUL with GPR. Dimension reduction means are also employed to reduce the
HFs series to figure out the most relevant part. Typical dimensionality reduction methods
size of HFs series to figure out the most relevant part. Typical dimensionality reduction
methods include Principal Component Analysis (PCA) [96], Independent Component
Analysis (ICA) [96] and autoencoder [96]. Moreover, the performance of raw HFs may not
be fully satisfactory due to the complexity of the underlying degradation process. To reduce
the possible deviation of the HF from the degradation process, the Box–Cox transformation
is sometimes used to enhance the linear relationship between the HF and the system’s
health status [99,100].
Table 3. Commonly used health factors of lithium–ion battery.
Classification Health Factors

Constant voltage charging time [93], voltage increment in the same time interval [94], time
Voltage curve relavant consumed for certain voltage increments [95], dQ/dV [97], dV/dQ [97], maximum slope of
voltage curve [95]
Constant current charging time [93], current increment in the same time interval [94], time
Current curve relavant
consumed for certain current increments [96], area under current curve [95]
Temperature increment in the same time interval [94], time consumed for certain
Temperature
temperature increments [95]
Other Wavelet packet energy entropy [98]
Typically, the NN, the regression method and their improvement are employed to
predict the RUL with an indicator of battery HFs. Reference [101] trained the RNN model
with constraint derived from prior physical knowledge to forecast the RUL. Reference [58]
employed a hybrid method with a PF and GPR to predict the aging trend sequence and
residual sequences. The proposed approach showed wide generality and reduced errors.
Reference [69] optimized the deep learning neural network (DLNN) with evolutionary
algorithms to predict the RUL.

The joint method is a combination of the model-based method and the data-driven
method and the combination of various data-driven methods that fuses the merits of
both methods. To improve prediction accuracy, reference [102] proposed an SVR-based
model with the approach of the normal particle swarm optimization (PSO). Reference [103]
proposed a joint method extreme learning machine (ELM), grey wolf optimization (GWO)
and differential evolution (DE). Reference [104] proposed a neural network-based method
combining LSTM with PSO and an attention mechanism for RUL prediction [105]. However,
all joint methods share the demerits of high computational complexity and uncertainty.
The RUL prediction methods are compared in Table 4.
Table 4. Comparison of lithium–ion battery RUL prediction methods.

Poor accuracy; not be able to consider
Empirical model Low calculation burden [3] influence of working conditions and
the environment [2]
Model-based method Low accuracy; low robustness under
Able to consider influence of working
Semi-empirical model dramatic working condition
conditions and environment
changes [1]
Large calculation amounts due to too
Electrochemical model Able to reveal the aging mechanism
many parameters [3]
Table 4. Cont.

Large calculation amounts; overfitting;
Does not require complex unable to reveal local RUL variation
Time series prediction method
mathematical models and highly dependent on training set
Data-driven method and data quality
Related to battery characteristics [4]
Feature-based method Heavy calculation burden [2]
Good accuracy for local RUL variation
Model-based method and
data-driven method
Joint method Relatively high reliability and Heavy calculation burden; high
Data-driven method and accuracy [4] complexity [2]
data-driven method
4. Conclusions
This paper summarized the techniques, models and algorithms used in lithium–ion
battery state estimation and RUL prediction in recent years and provides a comparison of
various methods. The actual operating conditions and environment of lithium–ion batteries
vary greatly. Batteries are highly complex, dynamic and nonlinear electrochemical systems,
which impose some challenges in practical application for the existing lithium–ion battery
SOC estimation methods and RUL prediction methods.
The engineering application capability of existing SOC estimation methods is poor.
It is difficult for an experimental method to estimate the SOC during battery operation,
and the application of the model-based method is limited because it relies heavily on the
accuracy of the battery model.
Some RUL prediction methods have low capability for generalization. In terms of the
empirical model and semi-empirical models, model accuracy is greatly affected by battery
parameters and working conditions, so they are difficult to generalize in real practice. In
terms of the data-driven method, the training data set is all the experimental data obtained
under a single working condition, which makes it difficult to use for covering complex
scenarios. The parameters of the data-driven method are usually fixed on most occasions,
hence it is easy to overfit when applied to other battery models and conditions.
(1) The balance between calculation time and accuracy is difficult to reach. Due to the
limitations in the number of computations required, most algorithms are not suitable
for online application. In addition, since batteries are usually integrated as modules,
calculations multiply in the real world which makes real-time application a challenge;
(2) There are only relevant studies on single batteries at present; however, factors such as
the inconsistency of batteries in battery strings in large-scale lithium battery systems
such as in an energy storage power station, will cause the original method to no
longer be applicable. At present, there is little research on the state estimation and
RUL prediction methods of battery packs.
These problems make most of the lithium–ion battery state estimation and the RUL
prediction methods difficult to promote on a large scale, but these challenges also show the future
direction for further research on state estimation and RUL prediction of lithium–ion battery:
(1) Develop the pack-level, cluster-level, system-level battery equivalent model to reduce
the amount of calculation required for state evaluation and RUL prediction;
(2) Enhance state estimation and RUL prediction methods with joint algorithms to com-
pensate for weaknesses and improve the algorithm’s capacity for generalization;
(3) Establish a multi-physical field coupling model for lithium–ion batteries to improve
the accuracy of battery status assessment and RUL prediction.
Author Contributions: Methodology, J.Z., Y.Z. and J.W.; software, Y.Z. and J.W.; supervision, M.L.;
validation, J.Z. and J.W.; visualization, M.L., B.Z. and J.W.; writing—original draft, J.Z. and J.W.;
writing—review and editing, J.Z., X.H. and C.L. All authors have read and agreed to the published
version of the manuscript.
Funding: This study was supported by the Huaneng Group Headquarters Science and Technology Project
of the Key Technology Research and System Development of Group Level Intelligence Operations Platform
Construction (HNKJ21-H52) and the Fault Diagnosis Technology Research and System Development of
Lithium–ion Battery Energy Storage Station Based on Mass Data (HNKJ21-H52-004).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: All data used during the study are available from the NASA’s offi-
cial website.
Conflicts of Interest: The authors declare no conflict of interest.
References
1. Sarmah, S.B.; Kalita, P.; Garg, A.; Niu, X.-D.; Zhang, X.-W.; Peng, X.; Bhattacharjee, D. A Review of State of Health Estimation of
Energy Storage Systems: Challenges and Possible Solutions for Futuristic Applications of Li-Ion Battery Packs in Electric Vehicles.
J. Electrochem. Energy Convers. Storage 2019, 16, 040801. [CrossRef]
2. Hasib, S.A.; Islam, S.; Chakrabortty, R.K.; Ryan, M.J.; Saha, D.K.; Ahamed, H.; Moyeen, S.I.; Das, S.K.; Ali, F.; Islam, R.; et al.
A Comprehensive Review of Available Battery Datasets, RUL Prediction Approaches, and Advanced Battery Management.
IEEE Access 2021, 9, 86166–86193. [CrossRef]
3. Elmahallawy, M.; Elfouly, T.; Alouani, A.; Massoud, A.M. A Comprehensive Review of Lithium-Ion Batteries Modeling, and State
of Health and Remaining Useful Lifetime Prediction. IEEE Access 2022, 10, 119040–119070. [CrossRef]
4. Su, N.K.H.; Juwono, F.H.; Wong, W.K.; Chew, I.M. Review on Machine Learning Methods for Remaining Useful Lifetime
Prediction of Lithium-ion Batteries. In Proceedings of the 2022 International Conference on Green Energy, Computing and
Sustainable Technology (GECOST), Miri Sarawak, Malaysia, 26–28 October 2022; pp. 286–292. [CrossRef]
5. Wang, Y.; Zuo, X. State of Charge Estimation Methods and Application Scenarios of Lithium-Ion Batteries. Power Syst. Autom.
2022, 32-1180/TP, 1–15. Available online: http://kns.cnki.net/kcms/detail/32.1180.TP.20220331.1418.002.html (accessed on
14 April 2022).
6. He, L.; Hu, M.; Shi, Q.; Ye, W. A staged state of charge estimation algorithm for lithium-ion batteries. Power Electron. Technol. 2020,
54, 8–11.
7. Wu, L.; Pang, H.; Jin, J.; Geng, Y.; Liu, K. A Review of State of Charge Estimation Methods for Lithium Ion Batteries Based on
Electrochemical Models. J. Electr. Technol. 2022, 37, 1703–1725. [CrossRef]
8. Wu, C.; Hu, W.; Meng, J.; Liu, Z.; Cheng, D. State of Charge Estimation of Lithium Ion Battery Based on Maximum Correlation
Entropy Extended Kalman Filter Algorithm. J. Electr. Technol. 2021, 36, 5165–5175. [CrossRef]
9. Li, C.; Xiao, F.; Fan, Y.; Yang, G.; Tang, X. State of Charge Estimation of Lithium Ion Battery Based on Gated Cyclic Unit Neural
Network and Huber-M Estimation Robust Kalman Filter Fusion Method. J. Electr. Technol. 2020, 35, 2051–2062. [CrossRef]
10. Wang, X.; Zheng, C.; Zhang, S.; Dai, M.; Xiao, W.; Chen, X. Summary of fusing algorithm research in estimation for state of charge
of battery. Sichuan Electr. Power Technol. 2021, 44, 43–46.
11. Li, Z.; Lu, L.; Ouyang, M. Comparison of Methods to Improve the Accuracy of Battery SOC Estimation by Ampere Hour
Integration. J. Tsinghua Univ. 2010, 50, 1293–1296.
12. Luo, Y.; Qi, P.; Huang, H.; Wang, J.; Wang, Y.; Li, P. Research on Estimation Method of Ampere Hour Integration SOC Based on
Capacity Correction. Automot. Eng. 2020, 42, 681–687.
13. Li, K.; Zhao, L.; Zhao, B.; Ke, H.; Li, J. Stream Ampere Hour Integral SOC Estimation Method Based on Frequent Item Statistics.
J. Chongqing Univ. Sci. Technol. 2022, 36, 19–27.
14. Guo, B.; Zhang, P.; Wang, W.; Wang, F. Study on SOC Estimation of Lithium Iron Phosphate Battery Based on OCV–SOC Curve
Cluster. Power Technol. 2019, 43, 1125–1128.
15. Huang, K.; Guo, Y.; Li, Z. State of Charge Estimation of Power Lithium Ion Batteries. Power Technol. 2018, 42, 1398–1401.
16. Blanke, H.; Bohlen, O.; Buller, S.; De Doncker, R.W.; Fricke, B.; Hammouche, A.; Linzen, D.; Thele, M.; Sauer, D.U. Impedance
Measurements on Lead-acid Batteries for State-of-charge, State-of-health and Cranking Capability Prognosis in Electric and
Hybrid Electric Vehicles. J. Power Sources 2005, 144, 418–425. [CrossRef]
17. Zenati, A.; Desprez, P.; Razik, H.; Rael, S. Impedance Measurements Combined with the Fuzzy Logic Methodology to Assess the
SOC and SOH of Lithium-ion Cells. In Proceedings of the 2010 IEEE Vehicle Power and Propulsion Conference (VPPC), Lille,
France, 1–3 September 2010.
18. Zeng, Q.; Zhang, X.; Fan, X. Discussion on Estimation Method of SOC of Electric Vehicle Power Battery. Electr. Meas. Instrum.
2014, 51, 76–84.
19. Available online: https://www.zhihu.com/question/434952446/answer/1628793249 (accessed on 15 December 2020).
20. Li, J.; Xiao, Y. Overview of Modeling Status of Lithium Ion Battery. Energy Storage Sci. Technol. 2022, 11, 697–703.
21. Wu, X.; Zhang, X. Parameter Identification of Second Order RC Equivalent Circuit Model of Lithium Battery. J. Nanjing Univ.
2020, 56, 754–761.
22. Park, J.H.; Kim, H.G.; Nho, E.C.; Chun, T.-W. Power conditioning system for a grid connected PV power generation using a
Quasi-Z-source inverter. IEEE J. Power Electron. 2010, 10, 79–84. [CrossRef]
23. Mao, H.; Wan, G.; Wang, L.; Zhang, Q. Estimation of Battery SOC Based on Modified Kalman Filter Algorithm. Power Technol.
2014, 38, 298–302.
24. Pang, H.; Guo, L.; Wu, L.; Min, J.; Liu, K. Improved Bipolarization Model of Lithium Ion Battery Considering the Effect of
Ambient Temperature and Its State of Charge Estimation. J. Electr. Technol. 2021, 36, 2178–2189.
25. Hu, X. SOC Estimation and Implementation of Power Lithium Battery Based on Traceless Kalman Filter. Master’s Thesis, Central
South University, Changsha, China, 2014.
26. Sun, D.; Chen, X. State of Charge Estimation of Lithium Battery Based on Discrete Sliding Mode Observer. Chin. J. Electr. Eng.
2015, 35, 185–191. [CrossRef]
27. Fei, Y.; Xie, C.; Tang, Z.; Zeng, C.; Quan, S. State Estimation of Lithium Battery Based on Square Root Unscented Kalman Filter.
Chin. J. Electr. Eng. 2017, 37, 4514–4520+4593. [CrossRef]
28. Wang, Y.; Li, L.; Ding, Q.; Liu, J.; Chen, P. Lithium-ion battery SOC estimation based on an improved adaptive extended Kalman
filter. In Proceedings of the 2021 IEEE 16th Conference on Industrial Electronics and Applications (ICIEA), Chengdu, China,
1–4 August 2021; pp. 417–421. [CrossRef]
29. Schwunk, S.; Armbruster, N.; Straub, S.; Kehl, J.; Vetter, M. Particle filter for state of charge and state of health estimation
forlithium-iron phosphate batteries. J. Power Sources 2013, 239, 705–710. [CrossRef]
30. Xie, C.; Fei, Y.; Zeng, C.; Fang, W. State Estimation of Vehicle Lithium Ion Battery Based on Traceless Particle Filter. J. Electr.
Technol. 2018, 33, 3958–3964. [CrossRef]
31. Chen, X.; Sun, D.; Chen, X. Modeling and Robust State of Charge Estimation of Lithium Ion Batteries. J. Electr. Technol. 2015,
30, 141–147.
32. Liu, C.Z.; Zhu, Q.; Li, L. A state of charge estimation method based on H-infinity observer for switched systems of lithium-ion
nickel-manganese-cobalt batteries. IEEE Trans. Ind. Electron. 2017, 64, 8128–8137. [CrossRef]
33. Chen, C.; Xiong, R.; Shen, W. A lithium-ion battery-in-the-loop approach to test and validate multiscale dual H infinity filters for
state-of-charge and capacity estimation. IEEE Trans. Power Electron. 2018, 33, 332–342. [CrossRef]
34. Da, Y.; Wan, Y.; He, W. An Adaptive Central Difference H-infinity Filter Based SOC Estimation for Lithium-ion Batteries with
Measurement Noise. In Proceedings of the 2022 41st Chinese Control Conference (CCC), Hefei, China, 25–27 July 2022; pp. 1402–1407.
[CrossRef]
35. Zhu, Q.; Li, L.; Hu, X.S. H-infinity-based nonlinear observer design for state of charge estimation of lithium-ion battery with
polynomial parameters. IEEE Trans. Veh. Technol. 2017, 12, 10853–10865. [CrossRef]
36. Zheng, Y.; Ouyang, M.; Han, X.; Lu, L.; Li, J. Investigating the error sources of the online state of charge estimation methods for
lithium-ion batteries in electric vehicles. J. Power Sources 2018, 377, 161–188. [CrossRef]
37. Hong, Y.; Chen, X. A Method for SOC Estimation of Lithium Battery Based on Improved Discrete Sliding Mode Observer.
Ind. Control Comput. 2018, 31, 67–68.
38. He, H.; Xiong, R. State of Charge Estimation of Lithium-Ion Power Battery Based on Sliding Mode Observer. J. Jilin Univ. 2011,
41, 623–628. [CrossRef]
39. Du, J. Research on Estimation Algorithm of State of Charge and Health of Lithium Ion Battery Based on Sliding Mode Observer.
Master’s Thesis, Harbin Institute of Technology, Harbin, China, 2018.
40. Doyle, M.; Fuller, T.F.; Newman, J. Modeling of galvanostatic charge and discharge of the lithium/polymer/insertion cell.
J. Electrochem. Soc. 1993, 140, 1526. [CrossRef]
41. Li, Z. Study on Electrochemical Modeling and Aging State Identification of Lithium-Ion Power Battery. Master’s Thesis, Beijing
University of Technology, Beijing, China, 2018. [CrossRef]
42. Ahmed, R.; Sayed, M.E.; Arasaratnam, I.; Tjong, J.; Habibi, S. Reduced-Order Electrochemical Model Parameters Identification and
State of Charge Estimation for Healthy and Aged Li-Ion Batteries—Part II: Aged Battery Model and State of Charge Estimation.
IEEE J. Emerg. Sel. Top. Power Electron. 2014, 2, 678–690. [CrossRef]
43. Zou, C.; Zhang, L.; Hu, X.; Wang, Z.; Wik, T.; Pecht, M. A review of fractional-order techniques applied to lithium-ion batteries,
lead-acid batteries, and supercapacitors. J. Power Sources 2018, 390, 286–296. [CrossRef]
44. Hussein, A.A. Capacity fade estimation in electric vehicles Li-ion batteries using artificial neural networks. IEEE Energy Convers.
Congr. Expo. 2013, 51, 677–681. [CrossRef]
45. Moura, S.J.; Argomedo, F.B.; Klein, R.; Mirtabatabaei, A.; Krstic, M. Battery State Estimation for a Single Particle Model with
Electrolyte Dynamics. IEEE Trans. Control Syst. Technol. 2017, 25, 453–468. [CrossRef]
46. Schmidt, A.P.; Bitzer, M.; Imre, Á.W.; Guzzella, L. Experiment-driven electrochemical modeling and systematic parameterization
for a lithium-ion battery cell. J. Power Sources 2010, 195, 5071–5080. [CrossRef]
47. Yin, A.; Zhang, W.; Zhao, H.; Jiang, H. Study on SOC Prediction of Lithium Iron Phosphate Battery Based on Neural Network.
J. Electron. Meas. Instrum. 2011, 25, 433–437. [CrossRef]
48. Hannan, M.A.; Lipu, M.S.H.; Hussain, A.; Saad, M.H.; Ayob, A. Neural Network Approach for Estimating State of Charge of
Lithium-Ion Battery Using Backtracking Search Algorithm. IEEE Access 2018, 6, 10069–10079. [CrossRef]
49. Zahid, T.; Xu, K.; Li, W.; Li, C.; Li, H. State of charge estimation for electric vehicle power battery using advanced machine
learning algorithm under diversified drive cycles. Energy 2018, 162, 871–882. [CrossRef]
50. Cai, X.; Li, B.; Wang, H.; Nie, L. Study on SOC Estimation of Power Battery Based on Neural Network Model. Electrom. Eng. 2015,
32, 128–132.
51. Xu, S.; Liu, Y.; Zhou, F. Research Progress of SOC Estimation of Lithium Ion Battery Based on RNN. Power Technol. 2021,
45, 263–269.
52. Ming, T.; Zhao, J.; Wang, X.; Wang, K. SOC Estimation of Lithium Battery under High Pulse Rate Based on Improved LSTM.
Power Syst. Prot. Control 2021, 49, 144–150.
53. Xu, C.Y.; Shen, J.Z.; Du, X.; Zhang, F. An intrusion detection system using a deep neural network with gated recurrent units.
54. Zheng, Y.; Wen, H.; Han, F.; Yang, X.; Zhang, J. Study on SOC Estimation of Power Battery Based on LSTM Neural Network.
Comput. Appl. Softw. 2020, 37, 78–81.
55. Wu, H.; Fan, Y.; Huang, H.; Fang, L.; Liu, J.; Chen, W.; Li, Z. SOC Estimation Method of Lithium-ion Batteries Based on
Relaxation Effect and GRU-RNN. In Proceedings of the 2022 IEEE 9th International Conference on Power Electronics Systems
and Applications (PESA), Hong Kong, China, 20–22 September 2022; pp. 1–6. [CrossRef]
56. Han, Y. Wide Temperature SOC Estimation of Lithium Ion Battery Based on Gated Cyclic Unit Neural Network. Master’s Thesis,
Harbin Institute of Technology, Harbin, China, 2021. [CrossRef]
57. Zhu, Y.; He, W.; Li, J.; Li, Y.; Li, P. SOC Estimation of Lithium Battery Based on Bi LSTM/Bi GRU Cyclic Neural Network.
Energy Storage Sci. Technol. 2021, 10, 1163–1176. [CrossRef]
58. Xiao, B.; Liu, Y.; Xiao, B. Accurate State-of-Charge Estimation Approach for Lithium-Ion Batteries by Gated Recurrent Unit with
Ensemble Optimizer. IEEE Access 2019, 7, 54192–54202. [CrossRef]
59. Luo, X.; Zhang, B.; Huang, X.; Hou, C. Estimation of Lithium Battery SOC Based on SVM. Power Technol. 2016, 40, 287–290.
60. Kecman, V. Support Vector Machines–An Introduction; Springer: Berlin/Heidelberg, Germany, 2005; pp. 1–47.
61. Anton, J.C.A.; Nieto, P.J.G.; Viejo, C.B.; Vilan, J.A.V. Support Vector Machines Used to Estimate the Battery State of Charge.
IEEE Trans. Power Electron. 2013, 28, 5919–5926. [CrossRef]
62. Dang, X.; Yan, L.; Jiang, H.; Wu, X.; Zhang, X. Research on SOC estimation Method of Power Battery Based on Online LS-SVM.
Power Technol. 2017, 41, 752–756.
63. Pao, Y.; Feng, C.; Ren, X.; Zhang, J.; Ma, C.; Shao, W. On-line Transient Stability Fault Screening Based on Support Vector Machine.
Power Syst. Autom. 2019, 43, 52–58.
64. Song, X.; Yang, F.; Wang, D.; Tsui, K.L. Combined CNN-LSTM Network for State-of-Charge Estimation of Lithium-Ion Batteries.
65. Li, R.; Chen, Y. SOC Estimation of Power Lithium Ion Battery Based on AFFDRLS-AHIF. J. Beijing Jiaotong Univ. 2020, 44, 129–135.
66. Zhang, S. A new method for lithium-ion battery’s SOH estimation and RUL prediction. In Proceedings of the 2018 13th IEEE
Conference on Industrial Electronics and Applications (ICIEA), Wuhan, China, 31 May–2 June 2018; pp. 2693–2697. [CrossRef]
67. Saha, B.; Kai, G.; Christophersen, J. Comparison of prognostic algorithms for estimating remaining useful life of batteries.
Trans. Inst. Meas. Control 2009, 31, 293–308. [CrossRef]
68. Cai, Y.; Chen, W.; Su, Y.; Jiang, K.; Huang, H. Review of methods for predicting residual life of lithium ion batteries. Chin. J. Power
Sources 2021, 45, 678–682.
69. Zhang, C.; Zhao, S.; He, Y. An Integrated Method of the Future Capacity and RUL Prediction for Lithium-Ion Battery Pack.
IEEE Trans. Veh.Technol. 2022, 71, 2601–2613. [CrossRef]
70. Zhang, L.; Mu, Z.; Sun, C. Remaining Useful Life Prediction for Lithium-Ion Batteries Based on Exponential Model and Particle
Filter. IEEE Access 2018, 6, 17729–17740. [CrossRef]
71. Cheng, L.; Wan, Y.; Zhou, Y.; Gao, D.W. Operational Reliability Modeling and Assessment of Battery Energy Storage Based on
Lithium-ion Battery Lifetime Degradation. J. Mod. Power Syst. Clean Energy 2022, 10, 1738–1749. [CrossRef]
72. Xue, Y.; Zhou, H.; Luo, Y.; Lam, J. Battery Degradation Modelling and Prediction with Combination of Machine Learning and
Semi-empirical Methods. In Proceedings of the 2022 12th International Conference on Power, Energy and Electrical Engineering
(CPEEE), Shiga, Japan, 25–27 February 2022; pp. 78–85. [CrossRef]
73. Rathore, K. An Empirical Capacity Degradation Modeling and Prognostics of Remaining Useful Life of Li-ion Battery using
Unscented Kalman Filter. In Proceedings of the 2018 8th IEEE India International Conference on Power Electronics (IICPE), Jaipur,
India, 13–15 December 2018; pp. 1–6. [CrossRef]
74. Zhu, Q.; Yu, X.; Wu, Q.; Xu, Y.; Chen, F.; Huang, R. Semi-empirical Degradation Model of Lithium-ion Battery with High Energy
Density. Energy Storage Sci. Technol. 2022, 11, 2324–2331. Available online: https://kns.cnki.net/kcms/detail/detail.aspx?doi=10
.19799/j.cnki.2095-4239.2021.0725 (accessed on 17 March 2022). [CrossRef]
75. Jônatas, M.; Pedro, T.J.; Carlos, A. Optimisation of an electric bus charging strategy considering a semi-empirical battery
degradation model and weather conditions. In Proceedings of the 2022 11th International Conference on Control, Automation
and Information Sciences (ICCAIS), Hanoi, Vietnam, 21–24 November 2022; pp. 298–303. [CrossRef]
76. Li, K.; Zhou, P.; Lu, Y.; Han, X.; Li, X.; Zheng, Y. Battery life estimation based on cloud data for electric vehicles. J. Power Sources
2020, 468, 228192. [CrossRef]
77. Chen, D.; Meng, J.; Huang, H.; Wu, J.; Liu, P.; Lu, J.; Liu, T. An empirical-data hybrid driven approach for remaining useful
lifeprediction of lithium-ion batteries considering capacity diving. Energy 2022, 245, 123222. [CrossRef]
78. Li, X.; Yuan, C.; Wang, Z. State of health estimation for Li-ion battery via partial incremental capacity analysis based on support
vector regression. Energy 2020, 203, 117852. [CrossRef]
79. Saurabh, S.; Darius, R.; Valentin, R.; Flynn, D.; Michael, P. Battery Stress Factor Ranking for Accelerated Degradation Test
Planning Using Machine Learning. Energies 2021, 14, 723. [CrossRef]
80. Dai, H.; Zhang, X.; Gu, W.; Wei, X.; Sun, Z. A Semi-Empirical Capacity Degradation Model of EV Li-Ion Batteries Based on Eyring
Equation. In Proceedings of the 2013 IEEE Vehicle Power and Propulsion Conference (VPPC), Beijing, China, 15–18 October 2013;
pp. 1–5. [CrossRef]
81. Wang, L.; Pan, C.; Liu, L.; Cheng, Y.; Zhao, X. On-board state of health estimation of LiFePO4 battery pack through differential
voltage analysis. Appl. Energy 2016, 168, 465–472. [CrossRef]
82. Wang, J.; Liu, P.; Hicks-Garner, J.; Sherman, E.; Soukiazian, S.; Verbrugge, M.; Tataria, H.; Musser, J.; Finamore, P. Cycle-life model
for graphite-LiFePO4 cells. J. Power Source 2011, 196, 3942–3948. [CrossRef]
83. Preetpal, S.; Che, C.; Cher, T.; Shyh-Chin, H. Semi-Empirical Capacity Fading Model for SoH Estimation of Li-Ion Batteries.
Appl. Sci. 2019, 9, 3012. [CrossRef]
84. Li, Z.; Lu, L.; Ouyang, M.; Xiao, Y. Modeling the capacity degradation of LiFePO4/graphite batteries based on stress coupling
analysis. J. Power Source 2011, 196, 9757–9766. [CrossRef]
85. Xu, B.; Oudalov, A.; Ulbig, A.; Andersson, G.; Kirschen, D.S. Modeling of lithium-ion battery degradation for cell life assessment.
IEEE Trans. Smart Grid 2018, 9, 1131–1140. [CrossRef]
86. Miao, Q.; Xie, L.; Cui, H.; Liang, W.; Pecht, M. Remaining useful life prediction of lithium-ion battery with unscented particle
filter technique. Microelectron. Reliab. 2013, 53, 805–810. [CrossRef]
87. Jiao, R.; Peng, K.; Dong, J. Remaining useful life prediction of Lithium-ion batteries based on conditional variational autoencoder-
sparticle filter. IEEE Trans. Instrum. Meas. 2020, 69, 8831–8843. [CrossRef]
88. Edge, J.S.; O’Kane, S.; Prosser, R.; Kirkaldy, N.D.; Patel, A.N.; Hales, A.; Ghosh, A.; Ai, W.; Chen, J.; Yang, J.; et al. Lithium ion
battery degradation: What you need to know. Phys. Chem. Chem. Phys. 2021, 23, 8200–8221. [CrossRef] [PubMed]
89. Chiodo, E.; De Falco, P.; Di Noia, L.P. Probabilistic Modeling of Li-Ion Battery Remaining Useful Life. IEEE Trans. Ind. Appl. 2022,
58, 5214–5226. [CrossRef]
90. Wu, Y.; Li, W.; Wang, Y.; Zhang, K. Remaining Useful Life Prediction of Lithium-Ion Batteries Using Neural Network and
Bat-Based Particle Filter. IEEE Access 2019, 7, 54843–54854. [CrossRef]
91. Zhou, Y.; Huang, M. Lithium-ion batteries remaining useful life prediction based on a mixture of empirical mode decomposition
and ARIMA model. Microelectron. Reliab. 2016, 65, 265–273. [CrossRef]
92. Li, D.; Yang, L. Remaining useful life prediction of lithium battery using convolutional neural network with optimized parameters.
In Proceedings of the 2020 5th Asia Conference on Power and Electrical Engineering (ACPEE), Chengdu, China, 4–7 June 2020;
93. Williard, N.; He, W.; Osterman, M.; Pecht, M. Comparative analysis of features for determining state of health in lithium-ion
batteries. Int. J. Progn. Health Manag. 2013, 4, 1–7. [CrossRef]
94. Ma, Q.; Wang, Y.; Yang, W.; Tao, B.; Zheng, Y. A Novel Health Index for Battery RUL Degradation Modeling and Prognostics. In
Proceedings of the 2019 IEEE 8th Data Driven Control and Learning Systems Conference (DDCLS), Dali, China, 24–27 May 2019;
95. Jia, J.; Liang, J.; Shi, Y.; Wen, J.; Pang, X.; Zeng, J. SOH and RUL Prediction of Lithium-Ion Batteries Based on Gaussian Process
Regression with Indirect Health Indicators. Energies 2020, 13, 375. [CrossRef]
96. Ren, L.; Zhao, L.; Hong, S.; Zhao, S.; Wang, H.; Zhang, L. Remaining Useful Life Prediction for Lithium-Ion Battery: A Deep
Learning Approach. IEEE Access 2018, 6, 50587–50598. [CrossRef]
97. Goyal, H.; Kikuchi, A.; Ueda, S.; Honkura, K.; Kawaji, J. Novel Multiple Parameter Optimization for Improving Accuracy of
Battery Ageing Model and Lifetime Prediction. In Proceedings of the IECON 2022–48th Annual Conference of the IEEE Industrial
Electronics Society, Brussels, Belgium, 17–20 October 2022; pp. 1–6. [CrossRef]
98. Chen, L.; Chen, J.; Wang, H.; Wang, Y.; An, J.; Yang, R.; Pan, H. Remaining Useful Life Prediction of Battery Using a Novel Indicator
and Framework With Fractional Grey Model and Unscented Particle Filter. IEEE Trans. Power Electron. 2020, 35, 5850–5859.
[CrossRef]
99. Liu, D.; Zhou, J.; Liao, H.; Peng, Y.; Peng, X. A Health Indicator Extraction and Optimization Framework for Lithium-Ion Battery
Degradation Modeling and Prognostics. IEEE Trans. Syst. Man Cybern. Syst. 2015, 45, 915–928. [CrossRef]
100. Zhang, Y.; Xiong, R.; He, H.; Pecht, M.G. Lithium-Ion Battery Remaining Useful Life Prediction With Box–Cox Transformation
and Monte Carlo Simulation. IEEE Trans. Ind. Electron. 2019, 66, 1585–1597. [CrossRef]
101. Huang, Z.; Zhou, X.; Gao, D.; Zhang, X.; Jiang, F.; Chen, B.; Yang, Y.; Wu, M.; Peng, J. A Novel Label-Free Supervision Learning
Method for Lithium-ion Battery RUL Prediction. In Proceedings of the 2019 IEEE Energy Conversion Congress and Exposition
(ECCE), Baltimore, MD, USA, 29 September–3 October 2019; pp. 3150–3156. [CrossRef]
102. Karmawijaya, M.I.; Haq, I.N.; Leksono, E.; Widyotriatmo, A. Development of Remaining Useful Life (RUL) Prediction of
Lithium-ion Battery Using Genetic Algorithm-Deep Learning Neural Network (GADNN) Hybrid Model. In Proceedings of the
2022 7th International Conference on Electric Vehicular Technology (ICEVT), Bali, Indonesia, 14–16 September 2022; pp. 13–19.
[CrossRef]
103. Xu, J.; Zhen, A.; Cai, Z.; Wang, P.; Gao, K.; Jiang, D. State of Health Diagnosis and Remaining Useful Life Prediction of Lithium-Ion
Batteries Based on Multi-Feature Data and Mechanism Fusion. IEEE Access 2021, 9, 85431–85441. [CrossRef]
104. Zhu, J.; Tan, T.; Wu, L.; Yuan, H. RUL Prediction of Lithium-Ion Battery Based on Improved DGWO-ELM Method in a Random
Discharge Rates Environment. IEEE Access 2019, 7, 125176–125187. [CrossRef]
105. Jiantao, Q.; Feng, L.; Yuxiang, M.; Jiaming, F. A Neural-Network-based Method for RUL Prediction and SOH Monitoring of
Lithium-Ion Battery. IEEE Access 2019, 7, 87178–87191. [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.

Review of State Estimation and Remaining Useful Li

Uploaded by

Copyright:

Available Formats

Review of State Estimation and Remaining Useful Li

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Review of State Estimation and Remaining Useful Li

Uploaded by

Copyright:

Available Formats

sustainability

Keywords: lithium–ion battery; SOC estimation; SOH estimation; RUL prediction

Sustainability 2023, 15, 5014. https://doi.org/10.3390/su15065014 https://www.mdpi.com/journal/sustainability

Figure 1. The relationship between SOC, SOH and RUL.

Open-circuit voltage method

Figure 2. Classification of lithium–ion battery SOC estimation methods.

2.2. The2. Experimental

Sustainability 2023, 15, 5014 4 of 22

2.2.3. Other Methods

2.3.2. SOC Estimation Method Based on the Electrochemical Model

2.4. Data-Driven Methods

2.4.1. Neural Networks and their Improvement Method

estimation based on a BP neural network. However, BP neural networks tend to easily

2.4.2. The Regression Analysis Method and Its Improvement

2.5. The Joint Method

2.5. The Joint Method

Classification Method Advantage Disadvantage

3. RUL Prediction Method

3.2. The Model-Based Method

Basic empirical model

Figure 8. Classification of RUL prediction methods for a lithium–ion battery.

polynomial model : Cmax = d1 n2 + d2 n + d3 (16)

3.2.2. The Semi-Empirical Model

3.2.3. The Electrical Model

3.2.4. The Electrochemical Model

3.3. The Data-Driven Method

3.3.1. The Time Series Prediction Method

Table 3. Commonly used health factors of lithium–ion battery.

Classification Health Factors

3.4. The Joint Method

Table 4. Comparison of lithium–ion battery RUL prediction methods.

Classification Method Advantage Disadvantage

Classification Method Advantage Disadvantage

You might also like