Review

Research Progress on Data-Driven Methods for Battery States

Estimation of Electric Buses
Dengfeng Zhao 1,2 , Haiyang Li 2 , Fang Zhou 2, * , Yudong Zhong 2 , Guosheng Zhang 1 , Zhaohui Liu 3
and Junjian Hou 2

1 Key Laboratory of Operation Safety Technology on Transport Vehicles, PRC, Research Institute of Highway,
Ministry of Transport, Beijing 100088, China; [email protected] (D.Z.); [email protected] (G.Z.)
2 College of Mechanical and Electrical Engineering, Zhengzhou University of Light Industry,
Zhengzhou 450002, China; [email protected] (H.L.); [email protected] (Y.Z.);
[email protected] (J.H.)
3 Yutong Bus Co., Ltd., Zhengzhou 450004, China; [email protected]
* Correspondence: [email protected]

Abstract: Battery states are very important for the safe and reliable use of new energy vehicles. The
estimation of power battery states has become a research hotspot in the development of electric buses
and transportation safety management. This paper summarizes the basic workflow of battery states
estimation tasks, compares, and analyzes the advantages and disadvantages of three types of data
sources for battery states estimation, summarizes the characteristics and research progress of the three
main models used for estimating power battery states such as machine learning models, deep learning
models, and hybrid models, and prospects the development trend of estimation methods. It can be
concluded that there are many data sources used for battery states estimation, and the onboard sensor
data under natural driving conditions has the characteristics of objectivity and authenticity, making it
the main data source for accurate power battery states estimation; Artificial neural network promotes
the rapid development of deep learning methods, and deep learning models are increasingly applied
in power battery states estimation, demonstrating advantages in accuracy and robustness; Hybrid
models estimate the states of power batteries more accurately and reliably by comprehensively
utilizing the characteristics of different types of models, which is an important development trend of
battery states estimation methods. Higher accuracy, real-time performance, and robustness are the
Citation: Zhao, D.; Li, H.; Zhou, F.;
development goals of power battery states estimation methods.
Zhong, Y.; Zhang, G.; Liu, Z.; Hou, J.
Research Progress on Data-Driven
Methods for Battery States
Keywords: battery states estimation method; data sources for battery state; machine learning model;
Estimation of Electric Buses. World deep learning model; hybrid model; data-driven method; electric bus
Electr. Veh. J. 2023, 14, 145. https://
doi.org/10.3390/wevj14060145

Academic Editor: Michael Fowler

1. Introduction
Received: 12 May 2023 The development of automobile electrification has effectively alleviated the oil energy
Revised: 23 May 2023 crisis and environmental pollution problems caused by fuel vehicles. Under the dual-
Accepted: 25 May 2023 wheel drive of market demand and policy support, the application scale of electric bus has
Published: 2 June 2023
rapidly increased. As the core energy storage component of electric vehicles, the states of a
power battery affect the driving safety performance of the vehicle [1]. Abnormal battery
states often lead to vehicle fire safety incidents. Due to the long mileage of the driving
Copyright: © 2023 by the authors.
route, electric buses need to be equipped with more power battery modules to meet daily
Licensee MDPI, Basel, Switzerland.
operational needs. At the same time, as electric buses transport more passengers, higher
This article is an open access article safety for power batteries and the entire vehicle is required. Battery states refer to the
distributed under the terms and working state of a battery during its service, mainly including state of charge (SOC) [2],
conditions of the Creative Commons state of health (SOH) [3], remaining useful life (RUL) [4], state of power (SOP) [5], state of
Attribution (CC BY) license (https:// energy (SOE) [6], state of safety SOS [7], etc. Accurate estimation of battery states, early
creativecommons.org/licenses/by/ detection of abnormalities, and timely warning and disposal are of great significance for
4.0/).

World Electr. Veh. J. 2023, 14, 145. https://doi.org/10.3390/wevj14060145 https://www.mdpi.com/journal/wevj

 World Electr. Veh. J. 2023, 14, x FOR PEER REVIEW 2 of 23
World Electr. Veh. J. 2023, 14, 145 2 of 22

ensuring the safe and reliable use of battery, thus prolonging service life and enhancing
ensuring the safe and
aging performance reliable and
of battery, use ensuring
of battery,the thus
safeprolonging service
use of electric life and enhancing
buses.
agingThe performance of battery, and ensuring the safe use of electric buses.
main goal of battery management is to estimate the power battery states accu-
The
rately, main goalguidance
providing of battery management
for is to estimate
safe use, maintenance, the power battery
replacement, states accurately,
and retirement of bat-
providing
teries. To achieve this goal, researchers have conducted a large amount ofofeffective
guidance for safe use, maintenance, replacement, and retirement batteries.re-To
achieve this goal, researchers have conducted a large amount of
search on the battery states estimation methods and arrived at fruitful achievements.effective research on the
battery states estimation
Researchers methodsdifferent
have designed and arrived at fruitful
workflows achievements.
to estimate the battery states [8,9].
Researchers
The basic workflow have designed
usually includesdifferent workflows
four steps: to estimate
data collection the battery states
and preprocessing, [8,9].
feature
The basic workflow usually includes four steps: data collection and
engineering, battery model construction, and application, as shown in Figure 1. Data ac- preprocessing, feature
engineering,
quisition andbattery model construction,
preprocessing is the first stepand application,
to focus as shown
on obtaining in Figure
the battery 1. data
states Data
acquisition and preprocessing is the first step to focus on obtaining
with the data acquisition system, and performing data preprocessing such as data clean- the battery states data
with the data acquisition system, and performing data preprocessing such
ing, data filtering, and regularization. The second step is feature engineering, which is to as data cleaning,
data
selectfiltering, and the
and extract regularization.
features of the The second step
collected data,isreduce
feature engineering,
the whichextract
data dimension, is to select
the
and extract the features of the collected data, reduce the data dimension,
data features closely related to the battery states, and avoid excessive data redundancy. extract the data
features
The main closely
methodsrelated to the
include batterycomponent
principal states, andanalysis,
avoid excessive
correlationdata redundancy.
coefficient The
analysis,
main methods include principal component analysis, correlation coefficient
and cosine similarity analysis, etc. The third step is to construct an estimation model, analysis, and
cosine
which is the core of the workflow, and to establish a mapping relationship between inputis
similarity analysis, etc. The third step is to construct an estimation model, which
the
datacore
andofoutput
the workflow,
data. Theand to establish
commonly usedamodels
mapping relationship
include machinebetween
learninginput
models,data and
deep
output data. The commonly used models include machine learning models,
learning models, and hybrid models. The final step is to apply the results of battery states deep learning
models,
estimation,and such
hybridas models.
abnormal The finalwarning,
states step is to triggering
apply the results of battery states
active intervention andestimation,
disposal
such as abnormal
procedures. states warning, triggering active intervention and disposal procedures.

Data sampling & preprocessing Feature engineering Model construction Application

 Data sampling  Feature reduction  Machine learning model  Evaluation result

Time  Feature selection ANN SOH
Voltage\Current  Correlation analysis SVM SOC
Temperature  Principal component GPR RUL
Insulation resistance analysis  Deep learning model SOT
Combustible gas CNN SOE
Pressure RNN et.al
CAN data LSTM  Application
 Data preprocessing GRU early warning
Cleaning  Hybrid model Intervention control
Filtering management
Normalizating

Figure1.
Figure 1. The
The basic
basic workflow
workflow of
of the
the battery
battery states
states estimation.
estimation.

While estimating
While estimatingthe thebattery
batterystates,
states,different
differentdata
datasources
sourceshave
havea asignificant
significant impact
impact on
on the
the estimation
estimation accuracy,
accuracy, and and high-qualitydata
high-quality dataare
areone
oneofofthe
the key
key factors
factors for
forconducting
conducting
data-driven battery
data-driven battery states
states estimation.
estimation. There
There are
are three
threecommon
commontypestypesof ofdata
datasources:
sources:test
test
data, simulation data, and natural driving data. Specific data includes
data, simulation data, and natural driving data. Specific data includes voltage, current, voltage, current,
tem-
temperature
perature [10–12],
[10–12], and and
CAN CANdatadata
[13].[13].
DengDeng et [10]
et al. al. [10] have
have estimated
estimated thethe battery
battery states
states by
by comprehensively capturing time series characteristic data of voltage
comprehensively capturing time series characteristic data of voltage and current. With and current. Withthe
the development
development of vehicle
of vehicle sensorsensor technology,
technology, 5G communication,
5G communication, and vehicle
and vehicle network-
networking tech-
ing technology,
nology, a massivea amount
massive of amount
batteryofstates
battery states
data datanatural
under under driving
natural conditions
driving conditions
has been
has beenand
recorded recorded
stored,and stored, important
providing providing input
important input for data-driven
for data-driven battery states battery states
estimation.
estimation.
Another key factor is to construct a battery states estimation model. The methods for
Another
building modelskeyarefactor is todivided
mainly construct a battery
into states estimation
two categories: model.
analytical The methods
model-based for
methods
building models are mainly divided into two categories: analytical
and data-driven model-based methods. The methods based on an analytical model mainly model-based methods
and data-driven
include model-based
electrochemical modelsmethods. The methods
and equivalent circuit based
models.onThe
an analytical model model
electrochemical mainlyis
include electrochemical models and equivalent circuit models. The electrochemical
based on the internal structure of the battery and simulates the complex chemical reaction model
is based on the
mechanism internal
inside structureThe
the battery. of the battery and
equivalent simulates
circuit modelthecharacterizes
complex chemical reac-
the battery
tion
by mechanism
analyzing its inside the characteristics
electrical battery. The equivalent circuit model
during working characterizes
and simulating the battery
characteristics
by analyzing
using its electrical It
circuit components. characteristics during working
has the characteristics andmodeling
of simple simulating characteristics
structure and low
using circuit components. It has the characteristics of simple modeling
computational cost [14,15]. The key models include the Rint model, Thevenin model, PNGV structure and low
model, GNL model, etc. [16]. The problem of the above method is that detailed physical
World Electr. Veh. J. 2023, 14, 145 3 of 22

structure parameters of the model need to be obtained, as well as the high nonlinearity and
strong coupling characteristics between each parameter, which result in low accuracy of the
estimation and difficulty in further significant improvement. Data-driven model is to treat
the battery as a black box, analyze hidden information and evolution rules from the external
feature parameters of the battery, and estimate the battery states by mining the hidden
feature information based on a large amount of dataset. Data-driven models involve
simulating battery states by end-to-end training of data, which has the characteristics
of simple modeling process, high estimation accuracy, and strong generalization ability.
However, this model needs a large amount of data for training [17]. In recent years, with
the development of internet of vehicles technology, a large amount of battery data has been
recorded and stored, laying the data foundation for data-driven models. These methods
are increasingly applied in power battery state models. This paper focuses on the research
progress of data-driven models.
Data-driven models include classical machine learning models, deep learning mod-
els, and hybrid models. In early battery states estimation research, classical machine
learning models are mainly used, and common models include artificial neural networks
(ANN) [18,19], support vector machine (SVM) [20,21], and Gaussian process regression
(GPR) [22,23], hidden Markov model (HMM) [24,25], random forest (RF) [26,27], fuzzy
control [28,29], autoregressive(AR) [30,31], relevance vector machine (RVM) [32,33], etc.
Although classic machine learning models can estimate battery states based on a small
number of data samples, the estimation quality relies on expert experience to manually
extract features, and the estimation accuracy is greatly affected by the selected features.
With the development of neural network technology, to further improve the accuracy
and robustness of battery states estimation, some scholars have begun to explore the use of
deep learning methods, such as convolutional neural networks (CNN) [34–36], recurrent
neural networks (RNN) [37,38], and other models. The deep learning models achieve high-
level abstract representation and modeling of data by constructing a depth map composed
of multiple processing layers and nonlinear and linear transformations. Compared with
the machine learning methods, deep learning methods can automatically extract features of
different depths from massive data, achieving end-to-end learning. They are not sensitive
to data noise, easy to understand, and they have good portability. When the data sample
size is sufficient, it can achieve higher estimation accuracy. However, the shortcomings of
deep learning models are the need for larger data samples and more training time to train
the model.
To further improve the accuracy and robustness of models and reduce training time,
some scholars have attempted to comprehensively utilize the characteristics of different
models to build hybrid models for battery states estimation [39–41]. For example, Song
used the feature extraction capability of CNN and the time series prediction capability of
RNN to try to build a hybrid model CNN-LSTM to estimate the SOC state of batteries,
extracted advanced spatial features from original data through CNN, and captured the
nonlinear relationship between SOC and measurable data such as current, voltage, and
temperature through LSTM. It has better tracking performance than the single model of
LSTM or CNN. The maximum average error of SOC estimation is less than 1.5%, and the
maximum root mean square error is less than 2%. The hybrid model has good application
prospects in the field of health estimation of batteries.
Researchers have summarized the research progress of power battery states estimation
technology. For example, Toughzaoui et al. [42] summarized the research status of battery
health status estimation and remaining life estimation, and Manoharan et al. [43] summa-
rized battery states estimation technology based on traditional machine learning models.
By analyzing the existing review literature, it was found that the existing literature mainly
analyzes some battery states based on machine learning models, and there is no analysis of
battery states estimation technology based on deep learning models and hybrid models.
The main purpose of this paper is to analyze the latest achievements in data-driven power
battery states estimation recently, summarize the main data sources and characteristics
World Electr. Veh. J. 2023, 14, x FOR PEER REVIEW 4 of 23

World Electr. Veh. J. 2023, 14, 145 hybrid models. The main purpose of this paper is to analyze the latest achievements in
4 of 22
data-driven power battery states estimation recently, summarize the main data sources
and characteristics of battery states, and compare and analyze the mainstream data-driven
models and their advantages and disadvantages. In addition, some other content was
of battery states, and compare and analyze the mainstream data-driven models and their
summarized, and the basic process for estimating the health status of power batteries was
advantages and disadvantages. In addition, some other content was summarized, and the
proposed. The future technological development trends of data sources and data-driven
basic process for estimating the health status of power batteries was proposed. The future
models for estimating the battery states were discussed.
technological development trends of data sources and data-driven models for estimating
The main contributions of this paper include: (1) Analyzing the main data sources
the battery states were discussed.
and their characteristics used to estimate battery states, providing guidance for subse-
The main contributions of this paper include: (1) Analyzing the main data sources and
quent data collection and application of power batteries; (2) Summarizing the construction
their characteristics used to estimate battery states, providing guidance for subsequent data
methods of
collection anddata-driven
applicationbattery states
of power estimation
batteries; models, providing
(2) Summarizing support for
the construction further
methods
research
of on model
data-driven construction
battery methods;models,
states estimation (3) Analyzing
providing the development trend of
support for further battery
research
states estimation technology to provide reference for future research on
on model construction methods; (3) Analyzing the development trend of battery states estimation meth-
ods.
estimation technology to provide reference for future research on estimation methods.
The structure
The structure of
of this
this paper
paper isisas
asfollows.
follows. The
The second
second part
part analyzes
analyzes thethe data
data sources
sources
and their characteristics. The third part discusses the model and characteristics
and their characteristics. The third part discusses the model and characteristics of data- of data-
drivenstatus
driven statusestimation
estimationofofpower
power batteries.
batteries. Finally,
Finally, thethe development
development trendtrend is analyzed
is analyzed and
and conclusions are presented.
conclusions are presented.

2. Data
2. Data for
for Battery
Battery States
States Estimation
Estimation
There are many data
There datasources
sourcesused
usedtotoestimate thethe
estimate battery states.
battery According
states. to the
According todata
the
acquisition
data method,
acquisition the main
method, data data
the main sources are divided
sources into three
are divided categories:
into three test data,
categories: sim-
test data,
ulation data,
simulation andand
data, natural driving
natural data,
driving as shown
data, in Figure
as shown 2. 2.
in Figure

Data for battery states estimation

Test data Simulation data Natural driving data

 Electricity related  Electrical performance  Battery related

Voltage simulation Voltage
Electric current  Temperature field Current
Resistance simulation Resistance
Capacitance  Battery state simulation  Vehicle related
 Temperature related  Others Velocity
 Others Mileage traveled
Gas composition  Enviroment related
pressure Temperature

Figure2.
Figure 2. Data
Data for
for battery
battery states
states estimation.
estimation.

2.1.
2.1. Test
Test Data
Data
Some
Some researchers
researchers use
use experimental
experimental data data toto estimate
estimate the
the states
states of
of batteries
batteries [44,45],
[44,45],
which can be further divided into the following categories: electrical data such
which can be further divided into the following categories: electrical data such as voltage, as voltage,
current,
current,resistance,
resistance,capacitance,
capacitance,SOC,
SOC,etc.;
etc.;heat
heatdata
datasuch
such as
astemperature;
temperature;otherotherdata
datasuch
such
as
as gas composition, pressure inside the battery box, etc. Some of these data can be directly
gas composition, pressure inside the battery box, etc. Some of these data can be directly
measured
measuredthrough
throughtesting
testingequipment,
equipment, such
suchas as
temperature,
temperature,which
which cancan
be directly measured
be directly meas-
through single point, multi-point, infrared imaging or ultrasound method.
ured through single point, multi-point, infrared imaging or ultrasound method. Some Some data
need to be obtained through indirect estimation methods, such as the resistance
data need to be obtained through indirect estimation methods, such as the resistance of of power
batteries. The characteristic
power batteries. of this type
The characteristic of data
of this typeis of
authenticity and objectivity,
data is authenticity but it requires
and objectivity, but
the use of professional equipment for collection.
it requires the use of professional equipment for collection.
2.2. Simulation Data
2.2. Simulation Data
Some researchers use simulation data to estimate the battery states [46,47]. For ex-
Some researchers use simulation data to estimate the battery states [46,47]. For exam-
ample, Sakile simulated the battery model in the MATLAB/Simulink environment and
ple, Sakile simulated the battery model in the MATLAB/Simulink environment and used
used simulation data to predict the SOC and RUL of the battery. The use of simulation to
collect data under various operating conditions, working environments, and meteorological
conditions has strong flexibility, but the main drawback is that the ability of simulation to
World Electr. Veh. J. 2023, 14, 145 5 of 22

reproduce and capture real operating conditions, working environments, and meteorologi-
cal conditions is limited, and simulation data may not be consistent with real operating
conditions data.

2.3. Natural Driving Data

With the development of on-board sensors, internet of vehicles, cloud platforms,
and other technologies, a large amount of natural driving data is stored and recorded.
Natural driving data have the characteristics of easy access to data, objective and realistic
status, large amount of data, and rich information. These data not only include battery
states data, but also include vehicle actual scene data and environmental meteorological
data [48]. Wang used real driving data from two electric buses to predict the battery
temperature during the charging phase of an electric bus [49]. With the development of
artificial intelligence and big data technology, natural driving data may become the main
source of data for estimating the status of batteries in engineering practice.
In summary, the three types of data sources used to estimate the battery states have
different characteristics, and the advantages and disadvantages of each type of data source
are analyzed in Table 1.

Table 1. The characteristics of data source.

Data Source Strengths Weakness

Special testing equipment is
required, and the testing process
The data are objective and requires a large amount of
Test Data truthful, and can collect extreme manpower and material
working condition data resources, with a long acquisition
cycle and high requirements for
the testing environment
Easy and convenient to obtain,
Data quality is greatly affected by
Simulation Data with good repeatability and no
model accuracy
environmental constraints
The data are real and objective, High data interference noise
Natural Driving Data easy to collect and cover a wide makes it difficult to obtain
range of scene conditions extreme scenario data

3. Data-Driven Model for Battery States Estimation

According to the characteristics of the model, the data-driven battery states estimation
World Electr. Veh. J. 2023, 14, x FOR PEER REVIEW 6 of 23
models can be divided into three categories: machine learning models, deep learning
models, and hybrid models, as shown in Figure 3.

Battery states estimation methods

Machine learning model Deep learning model Hybrid model

ANN CNN CNN-RNN

SVM RNN CNN-LSTM
GPR LSTM CNN-GRU
HMM GRU CNN-LSTM-GRU
Random Forest SRNN CNN-LSTM-DNN
Fuzzy control Bi-RNN IPSO-CNN-ILSTM
Auto regressive GNN Others
RVM LeNet
Others Others

Figure 3. Classification
Figure 3. Classification of
of data-driven
data-driven battery
battery states
states estimation
estimation models.
models.

3.1. Machine Learning Model

Machine learning refers to learning general rules from limited observation data and
utilizing these rules to predict and analyze unknown data. The model needs to first rep-
 GPR LSTM CNN-GRU
HMM GRU CNN-LSTM-GRU
Random Forest SRNN CNN-LSTM-DNN
Fuzzy control Bi-RNN IPSO-CNN-ILSTM
Auto regressive GNN Others
RVM LeNet
Others Others
World Electr. Veh. J. 2023, 14, 145 6 of 22
Figure 3. Classification of data-driven battery states estimation models.

3.1.
3.1.Machine
MachineLearning
LearningModel
Model
Machine
Machine learningrefers
learning referstotolearning
learninggeneral
generalrules
rulesfrom
fromlimited
limitedobservation
observationdata
dataand
and
utilizing
utilizing these rules to predict and analyze unknown data. The model needs torep-
these rules to predict and analyze unknown data. The model needs to first first
resent the data
represent as a set
the data as of features
a set and then
of features andinput
then these
inputfeatures into theinto
these features prediction classi-
the prediction
fier to predict
classifier the output
to predict results.results.
the output Its feature representation
Its feature mainlymainly
representation relies on manual
relies expe-
on manual
rience or feature
experience transformation
or feature methods
transformation for extraction,
methods and theand
for extraction, extracted featuresfeatures
the extracted have a
significant impact on
have a significant the recognition
impact accuracy
on the recognition of the model.
accuracy This paper
of the model. focusesfocuses
This paper on ana-
on
lyzing the progress
analyzing of estimating
the progress the battery
of estimating states
the battery using
states ANN,
using ANN,SVM, andand
SVM, GPR models.
GPR models.
(1)
(1) ANN
ANNmodel
model
ANN
ANN isan
is an information
information processing systemestablished
processing system establishedbased
basedononimitating
imitating the
the struc-
structure
ture
andand function
function of brain
of brain neural
neural networks.
networks. Artificial
Artificial neural
neural networks
networks have
have self-learning,
self-learning, self-
self-organizing, adaptive,
organizing, adaptive, andand strong
strong nonlinear
nonlinear function
function approximation
approximation capabilities,
capabilities, andand
has
has strong fault tolerance, which is suitable for complex nonlinear modeling
strong fault tolerance, which is suitable for complex nonlinear modeling problems with problems
with multiple
multiple related
related features.
features. The The
basicbasic structure
structure of ANN
of ANN includes
includes threethree layers:
layers: inputinput
layer,
layer, output layer, and hidden layer, as shown
output layer, and hidden layer, as shown in Figure 4.in Figure 4.

Input layer Hidden layer output layer

Figure
Figure4.4.The
Thebasic
basicstructure
structureofofan
anANN.
ANN.

ANNmodels
ANN modelshave
havesignificant
significantadvantages
advantagesininestimation
estimationaccuracy
accuracyand
androbustness,
robustness,and and
they have been widely applied in the field of battery states estimation. Researchers
they have been widely applied in the field of battery states estimation. Researchers have have
conductedextensive
conducted extensiveresearch
researchon onbuilding
buildingANN
ANNmodels
modelstotoestimate
estimatebattery
batterystates
states[50–55].
[50–55].
Forexample,
For example,Wang
Wangetetal.
al.[50]
[50]used
usedan anANN
ANNmodel
modeltotoestimate
estimatethe
thetemperature
temperaturechanges
changesofof
thelithium-ion
the lithium-ion batteries. They
They established
establishedtemperature
temperatureprediction
predictionmodels
modelsusing backprop-
using back-
agation neural
propagation networks
neural networks(BP-NN),
(BP-NN),radial basis
radial function
basis neural
function networks
neural (RBF-NN),
networks (RBF-NN), and
Elman neural networks (Elman-NN), respectively, and compared the
and Elman neural networks (Elman-NN), respectively, and compared the temperaturetemperature predic-
tion performance
prediction of different
performance neural
of different network
neural modeling
network techniques.
modeling The MSE
techniques. and MAE
The MSE and
values did not exceed 0.3. At the same time, it is noted that the Elman-NN model has good
adaptability and generalization ability, and has fast convergence speed. Bezha et al. [51]
proposed a method for estimating the internal impedance parameters of lithium-ion bat-
teries based on ANN, which achieves accurate estimation of the actual state of the battery
within 30 s, with a maximum error of less than 3%. The model has good universality.
Hussein et al. [52] used the ANN model to estimate the SOC of electric vehicle power
batteries, and the error was less than 3%. Jaliliantabar et al. [54] constructed an ANN model
to predict the SOT of lithium-ion batteries with the mean absolute percentage error (MAPE)
being about 0.331.
Overall, the ANN model has good performance in the process of battery state predic-
tion, the calculation process is fast and convenient, and the prediction result is relatively
accurate, suitable for all kinds of battery. ANN also has some weaknesses, such as the model
estimation accuracy being greatly affected by the training sample data, the large amount
of data is an important prerequisite for obtaining accurate estimates, and the prediction
ability of small sample data is poor. The parameters of ANN are complex, and it is easy
to fall into the local optimization of parameters in during training, resulting in overfitting.
 Overall, the ANN model has good performance in the process of battery state predic-
tion, the calculation process is fast and convenient, and the prediction result is relatively
accurate, suitable for all kinds of battery. ANN also has some weaknesses, such as the
model estimation accuracy being greatly affected by the training sample data, the large
amount of data is an important prerequisite for obtaining accurate estimates, and the pre-
World Electr. Veh. J. 2023, 14, 145 7 of 22
diction ability of small sample data is poor. The parameters of ANN are complex, and it
is easy to fall into the local optimization of parameters in during training, resulting in
overfitting. Due to the lack of clear methods for selecting network structures, the prior
Due to the lack
experience of clear methods
or comparison for selecting
of multiple modelsnetwork structures,
is required the prior
to determine the experience
final networkor
comparison
structure. of multiple models is required to determine the final network structure.
(2)
(2) SVM model
The SVM
The SVM model
model is is aa new
newlearning
learning machine
machine that
that maps
maps nonlinear
nonlinear functions
functions based
based onon
statistical learning theory. It maps nonlinear problems in low-dimensional space
statistical learning theory. It maps nonlinear problems in low-dimensional space to linear to linear
problems in
problems in high-dimensional space spacethrough
throughkernel
kernelfunction
functiontotocomplete
completethe modeling
the modeling of
complex
of complex nonlinear
nonlinearsystems,
systems,andandfind
findan
anappropriate
appropriate hyperplane
hyperplane to complete
complete accurate
accurate
classification of
classification of data.
data. Its
Its principle
principle is
is shown
shown in
in Figure
Figure5.5.

Low-dimensional space

High-dimensional space

Figure 5.
Figure 5. The
The principle
principle of
of SVM.
SVM.

Unlike ANN,
Unlike ANN, thethe SVM
SVM method
method has
has stricter
stricter mathematical
mathematical proof,
proof, lower
lower computational
computational
complexity, faster
complexity, faster convergence
convergence speed,
speed,and
andcancaneffectively
effectivelyprevent
preventlocal parameter
local parameteroptimi-
opti-
mization problems. SVM is not sensitive to the dimensions and
zation problems. SVM is not sensitive to the dimensions and variability variability of data, and is
is
suitable
suitablefor
forclassification
classificationand
andregression of of
regression complex
complexsmall sample
small data.
sample Additionally,
data. it hasit
Additionally,
strong generalization
has strong ability
generalization and high
ability estimation
and high accuracy
estimation [56]. Some
accuracy researchers
[56]. Some use SVM
researchers use
to estimate the states of battery, for example, Deng et al. [57] applied SVM to diagnose the
fault state of electric vehicle power batteries, with an accuracy of over 89%; Wang et al. [58]
used an SVM method to model the nonlinear dynamic characteristics of batteries based
on a small number of experimental data samples, with an estimated maximum relative
error of 3.61%; Chen et al. [59] constructed an SVM model to predict SOH online using
charging data, achieving an error of less than 2%; Li et al. [60] proposed a method to
indirectly estimate the RUL by using SVM model. Compared with ANN methods, the SVM
model has higher accuracy and shorter computational time, with a maximum error of 5%
in battery states estimation. Some researchers have also used a combination of SVM and
other methods to estimate battery states [61–65].
However, the SVM model still has some shortcomings in practical applications, for
instance, feature vector is difficult to measure and calculate, kernel parameters are difficult
to select, the model is highly dependent on cross-training and regularization methods and
sensitive to missing data during feature vector selection or training process.
(3) GPR model
The GPR model is a universal and resolvable non-parametric probability model that
uses a priori of Gaussian processes to conduct regression analysis on data. In theory, it
World Electr. Veh. J. 2023, 14, 145 8 of 22

can achieve universal approximation of any continuous function in compact space and has
been applied in the fields of time series analysis, image processing, and automatic control.
GPR has the advantage of low computational complexity in solving high-complexity
problems. Many researchers have constructed GPR battery states estimation models to
solve complex nonlinear problems during the process of electric state changes [22,23,66–77].
Wang et al. [22] established a battery states estimation model based on GPR, which has
prominent accuracy and robustness, with a maximum relative error within 2%. Liu et al. [23]
established a data-driven GPR model to predict the battery SOH, and achieved high
accuracy under the premise of small sample size input. Except for some individual point
estimation errors greater than 3%, most of them are less than 1.5%. Zhou et al. [66] designed
a cyclic GPR model with delayed feedback loop to estimate the status of batteries, which has
high accuracy and robustness, with an estimation error of 1.12%. Yang et al. [67] proposed
a GPR model based on the charging curve to estimate SOH. The model has good robustness
and reliability, and the estimation error of SOH is mostly less than 2%. Wang et al. [68]
proposed a data-driven integrated Gaussian process regression (GPR) model to estimate
SOH by comparing and analyzing the influence of different mean and kernel functions
on the estimation accuracy of GPR model, achieving mean absolute error (MAE) and root
mean square error (RMSE) of 1.7% and 2.41%, respectively. Pang et al. [69] proposed a
GPR model for estimating battery RUL, which has high estimation accuracy and achieves
battery RMSE less than 0.04.
In summary, GPR has the advantages of high model prediction accuracy and prob-
ability density prediction results. There are two main shortcomings. Firstly, due to the
inherent structure of the GPR model, the computational complexity is high when analyzing
large amounts of data; Secondly, the GPR model has more hyper-parameters, and the
hyper-parameter adjustment process is tedious during training.
In addition, some scholars have utilized other machine learning models, such as
particle filtering [78], Wavelet [79], Extra tree [80], Gradient boosting method [81], Linear
compression [82], KNN [83], etc., to estimate battery states and have achieved certain
results. A summary of machine learning models and their corresponding advantages and
disadvantages is shown in Table 2.

Table 2. Application and characteristic of machine learning model.

Battery States Application Strength Weakness

SOH [19–23,59,64,66– Used for electric health Used to estimate the battery Relying on manual experience or
68,71,73,76,78,79,81,82] state estimation states, learn general rules feature transformation methods to
from limited data, and predict extract features has a significant
Used for battery state of unknown data. Necessary to impact on the accuracy of model
SOC [18,52,55,72,74,75,77]
charge estimation represent the data as a set of estimation. The input and output
features and input them into dimensions of the model cannot
RUL [53,60,61,69,70,76,80,83] Used to estimate the the classifier to predict the be arbitrarily changed, and the
remaining battery life output results. It has certain sample length is fixed. Difficult to
Used to estimate other advantages in terms of handle ordinal numbers. When
Others accuracy and robustness the data sample size is large,
[50,51,54,56–58,62,63,65] status such as SOT, SOP,
SOS, etc. of the battery of estimation. overfitting may easily occur.

3.2. Deep Learning Model

The concept of deep learning originates from the research of artificial neural networks,
which use a processing mechanism of combining multiple hidden layers to stack and
processing output layer by layer, to transform the low-level feature representations that
is not closely related to the initial and target into higher-order abstract features that are
more closely related to the target, in order to discover distributed feature representations of
data. In recent years, this method has gradually been applied to the estimation of battery
states and has achieved good results in mapping battery states data to typical state specific
 3.2. Deep Learning Model
The concept of deep learning originates from the research of artificial neural net-
works, which use a processing mechanism of combining multiple hidden layers to stack
and processing output layer by layer, to transform the low-level feature representations
that is not closely related to the initial and target into higher-order abstract features that
World Electr. Veh. J. 2023, 14, 145 9 of 22
are more closely related to the target, in order to discover distributed feature representa-
tions of data. In recent years, this method has gradually been applied to the estimation of
battery states and has achieved good results in mapping battery states data to typical state
features
specific to estimate
features to battery states.
estimate Thestates.
battery common Thedeep learning
common deepmodels include
learning CNN,include
models RNN,
MLP [84,85], etc. This paper focuses on analyzing the state of the research on constructing
CNN, RNN, MLP [84,85], etc. This paper focuses on analyzing the state of the research on
battery state models
constructing battery using CNN and
state models RNN.
using CNN and RNN.
(1)
(1) CNN model
The CNN
The CNN model
model is is aa deep feed-forward
feed-forward neural
neural network
network model
model including
including convolution
convolution
operation inspired
operation inspiredby by the
the mechanism
mechanism of of biological
biological receptive
receptive field.
field. It is composed of multi-
layer networks,
layer networks, including
including input
input layer,
layer, convolution
convolution layer, layer, pooling
pooling layer,
layer, fully
fully connected
connected
layer,
layer, and
and output
output layer.
layer. The
The basic
basic structure
structure isis shown
shown in in Figure
Figure 6.6. Compared
Compared with with fully
fully
connected
connectedneural
neuralnetworks,
networks,CNN CNN automatically
automatically extracts
extractsthe the
salient features
salient of feature
features data
of feature
by changing the fully connected layer to the convolution layer and
data by changing the fully connected layer to the convolution layer and pooling layer, pooling layer, using
multiple convolution
using multiple poolingpooling
convolution operations, and thenand
operations, usesthen
mathematical statistics methods
uses mathematical statistics
or classifiers
methods to output to
or classifiers after the full
output afterconnection layer to complete
the full connection layer to the nonlinear
complete mapping
the nonlinear
from inputfrom
mapping to output.
input The CNN model
to output. The CNN utilizes
modela network
utilizesstructure
a network of sparse
structureconnections
of sparse
and parameter
connections andsharing to reduce
parameter theto
sharing complexity
reduce theofcomplexity
the model, of significantly
the model,reducing the
significantly
number of network weights, and has advantages such as automatic
reducing the number of network weights, and has advantages such as automatic feature feature extraction,
anti-noise
extraction,interference, and end-to-end
anti-noise interference, and learning.
end-to-end It islearning.
widely used in machine
It is widely usedvision, state
in machine
diagnosis, and other fields.
vision, state diagnosis, and other fields.

Fully connected layer

Convolutional layer

Convolutional layer

Concatenate layer

Output layer
Polling layer

Polling layer
Input layer

...

Figure 6.
Figure 6. The
The basic
basic structure
structure of
of CNN.
CNN.

CNN
CNN models havehave been
been applied
appliedininestimating
estimatingbattery
battery states
states and
and achieved
achieved many
many re-
research results[34–36,86–90].
search results [34–36,86–90].For
Forexample,
example,WeiWeiet
etal.
al. [34]
[34] constructed a CNN model
model and
and
trained
trained the
the model
model based
based on
on the
the battery
battery common
common dataset
dataset toto predict
predict the
the remaining
remaining life
life of
of
lithium-ion batteries. The life prediction results are superior to those of other existing
methods. Qian et al. [35] designed a 1D-CNN model architecture to estimate the battery
SOC using random segments of the charging curve as inputs. The models have good
robustness and accuracy and can accurately estimate battery SOC. Lu et al. [36] proposed a
CNN model for battery SOC estimation, which was used to estimate battery SOC based on
partial voltage data during battery discharge. It can accurately estimate battery SOC with
limited voltage data, and the MAPE is about 0.55%. Chemali et al. [86] proposed using
a CNN model driven by partial charging data to estimate battery SOH with an average
error of less than 0.8%. Sohn et al. [88] constructed a CNN model to extract features that
can reflect the dynamic changes in battery performance and accurately predict the battery
SOC. Shen et al. [89] constructed a deep convolutional neural network (DCNN) that can
accurately estimate battery SOC by using measurement data during charging, with higher
accuracy and robustness.
Some researchers utilize the feature extraction ability of CNN and combine it with
other machine learning methods for power battery state estimation [91,92]. For example,
Yang et al. [91] estimated the battery health status by building a CNN and random forest
hybrid model, which improved the estimation accuracy and robustness compared with
the single CNN model. In addition, some researchers have constructed CNN models
with various forms for battery health status research by coordinating with some adjust-
World Electr. Veh. J. 2023, 14, 145 10 of 22

ments and improvements, such as LeNet [93], ALexNet [94], VGGNet [95], ResNet [96],
DenseNet [97], etc.
In summary, the CNN models have high recognition accuracy and good robustness
and have achieved good application in battery states estimation field. However, as a
feedforward neural network, from the input to the output of the network from the bottom
to the top unidirectional connection, CNNs have the disadvantages that the input samples
are independent of each other, the output dimension is relatively fixed, and the output
only depends on the current input. At present, CNNs tend to have smaller convolutional
kernels, deeper network structures, fewer pooling layers, and gradually develop towards
fully connected networks.
(2) RNN model
The RNN model is a new type of neural network that takes sequence data as input
and realizes short-term memory capability through recursive loop units in the evolution
direction of sequence. Compared with CNN, the neurons of RNN can not only accept
information from other neurons, but also their own information. Through the network
parameter feedback mechanism, the important information of the network can be retained
and updated for a certain period of time, which presents significant advantages in the
modeling of time series problems. RNN has a loop network structure, connecting all nodes
in a chain manner. The simplified model is shown in Figure 7. RNN has the characteristics
of memory ability, parameter sharing, etc. It can theoretically approximate any nonlinear
dynamic system. It has certain advantages in learning the nonlinear characteristics of
World Electr. Veh. J. 2023, 14, x FOR PEER REVIEW 11 of 23
sequences, and it has been widely used in speech recognition, machine translation, and
other tasks.

Output layer
ht

ht

Hidden layer Delayer

xt ht-1

Input layer

Figure 7.
Figure 7. The
The basic
basic architecture
architectureof
ofRNN.
RNN.

Some
Some scholars
scholars have
have utilized
utilized RNN’s
RNN’s strong
strong ability
ability to
to map
map high-dimensional
high-dimensional and and
strongly
strongly nonlinear data to estimate battery states by constructing RNN
nonlinear data to estimate battery states by constructing RNN models,
models, which
which
have
have achieved
achieved good
good results
results [8,38,39,98–100].
[8,38,39,98–100]. For For example,
example, Catelani
Catelani et
et al.
al. [8]
[8] used
used anan RNN
RNN
model
model to to estimate
estimate the
the RUL
RUL of of lithium-ion
lithium-ion batteries
batteries with
with good
good accuracy.
accuracy. Feng
Feng et et al.
al. [38]
[38]
attempted
attempted to to construct
construct an an RNN
RNN framework
framework for for estimating
estimating battery
battery SOC,
SOC, which
which showed
showed
good
good estimation
estimationperformance
performancewith withRMSERMSE ofof
less than
less 1.29%.
than Hsieh
1.29%. et al.
Hsieh et [39] predicted
al. [39] the
predicted
discharge state of batteries by building an RNN model framework,
the discharge state of batteries by building an RNN model framework, with an error rate with an error rate of
less than 2%.
of less than 2%.
The
The RNN
RNNmodelmodelcan canonly
onlylearn
learninformation
informationthat is is
that close in in
close time, making
time, making it difficult
it difficult to
apply to sequence data that require long-term dependence. When
to apply to sequence data that require long-term dependence. When the input sequence the input sequence is
relatively long,
is relatively there
long, willwill
there be long-range
be long-range dependence
dependence problems caused
problems by gradient
caused vanishing
by gradient van-
and gradient explosion during the training process of RNN networks.
ishing and gradient explosion during the training process of RNN networks. The The most effective
most
way to address
effective way tothis issuethis
address is to introduce
issue a gatingamechanism,
is to introduce which iswhich
gating mechanism, calledisthe LSTM
called the
unit and the GRU unit.
LSTM unit and the GRU unit.
The LSTM unit updates the memory information of the unit by introducing a special-
ized memory unit that controls the input, memory, and output of information through
input gate, forget gate, and output gate. The basic structure is shown in Figure 8. Some
researchers have used LSTM to estimate battery states [101–108]. Yang et al. [101] estab-
World Electr. Veh. J. 2023, 14, 145 11 of 22

The LSTM unit updates the memory information of the unit by introducing a special-
ized memory unit that controls the input, memory, and output of information through
input gate, forget gate, and output gate. The basic structure is shown in Figure 8. Some re-
searchers have used LSTM to estimate battery states [101–108]. Yang et al. [101] established
an LSTM model to predict the battery SOH with error of less than 3%, and the model has
better accuracy and stability. Zhang et al. [102] built an LSTM model, which can accurately
World Electr. Veh. J. 2023, 14, x FOR PEER REVIEW 12 of 23
estimate the SOC and RUL of lithium-ion batteries. Park et al. [103] proposed an LSTM
model to estimate battery RUL, the MAPE of the model reached 0.47–1.88%.

xt xt ht-1
ht-1

Input gate Output gate

xt
it ot
ct

ht-1 ht’

ft
Forget gate

xt ht-1

Thebasic
Figure8.8.The
Figure basicstructure
structureof
ofan
anLSTM
LSTMunit.
unit.

UnlikeLSTM,
Unlike LSTM,the the GRU
GRU units
units simplify
simplify gatesgates by merging
by merging inputinput
gates gates and forget
and forget gates
into update gates and solve the long-range dependency problem by setting a reset gatea to
gates into update gates and solve the long-range dependency problem by setting re-
set gate to control the balance between input and forget. The basic structure is shown in
control the balance between input and forget. The basic structure is shown in Figure 9.
Figure 9. Researchers have applied GRU to estimate battery states [109–114], for example,
Researchers have applied GRU to estimate battery states [109–114], for example, Yang et
Yang et al. [109] used a GRU model to estimate the battery SOC by using current, voltage,
al. [109] used a GRU model to estimate the battery SOC by using current, voltage, and
and temperature data, with a maximum root mean square error of 3.5%. The model has
temperature data, with a maximum root mean square error of 3.5%. The model has good
good robustness. Guo et al. [110] used the GRU model to predict the RUL of lithium batter-
robustness. Guo et al. [110] used the GRU model to predict the RUL of lithium batteries
ies with different charging strategies, which can provide accurate prediction results under
with different charging strategies, which can provide accurate prediction results under
different charging strategies. The root mean square error of prediction can be controlled
different charging strategies. The root mean square error of prediction can be controlled
within 1%, and the prediction response speed is very fast. Lyu et al. [111] estimated the
within 1%, and the prediction response speed is very fast. Lyu et al. [111] estimated the
battery SOC based on a GRU model, noting that GRU outperforms LSTM and RNN in
battery SOC based on a GRU model, noting that GRU outperforms LSTM and RNN in
network performance and estimation accuracy.
network performance and estimation accuracy.
Some researchers have further improved the performance of recurrent neural networks
by increasing their depth, and used them for studying battery states estimation, such as
xt ht-1
Stacked Recurrent Neural Network (SRNN) [115–118], Bidirectional Recurrent Neural
Network (Bi-RNN) [119–123], and Graph Neural Network (GNN) [124,125], and have
applied them in battery states estimation and achieved high estimation accuracy.
The above are only some typical application cases, and there are still many successful
Update gate
application
xt
cases of RNN in building a health estimation model for power batteries [126–128],
zt
which will not be repeated.
ht’
ht

ht-1

ft

Reset gate
 temperature data, with a maximum root mean square error of 3.5%. The model has good
robustness. Guo et al. [110] used the GRU model to predict the RUL of lithium batteries
with different charging strategies, which can provide accurate prediction results under
different charging strategies. The root mean square error of prediction can be controlled
within 1%, and the prediction response speed is very fast. Lyu et al. [111] estimated the
World Electr. Veh. J. 2023, 14, 145 battery SOC based on a GRU model, noting that GRU outperforms LSTM and RNN 12 ofin
22
network performance and estimation accuracy.

xt ht-1

Update gate
xt zt

ht’
ht

ht-1

ft

Reset gate

xt ht-1

Thebasic
Figure9.9.The
Figure basicstructure
structureof
ofaaGRU
GRUunit.
unit.

It can be seen that RNN and its improved models can be used for estimating the
battery states, suitable for high-dimensional and big data sample learning, and can extract
deep spatiotemporal features. The error of battery states estimation is relatively low.
However, CNN requires a large amount of sample data and faces difficulties in obtaining
sample data. In summary, the models based on deep learning have achieved significant
application effects in the battery states estimation. The application fields and corresponding
characteristics are summarized in Table 3.

Table 3. Application and characteristics of CNN model.

Battery States Application Strength Weakness

SOH [86,87,91,101,102,104– Used for electric health Suitable for high-dimensional big Due to the constraints of the
106,114,116,117,122,125] state estimation data samples, reducing the model structure, the input and
SOC [35,36,38,39,89,93,95,98, number of parameters through output dimensions cannot be
Used for battery state of
99,109,111,113,115,119] weight sharing and aggregation, arbitrarily changed, and the
charge estimation
[120,124,128] automatically extracting features sample length is required to
and integrating them with be fixed. It is difficult to
RUL Used to estimate the
classifiers to achieve end-to-end process time series data, and
[34,92,96,103,107,110,112,118] remaining battery life
learning, with high accuracy and overfitting occurs easily when
Others [88,90,94,97,100,108, Used to estimate other resistance to noise interference. the data sample size is large.
121,123,126,127] status of the battery

3.3. Hybrid Model

A hybrid model refers to a high-precision model constructed by comprehensively
utilizing the different characteristics of different models. In recent years, hybrid mod-
els have attracted the attention of researchers. Some of them have put forward the
comprehensive use of hybrid model to estimate the battery state and achieved good
results [39–41,48,129–142]. For example, Song et al. [39] tried to build a hybrid model
CNN-LSTM to estimate the battery SOC by using the feature extraction capability of CNN
and the time series prediction capability of RNN, extracted advanced spatial features from
the original data through CNN, captured the nonlinear relationship between SOC and
measurable data such as current, voltage and temperature through LSTM, and obtained
better performance than the LSTM or CNN single model. The maximum average error
and the maximum root mean square error of SOC estimation was less than 1.5% and 2%,
respectively. Xu et al. [40] proposed that the hybrid model CNN-LSTM was added jump
 the battery energy state (SOE) by building a hybrid model framework consisting of LSTM
and CNN, with an estimation error of less than 3%. Zhang et al. [133] proposed a hybrid
model IPSO-CNN-ILSTM for estimating the battery RUL status. The mapping association
between fusion features and RUL was established through CNN and improved LSTM,
World Electr. Veh. J. 2023, 14, 145 and the improved particle swarm optimization (IPSO) was used to optimize the weights 13 of 22
and learning factor parameters of the mapping network to achieve a battery life estimation
result of about 0.7% MAPE, the estimation result of MAPE is better 5.0% than the CNN-
LSTM model. Zraibi et al. [134] estimated the battery RUL by constructing a hybrid model
connection (as shown in Figure 10), which solved the problem of neural network degra-
CNN-LSTM-DNN and comprehensively using the advantages of CNN, LSTM, and DNN.
dation caused by multi-layer LSTM, not only significantly improved the SOH estimation
The error and robustness of the hybrid model estimation are better than that of the single
accuracy, but also reduced the calculation amount of the model. RMSE was below 0.004
model. Yanwen et al. [135] used CNN to extract health status characteristics, mined data
on both the NASA and Oxford battery data sets. Ren et al. [41] estimated the battery RUL
time series features with local features of long and short-term memory (LSTM), and com-
using a hybrid CNN-LSTM model, the accuracy and RMSE estimated based on the hybrid
bined with a GRU cell to build a hybrid model, which significantly improved the accuracy
model were 94.97% and 5.03%, respectively, which were much better than the SVM model
of battery SOH estimation. In summary, hybrid models can selectively select models ac-
(corresponding
cording to targetresults were 81.77%
requirements and
and data 18.23%, and
resources respectively), and the
design learning modelofhas
networks good
differ-
generalization ability and robustness.
ent depths and widths, with higher accuracy and stronger robustness.

Input Data LSTM LSTM Output Dense

layer 1 LSTM LSTM ... LSTM

LSTM
Pooling layer 2 LSTM LSTM ... LSTM

... ... ...

...
...
Concatenate LSTM Output LSTM Output SOH

...

...
LSTM
layer 3 LSTM LSTM ... LSTM

CNN Layer LSTM Layer Output Layer

Figure 10. Classical hybrid model.
Figure 10. Classical hybrid model.
Chen et al. [129] proposed a hybrid model ELM-BSASVM composed of an extreme
learning
All machine and an SVM
in all, data-driven model
status to predict
estimation the battery
technology RUL. The
for power hybrid
batteries model has
is relatively
good robustness, and the RMSE value is 79.68% better than the SVM
mature at present, and different methods have their own advantages and applicable model. Zhao et al.sce-
[130]
and Liu et al. [131] explored a hybrid CNN and GRU model for battery
narios. Traditional machine learning models typically require extracting statistical fea- SOH estimation
and
turesvalidated
from data thebased
model
onby reconstructing
expert experience,feature
and thenseries
usingsamples on the
classifiers for Oxford battery
battery states
dataset. The average values of RMSE and MAE reached 0.582% and
estimation. The accuracy is greatly affected by manually selecting features. The deep 0.524%, respectively.
Wang et al.
learning [48] proposed
models have the acharacteristics
hybrid CNN of and LSTM model
automatically for predicting
extracting battery
distributed temper-
features
ature
from during
massivethe charging
data phaseend-to-end
and enabling of electric learning
buses, which was
of data. used
Their to predict in
applications short-term
battery
temperature changes
states estimation of batteries
methods in the future.
are becoming The model
increasingly popular,notwith
onlyhigh
has accuracy
excellent and
accu-
racy
greatand robustness,
robustness. Thebut alsomodel
hybrid reduces time and
consists space costs.
of multiple Meimodels,
different et al. [132] estimated
which can fullythe
battery energy
utilize the state (SOE)
advantages by building
of different models a hybrid model
to further framework
enhance consisting
generalization of LSTM
ability and
and CNN, with
recognition an estimation
robustness, and it error of less
has good than 3%.
research andZhang et al. [133]
application proposed
prospects. a hybrid
The perfor-
model IPSO-CNN-ILSTM for estimating the battery RUL status. The
mance characteristics of machine learning models, deep learning models, and hybrid mapping association
between fusion features and RUL was established through CNN and improved LSTM,
and the improved particle swarm optimization (IPSO) was used to optimize the weights
and learning factor parameters of the mapping network to achieve a battery life estima-
tion result of about 0.7% MAPE, the estimation result of MAPE is better 5.0% than the
CNN-LSTM model. Zraibi et al. [134] estimated the battery RUL by constructing a hybrid
model CNN-LSTM-DNN and comprehensively using the advantages of CNN, LSTM, and
DNN. The error and robustness of the hybrid model estimation are better than that of
the single model. Yanwen et al. [135] used CNN to extract health status characteristics,
mined data time series features with local features of long and short-term memory (LSTM),
and combined with a GRU cell to build a hybrid model, which significantly improved the
accuracy of battery SOH estimation. In summary, hybrid models can selectively select
models according to target requirements and data resources and design learning networks
of different depths and widths, with higher accuracy and stronger robustness.
All in all, data-driven status estimation technology for power batteries is relatively ma-
ture at present, and different methods have their own advantages and applicable scenarios.
Traditional machine learning models typically require extracting statistical features from
data based on expert experience, and then using classifiers for battery states estimation.
The accuracy is greatly affected by manually selecting features. The deep learning models
have the characteristics of automatically extracting distributed features from massive data
World Electr. Veh. J. 2023, 14, 145 14 of 22

and enabling end-to-end learning of data. Their applications in battery states estimation
methods are becoming increasingly popular, with high accuracy and great robustness. The
hybrid model consists of multiple different models, which can fully utilize the advantages
of different models to further enhance generalization ability and recognition robustness,
and it has good research and application prospects. The performance characteristics of
machine learning models, deep learning models, and hybrid models are summarized in
Table 4. Different models have different characteristics and need to select reasonably based
on goals, data resources, and other factors.

Table 4. Advantages and disadvantages of the models.

Methods Typical Model Advantages Disadvantages

Difficulty considering nonlinear
Requires small data samples, and
and complex time-varying
features are extracted based on expert
information, and low accuracy;
experience. Low computational
Parameters are complex and
SVM [20,57–65] complexity and fast convergence speed,
difficult to select; High
and strong training ability for small
dependence on cross training
sample data; Suitable for different battery
and regularization methods;
applications, good for diagnostic
Sensitive to missing data.
Need to use prior experience to
Machine learning A large amount of comprehensive sample determine the network structure
models data is required, and the estimation error of the mode; Needs lots of
is largely influenced by the training data; training data; Network
ANN [18,19,50–55] difficult to achieve online training; Strong parameters are complex, and
self-learning ability, high prediction the required training time and
accuracy, and good analysis effect on computing resources increase
nonlinear systems sharply with the increase of
data volume
High prediction accuracy, output in the
More hyper parameters,
GPR [22,23,66–77] form of probability prediction; High
complicated adjustment
computational complexity
Automatic feature extraction, seamless
connection with the classifier, high
accuracy, and it is greatly affected by the
Requires a large sample size of
correlation of feature data and the length
CNN [34–36,86–97] data, large amount of
of sample data; Simple control structure
calculation needed, complicated
and robust tracking performance in
uncertain environments, fast estimation,
Deep learning high accuracy
models
Accurate estimation of SOC, no initial
Basic RNN SOC needed, easily filter noise in data
[8,38,39,98–100] with the gating mechanism to consider Requires big data,
LSTM [101–108] the influence of time dimension, high large amount of calculation
GRU [109–114] accuracy, large computational power and needed, complicated
Others [115–128] resource occupation, and slow
convergence speed of multi-feature data
Considering the advantages of different
Hybrid models
CNN + RNN models, it has strong generalization
[39–41,48,129–142]
ability and good robustness

4. Future Development Trends

With the application of technologies such as artificial intelligence, network com-
munication, and advanced sensors, automobiles are accelerating towards electrification,
intelligence, and networking. Data-driven power battery states estimation methods have
broad application prospects in the field of battery safety management for electric buses. In
World Electr. Veh. J. 2023, 14, 145 15 of 22

terms of the future, power battery states estimation methods have shown the following
development trends.
(1) The accuracy and reliability of data-driven battery states estimation methods largely
depend on the quality and quantity of sample data, as well as the diversity of working
scenario. During the actual condition of electric vehicles, environmental conditions
such as road slope, climate, as well as abnormal voltage and temperature, can cause
the battery modules to deviate from their equilibrium state. Based on natural driving,
on-board sensor data have gradually become the main source of data for battery states
estimation. The use of onboard sensors and network technology to obtain battery
states data under actual scenarios has the advantages of objectivity, accuracy, reliability,
massive data, and low acquisition cost, which can provide reliable data support
for accurate estimation of battery states. Meanwhile, data migration learning, and
confrontation generation learning will provide new solutions for generating massive
training data samples [95,140,143–147]. Tang et al. [147] combined industrial data
with accelerated aging testing through migration-based machine learning to restore
large-scale battery aging datasets and reduce the cost of aging testing. Li et al. [93]
migrate the CNN model pre-trained on the large capacity battery data set to the
small target battery data set through the transfer learning technology to improve the
capacity estimation accuracy.
(2) The design of models is crucial for achieving accurate estimation of battery states. Fur-
ther development of battery states estimation models with higher accuracy, stronger
robustness, and better real-time performance is a research hotspot. The deep learning
model overcomes the problem of rapid decline in estimation accuracy and generaliza-
tion ability caused by changes in work scenarios and key feature parameter selection
in traditional machine learning models. By automatically learning and extracting
spatiotemporal features of battery states data under different work scenarios, it has the
characteristics of high accuracy and good robustness and can achieve a good balance
between accuracy and generalization ability. The data-driven model-based battery
states estimation method using deep learning will become one of the mainstream
methods, especially for the hybrid model, which can comprehensively utilize the
unique advantages of various models and use multiple methods for comprehensive
complementarity in different stages of battery states estimation tasks to improve the
accuracy of model. It has great value for further investigation.
(3) It is an important research direction to further improve the accuracy and robustness
of battery states estimation by considering multiple scenarios and multiple feature
parameter constraints, combined with advanced intelligent algorithms. Artificial
intelligence algorithms such as deep learning and transfer learning are widely used
in battery states estimation, and traditional battery states estimation methods are
being reshaped and upgraded. Under the influence of multiple factors such as break-
throughs in algorithm technology, powerful computing power, and massive data,
artificial intelligence has the ability to represent knowledge and make inferential
decisions at multiple levels, distributions, and tasks. In some fields, artificial intel-
ligence has reached or surpassed human level. However, in the scenario of power
battery applications, the challenges faced by deep human applications of deep learn-
ing are enormous due to the complexity and variability of the environment, the lack
of annotated data, and the difficulty in overcoming practical pain points.
(4) The accurate and real-time requirements in engineering applications have introduced
higher requirements for the software and hardware of battery management systems.
Some application scenarios require real-time and accurate estimation of battery states,
which poses new challenges to the performance of battery management systems.
In addition to low-cost high-computing-power microprocessors and high-precision
sensors, more in-depth research is also needed on the collection and processing of
on-board sensor data, multi-source high-dimensional heterogeneous data fusion,
efficient deep learning network architecture design, and high-performance hybrid
World Electr. Veh. J. 2023, 14, 145 16 of 22

model development. Software and hardware technologies complement each other to

further improve the accuracy, reliability, and real-time performance of battery states
estimation while reducing costs. How to deeply mine multisource, heterogeneous, and
massive vehicle sensor data information, and then use information fusion technology
to classify and process information at different levels, comprehensively utilize the
characteristics of different models, and more accurately identify the true status of
various batteries will become a research hotspot in the field of battery management
system technology.

5. Summary
The states estimation method of power batteries is an important research direction in
the field of new energy vehicles, it plays an important role in engineering fields such as
energy storage management and safety management. This paper summarizes the data-
driven battery states estimation methods including data sources and estimation models.
Firstly, the current research progress of battery states estimation methods is summarized.
Around the basic process of battery states estimation tasks, mainstream data sources
and typical models used for battery states estimation are analyzed and discussed. By
comparing and analyzing the advantages and disadvantages of three types of data sources,
it is found that there is rich information hidden in onboard sensor data and the natural
driving data can be used as the main data source for battery states estimation. Secondly,
the models for battery states estimation are divided into three categories: machine learning
models, deep learning models, and hybrid models. By analyzing the research progress
of various models, the paper notes that the deep learning models represented by CNN,
RNN, and hybrid models have advantages in accuracy, generalization ability, and other
aspects compared with traditional machine learning models. Hybrid models have received
significant attention from researchers, becoming the mainstream of data-driven battery
states estimation. Additionally, battery states estimation methods face multiple challenges
including higher accuracy, real-time performance, and robustness which are important
trends. Scholars need to continue to conduct in-depth and detailed theoretical and applied
research on data-driven battery states estimation methods, which has profound significance
for improving the safety of new energy vehicles and the healthy development of the
automobile industry.

Author Contributions: D.Z. put forward key issues and future work of the battery states estimation
methods; H.L. reviewed the deep learning methods; F.Z. improved the overall structure of the paper
and edited the paper; Y.Z. reviewed machine learning methods of battery states estimation; G.Z.
provided all tables and figures; Z.L. offered engineering application experience of battery states
estimation; J.H. reviewed fault diagnosis methods of batteries. All authors have read and agreed to
the published version of the manuscript.
Funding: This work was funded by the Opening Project of Key Laboratory of operation safety
technology on transport vehicles, Ministry of Transport, PRC (KFKT2022-05), partly funded by the
National Natural Science Foundation of China (62073298), and partly funded by Research Foundation
of Zhengzhou University of Light Industry (2021BSJJ021), and partly funded by Key Research and
Development Projects of Henan Province in 2022 (221111240200) and Major Science and Technol-
ogy Projects of Henan Province in 2022 (221100220200), and partly funded by Key Scientific and
Technological Project of Henan Province (222102240053).
Data Availability Statement: Data is not available due to private policy.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ding, H.; Wang, J.; Zhou, J.; Wang, C.; Lu, B. Building electrode skins for ultra-stable potassium metal batteries. Nat. Commun.
2023, 14, 2305. [CrossRef] [PubMed]
2. Takyi-Aninakwa, P.; Wang, S.; Zhang, H.; Appiah, E.; Bobobee, E.D.; Fernandez, C. A strong tracking adaptive fading-extended
Kalman filter for the state of charge estimation of lithium-ion batteries. Int. J. Energy Res. 2022, 46, 16427–16444. [CrossRef]
World Electr. Veh. J. 2023, 14, 145 17 of 22

3. Tang, X.; Gao, F.; Liu, K.; Liu, Q.; Foley, A.M. A Balancing Current Ratio Based State-of-Health Estimation Solution for Lithium-Ion
Battery Pack. IEEE Trans. Ind. Electron. 2021, 69, 8055–8065. [CrossRef]
4. 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]
5. Guo, R.; Shen, W. A data-model fusion method for online state of power estimation of lithium-ion batteries at high discharge rate
in electric vehicles. Energy 2022, 254, 124270. [CrossRef]
6. Chen, Y.; Yang, X.; Luo, D.; Wen, R. Remaining available energy prediction for lithium-ion batteries considering electrothermal
effect and energy conversion efficiency. J. Energy Storage 2021, 40, 102728. [CrossRef]
7. Cabrera-Castillo, E.; Niedermeier, F.; Jossen, A. Calculation of the state of safety (SOS) for lithium ion batteries. J. Power Sources
2016, 324, 509–520. [CrossRef]
8. Catelani, M.; Ciani, L.; Fantacci, R.; Patrizi, G.; Picano, B. Remaining Useful Life Estimation for Prognostics of Lithium-Ion
Batteries Based on Recurrent Neural Network. IEEE Trans. Instrum. Meas. 2021, 70, 1–11. [CrossRef]
9. You, M.; Liu, Y.; Chen, Z.; Zhou, X. Capacity Estimation of lithium battery based on charging data and Long Short-term Memory
Recurrent Neural Network. In Proceedings of the 2022 IEEE Intelligent Vehicles Symposium, Aachen, Germany, 4–9 June 2022;
pp. 230–234.
10. Deng, Z.; Hu, X.; Li, P.; Lin, X.; Bian, X. Data-Driven Battery State of Health Estimation Based on Random Partial Charging Data.
IEEE Trans. Power Electron. 2022, 37, 5021–5031. [CrossRef]
11. Zhang, J.; Jiang, Y.; Li, X.; Huo, M.; Luo, H.; Yin, S. An adaptive remaining useful life prediction approach for single battery with
unlabeled small sample data and parameter uncertainty. Reliab. Eng. Syst. Saf. 2022, 222, 108357. [CrossRef]
12. Cui, D.; Wang, Z.; Liu, P.; Wang, S.; Zhang, Z.; Dorrell, D.G.; Li, X. Battery electric vehicle usage pattern analysis driven by
massive real-world data. Energy 2022, 250, 123837. [CrossRef]
13. Geng, Z.; Thiringer, T. In situ key aging parameter determination of a vehicle battery using only CAN signals in commercial
vehicles. Appl. Energy 2022, 314, 118932. [CrossRef]
14. Wang, J.; Jia, Y.; Yang, N.; Lu, Y.; Shi, M.; Ren, X.; Lu, D. Precise equivalent circuit model for Li-ion battery by experimental
improvement and parameter optimization. J. Energy Storage 2022, 52, 104980. [CrossRef]
15. Guo, R.; Shen, W. A Review of Equivalent Circuit Model Based Online State of Power Estimation for Lithium-Ion Batteries in
Electric Vehicles. Vehicles 2021, 4, 1–29. [CrossRef]
16. Thakkar, R.R. Electrical equivalent circuit models of lithium-ion battery. In Management and Applications of Energy Storage Devices;
IntechOpen: London, England, 2021.
17. Finegan, D.P.; Zhu, J.; Feng, X.; Keyser, M.; Ulmefors, M.; Li, W.; Bazant, M.Z.; Cooper, S.J. The application of data-driven methods
and physics-based learning for improving battery safety. Joule 2021, 5, 316–329. [CrossRef]
18. Su, Z.; Li, X.; Qin, J.; Wenjie, D.; Ning, H. SOC estimation method of power battery based on BP artificial neural network. Energy
Storage Sci. Technol. 2019, 8, 868.
19. Sun, T.W.; Tsai, T.H. A battery management system with charge balancing and aging detection based on ANN. In Proceedings of
the 2021 IEEE In-ternational Symposium on Circuits and Systems (ISCAS), Daegu, Republic of Korea, 22–28 May 2021; pp. 1–4.
20. Feng, X.; Weng, C.; He, X.; Han, X.; Lu, L.; Ren, D.; Ouyang, M. Online State-of-Health Estimation for Li-Ion Battery Using Partial
Charging Segment Based on Support Vector Machine. IEEE Trans. Veh. Technol. 2019, 68, 8583–8592. [CrossRef]
21. Patil, M.A.; Tagade, P.; Hariharan, K.S.; Kolake, S.M.; Song, T.; Yeo, T.; Doo, S. A novel multistage Support Vector Machine based
approach for Li ion battery remaining useful life estimation. Appl. Energy 2015, 159, 285–297. [CrossRef]
22. Wang, Z.; Yuan, C.; Li, X. Lithium Battery State-of-Health Estimation via Differential Thermal Voltammetry With Gaussian
Process Regression. IEEE Trans. Transp. Electrif. 2020, 7, 16–25. [CrossRef]
23. Liu, J.; Liu, X. An improved method of state of health prediction for lithium batteries considering different temperature. J. Energy
Storage 2023, 63, 107028. [CrossRef]
24. Piao, C.; Li, Z.; Lu, S.; Jin, Z.; Cho, C. Analysis of Real-Time Estimation Method Based on Hidden Markov Models for Battery
System States of Health. J. Power Electron. 2016, 16, 217–226. [CrossRef]
25. Lai, X.; Huang, Y.; Gu, H.; Han, X.; Feng, X.; Dai, H.; Zheng, Y.; Ouyang, M. Remaining discharge energy estimation for lithium-ion
batteries based on future load prediction considering temperature and ageing effects. Energy 2021, 238, 121754. [CrossRef]
26. Li, Y.; Zou, C.; Berecibar, M.; Nanini-Maury, E.; Chan, J.C.-W.; Bossche, P.V.D.; Van Mierlo, J.; Omar, N. Random forest regression
for online capacity estimation of lithium-ion batteries. Appl. Energy 2018, 232, 197–210. [CrossRef]
27. Lipu, M.S.H.; Hannan, M.A.; Hussain, A.; Ansari, S.; Rahman, S.A.; Saad, M.H.M.; Muttaqi, K.M. Department of Electrical,
Electronic, and Systems Engineering, Universiti Kebangsaan Malaysia, Bangi, Malaysia Real-time state of charge estimation of
Lithium-ion batteries using optimized random forest regression algorithm. IEEE Trans. Intell. Veh. 2022, 8, 639–648. [CrossRef]
28. Malkhandi, S. Fuzzy logic-based learning system and estimation of state-of-charge of lead-acid battery. Eng. Appl. Artif. Intell.
2006, 19, 479–485. [CrossRef]
29. Zheng, W.; Xia, B.; Wang, W.; Lai, Y.; Wang, M.; Wang, H. State of Charge Estimation for Power Lithium-Ion Battery Using a
Fuzzy Logic Sliding Mode Observer. Energies 2019, 12, 2491. [CrossRef]
30. Feng, H.; Yan, H. State of health estimation of large-cycle lithium-ion batteries based on error compensation of autoregressive
model. J. Energy Storage 2022, 52, 104869. [CrossRef]
World Electr. Veh. J. 2023, 14, 145 18 of 22

31. Song, Y.; Liu, D.; Yang, C.; Peng, Y. Data-driven hybrid remaining useful life estimation approach for spacecraft lithium-ion
battery. Microelectron. Reliab. 2017, 75, 142–153. [CrossRef]
32. Chen, Z.; Shi, N.; Ji, Y.; Niu, M.; Wang, Y. Lithium-ion batteries remaining useful life prediction based on BLS-RVM. Energy 2021,
234, 121269. [CrossRef]
33. Liu, Y.; Zhao, G.; Peng, X. A lithium-ion battery remaining using life prediction method based on multi-kernel relevance vector
machine optimized model. Acta Electonica Sin. 2019, 47, 1285.
34. Wei, C.; Yue, S.; Bai, S. Remaining Useful Life Prediction Based on CNN Method. In 6th International Technical Conference on
Advances in Computing, Control and Industrial Engineering (CCIE 2021), Hangzhou, China, 15–16 January 2022; Springer: Singapore,
2022; pp. 309–316.
35. Qian, C.; Xu, B.; Chang, L.; Sun, B.; Feng, Q.; Yang, D.; Ren, Y.; Wang, Z. Convolutional neural network based capacity estimation
using random segments of the charging curves for lithium-ion batteries. Energy 2021, 227, 120333. [CrossRef]
36. Lu, Z.; Wang, B.; Tan, X.; Yang, F. Battery Capacity Estimation based on Convolutional Neural Network. In Proceedings of the
2022 Global Reliability and Prognostics and Health Management (PHM-Yantai), Yantai, China, 13–16 October 2022; pp. 1–6.
37. Feng, X.; Chen, J.; Zhang, Z.; Miao, S.; Zhu, Q. State-of-charge estimation of lithium-ion battery based on clockwork recurrent
neural network. Energy 2021, 236, 121360. [CrossRef]
38. Hsieh, Y.Z.; Tan, S.W.; Gu, S.L.; Jeng, Y.-L. Prediction of Battery Discharge States Based on the Recurrent Neural Network.
J. Internet Technol. 2020, 21, 113–120.
39. Song, X.; Yang, F.; Wang, D.; Tsui, K.-L. Combined CNN-LSTM Network for State-of-Charge Estimation of Lithium-Ion Batteries.
IEEE Access 2019, 7, 88894–88902. [CrossRef]
40. Xu, H.; Wu, L.; Xiong, S.; Li, W.; Garg, A.; Gao, L. An improved CNN-LSTM model-based state-of-health estimation approach for
lithium-ion bat-teries. Energy 2023, 276, 127585. [CrossRef]
41. Ren, L.; Dong, J.; Wang, X.; Meng, Z.; Zhao, L.; Deen, M.J. A Data-Driven Auto-CNN-LSTM Prediction Model for Lithium-Ion
Battery Remaining Useful Life. IEEE Trans. Ind. Inform. 2020, 17, 3478–3487. [CrossRef]
42. Toughzaoui, Y.; Toosi, S.B.; Chaoui, H.; Louahlia, H.; Petrone, R.; Le Masson, S.; Gualous, H. State of health estimation and
remaining useful life assessment of lithium-ion batteries: A comparative study. J. Energy Storage 2022, 51, 104520. [CrossRef]
43. Manoharan, A.; Begam, K.; Aparow, V.R.; Sooriamoorthy, D. Artificial Neural Networks, Gradient Boosting and Support Vector
Machines for electric vehicle battery state estimation: A review. J. Energy Storage 2022, 55, 105384. [CrossRef]
44. Sulzer, V.; Mohtat, P.; Aitio, A.; Lee, S.; Yeh, Y.T.; Steinbacher, F.; Khan, M.U.; Lee, J.W.; Siegel, J.B.; Stefanopoulou, A.G.; et al. The
challenge and opportunity of battery lifetime prediction from field data. Joule 2021, 5, 1934–1955. [CrossRef]
45. dos Reis, G.; Strange, C.; Yadav, M.; Li, S. Lithium-ion battery data and where to find it. Energy AI 2021, 5, 100081. [CrossRef]
46. Sakile, R.K.; Kumar, P.; Sinha, U.K. Modelling of Lithium-Ion Battery Using MATLAB/Simulink for Electric Vehicle Applications.
In Lecture Notes in Electrical Engineering Book Series; Springer: Berlin, Germany, 2020; pp. 291–299.
47. Yao, L.W.; Aziz, J.A.; Kong, P.Y.; Idris, N.R.N. Modeling of lithium-ion battery using MATLAB/simulink. In Proceedings of
the IECON 2013-39th Annual Conference of the IEEE Industrial Electronics Society, Vienna, Austria, 10–13 November 2013;
pp. 1729–1734.
48. Tao, Y.; Huang, M.; Chen, Y.; Yang, L. Orderly charging strategy of battery electric vehicle driven by real-world driving data.
Energy 2020, 193, 116806. [CrossRef]
49. Wang, K.; Wu, C.; Shi, J. Prediction of BatteryTemperature in Electric Bus Charging Stage Based on CNN-LSTM; SPIE: Washington, DC,
USA, 2023; Volume 12511, pp. 465–470.
50. Wang, Y.; Chen, X.; Li, C.; Yu, Y.; Zhou, G.; Wang, C.; Zhao, W. Temperature prediction of lithium-ion battery based on artificial
neural network model. Appl. Therm. Eng. 2023, 228, 120482. [CrossRef]
51. Bezha, M.; Nagaoka, N. An ANN model for estimating internal impedance of lithium-ion battery cell for industrial applica-tion.
In Proceedings of the 2018 21st International Conference on Electrical Machines and Systems (ICEMS), Jeju, Republic of Korea,
7–10 October 2018; pp. 2105–2110.
52. Hussein, A.A. Capacity Fade Estimation in Electric Vehicle Li-Ion Batteries Using Artificial Neural Networks. IEEE Trans. Ind.
Appl. 2014, 51, 2321–2330. [CrossRef]
53. Qin, W.; Lv, H.; Liu, C.; Nirmalya, D.; Jahanshahi, P. Remaining useful life prediction for lithium-ion batteries using particle filter
and artificial neural network. Ind. Manag. Data Syst. 2019, 120, 312–328. [CrossRef]
54. Jaliliantabar, F.; Mamat, R.; Kumarasamy, S. Prediction of lithium-ion battery temperature in different operating conditions
equipped with passive battery thermal management system by artificial neural networks. Mater. Today Proc. 2022, 48, 1796–1804.
[CrossRef]
55. Liu, X.; Dai, Y. Energy storage battery SOC estimate based on improved BP neural network. J. Phys. Conf. Ser. 2022, 2187, 012042.
[CrossRef]
56. Xu, J.; Han, J.; Nie, F.; Li, X. Multi-View Scaling Support Vector Machines for Classification and Feature Selection. IEEE Trans.
Knowl. Data Eng. 2019, 32, 1419–1430. [CrossRef]
57. Deng, F.; Bian, Y.; Zheng, H. Fault diagnosis for electric vehicle lithium batteries using a multi-classification support vector
machine. Electr. Eng. 2021, 104, 1831–1837. [CrossRef]
58. Junping, W.; Quanshi, C.; Binggang, C. Support vector machine based battery model for electric vehicles. Energy Convers. Manag.
2006, 47, 858–864. [CrossRef]
World Electr. Veh. J. 2023, 14, 145 19 of 22

59. Chen, Z.; Sun, M.; Shu, X.; Xiao, R.; Shen, J. Online State of Health Estimation for Lithium-Ion Batteries Based on Support Vector
Machine. Appl. Sci. 2018, 8, 925. [CrossRef]
60. Li, X.; Shu, X.; Shen, J.; Xiao, R.; Yan, W.; Chen, Z. An On-Board Remaining Useful Life Estimation Algorithm for Lithium-Ion
Batteries of Electric Vehicles. Energies 2017, 10, 691. [CrossRef]
61. Li, X.; Ma, Y.; Zhu, J. An online dual filters RUL prediction method of lithium-ion battery based on unscented particle filter and
least squares support vector machine. Measurement 2021, 184, 109935. [CrossRef]
62. De Klerk, M.L.; Saha, A.K. Performance analysis of DTC-SVM in a complete traction motor control mechanism for a battery
electric vehicle. Heliyon 2022, 8, e09265. [CrossRef] [PubMed]
63. Szabolcsi, R.; Menyhárt, J. Battery Voltage Limit Analysis with Support Vector Machine and Fuzzy Logic. Adv. Mil. Technol. 2017,
12, 21–32. [CrossRef]
64. Li, R.; Li, W.; Zhang, H.; Zhou, Y.; Tian, W. On-Line Estimation Method of Lithium-Ion Battery Health Status Based on PSO-SVM.
Front. Energy Res. 2021, 9, 693249. [CrossRef]
65. Yao, L.; Fang, Z.; Xiao, Y.; Hou, J.; Fu, Z. An Intelligent Fault Diagnosis Method for Lithium Battery Systems Based on Grid Search
Support Vector Machine. Energy 2020, 214, 118866. [CrossRef]
66. Zhou, Y.; Dong, G.; Tan, Q.; Han, X.; Chen, C.; Wei, J. State of health estimation for lithium-ion batteries using geometric
impedance spectrum features and recurrent Gaussian process regression. Energy 2023, 262, 125514. [CrossRef]
67. Yang, D.; Zhang, X.; Pan, R.; Wang, Y.; Chen, Z. A novel Gaussian process regression model for state-of-health estimation of
lithium-ion battery using charging curve. J. Power Sources 2018, 384, 387–395. [CrossRef]
68. Wang, J.; Deng, Z.; Yu, T.; Yoshida, A.; Xu, L.; Guan, G.; Abudula, A. State of health estimation based on modified Gaussian
process regression for lithium-ion batteries. J. Energy Storage 2022, 51, 104512. [CrossRef]
69. Pang, X.; Liu, X.; Jia, J.; Wen, J.; Shi, Y.; Zeng, J.; Zhao, Z. A lithium-ion battery remaining useful life prediction method based on
the incremental capacity analysis and Gaussian process regression. Microelectron. Reliab. 2021, 127, 114405. [CrossRef]
70. Xing, J.; Zhang, H.; Zhang, J. Remaining useful life prediction of–Lithium batteries based on principal component analysis and
improved Gaussian process regression. Int. J. Electrochem. Sci. 2023, 18, 100048. [CrossRef]
71. Zhao, J.; Xuebin, L.; Daiwei, Y.; Jun, Z.; Wenjin, Z. Lithium-ion battery state of health estimation using meta-heuristic optimization
and Gaussian process regression. J. Energy Storage 2023, 58, 106319. [CrossRef]
72. Deng, Z.; Hu, X.; Lin, X.; Che, Y.; Xu, L.; Guo, W. Data-driven state of charge estimation for lithium-ion battery packs based on
Gaussian process regression. Energy 2020, 205, 118000. [CrossRef]
73. Zhou, D.; Zheng, W.; Chen, S.; Fu, P.; Zhu, H.; Song, B.; Qu, X.; Wang, T. Research on state of health prediction model for lithium
batteries based on actual diverse data. Energy 2021, 230, 120851. [CrossRef]
74. Ozcan, G.; Pajovic, M.; Sahinoglu, Z.; Wang, Y.; Orlik, P.V.; Wada, T. Online battery state-of-charge estimation based on sparse
gaussian process regres-sion. In Proceedings of the 2016 IEEE Power and Energy Society General Meeting (PESGM), Boston, MA,
USA, 17–21 July 2016; pp. 1–5.
75. Chen, X.; Chen, X.; Chen, X. A novel framework for lithium-ion battery state of charge estimation based on Kalman filter Gaussian
process regression. Int. J. Energy Res. 2021, 45, 13238–13249. [CrossRef]
76. Feng, H.; Shi, G. SOH and RUL prediction of Li-ion batteries based on improved Gaussian process regression. J. Power Electron.
2021, 21, 1845–1854. [CrossRef]
77. Nagulapati, V.M.; Lee, H.; Jung, D.W.; Paramanantham, S.S.; Brigljevic, B.; Choi, Y.S.; Lim, H. A novel combined multi-battery
dataset based approach for enhanced prediction accuracy of data driven prognostic models in capacity estimation of lithium ion
batteries. Energy AI 2021, 5, 100089. [CrossRef]
78. Dong, G.; Chen, Z.; Wei, J.; Ling, Q. Battery Health Prognosis Using Brownian Motion Modeling and Particle Filtering. IEEE
Trans. Ind. Electron. 2018, 65, 8646–8655. [CrossRef]
79. Xu, J.; Mei, X.; Wang, X.; Fu, Y.; Zhao, Y.; Wang, J. A Relative State of Health Estimation Method Based on Wavelet Analysis for
Lithium-Ion Battery Cells. IEEE Trans. Ind. Electron. 2020, 68, 6973–6981. [CrossRef]
80. Paulson, N.H.; Kubal, J.; Ward, L.; Saxena, S.; Lu, W.; Babinec, S.J. Feature engineering for machine learning enabled early
prediction of battery lifetime. J. Power Sources 2022, 527, 231127. [CrossRef]
81. Huotari, M.; Arora, S.; Malhi, A.; Främling, K. Comparing seven methods for state-of-health time series prediction for the
lithium-ion battery packs of forklifts. Appl. Soft Comput. 2021, 111, 107670. [CrossRef]
82. Vilsen, S.B.; Stroe, D.-I. Battery state-of-health modelling by multiple linear regression. J. Clean. Prod. 2021, 290, 125700. [CrossRef]
83. Zhou, Y.; Huang, M.; Pecht, M. Remaining useful life estimation of lithium-ion cells based on k-nearest neighbor regression with
differential evolution optimization. J. Clean. Prod. 2019, 249, 119409. [CrossRef]
84. Sandhu, S.; Tyagi, R.; Talaie, E.; Srinivasan, S. Using neurocomputing techniques to determine microstructural properties in a
Li-ion battery. Neural Comput. Appl. 2022, 34, 9983–9999. [CrossRef]
85. Choi, C.; Park, S.; Kim, J. Uniqueness of multilayer perceptron-based capacity prediction for contributing state-of-charge
esti-mation in a lithium primary battery. Ain Shams Eng. J. 2023, 14, 101936. [CrossRef]
86. Chemali, E.; Kollmeyer, P.J.; Preindl, M.; Fahmy, Y.; Emadi, A. A Convolutional Neural Network Approach for Estimation of
Li-Ion Battery State of Health from Charge Profiles. Energies 2022, 15, 1185. [CrossRef]
World Electr. Veh. J. 2023, 14, 145 20 of 22

87. Li, P.; Zhang, Z.; Grosu, R.; Deng, Z.; Hou, J.; Rong, Y.; Wu, R. An end-to-end neural network framework for state-of-health
estimation and remaining useful life prediction of electric vehicle lithium batteries. Renew. Sustain. Energy Rev. 2021, 156, 111843.
[CrossRef]
88. Sohn, S.; Byun, H.E.; Lee, J.H. CNN-based Online Diagnosis of Knee-point in Li-ion Battery Capacity Fade Curve. IFAC Pap. 2022,
55, 181–185. [CrossRef]
89. Shen, S.; Sadoughi, M.; Chen, X.; Hong, M.; Hu, C. A deep learning method for online capacity estimation of lithium-ion batteries.
J. Energy Storage 2019, 25, 100817. [CrossRef]
90. Valant, C.J.; Wheaton, J.D.; Thurston, M.G.; McConky, S.P.; Nenadic, N. Evaluation of 1D CNN autoencoders for lithium-ion
battery condition assessment using synthetic data. Proc. Annu. Conf. PHM Soc. 2019, 11, 1–11. [CrossRef]
91. Yang, N.; Song, Z.; Hofmann, H.; Sun, J. Robust State of Health estimation of lithium-ion batteries using convolutional neural
network and random forest. J. Energy Storage 2022, 48, 103857. [CrossRef]
92. Ding, P.; Liu, X.; Li, H.; Huang, Z.; Zhang, K.; Shao, L.; Abedinia, O. Useful life prediction based on wavelet packet decomposition
and two-dimensional convolutional neural network for lithium-ion batteries. Renew. Sustain. Energy Rev. 2021, 148, 111287.
[CrossRef]
93. Li, Y.; Li, K.; Liu, X.; Wang, Y.; Zhang, L. Lithium-ion battery capacity estimation—A pruned convolutional neural network
approach assisted with transfer learning. Appl. Energy 2021, 285, 116410. [CrossRef]
94. Xie, J.; Xu, J.; Wei, Z.; Li, X. Fault isolating and grading for li-ion battery packs based on pseudo images and convolutional neural
network. Energy 2023, 263, 125867. [CrossRef]
95. El-Dalahmeh, M.; Al-Greer, M.; El-Dalahmeh, M.; Short, M. Time-frequency image analysis and transfer learning for capacity
pre-diction of lithium-ion batteries. Energies 2020, 13, 5447. [CrossRef]
96. Zhang, Q.; Yang, L.; Guo, W.; Qiang, J.; Peng, C.; Li, Q.; Deng, Z. A deep learning method for lithium-ion battery remaining useful
life prediction based on sparse segment data via cloud computing system. Energy 2021, 241, 122716. [CrossRef]
97. Din, N.U.; Zhang, L.; Yang, Y. Automated Battery Making Fault Classification Using Over-Sampled Image Data CNN Features.
Sensors 2023, 23, 1927. [CrossRef]
98. Zhao, R.; Kollmeyer, P.J.; Lorenz, R.D.; Jahns, T.M. A compact unified methodology via a recurrent neural network for accurate
modeling of lithium-ion battery voltage and state-of-charge. In Proceedings of the 2017 IEEE Energy Conversion Congress and
Exposition (ECCE), Cincinnati, OH, USA, 1–5 October 2017; pp. 5234–5241.
99. Wang, Z.; Li, X.; Wang, Y. State of Charge Estimation of Lithium-ion Battery Based on Improved Recurrent Neural Net-work.
J. Phys. Conf. Series 2021, 2109, 012005. [CrossRef]
100. You, G.W.; Park, S.; Oh, D. Diagnosis of electric vehicle batteries using recurrent neural networks. IEEE Trans. Ind. Electron. 2017,
64, 4885–4893. [CrossRef]
101. Yang, J.; Zou, L.; Wei, Y.; Yuan, P.; Zhou, C. Health Status Prediction of Lithium Battery Based on LSTM Model with Optimization
Algo-rithms. J. Phys. Conf. Series 2023, 2473, 012020. [CrossRef]
102. Zhang, L.; Ji, T.; Yu, S.; Liu, G. Accurate Prediction Approach of SOH for Lithium-Ion Batteries Based on LSTM Method. Batteries
2023, 9, 177. [CrossRef]
103. Park, K.; Choi, Y.; Choi, W.J.; Ryu, H.-Y.; Kim, H. LSTM-Based Battery Remaining Useful Life Prediction with Multi-Channel
Charging Profiles. IEEE Access 2020, 8, 20786–20798. [CrossRef]
104. Ma, Y.; Shan, C.; Gao, J.; Chen, H. A novel method for state of health estimation of lithium-ion batteries based on improved LSTM
and health indicators extraction. Energy 2022, 251, 123973. [CrossRef]
105. Wu, Y.; Xue, Q.; Shen, J.; Lei, Z.; Chen, Z.; Liu, Y. State of Health Estimation for Lithium-Ion Batteries Based on Healthy Features
and Long Short-Term Memory. IEEE Access 2020, 8, 28533–28547. [CrossRef]
106. Chen, Z.; Song, X.; Xiao, R.; Shen, J.; Xia, X. State of Health Estimation for Lithium-Ion Battery Based on Long Short Term Memory
Networks. DEStech Trans. Environ. Energy Earth Sci. 2019, 27855. [CrossRef] [PubMed]
107. Zhang, Y.; Xiong, R.; He, H.; Pecht, M.G. Long Short-Term Memory Recurrent Neural Network for Remaining Useful Life
Prediction of Lithium-Ion Batteries. IEEE Trans. Veh. Technol. 2018, 67, 5695–5705. [CrossRef]
108. Kara, A. A data-driven approach based on deep neural networks for lithium-ion battery prognostics. Neural Comput. Appl. 2021,
33, 13525–13538. [CrossRef]
109. Yang, F.; Li, W.; Li, C.; Miao, Q. State-of-charge estimation of lithium-ion batteries based on gated recurrent neural network.
Energy 2019, 175, 66–75. [CrossRef]
110. Guo, F.; Wu, X.; Liu, L.; Ye, J.; Wang, T.; Fu, L.; Wu, Y. Prediction of remaining useful life and state of health of lithium
batteries based on time series feature and Savitzky-Golay filter combined with gated recurrent unit neural network. Energy 2023,
270, 126880. [CrossRef]
111. Lyu, C.; Han, Y.; Guo, Q.; Wang, L.; Song, Y. State-of-charge estimation of lithium-ion batteries based on deep neural network. In
Proceedings of the 2020 Global Reliability and Prognostics and Health Management (PHM-Shanghai), Shanghai, China, 16–18
October 2020; pp. 1–7.
112. Rouhi Ardeshiri, R.; Ma, C. Multivariate gated recurrent unit for battery remaining useful life prediction: A deep learning
approach. Int. J. Energy Res. 2021, 45, 16633–16648. [CrossRef]
113. Chen, J.; Lu, C.; Chen, C.; Cheng, H.; Xuan, D. An Improved Gated Recurrent Unit Neural Network for State-of-Charge Estimation
of Lithium-Ion Battery. Appl. Sci. 2022, 12, 2305. [CrossRef]
World Electr. Veh. J. 2023, 14, 145 21 of 22

114. Ungurean, L.; Micea, M.V.; Cârstoiu, G. Online state of health prediction method for lithium-ion batteries, based on gated
re-current unit neural networks. Int. J. Energy Res. 2020, 44, 6767–6777. [CrossRef]
115. Bian, C.; He, H.; Yang, S. Stacked bidirectional long short-term memory networks for state-of-charge estimation of lithium-ion
batteries. Energy 2019, 191, 116538. [CrossRef]
116. Yayan, U.; Arslan, A.T.; Yucel, H. A Novel Method for SoH Prediction of Batteries Based on Stacked LSTM with Quick Charge
Data. Appl. Artif. Intell. 2021, 35, 421–439. [CrossRef]
117. Ardeshiri, R.R.; Liu, M.; Ma, C. Multivariate stacked bidirectional long short term memory for lithium-ion battery health
management. Reliab. Eng. Syst. Saf. 2022, 224, 108481. [CrossRef]
118. Liu, C.; Zhang, Y.; Sun, J.; Cui, Z.; Wang, K. Stacked bidirectional LSTM RNN to evaluate the remaining useful life of supercapacitor.
Int. J. Energy Res. 2021, 46, 3034–3043. [CrossRef]
119. Bian, C.; Yang, S.; Liu, J.; Zio, E. Robust state-of-charge estimation of Li-ion batteries based on multichannel convolutional and
bidirectional recurrent neural networks. Appl. Soft Comput. 2022, 116, 108401. [CrossRef]
120. Yang, B.; Wang, Y.; Zhan, Y. Lithium Battery State-of-Charge Estimation Based on a Bayesian Optimization Bidirectional Long
Short-Term Memory Neural Network. Energies 2022, 15, 4670. [CrossRef]
121. Zhang, H.; Tang, W.; Na, W.; Lee, P.-Y.; Kim, J. Implementation of generative adversarial network-CLS combined with bidirectional
long short-term memory for lithium-ion battery state prediction. J. Energy Storage 2020, 31, 101489. [CrossRef]
122. Zhu, C.; Gao, M.; He, Z.; Wu, H.; Sun, C.; Zhang, Z.; Bao, Z. State of health prediction for li-ion batteries with end-to-end deep
learning. J. Energy Storage 2023, 65, 107218. [CrossRef]
123. Wang, Z.; Luo, W.; Xu, S.; Yan, Y.; Huang, L.; Wang, J.; Hao, W.; Yang, Z. Electric Vehicle Lithium-Ion Battery Fault Diagnosis
Based on Multi-Method Fusion of Big Data. Sustainability 2023, 15, 1120. [CrossRef]
124. Wang, Z.; Yang, F.; Xu, Q.; Wang, Y.; Yan, H.; Xie, M. Capacity estimation of lithium-ion batteries based on data aggregation and
feature fusion via graph neural network. Appl. Energy 2023, 336, 120808. [CrossRef]
125. Yao, X.Y.; Chen, G.; Pecht, M.; Chen, B. A novel graph-based framework for state of health prediction of lithium-ion battery.
J. Energy Storage 2023, 58, 106437. [CrossRef]
126. Zhang, C.; Zhu, Y.; Dong, G.; Wei, J. Data-driven lithium-ion battery states estimation using neural networks and particle filtering.
Int. J. Energy Res. 2019, 43, 8230–8241. [CrossRef]
127. Veeraraghavan, A.; Adithya, V.; Bhave, A.; Akella, S. Battery aging estimation with deep learning. In Proceedings of the 2017
IEEE Transportation Electrification Conference (ITEC-India), Pune, India, 13–15 December 2017; pp. 1–4.
128. Li, S.; Ju, C.; Li, J.; Fang, R.; Tao, Z.; Li, B.; Zhang, T. State-of-Charge Estimation of Lithium-Ion Batteries in the Battery Degradation
Process Based on Recurrent Neural Network. Energies 2021, 14, 306. [CrossRef]
129. Chen, X.D.; Yang, H.Y.; Wun, J.S.; Wang, C.-H.; Li, L.-L. Life prediction of lithium-ion battery based on a hybrid model. Energy
Explor. Exploit. 2020, 38, 1854–1878. [CrossRef]
130. Zhao, J.; Zhu, Y.; Zhang, B.; Liu, M.; Wang, J.; Liu, C.; Zhang, Y. Method of Predicting SOH and RUL of Lithium-Ion Battery Based
on the Combination of LSTM and GPR. Sustainability 2022, 14, 11865. [CrossRef]
131. Liu, B.; Xu, J.; Xia, W. State-of-Health Estimation for Lithium-Ion Battery Based on an Attention-Based CNN-GRU Model with
Reconstructed Feature Series. Int. J. Energy Res. 2023, 2023, 8569161. [CrossRef]
132. Mei, P.; Huang, C.; Yang, D.; Chen, F.; Song, Q. A Vehicle-Cloud Collaborative Strategy for State of Energy Estimation based on
CNN-LSTM Networks. In Proceedings of the 2022 2nd International Conference on Computers and Automation, Paris, France,
18–20 August 2022; pp. 128–132.
133. Zhang, X.; Xu, Y.; Gong, Z. A feature fusion optimization algorithm for predicting the remaining useful life of lithium-ion batteries.
Energy Rep. 2023, 9, 142–153. [CrossRef]
134. Zraibi, B.; Okar, C.; Chaoui, H.; Mansouri, M. Remaining Useful Life Assessment for Lithium-Ion Batteries Using CNN-LSTM-
DNN Hybrid Method. IEEE Trans. Veh. Technol. 2021, 70, 4252–4261. [CrossRef]
135. Yanwen, D.A.I.; Aiqing, Y.U. Combined CNN-LSTM and GRU based health feature parameters for lithium-ion batteries SOH
estimation. Energy Storage Sci. Technol. 2022, 11, 1641.
136. Sun, S.; Sun, J.; Wang, Z.; Zhou, Z.; Cai, W. Prediction of Battery SOH by CNN-BiLSTM Network Fused with Attention Mechanism.
Energies 2022, 15, 4428. [CrossRef]
137. Zhu, Z.; Yang, Q.; Liu, X.; Gao, D. Attention-based CNN-BiLSTM for SOH and RUL estimation of lithium-ion batteries.
J. Algorithms Comput. Technol. 2022, 16, 17483026221130598. [CrossRef]
138. Li, D.; Yang, L. Remaining Useful Life Prediction of Lithium Battery Based on Sequential CNN–LSTM Method. J. Electrochem.
Energy Convers. Storage 2021, 18, 4050886. [CrossRef]
139. Cui, Z.; Kang, L.; Li, L.; Wang, L.; Wang, K. A hybrid neural network model with improved input for state of charge estimation of
lithium-ion battery at low temperatures. Renew. Energy 2022, 198, 1328–1340. [CrossRef]
140. Zou, G.; Yan, Z.; Zhang, C.; Song, L. Transfer learning with CNN-LSTM model for capacity prediction of lithium-ion batteries
under small sample. J. Phys. Conf. Series. 2022, 2258, 12042. [CrossRef]
141. Rastegarpanah, A.; Wang, Y.; Stolkin, R. Predicting the Remaining Life of Lithium-ion Batteries Using a CNN-LSTM Model. In
Proceedings of the 2022 8th International Conference on Mechatronics and Robotics Engineering (ICMRE), Munich, Germany,
10–12 February 2022; pp. 73–78.
World Electr. Veh. J. 2023, 14, 145 22 of 22

142. Tian, Y.; Wen, J.; Yang, Y.; Shi, Y.; Zeng, J. State-of-Health Prediction of Lithium-Ion Batteries Based on CNN-BiLSTM-AM. Batteries
2022, 8, 155. [CrossRef]
143. Chen, Z.; Shen, W.; Chen, L.; Wang, S. Adaptive online capacity prediction based on transfer learning for fast charging lithium-ion
batteries. Energy 2022, 248, 123537. [CrossRef]
144. Ma, G.; Xu, S.; Yang, T.; Du, Z.; Zhu, L.; Ding, H.; Yuan, Y. A Transfer Learning-Based Method for Personalized State of Health
Estimation of Lithium-Ion Batteries. IEEE Trans. Neural Netw. Learn. Syst. 2022, 1–11. [CrossRef]
145. Liu, Y.; Shu, X.; Yu, H.; Shen, J.; Zhang, Y.; Liu, Y.; Chen, Z. State of charge prediction framework for lithium-ion batteries
incorporating long short-term memory network and transfer learning. J. Energy Storage 2021, 37, 102494. [CrossRef]
146. Zhao, J.; Burke, A.F. Electric Vehicle Batteries: Status and Perspectives of Data-Driven Diagnosis and Prognosis. Batteries 2022,
8, 142. [CrossRef]
147. Tang, X.; Liu, K.; Li, K.; Widanage, W.D.; Kendrick, E.; Gao, F. Recovering large-scale battery aging dataset with machine learning.
Patterns 2021, 2, 100302. [CrossRef] [PubMed]

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.