Advaned Bms

Renewable and Sustainable Energy Reviews xxx (xxxx) xxx
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: http://www.elsevier.com/locate/rser
Advanced battery management strategies for a sustainable energy future:

Multilayer design concepts and research trends
Haifeng Dai a, b, Bo Jiang a, b, Xiaosong Hu c, *, Xianke Lin d, Xuezhe Wei a, b, Michael Pecht e
a
School of Automotive Studies, Tongji University, Shanghai, 201804, China
b
National Fuel Cell Vehicle & Powertrain System Research & Engineering Center, Shanghai, 201804, China
c
Department of Automotive Engineering, Chongqing University, Chongqing, 400044, China
d
Department of Automotive, Mechanical and Manufacturing Engineering, Ontario Tech University, Oshawa, ON, L1G 0C5, Canada
e
Center for Advanced Life Cycle Engineering, University of Maryland, College Park, MD, 20742, USA
A R T I C L E I N F O A B S T R A C T
Keywords: Lithium-ion batteries are promising energy storage devices for electric vehicles and renewable energy systems.
Lithium-ion batteries However, due to complex electrochemical processes, potential safety issues, and inherent poor durability of
Battery management technologies lithium-ion batteries, it is essential to monitor and manage batteries safely and efficiently. This study reviews the
Multilayer design concepts
development of battery management systems during the past periods and introduces a multilayer design ar
Safety and aging
chitecture for advanced battery management, which consists of three progressive layers. The foundation layer
Data and intelligence
focuses on the system physical basis and theoretical principle, the algorithm layer aims at providing a
comprehensive understanding of battery, and the application layer ensures a safe and efficient battery system
through sufficient management. A comprehensive overview of each layer is presented from both academic and
engineering perspectives. Future trends in research and development of next-generation battery management are
discussed. Based on data and intelligence, the next-generation battery management will achieve better safety,
performance, and interconnectivity.
generations: “no management”, “simple management”, and “advanced

management” [3], as shown in Fig. 1. The “no management” system is
1. Introduction
only suitable for early lead-acid batteries that have good anti-abuse
capabilities, and only monitors the battery terminal voltage for char
The need to reduce environmental pollution and dependence on
ge/discharge control. This generation of battery management has inef
nonrenewable fuels has contributed to the rapid growth of renewable
ficient control, requires time-consuming maintenance, and suffers from
energy systems (RESs) and electric vehicles (EVs) [1]. Batteries are
low utilization of battery energy. Compared with the “no management”
essential to RESs and EVs, as they are considered as a primary energy
system, the “simple management” system has not been significantly
storage source, and lithium-ion batteries (LIBs) are particularly prom
improved. The “simple management” system mainly focuses on
ising because of their excellent performance and continually decreasing
improving the detection circuit to ensure measurement accuracy and
price. However, LIBs are usually operated in a harsh environment and
reliability and can monitor external parameters (including voltage,
undergo all kinds of abusive conditions. Overloading, external heat
current and temperature) of each cell in a battery pack online and pre
transfer, excessively fast charging, and internal/external short circuit,
vent over-charging and over-discharging effectively. However, it has
can damage the battery life, and cause the abnormal temperature in
almost no algorithms or very simple ones if any, and it does not support
creases, leading to gas generation or even thermal runaway, combus
the online monitoring of internal states. Hence, the workload and
tion, and explosion. Therefore, batteries must be effectively managed to
complexity of battery maintenance are not reduced due to the inade
ensure their safety and prolong their service life. A battery management
quate understanding of battery internal states.
system (BMS) is needed to not only monitor all kinds of battery states,
With the increasing number of onboard batteries, advanced man
but also to ensure that battery performance meets the demands of the
agement is needed for battery modules. Advanced management systems
vehicle throughout the battery service life [2].
take different forms, including centralized systems and distributed
Battery management technologies have gone through three main
* Corresponding author.
E-mail address: [email protected] (X. Hu).
https://doi.org/10.1016/j.rser.2020.110480
Received 14 June 2020; Received in revised form 28 August 2020; Accepted 16 October 2020
1364-0321/© 2020 Elsevier Ltd. All rights reserved.
Please cite this article as: Haifeng Dai, Renewable and Sustainable Energy Reviews, https://doi.org/10.1016/j.rser.2020.110480
H. Dai et al. Renewable and Sustainable Energy Reviews xxx (xxxx) xxx
Abbreviations ML machine learning

MPC model predictive control
AC alternating current OCV open-circuit voltage
AHI ampere-hour integral P2D pseudo-two-dimensional
AI artificial intelligence PCM phase change material
ANN artificial neural network PF particle filter
BMS battery management system RES renewable energy system
BTMS battery thermal management system RUL remaining useful life
CC constant-current RUM remaining useful mileage
CCCV constant-current constant-voltage SAE Society of Automotive Engineers International
ECM equivalent circuit model SEI solid electrolyte interphase
EIS electrochemical impedance spectroscopy SOC state of charge
EKF extended Kalman filter SOE state of energy
EN European Norm SOH state of health
EV electric vehicle SOP state of power
FL fuzzy logic SOS state of safety
FSR functional safety requirement SOT state of temperature
GA genetic algorithm SPKF sigma-point Kalman filter
IEC International Electrotechnical Commission SVM support vector machine
ISO International Organization for Standardization UKF unscented Kalman filter
LIB lithium-ion battery UL Underwriter Laboratories
systems. The advanced management system focuses on improving the the energy society, which consists of numerous EVs, charging stations,
battery performance and the user’s driving experience, and enables the and power plants [5], and the features of this generation are data and
monitoring of battery system dynamics. Battery modeling and state intelligence. This generation of battery management focuses more on
estimation, thermal management, battery equalization, charging con battery lifecycle management, from manufacturing to recycling. With
trol, and fault diagnosis are all possible with the appropriate optimiza smart sensing and intelligent management, more accurate models and
tion algorithms and control strategies [4]. In the later development of advanced algorithms can be applied in battery management.
advanced management systems, battery safety and aging are also This literature review highlights the advanced battery management
considered. technologies in achieving high safety and long cycle life for high-
Advanced management represents the leading technologies in cur energy/density battery packs and discusses the outlook for the next-
rent battery management; however, there is still much room for further generation battery management. This review aims to fulfill the
development. In addition to the thousands of batteries already installed following objectives, which are also the original contributions of the
in vehicles, the batteries in the entire energy society are also the man paper as well: 1) conducts a bottom-up and multilayer structure analysis
agement objects of “next-generation management” systems. Next- of advanced battery management technology; 2) makes a comprehen
generation management will serve as the vital link between EVs and sive elaboration of battery management from both academic and
Fig. 1. The evolution of battery management technology.
2
engineering perspectives; 3) suggests a novel insight of the next battery internal). The application layer is the “brain” of the BMS and includes six
management based on smarting sensing, big data, blockchain and arti main tasks (safety management, aging management, balance manage
ficial intelligence (AI) technology. The remainder of this paper is orga ment, thermal management, charging management, and fault
nized as follows. The advanced battery management at multiple layers, diagnosis).
including the foundation layer, algorithm layer and application layer,
and the problems and challenges are explored in Section 2. Section 3 2.1. Foundation layer
discusses the trends of battery management technology development
based on data and intelligence. Some conclusions are provided in the The foundation layer serves as the cornerstone of battery manage
final section. ment technology and mainly involves the physical and theoretical
foundation. This layer is divided into six modules: system design, system
2. Advanced battery management verification, battery-related standards, data acquisition, data process
ing, and mechanism and simulation. The system design impacts the
The requirement decomposition from EVs to battery management is upper layers significantly; for instance, the performance of the micro
shown in Fig. 2, in which the arrow represents the requirement inheri processor affects the speed of algorithm execution and the design of the
tance path mainly related to battery management. The performance BMS hardware determines the equalization patterns. Sufficient verifi
requirements of EVs are decomposed into the performance requirements cation improves the reliability of the battery system, and battery-related
of the battery system, motor system, etc. Furthermore, battery man standards guide the application and testing of the battery system in
agement requirements are derived from the battery system’s perfor detail. Another indispensable requirement of the foundation layer is to
mance requirements, which focus on battery safety, aging, thermal, provide more external measurable data for the upper layer. The inves
charging, equalization, and fault management. tigation of the battery working and failure mechanisms provides the
The current advanced battery management technology can basically theoretical basis for battery modeling and management algorithm, and
meet the above requirements. Based on the hierarchical and modular the offline simulation based on battery data guides the online applica
concept of system development, the current advanced battery manage tion of the battery model and algorithm. The design of the foundation
ment technology can be divided into three levels according to bottom-up layer provides a technical basis for achieving safe and efficient battery
data flow: a foundation layer, an algorithm layer, and an application management technology.
layer, as shown in Fig. 3. The foundation layer focuses on the system
physical basis and theoretical principle. A BMS with excellent perfor 2.1.1. System design and verification
mance is achieved through detailed design and sufficient verification. In addition to the focus on BMS, it is necessary to take the internal
This layer provides the necessary measurement data for the upper layer structure of the battery system into account carefully. The composition
and offers battery mechanism and simulation as the basis for algorithm of such a battery system is shown in Fig. 4 (a). It usually includes a
and strategy design. Battery-related data plays an essential role, and the battery module, battery thermal management system (BTMS), BMS,
variety, sampling frequency, and complexity of data will determine the electrical sub-system, and mechanical sub-system.
execution of the upper-layer function. The algorithm and application Battery selection and evaluation is the first step of the module design,
layers focus more on the online application of battery management. The and battery type, dimensions, and performance parameters should be
algorithm layer is independent of battery type and, by abstraction and consistent with EV performance requirements. Proper sorting and
encapsulation, it can be adapted to a variety of batteries. The algorithm screening are necessary to reduce inconsistency between batteries.
layer is the core of a BMS. It performs battery modeling, parameter During the batteries grouping phase, their series and parallel modes
identification, estimation, and prediction of battery internal states. The should be determined. The BTMS must provide effective heat dissipation
primary task of the algorithm layer is to provide the application layer at high temperatures and avoid thermal runaway by adopting liquid or
with all important information about the battery (both external and air cooling. In addition, batteries should be warmed up at a low
Fig. 2. Performance requirements from EVs to battery management.
3
Fig. 3. Hierarchical architecture of the advanced battery management technology.
Fig. 4. Basic structural components of a battery system and development process for BMS: (a) composition of a battery system, including module, heat
exchanger, BMS, high-voltage system, and case; (b) the basic development process for the BMS at the system, software, and hardware levels.
temperature for optimal efficiency [6,7]. In high-voltage systems, the safe state rather than catastrophic failure due to random failure, system
electrical design and harness design are the main concerns, in which failure, or common-cause failure. The ISO 26262 standard is designed to
high-voltage faults must be detected rapidly. Such designs may include address functional safety issues and provide guidance for hardware and
an automatic cut-off function to prevent short circuits and electric software development. Fig. 4 (b) shows the basic development process
leakage. When designing the mechanical aspects of the system, the coupled with functional safety activities for the BMS, which mainly in
battery pack should have minimal weight and volume, and good shock cludes four phases: concept, system, hardware, and software phases.
and impact resistance. Except for the concept phase, all other phases consist of two main parts:
The battery system places very high requirements on the BMS, which design and verification.
is responsible for battery safety and performance. To ensure reliability in During the concept phase, the top-level requirements from the bat
automobile applications, detailed development and full verification are tery system are analyzed, and then the hazard analysis and risk assess
essential at different levels. Functional safety provides the system with a ment is conducted, which is to analyze the hazards of items and
4
formulate the safety goals. The functional safety requirements (FSRs) Table 1
will be determined to ensure the realization of safety goals. During the Major safety standards for EV batteries.
system phase, the FSRs are further refined into technical safety re Standards Scope of Standard Test items
quirements. The safety-related requirements and other non-safety- system application number
related requirements constitute the development requirements for the ISO Whole vehicle ISO 6469-1 On-board rechargeable
system phase. The system requirements will be decomposed and allo [9] energy storage system
cated to software and hardware, and then the design, verification, and ISO 6469-2 Vehicle operational safety
integration of software and hardware will be implemented. During the [10] means and protection against
failures
development of the software, layered software architecture is usually ISO 6469-3 Protection of persons against
adopted, which consists of the application software module responsible [11] electric shock
for algorithm and control strategy, and the basic software module ISO 6469-4 Post crash electrical safety
related to the BMS hardware. A layered software architecture helps to [12]
Battery packs or ISO 12405- Performance testing for
reduce the software complexity and ensure the normal operation of the
systems 4 [13] battery packs and systems
software. IEC Lithium-ion cell IEC 62660- Performance testing
1 [14]
2.1.2. Battery system standards IEC 62660- Reliability and abuse testing
The rapid development of EVs is inseparable from the support of laws 2 [15]
IEC 62660- Safety requirements and test
and standards. The safety standards for LIBs for EVs are shown in 3 [16] methods
Table 1. Currently, the international organizations that standardize EVs SAE Whole vehicle SAE J1766 Recommended practice for
consist of the International Organization for Standardization (ISO), the [17] electric, fuel cell and hybrid
International Electrotechnical Commission (IEC), and the Society of electric vehicle crash integrity
testing
Automotive Engineers International (SAE) [8]. ISO focuses on vehicle
Traction battery SAE J2464 Electric and hybrid electric
applications, and current standards related to LIBs include ISO 6469 systems [18] vehicle rechargeable energy
[9–12] and ISO 12405-4 [13]. Particularly, ISO 6469 provides a series of storage system safety and
requirements for the whole-vehicle safety of EVs. The performance abuse testing
testing methods formulated in ISO 12405-4 are to evaluate the basic SAE J2929 Electric and hybrid vehicle
[19] propulsion battery system
characteristics of battery systems, including performance, reliability,
safety standard - lithium-
and functionality, which helps vehicle manufacturers compare battery based rechargeable cells
systems provided by suppliers. The traction battery standards published EN Whole vehicle, EN 1987-1 Onboard energy storage
by the IEC include IEC 62660 family [14–16], which are mainly for component, and [20]
traction battery EN 1987-2 Functional safety and
performance testing, reliability and abuse testing, as well as the safety
[21] protection against failure
requirements of LIBs. The standards established by SAE are highly EN 1987-3 Protection of users against
authoritative and widely adopted by the automotive and other related [22] electrical hazards
industries. SAE J1766 [17] focuses on the crash integrity testing of EVs, U.S. Traction battery DOE/ID- Battery test manual for
while SAE J2464 [18] and SAE J2929 [19] are safety standards for systems 11069 [23] power-assist hybrid electric
vehicles
battery, especially the battery abuse testing requirements are presented
UL 2580 Batteries for use in EVs
in SAE J2464. The European Norm (EN) is the primary regional stan [24]
dard, referring to ISO and IEC standards. The standards concerning LIBs China Cells and modules GB/T Cycle life requirements and
are mainly EN 1987 series [20–22]. 31484 [25] test methods for traction
batteries of EV
At the same time, countries around the world are actively promoting
GB/T Safety requirements and test
their own national standards. Influential standards include the U.S. 31485 [26] methods for traction batteries
standards from U.S. Council for Automotive Research and Underwriter of EV
Laboratories (UL). German standard systems include German Institute GB/T Electrical performance
for Standardization and German Association of the Automotive Industry, 31486 [27] requirements and test
methods for traction batteries
Japanese standard systems include Japanese Industrial Standards and
of EV
Japan Electric Vehicle Association, and the Chinese standard system Battery packs or GB/T Test specification for high
includes the Standardization Administration of China. The battery test systems 31467.1 power applications
manual [23] published by the U.S. Council for Automotive Research has [28]
GB/T Test specification for high
been widely used to characterize the performance of energy storage
31467.2 energy applications
devices. UL 2580 [24] is applied to evaluate the safety of power battery [29]
systems under simulated abuse. In the field of LIBs, the Chinese standard GB/T Safety performance
systems are mainly concerned with standards for cells and modules. The 31467.3 requirements and test
standards include GB/T 31484 [25], GB/T 31485 [26], and GB/T 31486 [30] methods
BMS GB/T Technical requirements, test
[27], which regulate battery performance and safety. The GB/T 31467
38661 [31] methods and inspection rules
[28–30] standard has been established for battery packs and battery
systems, including test specifications for high power/energy applica
tions and safety-related test methods. Moreover, aiming at technical there is still a long way to go to establish a perfect standard system.
specifications of BMS for EVs, the GB/T 38661 [31] is constructed to
standardize the technical requirements, test methods and inspection 2.1.3. Data acquisition
rules. Most BMS algorithms use battery measurements as input; therefore,
With the rapid growth of the EVs and RESs, some deficiencies in the data acquisition technology is essential [32]. The accuracy and speed of
standard systems have been exposed. Updates of standards often lag data acquisition are also important factors affecting the performance of
behind technological development. As a result, the current LIB standards a BMS.
are not enough to cover the entire LIB industries. In addition, different Cell voltage, cell temperature, and load current are the typical
standards usually have different requirements, which will hinder the measurements collected by the BMS. At present, cell voltage data is
standardization and platformization of the battery system. Therefore, collected by specialized BMS measurement chips, which feature high
5
precision, fast transmission, and low cost. Commonly used integrated database with minimum cost. Using heuristic algorithms to repair data
chips can guarantee the voltage sampling accuracy within 2 mV [33]. may introduce errors [42]; thus, users usually have to be involved in the
Temperature data can be acquired by the thermistor or thermocouple repair process to avoid unwanted errors [43]. In some cases, value
installed on the battery. Current data can be acquired by Hall current confidence, instead of users, is used to guide this process [44].
sensors, fiber optic current sensors, or shunts [32]. ML techniques are also relevant to data cleaning. Krishnan [45]
Generally, two BMS architectures are available: a centralized BMS proposed a data cleaning approach, ActiveClean, to update both clean
and a distributed BMS, as shown in Fig. 5. For the centralized BMS, the data and ML models. Additionally, some research [46] has been pro
sampling units, including the cell voltage and temperature sampling posed to assess the remaining undetected errors in a dataset, thereby
units, together with other functional units, are placed on one controller evaluating the effectiveness of data cleaning.
unit. For the design of the distributed BMS, several cells are considered As a significant part of data handling, data cleaning greatly in
as one group, which is managed by a slave electronic control unit. The fluences data analysis. Furthermore, the cost should be considered when
slave unit measures the cell-related data (cell voltage and temperature). designing a data cleaning approach or system. Such systems should be
Several slave units will transfer these data to the master electronic specifically designed for individual cases. Recently, some researchers
control unit through the internal communication bus. In addition, the have proposed that EVs data processing can be achieved through the
master unit is responsible for other functions, including the execution of MapReduce feature of Hadoop, a programming model. Nevertheless,
BMS algorithm, control, and communication with other onboard there are only a few studies focusing on the cleaning of EV battery data.
controllers. It is important to build a data quality assessment and cleaning system for
In the future, more types of data related to battery management will battery data obtained from multiple sources. In addition, the computa
be measurable, such as battery stress and ultrasonic information. This tional complexity of data processing needs to be considered. This can be
novel data will innovate BMS measurement chips and system evaluated in two aspects: time complexity (the time spent in data pro
architecture. cessing) and space complexity (the storage cost for data processing).
Some researchers use simplified models or algorithms by sacrificing
2.1.4. Data processing and cleaning certain prediction accuracy to reduce computational complexity.
Large datasets from multiple sources are valuable for analysis and Finding a balance between accuracy and computational complexity will
machine learning (ML) and may provide new approaches to battery become a top priority.
management. However, real data, including battery data from multiple
sources, inevitably contain errors, such as missing values and mixed 2.1.5. Battery mechanisms and simulation
formats, which may reduce data quality and result in unreliable analyses Rechargeable LIBs provide excellent performance. A LIB consists of a
and decisions. Data cleaning can be applied to solve various data quality positive electrode, a separator, a negative electrode, and corresponding
problems, including duplicates, missing values, integrity constraint vi current collectors [47]. Lithium ions go back and forth between the
olations, and outliers [34]. Two data cleaning approaches (error negative electrode and positive electrode during the charging/di
detection and data repairing [35]) are introduced in detail as follows. scharging process. The basic electrode process during battery char
Error detection: A lot of work has been done to detect errors using ging/discharging is divided into electronic conduction, electrode
integrity constraints [36]. Although the integrity constraints approach reaction, diffusion in the solid phase, electrolyte phase lithium-ion
can capture errors that violate constraints, it is unable to distinguish transfer, and diffusion [48,49]. Battery operation involves not only
between correct and incorrect values. In addition, some data quality electrochemical behaviors but also thermal processes and performance
rules, including functional dependencies [37], inclusion dependencies degradation. Researches on battery working mechanisms contribute to
[38], conditional functional dependencies [39], and denial constraints the establishment of battery models, and battery models related to
[40], must be taken into account. Conditional functional dependencies different battery characteristics are presented in Section 2.2.1.1. In
can capture the inconsistencies that traditional functional dependencies addition to primary working mechanisms, the failure mechanisms,
may miss [39], while denial constraints generalize and enlarge the including aging, lithium plating, and thermal runaway, are also the fo
current class of constraints used for data cleaning [40]. Furthermore, cuses of the current research in battery management. Considerable
Wang [41] proposed that the use of fixing rules is efficient in identifying research efforts have been dedicated to understand failure mechanisms
incorrect values. and then to alleviate or avoid these failures [47,50–52]. In-depth study
Data repairing: Heuristic algorithms, along with user guidance, of battery failure mechanisms can help improve battery design and
value confidence, and fixing rules, are commonly applied. A large body provide guidance for scientific battery management.
of work has aimed to design a consistent database to replace the original Capacity loss and power loss are the main aging mechanisms for LIBs.
Fig. 5. The hierarchical architecture of a BMS: (a) centralized BMS; (b) distributed BMS.
6
Capacity loss can be further divided into the loss of lithium inventory parameter identification technique will be employed to determine the
and active material [53]. The former occurs when solid electrolyte unknown model parameters.
interphase (SEI) film grows on the negative electrode surface when LIBs
cycle. There are many causes for the loss of active material, including 2.2.1.1. Battery multi-domain modeling. The existing battery models
damage to the crystal structure of the electrode material and changes in consist of four categories: electrical, thermal, aging and coupled models,
the composition of the electrode material. Power loss is mainly caused which are shown in Fig. 6.
by the increase of interface contact resistance, increase of membrane Electrical models include three types: black box, equivalent circuit,
resistance of active materials, or reduction of inactive surface area. In and electrochemical models [61,62], as shown in Fig. 6 (a). The black
the actual operation of a battery, the factors mentioned above are often box models [62,63] describe a linear or nonlinear mapping function of
coupled and interact with each other, finally leading to battery perfor battery voltage response characteristics. It ignores internal mechanisms
mance and health degradation [54]. and has no physical meaning. Equivalent circuit models (ECMs) [64,65]
Current research on the mechanisms of lithium plating remains adopt simple electronic elements to describe battery electrical behav
conflicting. One opinion is that when lithium ions are inserted onto the iors, and the resistance-capacitance network based ECMs are widely
surfaces of graphite particles faster than the rate at which they can investigated. The ECMs contain relatively fewer parameters and are
diffuse inside those particles, lithium ions could accumulate on the widely used in online applications, but they lack physical significance
surface. When the lithium-ion concentration reaches saturation, lithium and can only provide macro-level physical quantities [66]. Electro
plating will occur [55]. Another opinion is that lithium plating occurs chemical models [67] use partial differential equations to characterize
when the negative electrode potential drops to the equilibrium potential the dynamics of electrodes and electrolytes based on internal mecha
of lithium deposition reaction [56]. nisms. They can accurately predict physical quantities but have very
As seen in recent EV incidents, thermal runaway is mainly caused by high computational cost. Hence, the traditional electrochemical models,
abuse conditions, which include mechanical, electrical, and thermal for example, pseudo-two-dimensional (P2D) models are not suitable for
abuses. All these conditions, as well as battery aging, will lead to various online application, and some simplified versions of the P2D models are
degrees of the internal short circuit, which generates heat [57]. The being developed [68].
abnormal heat generation raises the battery temperature and initiates Because temperature impacts battery life and safety, establishing an
side reactions, which release more heat. Ultimately, a accurate thermal model is critical. There are three types of thermal
heat-temperature-reaction loop is formed and cycles at an extremely models presented in Fig. 6 (b). Lumped parameter thermal models [69,
high temperature until the cell triggers thermal runaway [58]. 70] only consider the average temperature of the battery and are
The reaction mechanisms inside batteries are complex, and a applicable to low-rate charging/discharging. Dual-state thermal models
detailed description of the battery will make the battery model very [71,72] consider the core and surface temperatures of a battery, and are
complicated. Generally, these complex models are difficult for online conducive to real-time control and utilization. However, they assume
applications due to the limit of currently available computing ability, that all battery heat comes from the battery core, which will cause the
whereas simplified and approximate models can be used for online ap core temperature to be overestimated. Distributed thermal models [59,
plications. For the complex models, offline simulation can deepen the 73] over two or three spatial dimensions have high accuracy but involve
understanding of the battery and provide guidance for online applica multiple, coupled, partial differential equations that are difficult to
tion. Many researchers have conducted battery-related simulation implement directly.
studies based on battery mechanisms. For example, a three-dimensional Batteries age gradually during use, resulting in the attenuation in
thermal model in computational fluid dynamics based on a partial dif capacity and power. Aging models have been proposed to simulate the
ferential equation thermal model was established to simulate heat dynamics of battery aging. Fig. 6 (c) presents three aging mechanisms:
transfer processes and describe the internal temperature distribution of a lithium plating, film growth, and loss of active material [54,74]. The
battery [59]. A physics-based aging model was used to investigate the physics-based aging model [60,75] has high accuracy but requires
calendar aging and internal impedance of a battery cell using COMSOL complex calculations. The semi-empirical model [76] only provides a
software [60]. simple physical interpretation of battery aging mechanisms but has
moderate complexity and uses intuitive expressions. Lastly, empirical
2.2. Algorithm layer models [77] are mostly based on data and do not consider internal
mechanisms. The semi-empirical model and empirical model are
The optimal battery management technology requires a compre preferred in the prediction of battery remaining useful life (RUL).
hensive understanding of LIBs, not just making management decisions Temperature directly affects the battery electrical characteristics and
based on external measurements. Therefore, the primary task of the aging rates. High temperatures accelerate the growth of SEI films, while
algorithm layer is to understand the battery in multiple domains and at low temperatures tend to trigger lithium deposition. Some parameters of
multiple scales. Here, the multiple domains include battery electrical, the electrical characteristics also affect the heat generation of a battery.
thermal, aging, and further coupled domains, while multiple scales refer Degradation reduces capacity and increases internal resistance, which
to investigating the battery from different spatial dimensions (cell to changes the electrical and heat generation characteristics. Fig. 6 (d)
pack), dynamic dimensions (microsecond to year), and time dimensions shows a basic diagram of the electro-thermal-aging coupled process,
(present to future). This layer mainly consists of battery modeling, state where the electrical, thermal, and aging characteristics of batteries are
estimation, and state prediction. The control-oriented modeling and connected. Accordingly, coupled models have been established,
parameter identification should be the prerequisites for battery state including the electro-thermal coupled model [71,78] and
estimation and prediction, and upper layer battery management. Then, electro-thermal-aging coupled model [79]. The reaction mechanisms
the estimation and prediction of battery states can perceive current inside batteries are complex, and it is a non-trivial task to develop a
battery states and plan future use scenarios. Achieving the above func single, high-fidelity, yet computationally efficient model. The available
tions is the basis for the management strategies used in BMS. electrochemical model can accurately describe dynamic electrical re
actions but suffers from significant model-plant mismatch when con
2.2.1. Battery modeling fronted with large changes in battery temperature and health. Therefore,
Establishing a high-fidelity battery model is a critical challenge in the multi-domain coupled battery models are necessary to describe the
algorithm layer. The accuracy of a model directly affects the online state battery electrical, thermal, and aging behaviors simultaneously. Another
estimation and management strategies. The battery electrical, thermal, problem with the multi-domain coupled model is the different time
and aging characteristics determine the form of battery models, and the
7
Fig. 6. Battery models. Schematic illustration of (a) electrical models; (b) thermal models; (c) aging models; (d) coupled model.
scales. Usually, the time scale for the electrical model is fast, while the as improve the model accuracy.
aging model is relatively slow. The reasonable choice of the time scale
for different domain models will reduce the coupling complexity as well 2.2.1.2. Model parameter identification. Parameter identification is one
Fig. 7. Common classifications and methods of parameter identification for BMS. GITT: galvanostatic intermittent titration technique; PITT: potentiostatic
intermittent titration technique; LS: least squares; RLS: recursive least squares; RTLS: recursive total least squares.
8
of the key steps in battery modeling and contributes significantly to the (UKFs) [121–124], particle filters (PFs) [97,101,125,126] and
accuracy and reliability of battery models and state estimation. Many sigma-point Kalman filters (SPKFs) [127–129] to achieve the online
techniques have been proposed for parameter identification and can be estimation of battery states. The accuracy of these methods is dependent
classified into three categories: experimental measurement, offline on the precision of the model adopted, and they have the potential to be
identification, and online identification. These classifications and com used for real-time estimation as they have relatively low computational
mon methods of parameter acquisition in BMS are shown in Fig. 7. complexity. The latter methods are data-driven and use artificial neural
To obtain the parameters of an electrochemical model, experimental networks (ANNs) [100,130–134], support vector machines (SVMs)
measurement methods can be applied. Weppner and Huggins [80] [108,135–137], and fuzzy logic (FL) [138–142] to learn from training
proposed an electrochemical galvanostatic intermittent titration tech data for battery state estimation. The deep knowledge of battery
nique to obtain the diffusion coefficients and other kinetic properties. mechanisms is not necessary for these methods, and the accuracy of
Subsequently, Ho et al. [81] described a steady-state alternating current these methods depends strongly on the quantity and quality of input
(AC) technique to identify the important kinetic parameters of the battery data.
rate-limiting process. Methods such as the potentiostatic intermittent In addition to these common estimation methods, there are also
titration technique [82] are also used to measure battery electro estimation algorithms tailored to each battery state. For example, SOC
chemical parameters. These methods often provide accurate measure estimation can be achieved by the ampere-hour integral (AHI) method
ments of parameters that have physical meaning. However, it takes a [143–145], the open-circuit voltage (OCV) method [146–151], EIS and
substantial amount of time to measure these parameters, and precision internal resistance measurements [152–155], or multiple primary al
instruments are needed. gorithm fusion implementations [156,157]. SOP estimation can be
Offline parameter identification methods are preferred because they indirectly calculated from battery models based on estimates of other
have the potential to identify parameters accurately and quickly through states (SOT/SOC/SOH) [158,159], or directly estimated online based on
specific test conditions such as constant current discharge and hybrid interpolations in a map established from battery external characteristic
pulse power characteristics experiments. At present, optimization al parameters [160]. SOE estimation can be achieved with a novel and
gorithms, including particle swarm optimization [64] and genetic al simple method based on short-term cycling history [96]. SOT estimation
gorithms (GAs) [83–85], are widely used in offline identification. The can be achieved through impedance-based indirect estimation [161].
electrochemical impedance spectroscopy (EIS) approaches can also be SOH can be estimated directly with the empirical models proposed by
applied in the identification of ECM parameters [86]. Offline identifi various researchers [162–165], or indirectly by utilizing identified
cation methods can obtain model parameters only by directly measuring battery parameters such as available capacity and impedance [93,138,
the battery’s external characteristics (voltage, current, temperature, 166–170]. For SOS estimation, a qualitative analysis of battery
etc.). A shortcoming of experimental measurement and offline identifi safety-based hazard levels [106,107] is proposed, in which the hazard
cation is that the model parameters are constant at a specific battery level is determined by the severity and likelihood of hazards [106].
state. Nevertheless, some parameters may fluctuate due to changes in The state prediction of batteries mainly refers to RUL and RUM
battery states; therefore, using constant parameters identified offline prediction. Battery RUL means the remaining number of charging/dis
may lead to biased results. A practical solution is to update the param charging cycles, from the observation time to failure under certain
eters by interpolation or lookup table during the online application. operating conditions [110–115]. RUM refers to the remaining energy,
To maintain the model accuracy under broad operating conditions, which depends on battery state estimates. Changes in battery states,
online identification methods can identify model parameters in real parameters, and voltage output during future operations need to be
time. These methods include recursive least squares [87], moving win comprehensively considered [116]. Many researchers have conducted
dow least squares [88], and recursive total least squares methods [89]. comprehensive reviews on battery lifetime prediction [92,171–173].
State observers and filters such as dual extended Kalman filter (EKF) RUL prediction methods include empirical statistical and aging
[90] can also update model parameters in combination with online state performance-based methods [92], and the former is a direct prediction
estimation. Furthermore, in some research [91], the slowly varying while the latter is an indirect prediction. The empirical statistical
characteristics of model parameters are taken into account, and method includes the cycle number method, the AHI method and its
multi-scale observers are designed to update parameters at the macro associated weighting method, and the event-oriented aging accumula
scopic scale. With the continuous improvement of hardware and pro tion method. This method can only make rough estimates of battery RUL
cessors, it is becoming more feasible to apply online identification [174]. Aging performance-based methods include model-driven
algorithms, and the online acquisition of parameters may become methods based on battery aging mechanisms and data-driven methods
common. based on battery data and intelligent algorithms [92,171–173,175]. The
model-driven method is based on battery aging models and incorporates
2.2.2. State estimation and prediction adaptive PF [111,114,176–179] or UKF [180,181] algorithms to predict
State estimation and prediction of the battery system are critical RUL. Data-driven approaches are based on battery degradation data,
functions of the BMS and are the basis for strategy research. The related with SVM [182], relevance vector machine [115,183,184], Gaussian
research focuses on estimations of state of charge (SOC) [2,33,92–95], processing regression [185,186], ANN [187,188] and other ML algo
state of power (SOP) [92,93], state of energy (SOE) [96–101], state of rithms to learn from data and predict battery RUL. At present, a few
temperature (SOT) [71,102–105], state of health (SOH) [2,92,93], and research attempts have been made to predict RUM, and the existing
state of safety (SOS) [106–108], and their general definitions are sum methods can be divided into two types [116]: 1) indirect prediction
marized in Ref. [109]. Among these, the estimation of SOC is the most according to battery residual discharge energy and future average con
essential and well studied. SOS estimation is the current hot research sumption [189] and 2) direct prediction based on clustering-based
field but with great difficulty, which mainly focuses on the assessment of methods and Markov-Monte Carlo methods [190–192]. The above
battery safety status. Subsequently, to meet battery life and EV cruising state estimation and prediction methods for battery systems are sum
range requirements, there have also been studies on the prediction of marized in Table 2.
battery RUL and remaining useful mileage (RUM) [92,110–116]. Plenty of studies have investigated the algorithms for state estima
Existing state estimation methods have a lot in common and can be tion and prediction. Nevertheless, battery states are coupled and inter
separated into two main types: model-based methods combined with acted, and single-state estimation without considering the impact of
optimization algorithms, and data-driven methods combined with ML other states has limitations and cannot adapt to dynamic operating
algorithms [2,33,92,94]. The former is based on battery mechanism conditions and a complex external environment. Multi-state joint esti
models or ECMs using EKFs [99,102,117–120], unscented Kalman filters mation (at least two battery states) has attracted increasing attention,
9
Table 2 SOT estimation is at an intermediate time scale. Battery aging is a

Classification of battery state estimation and prediction methods. long-term process, and a long-term time scale is suitable for SOH esti
Categories States Unique Algorithms References Common mation. Lastly, the time scale for SOS estimation is adaptively adjustable
Algorithms because battery safety assessment needs to be dynamically adjusted
Estimation SOC AHI method [143–145] 1. Model-based according to other battery states. For the state prediction, RUL and RUM
OCV method [146–151] methods (EKF [99, prediction are in the future phase and depend on the estimation of SOH
EIS/internal [152–155] 102,117–120], and SOE. In addition to state estimation with different time scales, the
resistance UKF [121–124], state estimation and prediction at different spatial scales are needed.
measurement PF [97,101,125,
method 126],
Cell inconsistency during grouping and operation poses a huge chal
Hybrid method [156,157] SPKF [127–129]) lenge to state estimation, and there is still a huge technical gap. At
SOP Model-based indirect [158,159] 2. Data-driven present, only a few studies of multi-state estimation take cell inconsis
estimation methods (ANN tency into account, for example, the joint estimation of SOC and SOE
Map method based [160] [100,130–134],
considering inconsistency [199], and joint estimation of SOC and ca
on external SVM [108,
characteristic 135–137], pacity for series connected batteries [200,201].
parameters FL [138–142]) Future research on state estimation and prediction for next-
SOE Cycling history- [96] generation battery management will take three directions. The first di
based direct rection is the complete realization of multi-state joint estimation and
estimation
SOT Impedance-based [161]
prediction, considering multi-domain coupled characteristics and cell
indirect estimation inconsistency. The second direction is to introduce the intelligent based
SOH Empirical model- [162–165] on big data algorithms into state estimation and prediction of the battery
based method system. Details about this direction are introduced in Section 3. The
Indirect estimation [93,138,
third direction is to make battery state estimation and prediction algo
166–170]
SOS Hazard levels [106,107] rithms more robust, accurate, and fault-tolerant under complex working
Prediction RUL Empirical statistical [174] – environments.
method
Model-based method [111,114,
2.3. Application layer
176–181]
Data-driven method [115,
182–188] In practice, a complete BMS needs to deal with the multiple in
RUM Indirect prediction [189] fluences of operating conditions, the external environment, and its own
method internal problems. Therefore, advanced management strategies are
Direct prediction [190–192]
method
required to ensure that the system runs safely and efficiently. Therefore,
advanced management strategies are required to ensure the safe and
efficient running of the battery system. The application layer consists of
and preliminary progress has been achieved [109]. The co-estimation of safety management, thermal management, charging management,
battery SOC and SOH [193–195] is well studied because the update of equalization management, aging management, and fault diagnosis. Its
SOH plays an important part in improving the SOC estimation accuracy. primary goals are to optimize the thermal, electrical, aging, and safety
There are also some contributions to the co-estimation of SOC and SOP performance of the battery system.
[196,197], SOC and SOE [101], and SOC and SOT [198]. Fig. 8 shows a
diagram of multi-state joint estimation and prediction, in which several 2.3.1. Thermal management
states at different scales and phases interact with each other. The state To achieve excellent performance, battery systems need to operate
estimation focuses on the battery states in the present phase, and within a specific temperature range. Temperatures that are too low or
different states are with different time scales. SOC, SOP, and SOE esti too high will not only cause battery performance degradation, but also
mations based on battery electrical behaviors are performed in real time lead to irreversible damage. A BTMS adjusts the battery temperature to
due to the strong real-time requirements for dynamic characteristics. an appropriate range to achieve the desired battery performance, which
The battery thermal characteristic is relatively slower; therefore, the is to heat batteries in low temperatures and cool batteries that are too
Fig. 8. A diagram of battery multi-state joint

estimation and prediction. The diagram includes
the present and future phases, corresponding to the
state estimation and prediction. Four time scales
(real-time, intermediate, long-term, and adjustable
scales) are involved in the state estimation. Ts, battery
surface temperature; Ta, battery average temperature;
Ti, battery internal temperature; Pm, battery
maximum available power; Q, heat generation; Cm,
battery maximum available capacity; Re, battery
equivalent resistance. Reprinted with permission
from Ref. [109] (Copyright 2019, Elsevier).
10
hot, as well as rebalance the pack temperature distribution if the thermal performance is limited by their poor thermal conductivity of PCMs.
inconsistency occurs. Improving the thermal conductivity of composite PCMs is needed for
Thermal management-related technologies include BTMS system their wide application in battery systems, which has attracted increasing
design and BTMS online control strategy. The former provides sufficient research interests [210,211].
heat dissipation capability, while the latter controls the temperature Additional strategies for warming up batteries include resistance
range and distribution, as shown in Fig. 9 (a). After the battery pack is heating and internal heating. For resistance heating, metallic resistance
determined, the BTMS heating/cooling medium and assembly compo and positive temperature coefficient resistance can be used as heat
nents will be selected, which will affect the effectiveness of thermal sources to directly heat a battery system [212]. These resistances can be
management. Offline flow field design and simulation guide the BTMS placed at the bottom of a battery module or between every two indi
design. During online control, the battery temperature is measured by vidual batteries. Internal heating typically applies current such as direct
the BMS first, then the BMS sends a heating/cooling command to the current, pulse current, or AC to heat the inner of battery, thereby
BTMS to manage the pack temperature. Due to the influence of battery warming up the battery uniformly from low temperature [202,213,214].
temperature, the charge/discharge power limits are adjusted accord By optimizing the current parameters, better heating effects and less
ingly to protect the battery. damage to the battery can be achieved. Certain battery structures, such
The medium plays a significant part in a BTMS, and battery thermal as self-heating LIBs, can provide rapid heating [215]. Current
adjusting can be achieved by air, liquid, heat pipes, or phase change low-temperature management strategies heat batteries to a preset value.
materials (PCMs), as summarized in Fig. 9 (b) ~ (e). In some research, however, the battery maximum output power is
Air medium. A BTMS usually uses airflow across the battery surface adopted as an indicator of the heating process [216–219].
to transfer heat. For battery heating, the air is heated by a heater and In conclusion, BTMSs are essential for maintaining battery packs
circulated in an enclosed space by a fan [202]. Heat from the hot air is within the appropriate temperature range. Currently, the tradeoff be
transferred from the battery surface to its core. At high temperatures, tween weight, cost, complexity, and effectiveness is being considered in
cool air passes through the battery pack to remove the heat generated by BTMS designs.
it [206].
Liquid medium. Thermal management based on liquid is usually 2.3.2. Charging management
achieved by immersing the battery module in a heating/cooling fluid, or The charging time of LIBs is a pivotal limiting factor in the paradigm
by surrounding the module with jackets or plates containing such fluid shift from fossil-fueled vehicles to EVs. Simply increasing the charging
[207]. In the first case, the circulation of the liquid can improve the rate may cause serious safety issues, premature battery aging, and sig
thermal uniformity of the system. For the latter case, better hea nificant energy loss. LIBs have certain charging rate limits according to
ting/cooling can be achieved by optimizing the geometry of the jack their structure and electrochemical properties, internal dynamic states,
et/plate or the flow characteristics of the fluid [208]. and external conditions [220,221]. In addition to safety, charging time,
Heat pipes. Heat pipes are able to transfer heat by the latent heat of charge-induced health decay, and energy efficiency are important fac
evaporation and condensation of a coolant [205]. Heat pipes combined tors in battery charging. These factors can interact and conflict with each
with the former two kinds of media are often adopted in BTMSs [204, other during battery operation. Therefore, advanced charging manage
209]. ment must make an optimal tradeoff between these factors according to
PCMs. PCMs can absorb and release a certain amount of heat during the user requirements.
the melting and solidification processes. However, the heating/cooling Available charging strategies for LIBs can be categorized into three
Fig. 9. BTMS design and control technologies: (a) system design and online management strategies; (b) air medium: Reprinted with permission from Ref. [202]
(Copyright 2013, Elsevier); (c) liquid medium: Reprinted with permission from Refs. [203] (Copyright 2017, Elsevier); (d) heat pipes: Reprinted with permission
from Ref. [204] (Copyright 2015, Elsevier); (e) PCMs: Reprinted with permission from Refs. [205]. (Copyright 2016, Elsevier).
11
classifications: model-free, ECM-based, and physical model-based. The battery pack, and the difference in Coulomb efficiencies and self-
constant-current (CC) protocol and constant-current constant-voltage discharge rates of the cells. This will result in poor performance,
(CCCV) protocol are the most widely adopted model-free strategies in reduced total capacity, shortened life span, and even safety problems.
the industry, thanks to the simplicity of their design and implementa Hence, it is significant to investigate the equalization management of
tion. These two protocols are usually conservatively designed by sup LIBs to mitigate the performance differences between cells in the battery
pressing the initial current rate to protect battery health, which pack [236].
inevitably leads to a significant sacrifice in the charging rate. To alle Equalization management includes two research directions: equal
viate this problem, it is found that a fully discharged LIB can be charged ization strategy and equalization circuit topology. Equalization strate
with a large current in a short period without much battery degradation, gies are based on voltage [237], SOC [145,238], capacity [239], or a
and a boost-charging approach is proposed to shorten the charging time combination of these [240–242]. For the voltage-based strategies, the
[222]. Various multi-stage CC charging strategies were developed, goal is that all the individual cells’ voltage is consistent within a defined
where LIBs are first charged with a large current, then a sequence of range. This approach is commonly used in practical applications because
decreased current, which improves the tradeoff between charging time the control algorithm is easy to achieve and voltage is measured directly.
and battery degradation [223–227]. For example, the Taguchi method However, the equalization efficiency is limited by a single objective of
[223,224] is applied to optimize multi-stage CC charging strategies, voltage. The SOC-based strategy aims to make the individual cells’ SOC
including the number of stages, stage currents, and stage voltages. Liu consistent, which can maximize the battery pack capacity. The equal
et al. [227] used consecutive orthogonal arrays to find an optimal ization strategy based on battery capacity is built on the total capacity,
five-stage charging strategy and experimentally demonstrated that the rechargeable capacity, or releasable capacity. It is effective to reduce the
proposed method could significantly prolong battery lifetime and reduce charging time and increase the available capacity. However, battery
charging time. Well-developed CCCV families may improve charging SOC and capacity are not measurable and must be estimated. The ac
performance to some degree but, in essence, are heuristic, as battery curacy of these estimates is affected by many factors, such as the
dynamics and physical limits are not taken into account [221]. calculation method used, the model accuracy, the degree of aging, and
To overcome the drawbacks of model-free approaches, battery temperature.
models have been employed to develop advanced charging strategies. Equalization circuits are designed to achieve the energy consump
Model-based optimization and control algorithms are usually formu tion or transfer target given by the equalization strategy, in order to
lated to manipulate battery electrochemical reactions and then imple minimize the performance differences between each cell in battery
mented in an offline or online fashion. This process requires real-time packs. These circuits can be categorized by the type of energy con
knowledge of battery states. As stated in Section 2.2.2, most system sumption or transfer [243]. Energy consumption circuits remove the
states, such as core temperature and SOC, are unmeasurable. Therefore, excess charge from a fully charged battery until its equalization target
it is necessary to design state estimators for charging strategies. parameter is the same as that of the cell with the lowest charge or SOC
Well-designed ECMs are with relatively concise structures and can reference. Energy transfer circuits remove the charge from cells with
characterize battery electrical behaviors with accuracy [228], but are excessive charge and transfer it to cells with less charge [244]. Energy
intrinsically unable to describe the electrochemical and aging behaviors transfer circuits have different topologies depending on the components
[229]. A number of charging strategies synthesized from ECMs and that store the energy, such as capacitors, inductors, and other elements
optimization/control algorithms have been devised and can offer that control the switches or inverters. The common types of energy
improved performance [230,231]. However, without physical insights, transfer circuits are: chain-structures or switched-capacitors, trans
the solutions to these approaches are often sub-optimal. In more con formers, buck-boost or Cûk and other energy conversion circuits, and
crete terms, ECM-based charging strategies are conservative in some cascade multi-level half-bridge circuits [245]. Energy transfer methods
operating regions, resulting in underutilized battery capacity and have different types according to the energy flow between cell and pack
power, while they are aggressive in others, resulting in significant per [246].
formance and health degradation. Furthermore, safety cannot be guar At present, research into equalization circuits is relatively mature.
anteed with the associated fast-charging strategies. Energy-consuming circuits have been widely used in vehicles while
The gap between ECMs and non-damaging fast-charging strategies energy transfer circuits have not, due to their high cost and complexity
has motivated the use of physically meaningful battery models. It is of control. Therefore, research on equalization strategies is still the main
practical to use simplified physical models for real-time applications way to improve battery equalization. The focus is on achieving accurate
instead of the initial coupled electrochemical, thermal, and aging control under multiple constraints and establishing an equalization
models based on partial differential equations. Based on a simplified strategy that maximizes energy throughput during the battery’s service
electrochemical-thermal model, an open-loop optimum charging strat life.
egy was put forward to manage charging time and health-related con
straints [232]. Using the single-particle model in an isothermal case, a 2.3.4. Fault diagnosis
nonlinear model predictive control (MPC) for fast charging is presented Safe and effective battery system operation is guaranteed not only by
[233]. A one-step nonlinear constrained MPC charging strategy com its safety design, but also its safety management system. Therefore, fault
bined with a more complex battery model was presented in Refs. [234]. diagnosis is significant to improve the safety and reliability of a battery
Recently, by explicitly considering battery electrochemical, thermal, system. Based on hardware redundancy information, conventional
and aging dynamics via a reduced-order model, a linear-time-varying diagnosis methods involve measuring the same data via multiple hard
MPC is formulated to track SOC and SOH [235]. The developed ware components and making judgments based on simple majority
health-aware fast-charging strategy appears promising for significantly voting logic [247]. These methods may have the drawbacks of increased
reducing both charging time and battery degradation. The above studies cost, weight, volume, and power consumption, and decreased reliability
are valuable in the development of advanced battery charging man [248]. Currently, analytical redundancy based approaches have been
agement. Future research directions should focus on further improving widely employed in application [249], which mainly achieve these
computational efficiency, experimental model parameterization, and functions by analyzing redundant information of different hardware
algorithm validation. functions and comparing measurement output of the system model with
measurements from the sensors. These methods do not require addi
2.3.3. Equalization management tional hardware, but the uncertainty of the system, and the stability and
The performance parameters of each cell will become different after robustness of the model under various fault modes, should be considered
several cycles because of the uneven temperature distribution of the [250]. Detailed study and classification of diagnostic methods are
12
presented in Refs. [2,251]. The fault diagnosis method can be divided phenomenon. The information fusion method [273] tends to achieve a
into qualitative analysis methods and quantitative analysis methods, more reliable diagnosis by integrating and analyzing information from
and the quantitative analysis methods are further divided into analytical multiple sources. Some signal processing techniques, such as impedance
model methods and data-driven methods, which are shown in Fig. 10. spectroscopy [274–276], wavelet transform [277], and Shannon en
The common qualitative analysis methods can be divided into the tropy [278–280], have also made some progress in the fault diagnosis of
fault tree method [252], failure mode and effects analysis [253,254], LIBs, among which Shannon entropy is particularly applicable to
and the expert system method [255,256]. The first two methods abnormal fluctuations in a disordered system.
generally judge causality logically and perform early fault classification The aforementioned diagnostic methods have different advantages
and risk analysis. They have simple modeling, easy-to-understand re and disadvantages. Combining several optimal diagnostic algorithms
sults, and a wide range of applications. The expert system method uses may improve diagnostic performance. Data-driven intelligent algo
rich experience and heuristic knowledge instead of modeling. However, rithms are promising in fault diagnosis, and the deep-learning method
its diagnostic accuracy is affected by the quality of the expert’s experi combined with big data analysis is likely to become the favored diag
ence and knowledge. nostic method.
The analytical model-based approaches consist of state estimation,
parameter estimation, parity equation, and structural analysis theory 2.3.5. Safety and aging management
methods. A mathematical model established from the reconstructed Safety and aging are two major challenges of battery management, as
system and sensor measurements can be integrated to generate residual well as the primary concern of users for EVs. Battery safety issues are
information for diagnostic analysis [249]. Methods based on various directly related to human casualties, property loss, and environmental
observers and filters [257–260] are often used for state estimation and damage, while the aging issues affect the maintenance cost and service
parameter identification to strengthen the robustness of model-based life of EVs. Generally, safety management aims to protect the battery
methods [261,262]. The parity equation method [263] is not competi system from overuse, fire, explosion, leakage, and other hazards, and the
tive in nonlinear systems such as LIBs, and relatively few studies have purpose of battery aging management is to extend battery life. Battery
been conducted. Structural analysis theory [264–266] can analyze the safety and aging management are closely related and are located in the
isolability of various faults under the condition that the specific pa higher part of the application layer. The excellent operation of these two
rameters of the system are unknown, which remarkably reduces the management strategies requires the cooperation of the foundation layer,
workload of selecting the residual equation. However, this approach is algorithm layer, and other management strategies of the application
intensely dependent on the system’s analysis of the redundancy rela layer.
tionship. Combining structural analysis theory with other advanced Battery safety management focuses on thermal safety and high-
technologies [265] can achieve a better diagnostic result. voltage safety, and the realization of safety management requires sys
Recently, data-driven methods have received great attention, such as tem safety-related design, which can be divided into cell safety design,
ML [267,268], information fusion, and signal processing. These intelli mechanical safety design, electrical safety design, and thermal safety
gent methods are especially suitable for complex systems which are design, as summarized in Fig. 11.
difficult to establish accurate models. Learning from the historical data The safety design of cells is shown in Fig. 11 (a). The applications of
of a system and forming a nonlinear black-box mapping model, fault new separators such as inorganic ceramic fiber separators [281],
diagnosis can be achieved simply and quickly. ML algorithms widely paper-supported inorganic composite separators [282], and nano-silica
used in fault diagnosis include ANNs [269], SVMs [270,271], and FL modified polyimide nanofiber separators [283], can improve the
[272]. ANNs have good self-learning capabilities. SVMs are primarily wettability and thermal stability of separators. Temperature-sensitive
used for classification and regression, and are equally applicable to small electrodes can enhance the safety of cathode materials [284,285].
sample sizes. FL algorithms can characterize the membership of battery Electrolyte safety could be equally improved by adding an overcharge
faults, and establish the fuzzy relationship between the fault cause and protection additive [286] and a flame retardant [287]. Explosion-proof
Fig. 10. Categories of fault diagnosis methods used in LIBs. FMEA: failure mode and effects analysis.
13
Fig. 11. Safety designs in traction battery systems. (a) effective approaches to improve cell safety from materials and structures; (b) practical measures to
enhance mechanical safety; (c) main aspects to guarantee electrical safety; and (d) typical measures to improve thermal safety.
valves are used to prevent excessive internal pressure in cells. Fig. 12 shows the implementation strategies for battery safety and
For energy storage devices such as batteries, mechanical safety is aging management, in which the numbers in circles indicate the
essential. High-strength materials and sealing devices are used inside implementation order of safety and aging measures. In terms of safety
battery systems to protect batteries against penetration and water management, the system safety design and failure mechanism provide a
ingress, respectively. Fixation and damping designs can be adopted to sufficient physical and theoretical foundation for safety management, in
prevent the dislocation of battery modules and fatigue failure of parts. which the safety design contributes to physical protection and emer
The structural strength and bearing capacity of components can be gency cut-off, while research into the safety mechanisms of LIBs pro
improved to enhance collision protection. vides a theoretical foundation for the prevention and control of thermal
To ensure that the electrical system operates normally, electromag runaway [106,291,292]. The battery algorithm layer, together with
netic interference shielding can be installed. In addition to providing thermal, charging, and balance management, provides sufficient battery
reliable communication, early warning systems for leakage and high information to the fault diagnosis management, which helps diagnose
voltage are designed to enhance personal safety. Common approaches the battery from multiple aspects and levels. Accurate, fast, and robust
include high-voltage interlock and insulation resistance detection. Vi fault diagnosis algorithms are required, and an early fault warning and
bration or assembly defects can cause fractures or virtual connections at fault tolerance system [293] is needed to improve overall system safety.
the battery connections, which can result in increased resistance and After the system-related fault is detected, a fault indication is sent to
heat, thereby introducing safety risks [288–290]. Current battery safety management. Lastly, safety management makes corresponding
connection technologies include screw connections, welding, lock-type safety actions according to the severity of the fault. A minor fault may
connections, and glued connections. In practice, it is necessary to send a warning to the driver, and a fault with moderate severity will
consider the contact resistance and reliability of the connections contribute to system power limitation, while a severe fault leads to the
comprehensively. battery system emergency cut-off and high-voltage relay disconnection.
Temperature has a significant impact on battery safety. Internal short Compared with safety management that requires fast response, aging
circuit, external short circuit, overcharging, excessively fast charging, management is a long-term task in battery management. The effective
overloading, and external heat transfer can cause the temperature to management of battery aging is also built upon the foundation and al
increase, leading to gas generation or thermal runaway, combustion, gorithm layer. Firstly, the battery mechanisms related to aging will
and explosion. In addition to improving the heat resistance of battery provide a theoretical basis for aging management. Intelligent and
materials, a BTMS is required to control battery heat generation and reasonable battery thermal, charging, and balance management will
dissipation, a detailed discussion of which can be found in Section 2.3.1. help prolong the battery life. An appropriate temperature range will
14
Fig. 12. Battery safety management and aging management.
slow down the battery aging process, an intelligent charging strategy number of EVs, which brings the urgent need for battery lifecycle
will enable fast charging without damaging the battery, and a reason management. Despite the rapid development of vehicle-to-grid
able balance strategy will protect the weakest battery cell from failure. technology, it is still difficult for battery systems with current
Under reasonable aging management, the battery state prediction (RUL management techniques to consider the energy society. There
prediction) will update the prediction of the failure cycle. Based on the fore, battery management technologies that focus on battery
updated RUL, aging management will provide more optimized man lifecycle, as well as the energy society, are of vital importance.
agement strategies back to thermal, charging, and balance management, (3) Low intelligence of algorithms and low utilization of data:
and eventually closed-loop aging management will be constructed. Limited by the computing capability of the onboard controller,
the current advanced battery management algorithms, such as
2.4. Problems and challenges high-order filters and ML algorithms, are still difficult to apply
online. With the increasing requirements for battery management
Many technical and academic achievements have been made in performance, the algorithms and battery models used in the next-
advanced battery management technology, which basically meet the generation battery management will become more complicated
user’s demand for battery systems. However, problems with the safety, and well designed for battery life, safety, and performance.
energy density, performance, price, and environmental friendless of Obviously, the computing power of the current BMS controller
battery systems are still challenges for current advanced battery man cannot meet the demand. In addition, a lot of data during vehicle
agement. The main challenges are as follows: operation cannot be stored and processed by an onboard
embedded system, resulting in the low utilization of battery data.
(1) Limited battery-related measurements: The measurable physical Therefore, efficient and intelligent computing architecture will
quantities for batteries are limited to voltage, current, and tem bring great changes to next-generation battery management
perature. To better understand and effectively manage batteries, technology.
some advanced methods have been developed to estimate the
battery internal states, as discussed earlier. However, it is still 3. Trends in next-generation battery management technologies
challenging to obtain an accurate and comprehensive knowledge
of internal states through limited battery measurements. The EV users need batteries that are safe, affordable, and have excellent
development of battery sensing technology will help obtain data performance and long service life. Whether it is advanced battery
of higher dimensions and bring tremendous innovation for bat management or next-generation battery management technology, safety
tery modeling, state estimation, and management strategies. The and aging management are the top priorities. Unlike advanced man
next-generation battery management technology will focus not agement, next-generation battery management focuses on battery life
only on battery electrical and thermal-related measurement, but cycle management (from production, application, and maintenance to
also on battery electrochemical, mechanical, acoustical, and op recycling) [294]. “Data” and “intelligence” are the most distinctive
tical sensing. features of next-generation management. Fig. 13 shows the trends of
(2) Lack of battery lifecycle management: The current battery next-generation management technology, in which smart sensing,
management technology uses limited battery data (mainly during interconnectivity, and intelligence are the most important trends.
the vehicle application stage) to manage the battery, which leads Compared with previous battery management technology, the di
to a sub-optimal solution for the management algorithms due to mensions and amount of data in next-generation management have
the lack of lifecycle data. The secondary use of retired batteries changed significantly. Smart sensing will provide multi-dimensional
has attracted increasing attention due to the rapidly growing data, covering battery electrical, electrochemical, mechanical,
15
Fig. 13. Trends in next-generation multi-platform battery management technology: smart sensing, interconnectivity, and intelligence.
acoustical, and optical information, as well as data from cells to packs in data from the battery lifecycle and energy society. It will receive massive
EVs. The data amount reflects in the management of massive data from data from the vehicle platform first, and then store the data in a struc
the battery lifecycle and energy society. Data management technologies tured form. The large datasets will be helpful to improve the perfor
based on big data can achieve the optimization and management of mance of ML-based algorithms [296]. The powerful computing capacity
battery lifecycle across the stages of development, production, appli is responsible for the battery modeling and algorithm based on the
cation, sale, after-sale service, and decommissioning, while data man massive data. The complicated but high-fidelity multi-domain coupled
agement in the energy society can be achieved by blockchain battery models, together with complex and precise algorithms, can be
technology, which is a distributed information, point-to-point trans harnessed in the cloud platform [4]. Then, the cloud platform will send
mission, and recognition mechanism. With the rapid development of AI control commands and battery states with long-term information back to
technology, which has been widely used in many fields, battery man the vehicle platform. The information visualization is also an important
agement technology based on AI is rapidly emerging. AI-based battery part of the cloud platform, and is the belt between the whole manage
management can handle complicated battery models with high accu ment system and energy society operators [299,302]. The information
racy, as well as complex algorithms for multi-state estimation, and visualization will display the real-time sensing data and internal states,
eventually, achieve intelligent battery management. as well as the historical attenuation trajectory of batteries, which will
Next-generation battery management will be achieved through a help the operators to schedule corresponding maintenance measures.
distributed system, including the battery management on the vehicle The communication between vehicle platform and cloud platform is also
platform and the battery management on the cloud platform, which is of vital importance. Massive battery real-time data and early failure
also known as the cloud BMS. Many researchers have proposed the warning shall be transferred between the multi-platform, which requires
preliminary concept of multi-platform battery management [4,295, the communication to be with sufficiently low latency and wide data
296]. Advanced algorithms such as battery modeling [297], state esti bandwidth, to avoid time delay or packet losses [300]. During data
mation [298,299], hybrid energy storage management [300], communication, information security and privacy also shall be taken
cell-to-cell variation evaluation [301], and fault diagnosis [302] have into consideration [295,299,303].
been implemented on the prototype of cloud platform or with cloud
data. In Fig. 13, smart sensing is one of the key functions in battery
management on the vehicle platform. Besides this, the vehicle platform 3.1. Smart sensing
will also be responsible for the conventional functions of advanced
battery management. The vehicle platform will be in charge of onboard Limited by the development of battery sensors, the conventional
control and battery modeling/algorithm with low complexity and then measurement data are more related to the battery electrical and thermal
send high-precision measured data and real-time estimation to the cloud information (voltage, current, and temperature). The shortage of mea
platform. The cloud platform has three essential components: mass data surement cannot meet the requirements of next-generation management
storage and management, powerful computing capacity, and informa and hinder technological progress. Besides these traditional physical
tion visualization [302]. Mass data storage and management handle the quantities, some smart measurement information related to battery
electrochemical, mechanical, acoustical, and optical behaviors will
16
provide a more comprehensive understanding of the battery. The future mechanical information can be used to detect the irreversible volume
of battery smart sensing is summarized in Fig. 14. expansion caused by SEI growth [308] and to estimate battery SOC
Electrochemical sensing is employed to investigate the electrode [309,310] and SOH [311].
electrochemical process, which can be reflected in battery impedance The expansion and contraction of the battery upon cycling caused by
spectroscopy. Great achievements have been made in characterizing and lithium de-insertion and insertion will change not only the battery
modeling batteries using impedance spectroscopy in the laboratory, and volume but also the acoustic impedance, which brings a novel method
the direct onboard measurement of battery impedance has also been for battery sensing and detection: acoustical sensing. The ultrasonic
proved to be effective. Two approaches are promising for measuring diagnostics, as one of the non-destructive characterization methods for
battery impedance online. One approach requires the usage of an AC materials, have shown the preliminary effectiveness in internal gas
signal generator integrated in the vehicle, which can apply current with detection, electrical abuse detection [312], and state estimation. After
a specific frequency to the battery [304,305]. Therefore, the battery the transmitter generates the pulse signal, the peak height and delay
impedance at different frequencies is obtained according to the load time of the received wave show a good correlation with the battery SOC
current and measured voltage. The other approach is more flexible, and [313], also a close relationship with electrochemical charge/discharge
the signal processing methods, such as Fourier transformation and cycling and aging [314]. Similar to the above stress sensing, the
wavelet transformation [306], can directly acquire the battery imped generator and receiver of acoustical sensing are still in the laboratory
ance through the step or operating current over a wide frequency range. stage.
The lithium de-insertion and insertion behaviors are similar to Optical fiber sensors can convert the battery states into measurable
“breathing”, which causes changes in battery stress and deformation optical signals, which offer advantages of high speed, high precision,
during charging/discharging. Previous studies focused on the stress and robustness, and is considered as one of the most promising sensing
characteristic and modeling of the electrode [307]. For battery appli methods [315,316]. Two types of optical sensing show effectiveness in
cations, mechanical sensing is as important as electrical and thermal battery management, the external and internal optical sensors. A fiber
sensing; however, it is difficult to measure battery stress or deformation, Bragg grating sensor was attached on the battery surface to monitor the
especially for the onboard applications. At present, battery stress can be strain [317,318]; further, the influence of SOC and depth of discharging
acquired by placing the stress sensor or strain gauge on the battery; then, on strain was analyzed [317]. Compared with thermocouples, the
Fig. 14. Trends of smart sensing in next-generation battery management: (a) electrochemical impedance: Reprinted with permission from Ref. [305] (Copy
right 2020, IEEE); (b) mechanical stress: Reprinted with permission from Ref. [308] (Copyright 2019, Elsevier); (c) ultrasonic: Reprinted with permission from Refs.
[313] (Copyright 2017, Elsevier); (d) optical fiber: Reprinted with permission from Refs. [320] (Copyright 2019, Elsevier).
17
optical fiber sensors have better resolution and faster response rate in As shown in Fig. 15, at present, big data from traction batteries has
real time temperature monitoring during large-current conditions [319]. initially shown value in five stages of battery lifecycle: raw materials,
Because of the small size of the optical sensor, it can be embedded into battery manufacturing, vehicle application, battery maintenance, and
the battery to monitor more signals. Although the internal optical fiber decommissioning and recycling. The design of novel electrode and
sensing is more informative, it is much more difficult to implement than electrolyte materials can draw support from the amounts of research
external sensing. The seal integrity, influence to battery life and data, and the performance of the battery can be improved from the
compatibility with existing system designs of embedded optical fiber perspective of theoretical calculation and simulation [295]. A Genome
sensing are elaborated in detail [315], indicating that the onboard project for the development of electrolyte materials was proposed with a
application of embedded optical fiber sensors is promising. A hybrid novel big data technology, which accelerates the process of design and
sensing network consisting of fiber Bragg gratings and Fabry-Perot screen for electrolyte [323]. Battery manufacturers can use the data to
cavities is proposed and embedded inside the battery to monitor the guide the optimization of battery design and production processes, and
internal strain and temperature [320]. Moreover, a fiber-optic evanes improve battery safety performance. For vehicle applications, big data
cent wave sensor placed in the battery is used for early detection of can be used to support research and development, promote intelligent
significant capacity fade [321]. manufacturing, develop efficient onboard BMS, and achieve precision
Measurement is the foundation of battery management, and smart marketing in the sales phase. At present, big data-based battery man
sensing with multi-dimensional data will be significant to the innovation agement mainly focuses on the stage of the vehicle application. Some
of battery management technology. Not only the electrical and thermal- intelligent algorithms, combining with big data technology, have real
related information, but also the battery electrochemical, mechanical, ized battery life estimation [324], driving range estimation [325], and
acoustical, and optical relevant information will be employed in battery charging demand forecasting [322]. For maintenance centers, batteries
modeling, state estimation, and management. In addition to introducing can be managed in detail to provide personalized maintenance. Fault
multiple smart sensing to battery management, future trends in smart warnings, fault tracing [268], and battery life prediction can be pro
sensing will also focus on the transformation of external sensing to in vided through real-time data analysis. For decommissioned batteries, in
ternal sensing, the development of wireless sensing, and the integration the stage of large-scale pre-processing, big data is used to process
of multiple sensing. massive amounts of historical data, quickly identify the characteristic
parameters of each cell and module, evaluate the health status, and
realize the rapid separation of batched batteries. After decommissioned
3.2. High interconnectivity
batteries are put into use, historical data can also guide fault and safety
analysis. Meanwhile, it can be used to evaluate second-use life and
The production and application of batteries are inseparable from the
optimize management strategies to obtain the most significant economic
interaction, analysis, and utilization of data. The interaction of infor
benefit. When the battery has finished its second-use life, it can be dis
mation and the storage of data requires the support of big data and
assembled, and raw materials can be recycled. All information and data
blockchain technologies, and the effective analysis of data can bring
will be uploaded to the big data cloud computing platform. After the
considerable benefits to every part of the battery industry.
data has been processed, the platform will send information back to the
corresponding enterprises or users.
3.2.1. Application of big data technology to lifecycle data management
In the internet age, the generation of massive amounts of data has
3.2.2. Application of blockchain technology to energy society data
become the norm. Traction batteries generate a large amount of data
management
throughout their entire lifecycle of development, sale, use, maintenance,
Big data technology is used to manage the data during the battery
and recycling [295]. Based on cloud computing platforms, data from
lifecycle, and it essentially is longitudinal data management. The
batteries can be stored, analyzed, and mined by big data technology. A
numerous EVs, charging stations, and power plants constitute a huge
new service mode for the whole industry chain of alternative energy
energy society, and one battery system cannot be alone. Hence, lateral
vehicles can be explored. This is the trend of applying big data to
data management is of vital importance in next-generation battery
advanced battery management [322]. Currently, some big data centers
management. As the data management technology with point-to-point
have been put into use [268,278].
transmission and recognition mechanism, blockchain technology will
be effective in energy society data management.
Blockchain technology delivers data in real time. Unlike traditional
database technology that is only relevant within a single node, block
chain connects all nodes, and each node can see all transactions [326,
327]. The blockchain technology adopts the individual validity proof
and authorization, instead of the centralized application logic, to
implement constraints. Another advantage of blockchain-based data
bases is that the blockchain technology is highly fault-tolerant because
of the inherent redundancy of topology.
A distributed battery database system based on blockchain technol
ogy, which generates and stores data on battery usage and condition,
will be used to provide trustworthy, secure, and commercially viable full
lifecycle management of batteries as assets across a diverse range of
usages. Many companies, such as Filament and BlockApps Inc., have
contributed to developing a practical system for the full lifecycle man
agement of batteries involved in energy society. Battery swapping is a
convenient and fast EVs refueling method; however, the centralized data
system cannot guarantee the fair transaction between EVs and battery
swapping stations. A blockchain network with all battery’s operation
information during its lifecycle will solve trust lacking issues and can
take battery states, performance degradation, and battery depreciation
Fig. 15. Application of big data to the different battery industries. into account [328–330]. To facilitate EV users to seek suitable charging
18
station, a novel charging system based on consortium blockchain, which and RESs. First of all, the evolution of battery management technologies
integrates the charging stations from different new energy companies is given, considering the management scope and objective. Then, the
and governments, is proposed [331]. Besides, blockchain technology is hierarchical architecture of the advanced battery management is con
also innovatively applied to battery RUL prediction [332], which takes structed, for the first time, in terms of foundation, algorithm, and
advantage of the massive public battery data from EVs under different application layers. A comprehensive overview of each layer in advanced
driving conditions within the blockchain network, and can acquire more battery management is presented from both academic and engineering
RUL prognosis results. The energy society has recently become enor perspectives, and the main design and considerations of each layer are
mous enough, including the widespread power access [333], EV also presented and compared systematically. Finally, the prospects of
charging, and vehicle-to-grid application, which lead to the high the next-generation battery management technology based on data and
complexity of energy management due to massive data. The blockchain intelligence are discussed. The smarting sensing, the longitudinal and
technology is promising to realize this kind of data management. The lateral data management, and AI technology will forward the develop
blockchain technology also contributes to the realization of the ment of battery management and EVs.
distributed energy resources worldwide, including the solar panels,
smart appliances, and energy storage systems, which help to reduce
costs and increase reliability [334]. Declaration of competing interest
3.3. High intelligence The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
AI technology, as one of the most advanced technologies at present, the work reported in this paper.
is promising to be applied to next-generation battery management
technology, which can handle the complicated and accurate battery Acknowledgments
modeling, intelligent algorithms, and efficient control strategies. The
cloud platform can solve the issue that intelligent algorithms have not This work was supported in part by the NSF of China (Grant No.
realized in vehicle applications because of the limited onboard 51677136), NSF of China (Grant No. 51875054 and No. U1864212),
computing ability. With the help of the cloud platform shown in Fig. 13, Chongqing Natural Science Foundation for Distinguished Young
the next-generation battery management will be with high intelligence. Scholars (Grant No. cstc2019jcyjjq0010), Chongqing Science and
Currently, intelligence-based approaches have been increasingly Technology Bureau, China, and Natural Sciences and Engineering
used in battery modeling and state estimation. SVMs have been Research Council of Canada (RGPIN-2018-05471).
employed to model battery characteristics in the interests of deter
mining the nonlinear effects of current, charging state, and temperature Appendix A. Supplementary data
on battery characteristics [335,336]. Besides, a unified ANN-based
electro-thermal coupled modeling approach is proposed to describe Supplementary data to this article can be found online at https://doi.
the battery dynamic electro-thermal coupling behaviors [337]. These AI org/10.1016/j.rser.2020.110480.
technologies can help establish accurate battery models and improve the
battery models’ robustness [338]. By self-learning battery dynamic References
characteristics, accurate parameter identification can be realized
through optimization algorithms, and further to enhance the estimation [1] Rahimi-Eichi H, Ojha U, Baronti F, Chow M-Y. Battery management system: an
accuracy of battery states [339]. Deep neural network [340], overview of its application in the smart grid and electric vehicles. IEEE Ind
Electron Mag 2013;7:4–16.
genetic-fuzzy clustering [341], extreme learning machine [342] are [2] Lu L, Han X, Li J, Hua J, Ouyang M. A review on the key issues for lithium-ion
employed to battery state estimation, and prove reliability and accuracy. battery management in electric vehicles. J Power Sources 2013;226:272–88.
For battery RUL prediction, the recurrent neural network [343], [3] Jiang J, Zhang C. Fundamentals and applications of lithium-ion batteries in
electric drive vehicles. John Wiley & Sons; 2015.
Gaussian process regression [344], and other intelligent algorithms, are
[4] Hu X, Zou C, Zhang C, Li Y. Technological developments in batteries: a survey of
beginning to show effectiveness, and the urgent issue is how to enhance principal roles, types, and management needs. IEEE Power Energy Mag 2017;15:
the prediction accuracy with battery data from early aging phase [109]. 20–31.
[5] Quddus MA, Kabli M, Marufuzzaman M. Modeling electric vehicle charging
Intelligent thermal management of LIBs should be able to accurately
station expansion with an integration of renewable energy and Vehicle-to-Grid
predict temperature changes and distribution for a battery pack ac sources. Transport Res E Logist Transport Rev 2019;128:251–79.
cording to the external working environment and operating conditions. [6] Liu H, Wei Z, He W, Zhao J. Thermal issues about Li-ion batteries and recent
This can ensure that batteries always work in the optimum temperature progress in battery thermal management systems: a review. Energy Convers
Manag 2017;150:304–30.
range. In addition, timely diagnosis, early warning, and effective mea [7] Hu X, Zheng Y, Howey DA, Perez H, Foley A, Pecht M. Battery warm-up
sures can be taken to handle faults. Intelligent charging management methodologies at subzero temperatures for automotive applications: recent
will be able to manage many charging patterns, including quick advances and perspectives. Prog Energy Combust Sci 2020;77:100806.
[8] Ruiz V, Pfrang A, Kriston A, Omar N, Van den Bossche P, Boon-Brett L. A review
charging, slow charging, and storage charging, according to different of international abuse testing standards and regulations for lithium ion batteries
optimization objectives. Intelligent balancing management of batteries in electric and hybrid electric vehicles. Renew Sustain Energy Rev 2018;81:
would accurately judge inconsistencies within a battery pack, optimize 1427–52.
[9] ISO 6469-1 Electrically propelled road vehicles - safety specifications - Part 1:
the usage range of battery SOC, and maximize power throughput ac rechargeable energy storage system (RESS). 2019.
cording to the aging rate of cells in different SOC regions. [10] ISO 6469-2 Electrically propelled road vehicles - safety specifications - Part 2:
In the future, AI will gradually infiltrate battery management at all vehicle operational safety. 2018.
[11] ISO 6469-3 Electrically propelled road vehicles - safety specifications - Part 3:
levels. Intelligent mechanism analysis at the foundation level, intelligent
electrical safety. 2018.
modeling and estimation at the algorithm level, and intelligent safety, [12] ISO 6469-4 Electrically propelled road vehicles - safety specifications - Part 4:
charging, balancing, and fault diagnosis strategies at the application post crash electrical safety. 2015.
[13] ISO 12405-4 Electrically propelled road vehicles - test specification for lithium-
level will be realized.
ion traction battery packs and systems - Part 4: performance testing. 2018.
[14] IEC 62660-1 Secondary lithium-ion cells for the propulsion of electric road
4. Conclusions vehicles - Part 1: performance testing. 2018.
[15] IEC 62660-2 Secondary lithium-ion cells for the propulsion of electric road
vehicles - Part 2: reliability and abuse testing. 2018.
This paper presents the cutting-edge battery management technol [16] IEC 62660-3 Secondary lithium-ion cells for the propulsion of electric road
ogies in achieving satisfactory safety and long service life for LIBs in EVs vehicles - Part 3: safety requirements. 2016.
19
[17] SAE J1766 Recommended Practice for Electric. Fuel cell and hybrid electric [53] Zhu J, Dewi Darma MS, Knapp M, Sørensen DR, Heere M, Fang Q, et al.
vehicle crash integrity testing. 2014. Investigation of lithium-ion battery degradation mechanisms by combining
[18] SAE J2464 electric and hybrid electric vehicle rechargeable energy storage differential voltage analysis and alternating current impedance. J Power Sources
system (RESS) safety and abuse testing. 2009. 2020:448.
[19] SAE J2929 safety standard for electric and hybrid vehicle propulsion battery [54] Wang T. Capacity-loss prediction and State-of-health estimation in lithium-ion
systems utilizing lithium-based rechargeable cells. 2013. batteries [Doctoral dissertation]. Harbin Institute of Technology; 2016.
[20] EN 1987-1 Electrically propelled road vehicles - specific requirements for safety - [55] Purushothaman BK, Landau U. Rapid charging of lithium-ion batteries using
Part 1: on-board energy storage. 1997. pulsed currents. J Electrochem Soc 2006;153:A533–42.
[21] EN 1987-2 Electrically propelled road vehicles - specific requirements for safety - [56] Legrand N, Knosp B, Desprez P, Lapicque F, Raël S. Physical characterization of
Part 2: functional safety means and protection against failures. 1998. the charging process of a Li-ion battery and prediction of Li plating by
[22] EN 1987-3 Electrically propelled road vehicles - specific requirements for safety - electrochemical modelling. J Power Sources 2014;245:208–16.
Part 3: protection of users against electrical hazards. 1998. [57] Guo R, Lu L, Ouyang M, Feng X. Mechanism of the entire overdischarge process
[23] DOE/ID-11069 FreedomCAR battery test manual for power-assist hybrid electric and overdischarge-induced internal short circuit in lithium-ion batteries. Sci Rep
vehicles. 2003. 2016;6:30248.
[24] UL 2580 batteries for use in electric vehicles. 2020. [58] Feng X, Zheng S, He X, Wang L, Wang Y, Ren D, et al. Time sequence map for
[25] GB/T 31484 Cycle life requirements and test methods for traction battery of interpreting the thermal runaway mechanism of lithium-ion batteries with
electric vehicle. 2015. LiNixCoyMnzO2 cathode. Front Energy Res 2018;6:00126.
[26] GB/T 31485 Safety requirements and test methods for traction battery of electric [59] Gümüşsu E, Ö Ekici, Köksal M. 3-D CFD modeling and experimental testing of
vehicle. 2015. thermal behavior of a Li-Ion battery. Appl Therm Eng 2017;120:484–95.
[27] GB/T 31486 Electrical performance requirements and test methods for traction [60] Zavalis TG, Klett M, Kjell MH, Behm M, Lindström RW, Lindbergh G. Aging in
battery of electric vehicle. 2015. lithium-ion batteries: model and experimental investigation of harvested LiFePO4
[28] GB/T 31467.1 Lithium-ion traction battery pack and system for electric vehicles - and mesocarbon microbead graphite electrodes. Electrochim Acta 2013;110:
Part 1: test specification for high power applications. 2015. 335–48.
[29] GB/T 31467.2 Lithium-ion traction battery pack and system for electric vehicles - [61] Seaman A, Dao TS, Mcphee J. A survey of mathematics-based equivalent-circuit
Part 2: test specification for high energy applications. 2015. and electrochemical battery models for hybrid and electric vehicle simulation.
[30] GB/T 31467.3 Lithium-ion traction battery pack and system for electric vehicles - J Power Sources 2014;256:410–23.
Part 3: safety requirements and test methods. 2015. [62] Johnson VH. Battery performance models in ADVISOR. J Power Sources 2002;
[31] GB/T 38661 Technical specifications of battery management system for electric 110:321–9.
vehicles. 2020. [63] Wang J, Chen Q, Cao B. Support vector machine based battery model for electric
[32] Xu D. Analysis of data acquisition method for power battery management system vehicles. Energy Convers Manag 2006;47:858–64.
on electric vehicles. Auto Electric Parts 2017:5–8. [64] Hu X, Li S, Peng H. A comparative study of equivalent circuit models for Li-ion
[33] Zheng Y, Ouyang M, Han X, Lu L, Li J. Investigating the error sources of the online batteries. J Power Sources 2012;198:359–67.
state of charge estimation methods for lithium-ion batteries in electric vehicles. [65] Xu J, Mi CC, Cao B, Cao J. A new method to estimate the state of charge of
J Power Sources 2018;377:161–88. lithium-ion batteries based on the battery impedance model. J Power Sources
[34] Chu X, Ilyas IF, Krishnan S, Wang J. Data cleaning: overview and emerging 2013;233:277–84.
challenges. In: Proceedings of the 2016 international conference on management [66] Chaturvedi NA, Klein R, Christensen J, Ahmed J, Kojic A. Algorithms for
of data; 2016. p. 2201–6. advanced battery-management systems. IEEE Contr Syst Mag 2010;30:49–68.
[35] Tang N. Big data cleaning. In: Web technologies and applications 16th Asia- [67] Doyle M, Fuller TF, Newman J. Modeling of galvanostatic charge and discharge of
Pacific web conference. Springer International Publishing; 2014. p. 13–24. the lithium/polymer/insertion cell. J Electrochem Soc 1993;140:1526–33.
[36] Fan W. Dependencies revisited for improving data quality. In: Proceedings of the [68] Jokar A, Rajabloo B, Désilets M, Lacroix M. Review of simplified Pseudo-two-
twenty-seventh ACM sigmod-sigact-sigart symposium on principles of database Dimensional models of lithium-ion batteries. J Power Sources 2016;327:44–55.
systems; 2008. p. 159–70. [69] Thomas KE, Newman J. Heats of mixing and of entropy in porous insertion
[37] Beskales G, Ilyas IF, Golab L. Sampling the repairs of functional dependency electrodes. J Power Sources 2003;119:844–9.
violations under hard constraints. Proceedings of the VLDB Endowment 2010;3: [70] Kim Y, Mohan S, Siegel JB, Stefanopoulou AG. Maximum power estimation of
197–207. lithium-ion batteries accounting for thermal and electrical constraints. In: ASME
[38] Bohannon P, Fan W, Flaster M, Rastogi R. A cost-based model and effective 2013 dynamic systems and control conference. ASME; 2013.
heuristic for repairing constraints by value modification. In: Proceedings of the [71] Lin X, Perez HE, Mohan S, Siegel JB, Stefanopoulou AG, Ding Y, et al. A lumped-
2005 ACM SIGMOD international conference on Management of data; 2005. parameter electro-thermal model for cylindrical batteries. J Power Sources 2014;
p. 143–54. 257:1–11.
[39] Fan W, Geerts F, Jia X, Kementsietsidis A. Conditional functional dependencies [72] Forgez C, Do DV, Friedrich G, Morcrette M, Delacourt C. Thermal modeling of a
for capturing data inconsistencies. ACM Trans Database Syst 2008;33:1–48. cylindrical LiFePO 4/graphite lithium-ion battery. J Power Sources 2010;195:
[40] Chu X, Ilyas IF, Papotti P. Holistic data cleaning: putting violations into context. 2961–8.
In: 2013 IEEE 29th international conference on data engineering (ICDE). IEEE; [73] Richardson RR, Zhao S, Howey DA. On-board monitoring of 2-D spatially-
2013. p. 458–69. resolved temperatures in cylindrical lithium-ion batteries: Part I. Low-order
[41] Wang J, Tang N. Towards dependable data repairing with fixing rules. In: thermal modelling. J Power Sources 2016;326:377–88.
Proceedings of the 2014 ACM SIGMOD international conference on management [74] Liu Q, Du C, Shen B, Zuo P, Cheng X, Ma Y, et al. Understanding undesirable
of data; 2014. p. 457–68. anode lithium plating issues in lithium-ion batteries. RSC Adv 2016;6:88683–700.
[42] Fan W, Li J, Ma S, Tang N, Yu W. Towards certain fixes with editing rules and [75] Christensen J, Newman J. Effect of anode film resistance on the charge/discharge
master data. Vldb Journal 2012;21:213–38. capacity of a lithium-ion battery. J Electrochem Soc 2003;150:A1416–20.
[43] Yakout M, Elmagarmid AK, Neville J, Ouzzani M, Ilyas IF. Guided data repair. [76] Cordoba-Arenas A, Onori S, Guezennec Y, Rizzoni G. Capacity and power fade
Proceedings of the VLDB Endowment 2011;4:279–89. cycle-life model for plug-in hybrid electric vehicle lithium-ion battery cells
[44] Cong G, Fan W, Geerts F, Jia X, Ma S. Improving data quality: consistency and containing blended spinel and layered-oxide positive electrodes. J Power Sources
accuracy. In: 33rd international conference on very large data bases; 2007. 2015;278:473–83.
p. 315–26. [77] He W, Williard N, Osterman M, Pecht M. Prognostics of lithium-ion batteries
[45] Krishnan S, Wang J, Wu E, Franklin M, Goldberg K. ActiveClean: interactive data based on Dempster–Shafer theory and the Bayesian Monte Carlo method. J Power
cleaning for statistical modeling. In: 42nd international conference on very large Sources 2011;196:10314–21.
data bases; 2016. p. 948–59. [78] Fang W, Ramadass P, Zhang Z. Study of internal short in a Li-ion cell-II.
[46] Chung Y, Krishnan S, Kraska T. A data quality metric (DQM): how to estimate the Numerical investigation using a 3D electrochemical-thermal model. J Power
number of undetected errors in data sets. In: 43rd international conference on Sources 2014;248:1090–8.
very large data bases. ACM; 2016. p. 1094–105. [79] Prada E, Domenico DD, Creff Y, Bernard J, Sauvant-Moynot V, Huet F.
[47] Duan J, Tang X, Dai H, Yang Y, Wu W, Wei X, et al. Building safe lithium-ion A simplified electrochemical and thermal aging model of LiFePO4-graphite Li-ion
batteries for electric vehicles: a review. Electrochem Energy Rev 2019;3:1–42. batteries: power and capacity fade simulations. J Electrochem Soc 2013;160:
[48] Zhu J, Sun Z, Wei X, Dai H. A new electrochemical impedance spectroscopy A616–28.
model of a high-power lithium-ion battery. RSC Adv 2014;4:29988–98. [80] Weppner W, Huggins RA. Determination of the kinetic parameters of mixed-
[49] Li W, Cao D, Jöst D, Ringbeck F, Kuipers M, Frie F, et al. Parameter sensitivity conducting electrodes and application to the system Li3Sb. J Electrochem Soc
analysis of electrochemical model-based battery management systems for 1977;124:1569–78.
lithium-ion batteries. Appl Energy 2020;269:115104. [81] Ho C, Raistrick ID, Huggins RA. Application of AC techniques to the study of
[50] Xiong R, Pan Y, Shen W, Li H, Sun F. Lithium-ion battery aging mechanisms and lithium diffusion in tungsten trioxide thin films. J Electrochem Soc 1980;127:
diagnosis method for automotive applications: recent advances and perspectives. 343–50.
Renew Sustain Energy Rev 2020;131:110048. [82] Deiss E. Spurious potential dependence of diffusion coefficients in Li + insertion
[51] Han X, Lu L, Zheng Y, Feng X, Li Z, Li J, et al. A review on the key issues of the electrodes measured with PITT. Electrochim Acta 2002;47:4027–34.
lithium ion battery degradation among the whole life cycle. eTransportation [83] Brondani MDF, Sausen ATZR, Sausen PS, Binelo MO. Parameter estimation of
2019;1:100005. lithium ion polymer battery mathematical model using genetic algorithm.
[52] Feng X, Ouyang M, Liu X, Lu L, Xia Y, He X. Thermal runaway mechanism of Comput Appl Math 2018;37:296–313.
lithium ion battery for electric vehicles: a review. Energy Storage Mater 2018;10:
246–67.
20
[84] Xu J, Cao B, Chen Z, Zou Z. An online state of charge estimation method with [113] Hu C, Ye H, Jain G, Schmidt C. Remaining useful life assessment of lithium-ion
reduced prior battery testing information. Int J Electr Power Energy Syst 2014;63: batteries in implantable medical devices. J Power Sources 2018;375:118–30.
178–84. [114] Zhang X, Miao Q, Liu Z. Remaining useful life prediction of lithium-ion battery
[85] Xiao R, Shen J, Li X, Yan W, Pan E, Chen Z. Comparisons of modeling and state of using an improved UPF method based on MCMC. Microelectron Reliab 2017;75:
charge estimation for lithium-ion battery based on fractional order and integral 288–95.
order methods. Energies 2016;9:9030184. [115] Zhou J. Research on lithium-ion battery remaining useful life estimation with
[86] Nelatury SR, Singh P. Equivalent circuit parameters of nickel/metal hydride relevance vector machine [Doctoral dissertation]. Harbin Institute of Technology;
batteries from sparse impedance measurements. J Power Sources 2004;132: 2013.
309–14. [116] Liu G. Prediction of battery remaining discharge energy oriented for remaining
[87] Rahimi-Eichi H, Chow MY. Adaptive parameter identification and State-of-Charge driving range estimation of electric vehicles [Doctoral dissertation]. Tsinghua
estimation of lithium-ion batteries. In: 38th annual conference on IEEE industrial University; 2015.
electronics society. IEEE; 2012. p. 4012–7. [117] Chen Z, Fu Y, Mi CC. State of charge estimation of lithium-ion batteries in electric
[88] Rahimi-Eichi H, Baronti F, Chow MY. Online adaptive parameter identification drive vehicles using extended kalman filtering. IEEE Trans Veh Technol 2013;62:
and state-of-charge coestimation for lithium-polymer battery cells. IEEE Trans Ind 1020–30.
Electron 2014;61:2053–61. [118] Kim J, Cho BH. State-of-Charge estimation and state-of-health prediction of a Li-
[89] Wei Z, Zou C, Leng F, Soong BH, Tseng KJ. Online model identification and state ion degraded battery based on an EKF combined with a per-unit system. IEEE
of charge estimate for lithium-ion battery with a recursive total least squares- Trans Veh Technol 2011;60:4249–60.
based observer. IEEE Trans Ind Electron 2018;65:1336–46. [119] Xiong R, He H, Sun F, Liu X, Liu Z. Model-based state of charge and peak power
[90] Kim J, Lee S, Cho BH. Complementary cooperation algorithm based on DEKF capability joint estimation of lithium-ion battery in plug-in hybrid electric
combined with pattern recognition for SOC/capacity estimation and SOH vehicles. J Power Sources 2013;229:159–69.
prediction. IEEE Trans Power Electron 2011;27:436–51. [120] He H, Xiong R, Zhang X, Sun F, Fan JX. State-of-Charge estimation of the lithium-
[91] Xiong R, Sun F, Chen Z, He H. A data-driven multi-scale extended Kalman ion battery using an adaptive extended kalman filter based on an improved
filtering based parameter and state estimation approach of lithium-ion polymer thevenin model. IEEE Trans Veh Technol 2011;60:1461–9.
battery in electric vehicles. Appl Energy 2014;113:463–76. [121] Sun F, Hu X, Zou Y, Li S. Adaptive unscented Kalman filtering for state of charge
[92] Waag W, Fleischer C, Sauer DU. Critical review of the methods for monitoring of estimation of a lithium-ion battery for electric vehicles. Energy 2011;36:3531–40.
lithium-ion batteries in electric and hybrid vehicles. J Power Sources 2014;258: [122] He W, Williard N, Chen C, Pecht M. State of charge estimation for electric vehicle
321–39. batteries using unscented kalman filtering. Microelectron Reliab 2013;53:840–7.
[93] Plett GL. Extended Kalman filtering for battery management systems of LiPB- [123] Liu SL, Cui NX, Zhang CH. State of health estimation of lithium-ion battery based
based HEV battery packs: Part 3. State and parameter estimation. J Power Sources on AUKF. Power Electronics 2017;51:122–4.
2004;134:277–92. [124] Zhang W, Shi W, Ma Z. Adaptive unscented Kalman filter based state of energy
[94] Hannan MA, Lipu MSH, Hussain A, Mohamed A. A review of lithium-ion battery and power capability estimation approach for lithium-ion battery. J Power
state of charge estimation and management system in electric vehicle Sources 2015;289:50–62.
applications: challenges and recommendations. Renew Sustain Energy Rev 2017; [125] He Y, Liu XT, Zhang CB, Chen ZH. A new model for State-of-Charge (SOC)
78:834–54. estimation for high-power Li-ion batteries. Appl Energy 2013;101:808–14.
[95] Zhang X, Zhang W, Li H. Review on state of charge estimation methods for Li-ion [126] Bi J, Zhang T, Yu H, Kang Y. State-of-health estimation of lithium-ion battery
batteries. Trans Electr Electron Mater 2017;18:136–40. packs in electric vehicles based on genetic resampling particle filter. Appl Energy
[96] Barai A, Uddin K, Widanalage WD, Mcgordon A, Jennings P. The effect of average 2016;182:558–68.
cycling current on total energy of lithium-ion batteries for electric vehicles. [127] Plett GL. Sigma-point Kalman filtering for battery management systems of LiPB-
J Power Sources 2016;303:81–5. based HEV battery packs: Part 2. Simultaneous state and parameter estimation.
[97] Dong G, Chen Z, Wei J, Zhang C, Wang P. An online model-based method for state J Power Sources 2006;161:1369–84.
of energy estimation of lithium-ion batteries using dual filters. J Power Sources [128] Li J, Barillas JK, Guenther C, Danzer MA. A comparative study of state of charge
2016;301:277–86. estimation algorithms for LiFePO 4 batteries used in electric vehicles. J Power
[98] Li X, Fan G, Pan K, Wei G, Zhu C, Rizzoni G, et al. A physics-based fractional order Sources 2013;230:244–50.
model and state of energy estimation for lithium ion batteries. Part I: model [129] Michel PH, Heiries V. An adaptive sigma point kalman filter hybridized by
development and observability analysis. J Power Sources 2017;367:187–201. support vector machine algorithm for battery SoC and SoH estimation. In: 2015
[99] Li X, Pan K, Fan G, Lu R, Zhu C, Rizzoni G, et al. A physics-based fractional order IEEE 81st vehicular technology conference (VTC spring). IEEE; 2015. p. 1–7.
model and state of energy estimation for lithium ion batteries. Part II: parameter [130] Cai C, Du D, Liu Z, Ge J. State-of-charge (SOC) estimation of high power Ni-MH
identification and state of energy estimation for LiFePO 4 battery. J Power rechargeable battery with artificial neural network. In: Proceedings of the 9th
Sources 2017;367:202–13. international conference on neural information processing. IEEE; 2002. p. 824–8.
[100] Liu X, Wu J, Zhang C, Chen Z. A method for state of energy estimation of lithium- [131] Chang WY. Estimation of the state of charge for a LFP battery using a hybrid
ion batteries at dynamic currents and temperatures. J Power Sources 2014;270: method that combines a RBF neural network, an OLS algorithm and AGA. Int J
151–7. Electr Power Energy Syst 2013;53:603–11.
[101] Wang Y, Zhang C, Chen Z. A method for joint estimation of state-of-charge and [132] Bai G, Wang P, Hu C, Pecht M. A generic model-free approach for lithium-ion
available energy of LiFePO4 batteries. Appl Energy 2014;135:81–7. battery health management. Appl Energy 2014;135:247–60.
[102] Richardson RR, Howey DA. Sensorless battery internal temperature estimation [133] Andre D, Nuhic A, Soczka-Guth T, Sauer DU. Comparative study of a structured
using a kalman filter with impedance measurement. IEEE Trans Sustainable neural network and an extended Kalman filter for state of health determination of
Energy 2015;6:1190–9. lithium-ion batteries in hybrid electricvehicles. Eng Appl Artif Intell 2013;26:
[103] Kim Y, Siegel JB, Stefanopoulou AG. A computationally efficient thermal model of 951–61.
cylindrical battery cells for the estimation of radially distributed temperatures. In: [134] Wang QK, He YJ, Shen JN, Ma ZF, Zhong GB. A unified modeling framework for
2013 American control conference. IEEE; 2013. p. 698–703. lithium-ion batteries: an artificial neural network based thermal coupled
[104] Lin X, Perez HE, Siegel JB, Stefanopoulou AG, Li Y, Anderson RD, et al. Online equivalent circuit model approach. Energy 2017;138:118–32.
parameterization of lumped thermal dynamics in cylindrical lithium-ion batteries [135] Hansen T, Wang CJ. Support vector based battery state of charge estimator.
for core temperature estimation and health monitoring. IEEE Trans Contr Syst J Power Sources 2005;141:351–8.
Technol 2013;21:1745–55. [136] Klass V, Behm M, Lindbergh G. A support vector machine-based state-of-health
[105] Zhang C, Li K, Deng J, Song S. Improved realtime state-of-charge estimation of estimation method for lithium-ion batteries under electric vehicle operation.
LiFePO4 battery based on a novel thermoelectric model. IEEE Trans Ind Electron J Power Sources 2014;270:262–72.
2017;64:654–63. [137] Andre D, Appel C, Soczka-Guth T, Sauer DU. Advanced mathematical methods of
[106] Cabrera-Castillo E, Niedermeier F, Jossen A. Calculation of the state of safety SOC and SOH estimation for lithium-ion batteries. J Power Sources 2013;224:
(SOS) for lithium ion batteries. J Power Sources 2016;324:509–20. 20–7.
[107] Ashtiani CN. Analysis of battery safety and hazards’ risk mitigation. In: Battery [138] Salkind AJ, Fennie C, Singh P, Atwater T, Reisner DE. Determination of state-of-
safety and abuse tolerance - 212th ECS meeting; 2008. p. 1–11. charge and state-of-health of batteries by fuzzy logic methodology. J Power
[108] Li W, Zhu J, Xia Y, Gorji MB, Wierzbicki T. Data-driven safety envelope of Sources 1999;80:293–300.
lithium-ion batteries for electric vehicles. Joule 2019;3:2703–15. [139] Zenati A, Desprez P, Razik H. Estimation of the SOC and the SOH of li-ion
[109] Hu X, Feng F, Liu K, Zhang L, Xie J, Liu B. State estimation for advanced battery batteries, by combining impedance measurements with the fuzzy logic inference.
management: key challenges and future trends. Renew Sustain Energy Rev 2019; In: IECON 2010 - 36th annual conference on IEEE industrial electronics society.
114:109334. IEEE; 2010. p. 1773–8.
[110] Ng SSY, Xing Y, Tsui KL. A naive Bayes model for robust remaining useful life [140] Awadallah MA, Venkatesh B. Accuracy improvement of SOC estimation in
prediction of lithium-ion battery. Appl Energy 2014;118:114–23. lithium-ion batteries. J Storage Mater 2016;6:95–104.
[111] Mo B, Yu J, Tang D, Liu H. A remaining useful life prediction approach for [141] Zenati A, Desprez P, Razik H, Rael S. A methodology to assess the State of Health
lithium-ion batteries using Kalman filter and an improved particle filter. In: 2016 of lithium-ion batteries based on the battery’s parameters and a Fuzzy Logic
IEEE international conference on prognostics and health management (ICPHM). System. In: 2012 IEEE international electric vehicle conference. IEEE; 2012.
IEEE; 2016. p. 1–6.
[112] Rigamonti M, Baraldi P, Zio E, Roychoudhury I, Goebel K, Poll S. Ensemble of [142] Gan L, Yang F, Shi YF, He HL. Lithium-ion battery state of function estimation
optimized echo state networks for remaining useful life prediction. based on fuzzy logic algorithm with associated variables. In: 2017 3rd
Neurocomputing 2018;281:121–38. international conference on energy, environment and materials science. IOP
Publishing; 2017. p. 1–7.
21
[143] Thele M, Bohlen O, Sauer DU, Karden E. Development of a voltage-behavior [174] Liu D, Zhou J, Guo L, Peng Y. Survey on lithium-ion battery health assessment and
model for NiMH batteries using an impedance-based modeling concept. J Power cycle life estimation. Chin J Sci Instrum 2015;36:1–16.
Sources 2008;175:635–43. [175] Luo WL, Zhang LQ, Lv C, Wang LX. Review on foreign status of life prediction of
[144] Feng F, Lu R, Zhu C. A combined state of charge estimation method for lithium- lithium-ion batteries. Journal of Power Supply 2013;11:140–4.
ion batteries used in a wide ambient temperature range. Energies 2014;7: [176] Dong H, Jin X, Lou Y, Wang C. Lithium-ion battery state of health monitoring and
3004–32. remaining useful life prediction based on support vector regression-particle filter.
[145] Feng F, Lu R, Zhu C. Equalisation strategy for serially connected LiFePO4 battery J Power Sources 2014;271:114–23.
cells. IET Electr Syst Transp 2016;6:246–52. [177] Wang D, Yang F, Tsui K-L, Zhou Q, Bae SJ. Remaining useful life prediction of
[146] Pop V, Bergveld HJ, Danilov D, Regtien PPL, Notten PHL. Battery management lithium-ion batteries based on spherical cubature particle filter. IEEE Trans
systems - accurate state-of-charge indication for battery-powered applications. Instrum Meas 2016;65:1282–91.
Springer; 2008. [178] Miao Q, Xie L, Cui H, Liang W, Pecht M. Remaining useful life prediction of
[147] He H, Zhang X, Xiong R, Xu Y, Guo H. Online model-based estimation of state-of- lithium-ion battery with unscented particle filter technique. Microelectron Reliab
charge and open-circuit voltage of lithium-ion batteries in electric vehicles. 2013;53:805–10.
Energy 2012;39:310–8. [179] Liu Y, Liu Z, Song C. Improved particle filter algorithm for RUL prediction of
[148] Rozaqi L, Rijanto E. SOC estimation for Li-ion battery using optimum RLS method lithium-ion batteries. J Jilin Univ 2018;36:173–7.
based on genetic algorithm. In: 2016 8th international conference on information [180] Chang Y, Fang H, Zhang Y. A new hybrid method for the prediction of the
technology and electrical engineering (ICITEE). IEEE; 2016. p. 1–4. remaining useful life of a lithium-ion battery. Appl Energy 2017;206:1564–78.
[149] Aylor JH, Thieme A, Johnson BW. A battery state-of-charge indicator for electric [181] Zheng X, Fang H. An integrated unscented kalman filter and relevance vector
wheelchairs. IEEE Trans Ind Electron 1992;39:398–409. regression approach for lithium-ion battery remaining useful life and short-term
[150] Hoening S, Palanisamy TG, Singh H. Method for determining state of charge of a capacity prediction. Reliab Eng Syst Saf 2015;144:74–82.
battery by measuring its open circuit voltage. 2002. [182] Patil MA, Tagade P, Hariharan KS, Kolake SM, Song T, Yeo T, et al. A novel
[151] Feng F, Lu R, Wei G, Zhu C. Online estimation of model parameters and state of multistage Support Vector Machine based approach for Li ion battery remaining
charge of LiFePO4 batteries using a novel open-circuit voltage at various ambient useful life estimation. Appl Energy 2015;159:285–97.
temperatures. Energies 2015;8:2950–76. [183] Liu D, Zhou J, Pan D, Peng Y, Peng X. Lithium-ion battery remaining useful life
[152] Blanke H, Bohlen O, Buller S, Doncker RWD, Fricke B, Hammouche A, et al. estimation with an optimized Relevance Vector Machine algorithm with
Impedance measurements on lead–acid batteries for state-of-charge, state-of- incremental learning. Measurement 2015;63:143–51.
health and cranking capability prognosis in electric and hybrid electric vehicles. [184] Zhang C, He Y, Yuan L, Xiang S, Wang J. Prognostics of lithium-ion batteries
J Power Sources 2005;144:418–25. based on wavelet denoising and DE-RVM. Comput Intell Neurosci 2015;2015:
[153] Bundy K, Karlsson M, Lindbergh G, Lundqvist A. An electrochemical impedance 918305.
spectroscopy method for prediction of the state of charge of a nickel-metal [185] Li L, Wang P, Chao K-H, Zhou Y, Xie Y. Remaining useful life prediction for
hydride battery at open circuit and during discharge. J Power Sources 1998;72: lithium-ion batteries based on Gaussian processes mixture. PloS One 2016;11:
118–25. 0163004.
[154] Hammouche A, Karden E, Doncker RWD. Monitoring state-of-charge of Ni–MH [186] Zhang Y, Liu D, Yu J, Peng Y, Peng X. EMA remaining useful life prediction with
and Ni–Cd batteries using impedance spectroscopy. J Power Sources 2004;127: weighted bagging GPR algorithm. Microelectron Reliab 2017;75:253–63.
105–11. [187] Liu J, Saxena A, Goebel K, Saha B, Wang W. An adaptive recurrent neural network
[155] Ranjbar AH, Banaei A, Khoobroo A, Fahimi B. Online estimation of state of charge for remaining useful life prediction of lithium-ion batteries. In: Annual conference
in Li-ion batteries using impulse response concept. IEEE Trans Smart Grid 2012;3: of the prognostics and health management society; 2010.
360–7. [188] Liu K, Shang Y, Ouyang Q, Widanage WD. A data-driven approach with
[156] Verbrugge M, Tate E. Adaptive state of charge algorithm for nickel metal hydride uncertainty quantification for predicting future capacities and remaining useful
batteries including hysteresis phenomena. J Power Sources 2004;126:236–49. life of lithium-ion battery. IEEE Trans Ind Electron 2020. https://doi.org/
[157] Li Z, Lu L, Ouyang M. Comparison of methods for improving SOC estimation 10.1109/TIE.2020.2973876.
accuracy through an ampere-hour integeration approach. J Tsinghua Univ 2010; [189] Chen Q, Sun F, Zhu J. Modern electric vehicle technology. Beijing Institute of
50:1293–6. Technology Press; 2002.
[158] Meissner E, Richter G. Battery monitoring and electrical energy management: [190] Gong Q, Midlam-Mohler S, Marano V, Rizzoni G, Guezennec Y. Statistical analysis
precondition for future vehicle electric power systems. J Power Sources 2003; of PHEV fleet data. In: 2010 IEEE vehicle power and propulsion conference. IEEE;
116:79–98. 2010. p. 1–6.
[159] Waag W, Fleischer C, Sauer DU. Adaptive on-line prediction of the available [191] Oliva JA, Weihrauch C, Bertram T. A model-based approach for predicting the
power of lithium-ion batteries. J Power Sources 2013;242:548–59. remaining driving range in electric vehicles. In: Proceedings of the annual
[160] Kim DY, Jung DY. Method of estimating maximum output of battery for hybrid conference of the prognostics and health management society; 2013. p. 438–48.
electric vehicle. 2009. [192] Yu H, Tseng F, McGee R. Driving pattern identification for EV range estimation.
[161] Richardson RR, Ireland PT, Howey DA. Battery internal temperature estimation In: 2012 IEEE international electric vehicle conference. IEEE; 2012.
by combined impedance and surface temperature measurement. J Power Sources [193] Hu X, Yuan H, Zou C, Li Z, Zhang L. Co-estimation of state of charge and state of
2014;265:254–61. health for lithium-ion batteries based on fractional-order calculus. IEEE Trans Veh
[162] Bloom I, Cole BW, Sohn JJ, Jones SA, Polzin EG, Battaglia VS, et al. An Technol 2018;67:10319–29.
accelerated calendar and cycle life study of Li-ion cells. J Power Sources 2001; [194] Jiang B, Dai H, Wei X, Xu T. Joint estimation of lithium-ion battery state of charge
101:238–47. and capacity within an adaptive variable multi-timescale framework considering
[163] Matsushima T. Deterioration estimation of lithium-ion cells in direct current current measurement offset. Appl Energy 2019;253:113619.
power supply systems and characteristics of 400-Ah lithium-ion cells. J Power [195] Li X, Wang Z, Zhang L. Co-estimation of capacity and state-of-charge for lithium-
Sources 2009;189:847–54. ion batteries in electric vehicles. Energy 2019;174:33–44.
[164] Spotnitz R. Simulation of capacity fade in lithium-ion batteries. J Power Sources [196] Xiong R, Sun F, He H, Nguyen TD. A data-driven adaptive state of charge and
2003;113:72–80. power capability joint estimator of lithium-ion polymer battery used in electric
[165] Wang J, Liu P, Hicks-Garner J, Sherman E, Soukiazian S, Verbrugge M, et al. vehicles. Energy 2013;63:295–308.
Cycle-life model for graphite-LiFePO4 cells. J Power Sources 2011;196:3942–8. [197] Lai X, He L, Wang S, Zhou L, Zhang Y, Sun T, et al. Co-estimation of state of
[166] Chiang YH, Sean WY. Dynamical estimation of State-of-Health of batteries by charge and state of power for lithium-ion batteries based on fractional variable-
using adaptive observer. In: 2009 2nd international conference on power order model. J Clean Prod 2020;255:120203.
electronics and intelligent transportation system (PEITS). IEEE; 2009. p. 110–5. [198] Feng F, Teng S, Liu K, Xie J, Xie Y, Liu B, et al. Co-estimation of lithium-ion
[167] Chiang YH, Sean WY, Ke JC. Online estimation of internal resistance and open- battery state of charge and state of temperature based on a hybrid
circuit voltage of lithium-ion batteries in electric vehicles. J Power Sources 2011; electrochemical-thermal-neural-network model. J Power Sources 2020;455:
196:3921–32. 227935.
[168] Eddahech A, Briat O, Bertrand N, Delétage JY, Vinassa JM. Behavior and state-of- [199] Fang Q, Wei X, Dai H. A remaining discharge energy prediction method for
health monitoring of Li-ion batteries using impedance spectroscopy and recurrent lithium-ion battery pack considering SOC and parameter inconsistency. Energies
neural networks. Int J Electr Power Energy Syst 2012;42:487–94. 2019;12:12060987.
[169] Nuhic A, Terzimehic T, Soczka-Guth T, Buchholz M, Dietmayer K. Health [200] Li J, Greye B, Buchholz M, Danzer MA. Interval method for an efficient state of
diagnosis and remaining useful life prognostics of lithium-ion batteries using charge and capacity estimation of multicell batteries. J Storage Mater 2017;13:
data-driven methods. J Power Sources 2013;239:680–8. 1–9.
[170] Shahriari M, Farrokhi M. Online state-of-health estimation of VRLA batteries [201] Yang C, Wang X, Fang Q, Dai H, Cao Y, Wei X. An online SOC and capacity
using state of charge. IEEE Trans Ind Electron 2013;60:191–202. estimation method for aged lithium-ion battery pack considering cell
[171] Lipu MSH, Hannan MA, Hussain A, Hoque MM, Ker PJ, Saad MHM, et al. A review inconsistency. J Storage Mater 2020;29:101250.
of state of health and remaining useful life estimation methods for lithium-ion [202] Ji Y, Wang CY. Heating strategies for Li-ion batteries operated from subzero
battery in electric vehicles: challenges and recommendations. J Clean Prod 2018; temperatures. Electrochim Acta 2013;107:664–74.
205:115–33. [203] Sefidan AM, Sojoudi A, Saha SC. Nanofluid-based cooling of cylindrical lithium-
[172] Meng H, Li Y-F. A review on prognostics and health management (PHM) methods ion battery packs employing forced air flow. Int J Therm Sci 2017;117:44–58.
of lithium-ion batteries. Renew Sustain Energy Rev 2019;116:109405. [204] Wang Q, Jiang B, Xue QF, Sun HL, Li B, Zou HM, et al. Experimental investigation
[173] Li Y, Liu K, Foley AM, Zülke A, Berecibar M, Nanini-Maury E, et al. Data-driven on EV battery cooling and heating by heat pipes. Appl Therm Eng 2015;88:54–60.
health estimation and lifetime prediction of lithium-ion batteries: a review.
Renew Sustain Energy Rev 2019;113:109254.
22
[205] Wang Q, Jiang B, Li B, Yan Y. A critical review of thermal management models [237] Lee YS, Cheng MW. Intelligent control battery equalization for series connected
and solutions of lithium-ion batteries for the development of pure electric lithium-ion battery strings. IEEE Trans Ind Electron 2005;52:1297–307.
vehicles. Renew Sustain Energy Rev 2016;64:106–28. [238] Dong B, Han Y. A new architecture for battery charge equalization. In: 2011 IEEE
[206] Kelly KJ, Mihalic M, Zolot M. Battery usage and thermal performance of the energy conversion congress and exposition. IEEE; 2011. p. 928–34.
toyota prius and honda insight during chassis dynamometer testing. In: The [239] Einhorn M, Guertlschmid W, Blochberger T, Kumpusch R, Permann R, Conte FV,
seventeenth battery conference on applications and advances. IEEE; 2002. et al. A current equalization method for serially connected battery cells using a
p. 247–52. single power converter for each cell. IEEE Trans Veh Technol 2011;60:4227–37.
[207] Jaguemont J, Boulon L, Dubé Y. A comprehensive review of lithium-ion batteries [240] Manenti A, Abba A, Merati A, Savaresi SM, Geraci A. A new BMS architecture
used in hybrid and electric vehicles at cold temperatures. Appl Energy 2016;164: based on cell redundancy. IEEE Trans Ind Electron 2011;58:4314–22.
99–114. [241] Qian H, Zhang J, Lai JS, Yu W. A high-efficiency grid-tie battery energy storage
[208] Huo Y, Rao Z, Liu X, Zhao J. Investigation of power battery thermal management system. IEEE Trans Power Electron 2011;26:886–96.
by using mini-channel cold plate. Energy Convers Manag 2015;89:387–95. [242] Yarlagadda S, Hartley TT, Husain I. A battery management system using an active
[209] Kim J, Oh J, Lee H. Review on battery thermal management system for electric charge equalization technique based on a DC/DC converter topology. IEEE Trans
vehicles. Appl Therm Eng 2019;149:192–212. Ind Appl 2013;49:2720–9.
[210] Khateeb SA, Farid MM, Selman JR, Al-Hallaj S. Design and simulation of a [243] Daowd M, Omar N, Bossche PVD, Mierlo JV. Passive and active battery balancing
lithium-ion battery with a phase change material thermal management system for comparison based on MATLAB simulation. In: 2011 IEEE vehicle power and
an electric scooter. J Power Sources 2004;128:292–307. propulsion conference. IEEE; 2011. p. 1–7.
[211] Zhang D, Tian S, Xiao D. Experimental study on the phase change behavior of [244] Bouchhima N, Schnierle M, Schulte S, Birke KP. Active model-based balancing
phase change material confined in pores. Sol Energy 2007;81:653–60. strategy for self-reconfigurable batteries. J Power Sources 2016;322:129–37.
[212] Jin X, Li JQ, Zhang CN, Wu PE. Researches on modeling and experiment of Li-ion [245] Qi J, Lu DDC. Review of battery cell balancing techniques. In: 2014 Australasian
battery PTC self-heating in electric vehicles. In: Applied energy symposium and universities power engineering conference (AUPEC). IEEE; 2014. p. 1–6.
forum: low - carbon cities and urban energy systems; 2016. p. 62–7. [246] Gallardo-Lozano J, Romero-Cadaval E, Milanes-Montero MI, Guerrero-
[213] Zhang J, Ge H, Li Z, Ding Z. Internal heating of lithium-ion batteries using Martinez MA. Battery equalization active methods. J Power Sources 2014;246:
alternating current based on the heat generation model in frequency domain. 934–49.
J Power Sources 2015;273:1030–7. [247] Nelson VP. Fault-tolerant computing: fundamental concepts. Computer 1990;23:
[214] Zuñiga M, Jaguemont J, Boulon L, Dubé Y. Heating lithium-ion batteries with 19–25.
bidirectional current pulses. In: 2015 IEEE vehicle power and propulsion [248] Frank PM. Fault-diagnosis in dynamic-systems using analytical and knowledge-
conference (VPPC). IEEE; 2015. based redundancy - a survey and some new results. Automatica 1990;26:459–74.
[215] Wang CY, Xu T, Ge S, Zhang G, Yang XG, Ji Y. A fast rechargeable lithium-ion [249] Hwang I, Kim S, Kim Y, Seah CE. A survey of fault detection, isolation, and
battery at subfreezing temperatures. J Electrochem Soc 2016;163:A1944–50. reconfiguration methods. IEEE Trans Contr Syst Technol 2010;18:636–53.
[216] Mohan S, Kim Y, Stefanopoulou AG. Energy-Conscious warm-up of Li-ion cells [250] Patton RJ. Fault detection and diagnosis in aerospace systems using analytical
from subzero temperatures. IEEE Trans Ind Electron 2016;63:2954–64. redundancy. Comput Contr Eng J 1991;2:127–36.
[217] Mohan S, Siegel J, Stefanopoulou AG, Castanier M, Ding Y. Synthesis of an [251] Wu C, Zhu C, Ge Y, Zhao Y. A review on fault mechanism and diagnosis approach
energy-optimal self-heating strategy for Li-ion batteries. In: 2016 IEEE 55th for Li-ion batteries. J Nanomater 2015;2015:631263.
conference on decision and control (CDC). IEEE; 2016. p. 1589–94. [252] Levy SC. Safety and reliability considerations for lithium batteries. J Power
[218] Mohan SN. Control of lithium-ion battery warm-up from sub-zero temperatures Sources 1997;68:75–7.
[Doctoral dissertation]. University of Michigan; 2017. [253] Held M, Brönnimann R. Safe cell, safe battery? Battery fire investigation using
[219] Mohan S, Kim Y, Stefanopoulou AG, Ding Y. On the warmup of Li-ion cells from FMEA, FTA and practical experiments. Microelectron Reliab 2016;64:705–10.
sub-zero temperatures. In: 2014 American control conference. IEEE; 2014. [254] Hendricks C, Williard N, Mathew S, Pecht M. A failure modes, mechanisms, and
p. 1547–52. effects analysis (FMMEA) of lithium-ion batteries. J Power Sources 2015;297:
[220] Ahmed S, Bloom I, Jansen AN, Tanim T, Dufek EJ, Pesaran A, et al. Enabling fast 113–20.
charging – a battery technology gap assessment. J Power Sources 2017;367: [255] Liu W. Research and implementation of failure diagnosis expert system of battery
250–62. pack [master dissertation]. Hunan University; 2005.
[221] Zou C, Hu X, Wei Z, Wik T, Egardt B. Electrochemical estimation and control for [256] Wu J. Design of remote monitoring system and study on fault diagnosis method
lithium-ion battery health-aware fast charging. IEEE Trans Ind Electron 2018;65: for electric vehicle [master dissertation]. Jilin University; 2011.
6635–45. [257] Liu Z, He H. Model-based sensor fault diagnosis of a lithium-ion battery in electric
[222] Notten PHL, JHGOh Veld, Beek JRGv. Boostcharging Li-ion batteries: a vehicles. Energies 2015;8:6509–27.
challenging new charging concept. J Power Sources 2005;145:89–94. [258] Liu Z, He H. Sensor fault detection and isolation for a lithium-ion battery pack in
[223] Liu Y-H, Luo Y-F. Search for an optimal rapid-charging pattern for Li-ion batteries electric vehicles using adaptive extended Kalman filter. Appl Energy 2017;185:
using the Taguchi approach. IEEE Trans Ind Electron 2010;57:3963–71. 2033–44.
[224] Vo TT, Chen X, Shen W, Kapoor A. New charging strategy for lithium-ion batteries [259] Yang R, Xiong R, He H, Chen Z. A fractional-order model-based battery external
based on the integration of Taguchi method and state of charge estimation. short circuit fault diagnosis approach for all-climate electric vehicles application.
J Power Sources 2015;273:413–22. J Clean Prod 2018;187:950–9.
[225] Zou C, Manzie C, Nešić D. A framework for simplification of PDE-based lithium- [260] Kim D, Koo K, Jeong JJ, Goh T, Kim SW. Second-order discrete-time sliding mode
ion battery models. IEEE Trans Contr Syst Technol 2016;24:1594–609. observer for state of charge determination based on a dynamic resistance Li-ion
[226] Zou C, Manzie C, Nešić D. PDE battery model simplification for charging strategy battery model. Energies 2013;6:5538–51.
evaluation. In: 2015 10th Asian control conference (ASCC). IEEE; 2015. p. 1–6. [261] Isermann R. Model-based fault-detection and diagnosis – status and applications.
[227] Liu Y-H, Hsieh C-H, Luo Y-F. Search for an optimal five-step charging pattern for Annu Rev Contr 2005;29:71–85.
Li-ion batteries using consecutive orthogonal arrays. IEEE Trans Energy Convers [262] Suozzo C, Onori S, Rizzoni G. Model-based fault diagnosis for NiMH battery. In:
2011;26:654–61. ASME 2008 dynamic systems and control conference. ASME; 2008. p. 1075–81.
[228] Hu X, Li S, Peng H, Sun F. Charging time and loss optimization for LiNMC and [263] Marcicki J, Onori S, Rizzoni G. Nonlinear fault detection and isolation for a
LiFePO4 batteries based on equivalent circuit models. J Power Sources 2013;239: lithium-ion battery management system. In: ASME 2010 dynamic systems and
449–57. control conference. ASME; 2010. p. 607–14.
[229] Zou C, Manzie C, Nešić D, Kallapur AG. Multi-time-scale observer design for state- [264] Liu Z, He H, Ahmed Q, Rizzoni G. Structural analysis based fault detection and
of-charge and state-of-health of a lithium-ion battery. J Power Sources 2016;335: isolation applied for A lithium-ion battery pack. IFAC-PapersOnLine. 2015;48:
121–30. 1465–70.
[230] Zou C, Hu X, Wei Z, Tang X. Electrothermal dynamics-conscious lithium-ion [265] Liu Z, Ahmed Q, Zhang J, Rizzoni G, He H. Structural analysis based sensors fault
battery cell-level charging management via state-monitored predictive control. detection and isolation of cylindrical lithium-ion batteries in automotive
Energy 2017;141:250–9. applications. Contr Eng Pract 2016;52:46–58.
[231] Perez HE, Hu X, Dey S, Moura SJ. Optimal charging of Li-ion batteries with [266] Liu Z, Ahmed Q, Rizzoni G, He H. Fault detection and isolation for lithium-ion
coupled electro-thermal-aging dynamics. IEEE Trans Veh Technol 2017;66: battery system using structural analysis and sequential residual generation. In:
7761–70. ASME 2014 dynamic systems and control conference. ASME; 2014.
[232] Perez HE, Dey S, Hu X, Moura SJ. Optimal charging of Li-ion batteries via a single [267] Zhao Y, Liu P, Wang Z, Hong J. Electric vehicle battery fault diagnosis based on
particle model with electrolyte and thermal dynamics. J Electrochem Soc 2017; statistical method. In: 8th international conference on applied energy. ICAE;
164:A1679–87. 2017. p. 2366–71.
[233] Liu J, Li G, Fathy HK. An extended differential flatness approach for the health- [268] Zhao Y, Liu P, Wang Z, Zhang L, Hong J. Fault and defect diagnosis of battery for
conscious nonlinear model predictive control of lithium-ion batteries. IEEE Trans electric vehicles based on big data analysis methods. Appl Energy 2017;207:
Contr Syst Technol 2017;25:1882–9. 354–62.
[234] Klein R, Chaturvedi NA, Christensen J, Ahmed J, Findeisen R, Kojic A. Optimal [269] Balaban E, Saxena A, Bansal P, Goebel KF, Curran S. Modeling, detection, and
charging strategies in lithium-ion battery. In: Proceedings of the 2011 American disambiguation of sensor faults for aerospace applications. IEEE Sensor J 2009;9:
control conference. IEEE; 2011. p. 382–7. 1907–17.
[235] Zou C, Manzie C, Nešić D. Model predictive control for lithium-ion battery [270] Widodo A, Shim M-C, Caesarendra W, Yang B-S. Intelligent prognostics for
optimal charging. IEEE ASME Trans Mechatron 2018;23:947–57. battery health monitoring based on sample entropy. Expert Syst Appl 2011;38:
[236] Hoque MM, Hannan MA, Mohamed A, Ayob A. Battery charge equalization 11763–9.
controller in electric vehicle applications: a review. Renew Sustain Energy Rev
2017;75:1363–85.
23
[271] Park JI, Baek SH, Jeong MK, Bae SJ. Dual features functional support vector [302] Kim T, Makwana D, Adhikaree A, Vagdoda J, Lee Y. Cloud-based battery
machines for fault detection of rechargeable batteries. IEEE Trans Syst Man condition monitoring and fault diagnosis platform for large-scale lithium-ion
Cybern C Appl Rev 2009;39:480–5. battery energy storage systems. Energies 2018;11:11010125.
[272] Singh A, Izadian A, Anwar S. Fault diagnosis of Li-Ion batteries using multiple- [303] Singh S, Weeber M, Birke KP, Sauer A. Development and utilization of a
model adaptive estimation. In: 39th annual conference of the IEEE industrial framework for data-driven life cycle management of battery cells. Procedia Manuf
electronics society. IEEE; 2013. p. 3524–9. 2020;43:431–8.
[273] Hu C, He R, Wang R, Yu J. Fault prediction and fault-tolerant of lithium-ion [304] Dam SK, John V. High resolution converter for battery impedance spectroscopy.
batteries temperature failure for electric vehicle. In: 2012 third international IEEE Trans Ind Appl 2018;54:1502–12.
conference on digital manufacturing & automation. IEEE; 2012. p. 410–3. [305] Wang X, Wei X, Chen Q, Dai H. A novel system for measuring alternating current
[274] Wu C, Zhu C, Sun J, Ge Y. A synthesized diagnosis approach for lithium-ion impedance spectra of series-connected lithium-ion batteries with a high-power
battery in hybrid electric vehicle. IEEE Trans Veh Technol 2017;66:5595–603. dual active bridge converter and distributed sampling units. IEEE Trans Ind
[275] Socher S, Jehle C, Potthoff U. Improving the functional safety of automotive Electron 2020. https://doi.org/10.1109/TIE.2020.3001841.
batteries using in-situ impedance spectroscopy. In: 6th transport research arena; [306] Hoshi Y, Yakabe N, Isobe K, Saito T, Shitanda I, Itagaki M. Wavelet
2016. p. 3661–6. transformation to determine impedance spectra of lithium-ion rechargeable
[276] Kaypmaz TC, Tuncay RN. An advanced cell model for diagnosing faults in battery. J Power Sources 2016;315:351–8.
operation of Li-ion Polymer batteries. In: 2011 IEEE vehicle power and propulsion [307] Xie H, Zhang Q, Song H, Shi B, Kang Y. Modeling and in situ characterization of
conference. IEEE; 2011. p. 1–5. lithiation-induced stress in electrodes during the coupled mechano-electro-
[277] Kim J, Cho BH. An innovative approach for characteristic analysis and state-of- chemical process. J Power Sources 2017;342:896–903.
health diagnosis for a Li-ion cell based on the discrete wavelet transform. J Power [308] Louli AJ, Ellis LD, Dahn JR. Operando pressure measurements reveal solid
Sources 2014;260:115–30. electrolyte interphase growth to rank Li-ion cell performance. Joule 2019;3:
[278] Li X, Wang Z. A novel fault diagnosis method for lithium-Ion battery packs of 745–61.
electric vehicles. Measurement 2018;116:402–11. [309] Dai H, Yu C, Wei X, Sun Z. State of charge estimation for lithium-ion pouch
[279] Sun Z, Liu P, Wang Z. Real-time fault diagnosis method of battery system based on batteries based on stress measurement. Energy 2017;129:16–27.
Shannon entropy. In: 8th international conference on applied energy. ICAE; 2017. [310] Hickey R, Jahns TM. Measuring individual battery dimensional changes for state-
p. 2354–9. of-charge estimation using strain gauge sensors. In: 2019 IEEE energy conversion
[280] Wang Z, Hong J, Liu P, Zhang L. Voltage fault diagnosis and prognosis of battery congress and exposition (ECCE). IEEE; 2019. p. 2460–5.
systems based on entropy and Z -score for electric vehicles. Appl Energy 2017; [311] Cannarella J, Arnold CB. State of health and charge measurements in lithium-ion
196:289–302. batteries using mechanical stress. J Power Sources 2014;269:7–14.
[281] Zhao X, Zhang Z, Yang S, Liang G. Inorganic ceramic fiber separator for [312] Oca L, Guillet N, Tessard R, Iraola U. Lithium-ion capacitor safety assessment
electrochemical and safety performance improvement of lithium-ion batteries. under electrical abuse tests based on ultrasound characterization and cell
Ceram Int 2017;43:14775–83. opening. J Storage Mater 2019;23:29–36.
[282] Wang Z, Xiang H, Wang L, Xia R, Nie S, Chen C, et al. A paper-supported inorganic [313] Gold L, Bach T, Virsik W, Schmitt A, Müller J, Staab TEM, et al. Probing lithium-
composite separator for high-safety lithium-ion batteries. J Membr Sci 2018;553: ion batteries’ state-of-charge using ultrasonic transmission – concept and
10–6. laboratory testing. J Power Sources 2017;343:536–44.
[283] Wang Y, Wang S, Fang J, Ding L-X, Wang H. A nano-silica modified polyimide [314] Ladpli P, Kopsaftopoulos F, Chang F-K. Estimating state of charge and health of
nanofiber separator with enhanced thermal and wetting properties for high safety lithium-ion batteries with guided waves using built-in piezoelectric sensors/
lithium-ion batteries. J Membr Sci 2017;537:248–54. actuators. J Power Sources 2018;384:342–54.
[284] Peng P, Jiang F. Thermal safety of lithium-ion batteries with various cathode [315] Raghavan A, Kiesel P, Sommer LW, Schwartz J, Lochbaum A, Hegyi A, et al.
materials: a numerical study. Int J Heat Mass Tran 2016;103:1008–16. Embedded fiber-optic sensing for accurate internal monitoring of cell state in
[285] Qian G, Wang L, Shang Y, He X, Tang S, Liu M, et al. Polyimide binder: a facile advanced battery management systems part 1: cell embedding method and
way to improve safety of lithium ion batteries. Electrochim Acta 2016;187:113–8. performance. J Power Sources 2017;341:466–73.
[286] Noelle DJ, Shi Y, Wang M, Le AV, Qiao Y. Aggressive electrolyte poisons and [316] Ganguli A, Saha B, Raghavan A, Kiesel P, Arakaki K, Schuh A, et al. Embedded
multifunctional fluids comprised of diols and diamines for emergency shutdown fiber-optic sensing for accurate internal monitoring of cell state in advanced
of lithium-ion batteries. J Power Sources 2018;384:93–7. battery management systems part 2: internal cell signals and utility for state
[287] Nv Aspern, Röser S, Rezaei Rad B, Murmann P, Streipert B, Mönnighoff X, et al. estimation. J Power Sources 2017;341:474–82.
Phosphorus additives for improving high voltage stability and safety of lithium [317] Peng J, Zhou X, Jia S, Jin Y, Xu S, Chen J. High precision strain monitoring for
ion batteries. J Fluor Chem 2017;198:24–33. lithium ion batteries based on fiber Bragg grating sensors. J Power Sources 2019;
[288] Liu G, Ouyang M, Lu L, Li J, Han X. Analysis of the heat generation of lithium-ion 433:226692.
battery during charging and discharging considering different influencing factors. [318] Xu Z, Cao B, Zhou Q, Xu J, Liu M. Structure design, Prototype manufacture and
J Therm Anal Calorim 2014;116:1001–10. Performance analysis of a fiber grating sensor for strain measurement of batteries.
[289] Feng X, Sun J, Ouyang M, He X, Lu L, Han X, et al. Characterization of large In: 2019 IEEE 4th advanced information technology, electronic and automation
format lithium ion battery exposed to extremely high temperature. J Power control conference (IAEAC). IEEE; 2019. p. 2188–92.
Sources 2014;272:457–67. [319] Nascimento M, Ferreira MS, Pinto JL. Real time thermal monitoring of lithium
[290] Han X, Ouyang M, Lu L, Li J. Cycle life of commercial lithium-ion batteries with batteries with fiber sensors and thermocouples: a comparative study.
lithium titanium oxide anodes in electric vehicles. Energies 2014;7:4895–909. Measurement 2017;111:260–3.
[291] Lisbona D, Snee T. A review of hazards associated with primary lithium and [320] Nascimento M, Novais S, Ding MS, Ferreira MS, Koch S, Passerini S, et al. Internal
lithium-ion batteries. Process Saf Environ Protect 2011;89:434–42. strain and temperature discrimination with optical fiber hybrid sensors in Li-ion
[292] Abada S, Marlair G, Lecocq A, Petit M, Sauvant-Moynot V, Huet F. Safety focused batteries. J Power Sources 2019;410–411:1–9.
modeling of lithium-ion batteries: a review. J Power Sources 2016;306:178–92. [321] Ghannoum A, Nieva P. Graphite lithiation and capacity fade monitoring of
[293] Zhang J, Rizzoni G. Functional safety of electrified vehicles through model-based lithium ion batteries using optical fibers. J Storage Mater 2020;28:101233.
fault diagnosis. In: 4th IFAC workshop on engine and powertrain control. [322] Arias MB, Bae S. Electric vehicle charging demand forecasting model based on big
Simulation and Modeling; 2015. p. 454–61. data technologies. Appl Energy 2016;183:327–39.
[294] Li J, Qiao Z, Simeone A, Bao J, Zhang Y. An activity theory-based analysis [323] Qu X, Jain A, Rajput NN, Cheng L, Zhang Y, Ong SP, et al. The Electrolyte Genome
approach for end-of-life management of electric vehicle batteries. Resour Conserv project: a big data approach in battery materials discovery. Comput Mater Sci
Recycl 2020;162:105040. 2015;103:56–67.
[295] Yang S, He R, Zhang Z, Cao Y, Gao X, Liu X. CHAIN: cyber hierarchy and [324] Li K, Zhou P, Lu Y, Han X, Li X, Zheng Y. Battery life estimation based on cloud
interactional network enabling digital solution for battery full-lifespan data for electric vehicles. J Power Sources 2020;468:228192.
management. Matter 2020;3:27–41. [325] Fetene GM, Kaplan S, Mabit SL, Jensen AF, Prato CG. Harnessing big data for
[296] Wu B, Widanage WD, Yang S, Liu X. Battery digital twins: perspectives on the estimating the energy consumption and driving range of electric vehicles.
fusion of models, data and artificial intelligence for smart battery management Transport Res Transport Environ 2017;54:1–11.
systems. Energy and AI 2020;1:100016. [326] Pop C, Cioara T, Antal M, Anghel I, Salomie I, Bertoncini M. Blockchain based
[297] Li S, Li J, Hanxiao W. Big data driven Lithium-ion battery modeling method: a decentralized management of demand response programs in smart energy grids.
Cyber-Physical System approach. In: 2019 IEEE international conference on Sensors 2018;18.
industrial cyber physical systems. ICPS); 2019. [327] Di Silvestre ML, Gallo P, Guerrero JM, Musca R, Riva Sanseverino E, Sciumè G,
[298] Li S, He H, Li J. Big data driven lithium-ion battery modeling method based on et al. Blockchain for power systems: current trends and future applications.
SDAE-ELM algorithm and data pre-processing technology. Appl Energy 2019;242: Renew Sustain Energy Rev 2020;119:109585.
1259–73. [328] Hua S, Zhou E, Pi B, Sun J, Nomura Y, Kurihara H. Apply blockchain technology
[299] Li W, Rentemeister M, Badeda J, Jöst D, Schulte D, Sauer DU. Digital twin for to electric vehicle battery refueling. In: Proceedings of the 51st Hawaii
battery systems: cloud battery management system with online state-of-charge international conference on system Sciences; 2018.
and state-of-health estimation. J Storage Mater 2020;30:101557. [329] Florea BC. Electric vehicles battery management network using blockchain IoT.
[300] Hou J, Song Z. A hierarchical energy management strategy for hybrid energy In: 2020 IEEE international conference on automation, quality and testing,
storage via vehicle-to-cloud connectivity. Appl Energy 2020;257:113900. robotics (AQTR). IEEE; 2020.
[301] Lu Y, Li K, Han X, Feng X, Chu Z, Lu L, et al. A method of cell-to-cell variation [330] Florea BC, Taralunga DD. Blockchain IoT for smart electric vehicles battery
evaluation for battery packs in electric vehicles with charging cloud data. management. Sustainability 2020;12:12103984.
eTransportation 2020:100077. [331] Fu Z, Dong P, Ju Y. An intelligent electric vehicle charging system for new energy
companies based on consortium blockchain. J Clean Prod 2020;261:121219.
24
[332] Hu X, Xu L, Lin X, Pecht M. Battery lifetime prognostics. Joule 2020;4:310–46. [339] Chen Z, Mi CC, Fu Y, Xu J, Gong X. Online battery state of health estimation based
[333] Miglani A, Kumar N, Chamola V, Zeadally S. Blockchain for internet of energy on Genetic Algorithm for electric and hybrid vehicle applications. J Power
management: review, solutions, and challenges. Comput Commun 2020;151: Sources 2013;240:184–92.
395–418. [340] Chemali E, Kollmeyer PJ, Preindl M, Emadi A. State-of-charge estimation of Li-ion
[334] Mokhtari Melissa. A new approach to Decentralized energy. 2017. https://me batteries using deep neural networks: a machine learning approach. J Power
dium.com/blockchain-at-berkeley/a-new-approach-to-decentralized-energy-5ab Sources 2018;400:242–55.
2b5460fa0. [Accessed 26 September 2017]. [341] Hu X, Li SE, Yang Y. Advanced machine learning approach for lithium-ion battery
[335] Junping W, Quanshi C, Binggang C. Support vector machine based battery model state estimation in electric vehicles. IEEE Trans Transp Electrif 2016;2:140–9.
for electric vehicles. Energy Convers Manag 2006;47:858–64. [342] Densmore A, Hanif M. Determining battery SoC using electrochemical impedance
[336] Klass V, Behm M, Lindbergh G. Capturing lithium-ion battery dynamics with spectroscopy and the extreme learning machine. In: 2015 IEEE 2nd international
support vector machine-based battery model. J Power Sources 2015;298:92–101. future energy electronics conference (IFEEC). IEEE; 2015. p. 1–7.
[337] Wang Q-K, He Y-J, Shen J-N, Ma Z-F, Zhong G-B. A unified modeling framework [343] Zhang Y, Xiong R, He H, Pecht MG. Long short-term memory recurrent neural
for lithium-ion batteries: an artificial neural network based thermal coupled network for remaining useful life prediction of lithium-ion batteries. IEEE Trans
equivalent circuit model approach. Energy 2017;138:118–32. Veh Technol 2018;67:5695–705.
[338] Wang Y, Tian J, Sun Z, Wang L, Xu R, Li M, et al. A comprehensive review of [344] Liu K, Li Y, Hu X, Lucu M, Widanage WD. Gaussian process regression with
battery modeling and state estimation approaches for advanced battery automatic relevance determination kernel for calendar aging prediction of
management systems. Renew Sustain Energy Rev 2020;131:110015. lithium-ion batteries. IEEE Trans Ind Inf 2020;16:3767–77.
25

Advaned Bms

Uploaded by

Copyright:

Available Formats

Advaned Bms

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advaned Bms

Uploaded by

Copyright:

Available Formats

Renewable and Sustainable Energy Reviews xxx (xxxx) xxx

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

Advanced battery management strategies for a sustainable energy future:

generations: “no management”, “simple management”, and “advanced

Abbreviations ML machine learning

Fig. 1. The evolution of battery management technology.

Fig. 2. Performance requirements from EVs to battery management.

Fig. 3. Hierarchical architecture of the advanced battery management technology.

Table 2 SOT estimation is at an intermediate time scale. Battery aging is a

Fig. 8. A diagram of battery multi-state joint

Fig. 12. Battery safety management and aging management.

You might also like