SchoolofInformationEngineering 7420190042 WahhabMdAbdul ShunliWang

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/348150522
Master's Degree Final Defense PPT
Presentation · January 2020

DOI: 10.13140/RG.2.2.34662.65608/1
CITATIONS READS
0 2,350
2 authors:
Md Abdul Wahhab Shunli Wang

Pegatron Southwest University of Science and Technology
24 PUBLICATIONS 0 CITATIONS 5,011 PUBLICATIONS 2,463 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
001_JCYJS_BMS View project
Research on Ultra-Wide Spectrum Terahertz Source Space Power Synthesis Technology and Radiation Characteristics (No.11176032) View project
All content following this page was uploaded by Md Abdul Wahhab on 12 February 2021.
The user has requested enhancement of the downloaded file.

学术型硕士研究生学位论文
ACADEMIC MASTER DISSERTATION
(Master of Science in Control Science and Engineering)
论文题目 Thesis Title Accurate power state

estimation situated by extended Kalman filtering
and equivalent circuit modeling of high energy
lithium-ion batteries
作者姓名 Name Wahhab Md Abdul

学号 Student’s No. 7420190042
2021 年year 01 月month

Classification code: School code: 10619
UDC: Security classification:
西南科技大学硕士学位论文

Thesis Title: Accurate power state estimation situated by extended

Kalman filtering and equivalent circuit modeling of high energy
lithium-ion batteries
Candidate Name: Wahhab Md Abdul
Major：Control Science and Engineering
Supervisor Name and Title: Wang Shunli, Associate Professor
Name of School: School of Information Engineering
Date of Dissertation Defense

2020 Year 12 Month 24 Day
I
分类号：学校代码: 10619
UDC: 密级:
西南科技大学学术型硕士学位论文

(控制科学与工程理学硕士)
论文题目: 基于等效电路模型和扩展卡尔曼滤波算法的高比能
锂离子电池准确功率状态估计
作者:Wahhab Md Abdul
学科(专业)或领域名称:控制科学与工程
指导教师姓名、职称: 王顺利，副教授
学生所属培养单位:信息工程学院
论文答辩日期
2020年 12月 24日
II
西南科技大学学术型硕士学位论文

Topic
话题
Accurate power state estimation situated by extended
Kalman filtering and equivalent circuit modeling of high
energy lithium-ion batteries
Candidate 申请人: Wahhab Md Abdul
Supervisor 导师: Wang Shunli, Associate Professor
January 2021
2021 年 01 月
III
摘要
摘要
锂离子电池（Lithium-ion Battery, LIB）因具有能量密度高，体积小和可循环使用
等优点，已逐渐成为新能源汽车（Energy Vehicles, EVs）的主要供能设备。经研究表
明，对锂电池合理的使用可延长其使用寿命，达到节能减排的目的。锂离子电池的内
部结构复杂，其荷电状态（SOC）的估算会受到许多条件的影响，例如放电电流，自
放电效应，内部温度，外部环境温度和电池老化等因素。这使得其荷电状态难以准确
估计。然而，准确的 SOC 估算是电动汽车（EVs）中使用的能量控制技术的象征性指
标，可以反映其剩余容量，对电动汽车的稳定高效运行具有重要意义。通过准确的剩
余容量，可以规划电池使用策略，使锂电池的工作状态达到最佳。本文确定了动力锂
离子电池精确 SOC 估计的以下事项。
1、选择高阶电池模型进行状态表征。
2、运用数学理论推导分析建立空间状态方程。
3、在不同环境温度条件下进行混合动力脉冲能力特性（HPPC）实验。
4、运用一种扩展卡尔曼滤波算法及其迭代计算方法已解决电池的非线性问题。
5、将在 Simulink 中构建的模型与高阶模型进行比较，已验证模型精确性。
6、通过 S-function 实现算法并验证模型参数辨识精度。
选择根据电池生产参数提取的高阶等效电路模型，对该模型通过与其欧姆内阻串联标
准 2 阶 RC 进行改进。通过混合动力脉冲能力特性测试，对等效电路模型中的 2 阶 RC
值进行参数辨识。为锂电池 SOC 估算提供模型基础。基于开路电压与荷电状态的时域
关系进行参数拟合，并运用扩展的卡尔曼滤波算法。仿真结果表明，该优化方法可以
显着提高 SOC 估计精度，使得电压仿真误差仅为 0.051V。并通过扩展卡尔曼滤波算
法将初始误差降低到小于 0.029V。
关键词：锂离子电池；高阶动态建模；开路电压；参数提取；HPPC 测试；扩展卡尔
曼滤波算法。
论文类型：应用研究
IV
ABSTRACT
ABSTRACT
The lithium-ion (Li-ion) battery has gradually become the preferred power supply for new
energy vehicles for the advantages of high energy, small size, and recharge-ability. Over time,
it is observed that the reasonable use of the battery can serve to prolong its service lifespan,
energy-saving, emission reduction, and unnecessary losses. The internal structure of the Li-
ion battery is complex, so the level for the state of charge (SOC) is affected by various complex
factors such as discharging current, self-discharge effect, internal temperature, external
environment temperature, and battery aging. It makes the state value difficult to be estimated
accurately. However, an accurate SOC estimation charging status is a symbolic indication of
the energy control technique used in electric vehicles (EVs), which can reflect its residual
capacity. Through the accurate residual capacity, the battery application strategy can be
planned to realize the battery operation of the best condition. In this thesis following matters
are determine for the precise SOC estimation of power Li-ion battery.
1. The high-order battery model is picked.

2. State-space factors are analyzed by mathematically derivate.
3. The HPPC experiment is carried out at ambient temperature.
4. To deal with non-linear behavior, an extended Kalman filter algorithm is presented
with its mathematical subscription.
5. To verify model accuracy, the constructed model in Simulink compared with the
high-order model.
6. The algorithm implemented by S-function and verify parameters.
In the detailed, the batteries productive parameter extraction is picked a high-order equivalent
circuit modeling, which has been extended by standard 2-RC pairs in series with its ohmic
internal resistance. The RC (resistance and capacitance) pairs values have been calculated
mathematically by solving the two states high-order circuit working conditions. Furthermore,
the hybrid pulse power characterization (HPPC) experiment helps to parameter identification
which is proposed to estimate the SOC. Then the calculation is performed by open-circuit
voltage (OCV) and an algorithm extended Kalman filter (EKF) is established. The simulation
results showed 0.051V that the optimization method can improve the SOC estimation
precision remarkably. In the extended calculation process, the iterative prediction and
correction strategies are introduced into the test, and the initial error is reduced by less than
0.029V convergence by an extended Kalman filtering algorithm.
KEYWORDS: Li-ion batteries; high-order modeling; open-circuit voltage; parameter

extraction; HPPC test; extended Kalman filter.
TYPE OF THESIS: Application Research
V
Thesis originality statement
独创性声明
本人声明所呈交的论文是我个人在导师指导下进行的研究工作及取得的研究
成果。尽我所知，除了文中特别加以标注和致谢的地方外，论文中不包含其他
人已经发表或撰写过的研究成果。与我一同工作的同志对本研究所做的任何贡
献均已在论文中作了明确的说明并表示了谢意。本人依法享有和承担由此论文
所产生的权利和责任。
Thesis originality statement
I solemnly declare that the paper submitted is personal research work which has been
carried out under the guidance of my supervisor. As far as I know, except where specifically
noted and appreciated the in text, the paper does not contain the research results published or
written by others, nor does it contain materials used for obtaining degrees or certificates from
the Southwest University of Science and Technology or other educational institutions. As
working with my research team I have made a clear statement, and thanked the contribution
of learning. I enjoy and bear the rights and responsibilities arising from the thesis in accordance
with the law.
论文作者签名：日期：2020 月 12 日 30
Graduate Signature: Date：year / month / day
VI
Statement on the use of the License
学位论文版权使用授权声明
本人完全了解西南科技大学有关保留、使用学位论文的规定，即：研究生在
西南科技大学学习和工作期间论文工作的知识产权单位属于西南科技大学。同
意授权西南科技大学保留并向国家有关部门或机构送交该论文的印刷本和电子
版本，允许该论文被查阅和借阅。同意授权西南科技大学可以将本学位论文的
全部或部分内容编入有关数据库进行检索，可以采用影印、缩印、扫描或数字
化等其他复制手段保存和汇编本学位论文。保密论文在解密后遵守此规定。
Statement on the use of the License
I fully understand the southwest university of science technology rules and regulation
for the use of paper, namely: The intellectual property unit that the graduate student research
works while studying and working at Southwest university of science and technology belongs
to the university. I am agree to authorize Southwest university of science and technology to
retain and submit printed and electronic versions of the paper to the relevant state departments
or agencies, allowing the paper to be consulted and borrowed. And also I authorized the
university to compile all or part of this Thesis into relevant database for retrieval. The Thesis
can be saved and compiled by other means of reproduction, such as photocopying, compact
printing, scanning or digitization. The confidential Thesis shall comply with this regulation
after being declassified.
论文作者签名：日期：2020 年 12 月 30
Graduate Signature： Date: year / month / day
指导导师签名：日期:2020 年 12 月 30
Supervisor signature： Date: year / month / day
VII
Catalogue
目录 Catalogue
CHAPTER 1 INTRODUCTION ...................................................................................... 1
1.1 Thesis Motivation ..................................................................................................... 1
1.2 Thesis Objective ....................................................................................................... 1
1.3 Problem Formulation ................................................................................................ 3
1.4 Thesis Organization .................................................................................................. 4
CHAPTER 2 MATHEMATICAL ANALYSIS .............................................................. 6
2.1 Operating Characteristic of Li-ion battery ............................................................... 6
2.1.1 Operating principle Li-ion batteries ..................................................................... 6
2.1.2 Prolong the life cycles .......................................................................................... 7
2.1.3 Critical Parameters Description............................................................................ 8
2.2 Critical Indicators for State of Charge Estimation ................................................... 9
2.2.1 Temperature Effect on Li-ion Batteries................................................................ 9
2.2.2 Rate of Charge and Discharge .............................................................................. 9
2.2.3 Aging Degree of Battery .................................................................................... 10
2.3 State of Charge Estimation Strategy ...................................................................... 10
2.3.1 Characterization State of Charge Estimation ..................................................... 10
2.3.2 Transformation of Coordinate ............................................................................ 11
2.3.3 Algorithm for Binary Iteration ........................................................................... 12
2.4 The Iterative filtering algorithm strategy ............................................................... 12
2.4.1 Kalman Filtering Expansion Procedure ............................................................. 13
2.4.2 Extended Kalman Filtering Expansion Procedure ............................................. 14
2.4.3 Estimation Model Construction.......................................................................... 16
2.5 Summary of this chapter......................................................................................... 18
CHAPTER 3 MODEL BUILDING AND REALIZATION ........................................ 19
3.1 The Battery Modeling Method Approaches ........................................................... 19
3.1.1 Comparison of The Batteries Modeling Systems ............................................... 19
3.1.2 Electrochemical Battery Model .......................................................................... 21
3.1.3 Equivalent Circuit Battery Model ...................................................................... 21
3.2 Evaluation of Equivalent Circuit Model ................................................................ 22
3.2.1 Internal Resistance Modeling Description ......................................................... 22
3.2.2 Thevenin Electrical Battery Modeling Description............................................ 23
3.2.3 PNGV Electrical Battery Modeling Description ................................................ 24
3.3 Establishment of High-Order Battery Model ......................................................... 24
3.3.1 High-Order Electrical Modeling ........................................................................ 25
3.3.2 Ohmic Internal Resistance Methods................................................................... 26
3.3.3 State Space Iteration Process .............................................................................. 27
3.4 Model Parameter Description ................................................................................. 29
3.4.1 Ampere-Hour Counting (Ah) Method ................................................................ 29
VIII
Catalogue
3.4.2 Exponential Curve Fitting Technique ................................................................ 30

CHAPTER 4 EXPERIMENTAL VERIFICATION .................................................... 33
4.1 Build Experiment Platform and Data Acquisition ................................................. 33
4.1.1 Real-Time Platform Implementation.................................................................. 33
4.1.2 Hybrid Pulse Power Characteristic Test ............................................................. 34
4.1.3 Capacity Tracking Experiments ......................................................................... 36
4.2 Model Parameter Identification .............................................................................. 37
4.2.1 Parameters Extraction Results ............................................................................ 37
4.2.2 Open-Circuit Voltage Characterization .............................................................. 40
4.3 Detection of High-Order Model Verification ......................................................... 42
4.3.1 High-order Model Validation Strategies ............................................................ 42
4.3.2 Corroboration of Model Voltage Parameters ..................................................... 43
4.4 SOC Estimation Effect Based on EKF ................................................................... 45
4.4.1 Simulation Interfacing Process ........................................................................... 45
4.4.2 The approach of Algorithm Verification Strategy ............................................. 48
CHAPTER 5 CONCLUSION AND RESULT .............................................................. 50
ACKNOWLEDGMENT ...................................................................................................... 51
REFERENCES ..................................................................................................................... 52
IX
List of Figures
List of Figures
Figures List Page No
Figure 1-1: Essential research contents of the thesis.............................................................. 2
Figure 1-2: Major working procedure for SOC estimation approach ..................................... 3
Figure 2-1: Schematic structure of a Li-ion battery ................................................................ 6
Figure 2-2: Voltage variation of SOC estimation waveform ................................................. 11
Figure 2-3: Iterative SOC estimation technique .................................................................... 12
Figure 2-4: Flowchart process of the Kalman Filtering algorithm........................................ 13
Figure 2-5: The processing and updating processes of EKF ................................................. 16
Figure 2-6: Double unscented transform Extended Kalman Filtering process ..................... 16
Figure 2-7: Updates iteration calculation of EKF ................................................................. 17
Figure 2-8: Overview of mathematical analysis and algorithm ............................................ 18
Figure 3-1: The classifications of battery modeling methods ............................................... 19
Figure 3-2: Connection procedure of the modeling system .................................................. 20
Figure 3-3: Schematic diagram of Rint model ....................................................................... 22
Figure 3-4: Schematic diagram of the Thevenin model ........................................................ 23
Figure 3-5: Schematic diagram of the PNGV model ............................................................ 24
Figure 3-6: Essential steps for selecting appropriate models ................................................ 25
Figure 3-7: Schematic diagram of the high-order modeling ................................................. 25
Figure 3-8: Relation between ohmic resistance with SOC ................................................... 27
Figure 3-9: Working steps for identification parameter ........................................................ 28
Figure 3-10: Curve fitting method for identification parameters .......................................... 31
Figure 3-11: Overview of battery modeling system and techniques ..................................... 31
Figure 4-1: The schematics of the battery test platform........................................................ 33
Figure 4-2: Hybrid pulse characteristic test procedure ......................................................... 35
Figure 4-3: Complete current and voltage discharge profile during the HPPC test .............. 35
Figure 4-4: Voltage and Current transition waveform .......................................................... 36
Figure 4-5: Capacity and Energy transition waveform ......................................................... 37
Figure 4-6: High-order resistance fitting waveform ............................................................. 39
Figure 4-7: High-order resistance fitting waveform ............................................................. 40
Figure 4-8: Open circuit voltage and state of charge calibration .......................................... 40
Figure 4-9: Specific open-circuit voltage at different SOC .................................................. 41
Figure 4-10: Top view structure of terminal voltage verification model .............................. 42
Figure 4-11: Inner structure of ampere-hour (Ah) integral model ....................................... 43
Figure 4-12: The central part of inside the high-order model in Simulink ........................... 43
Figure 4-13: Waveform of simulation comparison ............................................................... 44
Figure 4-14: Error indication of simulation result ................................................................ 44
Figure 4-15: Structure of high-order model with EKF based simulation approach .............. 45
Figure 4-16: EKF estimation results with SOC .................................................................... 48
Figure 4-17: EKF estimation error with SOC ....................................................................... 48
Figure 4-18: Overview of experimental verification and verification .................................. 49
X
List of Tables
List of Tables
Tables List Page
Table 2-1: Life cycle feature of the depth of discharge........................................................... 7
Table 2-2: The major parameters of Li-ion batteries ............................................................... 8
Table 3-1: Advantages and disadvantages of the modeling methods.................................... 20
Table 4-1: Key function of battery parameters...................................................................... 34
Table 4-2: Discharge-profile criteria under specific SOC levels .......................................... 39
XI
CHAPTER 1 INTRODUCTION
1.1 Thesis Motivation
The global attention in the direction of battery electric vehicles (EVs) is gaining much
popularity as they are the sustainable mode of transportation and the alternatives of the
international combustion engine (ICE) based vehicles [1]. The EVs effectively overcome the
fossil fuel crisis and environmental pollution, considering the major hurdle for the automobile
sector [2]. As the only power supply in pure EVs, the pack capacity is of great importance. In
this case, the superiority of high specific energy and power Li-ion batteries is widely used in
EVs [3]. So, the Li-ion batteries have increased in popularity since commercialized in 1991.
Due to their advantages such as lightweight, fast charging, high energy density, low self-
discharge, and long lifespan [4]. The usable energy classified as the state of charge (SOC) is
identical to that of a vehicles fuel gauge operated by a combustion engine [5]. The safe
operation of batteries and efficient energy management strategy and the SOC accuracy
estimation is essential for battery management systems (BMS) [6]. However, the evaluation
of battery SOC is challenging in uncertain and complex EVs environments [7]. Therefore, to
manage the Li-ion more efficiently and improve battery performance, it is necessary to obtain
the inner state parameters and accurately make an accurate SOC [8]. The SOC is not directly
measurable and requires a particular type of method to be estimated but often uses some
specific models such as the Empirical model, the equivalent circuit model, and the
electrochemical model [9]. Deterioration of battery cells leading to reduced performance is a
problem that limits battery life.
Moreover, Equivalent circuit modeling (ECM) batteries have been investigated

specifically for applications such as BMS development and vehicle power management
control [10]. A good battery model can estimate the battery storage power details and the
voltage response to the load [11]. The ECMs can describe those characteristics for the Li-ion
batteries [12]. In some cases, modeling the side reactions is required in terms of battery losses,
which can also be realized with the ECMs [13]. A novel approach for identifying parameters
requires a high-order equivalent circuit proposed in this literature with improved mathematical
analysis. After the extended Kalman filter (EKF) algorithm is proposed to estimate the SOC
estimation in the real-time solid data. The validation results show that the proposed methods
are a good performance in estimation accuracy and uncertainty of the parameters.
1.2 Thesis Objective
Electrical vehicles powered by Li-ion batteries are more environmentally friendly than
gasoline-powered ones, and this kind of environmentally friendly energy is gradually widely
used in various fields. In such high demand, the battery state is significant because it can be
estimated. To obtain to support the EVs driving distance, the influence of such factors as long-
1
西南科技大学（硕士）学位论文 / Southwest University of Science and Technology Master's Thesis
term use can lead to a measurement error cumulate expansion due to the voltage, resistance,
and temperature [14]. The highly nonlinear characteristic also brings to the battery state
estimation lots of difficulties. Because the power batteries operating characteristics are highly
nonlinear, it is necessary to establish an accurate battery model to estimate the accurate state.
To solve this issue, several algorithms are proposed and used to estimate the battery state,
including the current integration, open-circuit voltage, Kalman filtering algorithms, and
Neural Network [15]. The current integral method neglects the influence of the battery self-
discharging current rate, aging degree, and charge-discharge current rate on the battery state.
Long-term use leads to the accumulation and expansion of measurement error, so it is
necessary to introduce relevant correction factors to correct the accumulated error. However,
to aims SOC estimation desire accurately, the following factors are determined and the core
research objective
1. An accurate state of charge (SOC) estimation.
2. Understand the dynamic characteristics of battery behavior.

3. Investigate the internal reaction and temperature.
4. Monitoring and control charge and discharge flow.
5. Make an effective BMS for power Li-ion batteries.
In this thesis, the above objects are analyzed in various ways to determine a Li-ion ternary
battery for EVs as the research objective. The entire process is carried out by the offline
process, and the process is mainly established in the laboratory environment and use room
temperature. To organize and established the aims for the thesis, the following in Figure 1-1
is chosen essential contents of the thesis.
Ampere Hour Method
High-order Equivalent
Circuit Model
Mathematical Analysis
Test Platform Design and

Thesis Implementation
Contents
Hybrid Pulse Test and
Parameter Extraction
Simulation Model Design

and Results Analysis
Establishment of Taylor
Expansion (EKF)
Figure 1-1: Essential research contents of the thesis
2
To perform and validation of the research objective, the above thesis contents help to carry
out the necessary procedure. The Ampere hour method commonly known as the coulomb
counting method is to perform the first stage. Then similarly model selection, state-space
equation, and so on. The high-order improved equivalent circuit model is used as the
significant model to get the parameters. Furthermore, using those parameters used for SOC
estimation, which commonly defined as the percentage of the maximum possible charge
present inside a rechargeable battery from 0% to 100%. The detailed thesis contents are briefly
introduced in Figure 1-2.
Measures the
Ampere Hour Describe Indicate the SOC
discharging
(Ah) Strategy remaining energy ratio
current
“CT-4616- TCP/IP port

Experimental Collaborate with
5V100” as a connect with host
Platform Design host computer
equipment computer
Selected ternary Input expected

Pules Test After finish save
battery as a test parameters into
(HPPC) the real-time data
object BTS-7.6 software
Parameters
Equivalent Picked up a High- Investigate the
identify based on
Circuit Model order Model internal character
formulas
Extended State-space model

Linearization and
Kalman Filter observation by
Prediction of SoC
Update with SOC
Algorithm Taylor series
Simulation Design a
Design a Write code in
simulation model
Design in based on High-
Simulink model Mathlab based on
Mathlab based on EKF EKF algorithm
order model
Figure 1-2: Major working procedure for SOC estimation approach
In the above Figure 1-2 demonstrated the following steps by steps the working process has
carried out. However, various experimental methods are integrated, such as the hybrid charge-
discharge, constant-voltage charging with different multipliers, constant discharge, and cyclic
discharge shelved experiments. It is used to study the Li-ion battery operating characteristics
and analyze the characteristics of the battery response. Besides, its long-term use accuracy in
estimating the state of the same battery group can be greatly reduced. The complex effect of
factors such as battery temperature, self-discharging current rate, and the varying degree of
battery aging that those contents have been described elaborately with its internal formulation.
1.3 Problem Formulation
Electric vehicles (EVs) and hybrid vehicles (HVs), owing to the benefits of reduced
emission and energy conservation, are recognized worldwide as one of the significant growth
paths of automotive industries. The Li-ion batteries are the necessary power source for EVs
due to their high energy and energy density. Therefore, it uses rechargeable batteries to
3
maintain permanent energy supply—various types of rechargeable industrial batteries, such as

lead-acid, Ni-MH, Ni-Cd, and Li-ion. However, Li-ion battery production seems to be more
suitable than other batteries. In such cases high power density, high voltages, critical
load/unload cycles, and safety are useful features. The battery charging status must also be
controlled explicitly for the battery SOC estimation. The battery management system (BMS)
configured algorithms to allow better utilization and long battery life.
The crucial role of the BMS is to control the remaining power. Accuracy in the
measurement of the SOC makes it easy for the battery to control adequate electricity, prevents
irreversible damage to the internal battery device, and ensures maximum use. Many SOC
estimation approaches are available for electrochemical accounting, model-based, and black-
box often referred to as data-oriented. It aims to develop a solution that honors precision and
versatility as target the Li-ion batteries deployment on the BMS. In this idea, an effective SOC
estimation algorithm is based on a coulomb-counting algorithm, an accounting approach. It
was using a piecewise linear interaction mapping between SOC and OCV to solve this
downside approach and increase its accuracy.
1.4 Thesis Organization
The thesis includes the chapters listed below,
Chapter 1 : INTRODUCTION – The research motivation and objective of Li-ion battery

technology for state of charge (SOC) is elaborately discussed in the section. At the time, the
problem formulation research objective and implementation technique of this thesis are
broadly explained.
Chapter 2 : MATHEMATICAL ANALYSIS – The chapter broadly discussed mathematical

analysis due to state of charge estimation approaches for the Li-ion batteries. This section
included the essential characteristic of Li-ion batteries. The critical factors of the state of
charge (SOC) estimation approach are discussed briefly. The SOC estimation technique with
coordinate transformation and the binary iterative algorithm presented in the third part.
Furthermore, the Kalman filter and extended Kalman filtering algorithm were also introduced
mathematically in the portion.
Chapter 3 : MODEL BUILDING AND REALIZATION – The various types of battery

models in details are introduced in the chapter with its mathematical analysis. There are three
types of battery modeling with high-order battery modeling, which is the core topic in this
section. The mathematical description has briefly been described in part. Also, parameter
description techniques are widely present, and this segment morality that disported every
notable representation is also introduced in the chapter.
Chapter 4 : EXPERIMENTAL VERIFICATION – The experimental platform design, data

collection tactics with HPPC, capacity test, parameter identification results, and simulation
4
verification experiments were extensively incorporated in this chapter. After then, compared
with the real SOC and model SOC is verified using the high-order equivalent circuit is
presented. Furthermore, the extended Kalman filter (EKF) algorithm is established with
mathematical description and model building in simulation to get the accurate SOC also
discussed in the chapter.
Chapter 5 : CONCLUSION AND RESULT – Summarized the entire graduation thesis design
work, reviewed the all contents and effect of the entire design and the difficulties encountered
in the design process, and proposed some improvement measures.
5
CHAPTER 2 MATHEMATICAL ANALYSIS
2.1 Operating Characteristic of Li-ion battery
In global high-tech growth, the Li-ion battery industry is an excellent direction. The
benefits of Li-ion high oil, high electricity, high conversion rates, long life cycles, and no
emissions. Thanks to its good electrochemical durability, high density of fuel, long service life,
and lack of maintenance, often used in electric cars and various power storage systems. The
Li-ion batteries currently offer a wider variety of applications. It primarily encompasses five
areas: transport, power storage, mobile communications, modern energy storage, and military
aerospace. It will substitute oil with power, minimize emissions of greenhouse gas, and store
electricity from the grid for unnecessary use of electric cars.
2.1.1 Operating principle Li-ion batteries
The batteries referred to in this article are all Li-ion batteries. The positive electrode
substrate of a Li-ion battery is a lithium-alloy metal oxide [16]. Its negative electrode content
is graphite. The charging and discharging phase understand a chemical reaction and the
essential response and working process of Li-ion in Figure 2-1.
Load e-
+ -
Positive Negative
e- electrode
electrode Power
Shell
Charge
Li+
Discharge
Electrolyte
CoO2 Electrolyte C
Septum
Figure 2-1: Schematic structure of a Li-ion battery
In Figure 2-1 shows that the Li-ion metal oxide positive electrode material is emitted during
the batteries charging phase. The lithium-cobalt metal oxide transfer is incorporated in the
negative electrodes coated framework via the electrolyte and the separator, and the carbon
substrate. The positive electrode enters a Li-ion rich, prosperous state at this stage. In contrast,
the negative electrode enters a Li-ion lean state [17]. During the discharge process, Li-ion can
be released from the negative electrode carbon material layer. The carbon content coating
would be deposited in the negative electrode crystal lattice since going through the separator
and the electrolyte [18]. A positive electrode is progressing towards a Li-rich state. A negative
electrode is progressing towards a lithium-poor state. The chemical quality of the electrode
6
continues to recover. The complete equation of the reaction is positive and negative Li-ion
electrodes. Each battery is shown in the formula (2-1).
positive：LiCoO ⎯⎯⎯ ⎯⎯⎯⎯ch arg e

⎯→ Li1− xCoO2 + xLi + xe −
 2 disch arg e
 − ⎯⎯⎯⎯ ch arg e
→ LixC6
 Negative : xLi + xe + 6C ⎯⎯⎯ ⎯ (2-1)

disch arg e
⎯⎯⎯⎯
All : LiCoO2 + 6C ⎯⎯⎯
ch arg e
⎯→ Li1− xCoO2 + LixC6
 disch arg e
Wherein formula (2-1), it can be shown that the internal resistance, polarization, aging, and
other factors are added in the process of the Li-ion battery. The embedding and unpacking
going through the separator, and traveling in the electrolyte can impact the calculation of SOC
[19]. These considerations and the SOC estimate need to be addressed in the model
development process.
2.1.2 Prolong the life cycles
The electrodes design can also be strengthened. The batteries capacity can be increased
by testing the nature and scale of materials used for batteries construction. The chemical
reactions at electrodes take place inside the Li-ion batteries. The electrodes internal resistance
can decrease the heat produced during use. The cell potential and lifespan can also be increased
[20]. Electrode sizes should be minimized to increase performance. It was discovered that
there were limitations to the reduction, such that if the electrode thickness had fallen below.
Then the batteries could no longer meet the necessary energy requirements [21]. There are
additional trade-offs for the thinner content size, including rising processing costs. The
electrode scale can be included as the primary design element in the whole devices overall
design presented in Table 2-1.
Table 2-1: Life cycle feature of the depth of discharge
Depth of Discharge Discharge Cycles
100 300-600
80 400-900
60 600-900
40 1000-3000
20 2000-9000
10 6000-15000
In the design of the electrodes, another region of development is made. By changing the
materials as made, ion journeys may be optimized to maximize overall conductivity and
thermal capacity on the transport distances or routes. In certain situations, the additional
7
advantage of these improvements was to increase the mechanical quality and the heat
efficiency or overall power of the mechanical. Gold, graphene, yttrium, and zinc are used in
the components applied to the electrodes [22]. Several various influences influence this
mechanism, and further study is underway.
2.1.3 Critical Parameters Description
There are unknown safety risks in using Li-ion batteries. It is essential to consider Li-ion
batteries critical criteria to prevent explosion and fire caused by Li-ion batteries failure [23].
To protect against unwanted circumstances, it is obligatory to measures some critical action.
To observe the critical situation, in Table 2-2 parameters are necessary to take care of the high-
power Li-ion for the electric vehicles.
Table 2-2: The major parameters of Li-ion batteries
Parameters Name Parameter definition
The amount of active material determines the batteries capacity,

Battery Capacity usually expressed in Ah. For example, 1000mAh can be converted
into a charge of about 3600C at a current of 1A for 1h.
The possible differentiation throughout the batteries between

Nominal Voltage
positive and negative electrodes is the nominal batteries voltage.
The active material on the polar plate has reached saturation when
End of Charge
the rechargeable batteries are fully charged. It then begins to
(EOC) Voltage
charge, and the batteries voltage will not increase.
When the batteries are fully discharged, the end-of-discharge

Discharge end
voltage corresponds to the lowest voltage permissible. Discharge
Voltage
end voltage is compared to the rate of discharge.
The amount of the batteries overall power expended when it is not

Self-discharge Rate used for any time. Generally, at an average temperature, the self-
discharge rate of Li-ion batteries is 5%-8%.
In the above Table 2-2, show that five critical parameters are deeply affected in the
researchable Li-ion batteries. The battery capacity is a significant factor for storage capability,
which is expressed in milliampere-hour/mAh. The nominal voltage is the rated voltage which
is determined by the manufacturer. The end of charge voltage when batteries charge is almost
near to finish then has appeared. While the batteries are completely discharged, and the EOD
voltage associate with the lowest voltage permissible. The self-discharge ratio is the amount
of the batteries overall power and the rate is obtained 5%-8%.
8
2.2 Critical Indicators for State of Charge Estimation
State of Charge (SOC) estimation is an obligatory part of precise state estimation for the
Li-ion batteries. However, some critical influences are battery-related to estimating factors.
This part of several influences is discussed elaborately in the portion.
2.2.1 Temperature Effect on Li-ion Batteries
As the outside temperature influences the Li-ion batteries body electrochemical reaction,
the SOC evaluation accuracy will still be impaired. The higher the temperature and the higher
the degree of electrochemical reaction and the batteries energy will also rise [24]. These factors
are creating much high-temperature gas, destroys the batteries, speeds up the batteries aging,
and affects the batteries lives. However, only by seeking a suitable temperature will the
batteries output be taken into full play, and the batteries charged and discharged efficiently
[25].
The battery size (how many amp-hours it can hold) is decreased as the temperature
decreases and increases as the temperature rises. That is why the car battery fails on a cold
winter morning, even though it worked the previous day perfectly. Suppose the batteries spend
most of the year shivering in the cold. In that case, the decreased power must be taken into
consideration when sizing the machine batteries. The typical value for batteries is 25 degrees
C (about 77 degrees F) at room temperature. Approximately -22 degrees F (-30 C) the battery
power of Ah is reduced to 50 percent. At freezing, power is decreased by 20%. Capacity is
improved at higher temperatures at 122 degrees F. The capacity of the battery will be around
12 percent higher.
2.2.2 Rate of Charge and Discharge
The charge and discharge current are the charging rate and discharge rate for the Li-ion
batteries. In the same period, the charge and discharge rates for the Li-ion batteries will change
subsequently. The charge quantity which the batteries will release is usually inversely
proportional to the loading and discharge rate [26, 27].
The C-rates is to regulate the charge and discharge rates of the battery. The battery
capacity is usually rated at 1C, which means that the fully charged battery rated at 1Ah can
supply 1A for one hour. The same battery discharging at 0.5C should deliver 500mA for two
hours, and 2C should deliver 2A for 30 minutes. Losses at accelerated discharge decrease the
time of discharge, and these losses also impact the time of loading. The C-rate of 1C is often
referred to as one-hour releases such as the two-hour release of 0.5C or C/2 and the five-hour
release of 0.2C or C/5. Any high-performance low-stress batteries can be charged and
discharged above 1C. Table 1 indicates the typical time frames at various C-rates.
9
2.2.3 Aging Degree of Battery
The batteries aging is due to changes in the batteries internal structure, the increase in
temperature, and the batteries dropping. The constant charging and discharge will lead to a
change in the batteries internal structure, influencing the batteries degree of aging [28]. During
the battery degradation phase, the SOC value will also decline, which will ultimately lead to
the batter's inability to work. Generally speaking, the battery is charging, discharge mode, the
number of cycles, and discharge depth can affect the batteries aging [29].
The everyday use of batteries in HEVs implies a paradigm of accurate battery aging and
battery life. The aging of batteries can be divided into two parts: the aging of calendars and
cycle one. Calendar aging refers to the endless amount of lost storage ability. In other words,
battery capacity is responsible for the deterioration. Cycle aging is related to the influence of
cycles (charge or discharge) for battery use times. It occurs when the battery is either loaded
or unloaded. This is directly due to the level, the mode of use the temperature, and current
battery specifications. As a result, its performance steadily deteriorates during battery life
because its electrochemical components are deteriorating and the performance of EVs and fuel
efficiency deteriorates.
2.3 State of Charge Estimation Strategy
The state of charge (SOC) is primarily used to describe the remaining capacity of the
batteries. The batteries residual capacity corresponds to the rated capacity under the same
circumstances following the batteries discharge at a given discharge rate. It is typically
expressed as a percentage (%) [30]. The range of its value is 0~1. When SOC=0, this means
that the battery is completely discharged. When SOC=1 is used, this means that the batteries
are fully charged.
2.3.1 Characterization State of Charge Estimation
The state of charge (SOC) represents the amount of the power amount left in the battery,
which is the ratio of the remaining battery amount and the rated capacity under a particular
discharge condition. It is a relative quantity, expressed as a percentage with a value range of
0% - 100%. Consequently, it is the residual battery capacity of the present time point [31]. It
cannot be measured directly but only indirectly through other external battery parameters,
which plays an essential role in battery health evaluation. The calculation formula is defined
as the ratio of the current residual capacity to capacity in its fully charged state when the
battery is fully shelved at a particular time point as formula (2-2).
SOCt = ( Zt Zn ) 100% (2-2)
Wherein, Zt is the remaining battery capacity at the time point of t, and Zn is the rated capacity
of the Li-ion battery. The SOC range is considered between 0% to 100%. When it is 100%,
10
the battery is fully charged, and when it is 0%, the battery is empty [32]. Among them, the
discharge capacity can be described as the current integral of time. It is considering that the
Li-ion battery is affected by various factors in the charge-discharge process, such as the
ambient temperature, experimental current, and cycling number. The SOC of batteries cannot
be estimated accurately by the formulation in (2-2). Therefore, a more precise approximation
Formula, as shown in formula (2-3), is proposed based on the formula.
SOC = (1 −  Z I Zn ) 100% (2-3)
Wherein, Zt means the discharged battery power for the parameter of I.ζis the battery
efficiency parameter, which can be obtained through hybrid pulse experimental data. The
method of measuring the remaining battery power has always been the focus and
sophistication of the battery control system. To increase the Li-ion batteries service life and
ensure optimum discharge performance [33], it is crucial to calculate the batteries remaining
energy with high precision. Consequently, the Coulombic efficiency is used to describe the
discharging characteristic, which refers to the ratio of the discharge capacity and the charging
capacity for the same cycle process.
2.3.2 Transformation of Coordinate
One concept is to exchange the coordinate system's horizontal and vertical coordinates to
transform the open-circuit voltage into an independent variable. The SOC is also a dependent
variable until modifying the fitting curve and polynomial equation through discrete points [34,
35]. The curve fitting is done to achieve an open-circuit voltage curve and its polynomial
functional relationship. This way is possible since it can be realized without considering the
nonlinear curve fitting degree. An open-circuit and closed-circuit voltage difference can be
defined as seen in Figure 2-2.
10
4.14 UOC 4.14 U
3.96 8
3.96
U (V)
U (V)
I(A)
3.78 3.78 6
3.60 3.60 4
3.42 3.42 2
0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
S1(%) S1(%)
(a): Relation between SOC with Uoc waveform (b): Voltage and current waveform in phase
Figure 2-2: Voltage variation of SOC estimation waveform
The dynamic non-linear relationship between the OCV and SOC is presented in Figure 2-2(a).
The direct polynomial fitting effect is sufficient because OCV is connected to the batteries
internal states [36]. The transfer coordinates a significant difference after correction in the
same order, as seen in Figure 2-2(b) as can be seen from the contrast of the fitting effect before
and after a coordinated transition. The approach is not extremely feasible for direct coordinated
11
transformation, except if the predicted result has been reached under such conditions. The
fitting curve converges and it’s may only refer to a particular category or even a batch of
batteries and not widely right. Moreover, the SOC induces a subsequent change in the OCV in
terms of the Li-ion batteries functional theory, rather than the latter being the first to be
calculated.
2.3.3 Algorithm for Binary Iteration
The implementation process of an equivalent circuit modeling for a battery is the basis
for a detailed battery status determination. The model of the battery will adequately define the
relation between the condition and different battery parameters. Meanwhile, the effect on
temperature, age, and other variables should be taken into consideration. As a research goal, a
ternary Li-ion battery is considered. The Kalman filtering algorithm is generalized, and the
status calculation can be carried out, as is seen in Figure 2-3.
Start
Terminal voltage Initial S

Y
S value at T=k
Calculate the S estimation value at T=k+1 Current
Calculate the terminal voltage at T=k+1
Terminal voltage Calculate terminal voltage error
Calculate corrected gain Filter gain
Output S Corrected S value at T=k+1

N
Stop?
End
Figure 2-3: Iterative SOC estimation technique
The battery model parameters are tested for different SOC points, and an accurate battery
equivalent model is established utilizing a complete composite pulse-power experiment. On
this basis, the algorithm is used to approximate the value of the state. To establish the algorithm
based on high order equivalent model circuit used to evaluate the exact battery characterization.
2.4 The Iterative filtering algorithm strategy
As a selectable model-based battery system, the extended Kalman filter (EKF) can handle
the battery well. The Li-ion battery carries a non-linearity behavior so that EKF can develop
an estimation with its algorithm.
12
2.4.1 Kalman Filtering Expansion Procedure
There are several widely used techniques for calculating states for the dynamics. The
mistake of the ampere-hour integral process is immense [37]. The open-circuit voltage
approach is primarily used for off-line estimation. High-precision estimation approaches
include neural networks, fuzzy inference, and Kalman filtering methods. The extended
Kalman filtering algorithm will linearize the Taylor series expansion of first-order non-linear
functions that ignore the remaining high-order terms. The transform nonlinear problems with
linear problems, which can contribute to higher accuracy [38].
The calculation method requires an updating of time and measurement. The time change,
also known as the forecast, is a method for forecasting current state variables and providing a
prior approximation of the next time point [39]. The measurement-update process, also known
as the adjustment process is the feedback process on the measured values and the variances
correction.
The calculation method requires an updating of time and measurement. The time update,
also known as the forecast, predicts the current state variables and provides a prior estimate of
the next time point. The measurement-update process, also known as the correction process,
is a feedback process on the observed values and the deviation correction. The process is
represented as seen in its process is described as shown in Figure 2-4.
Initial state value
S value at k
Filter input Calculate S value at k + 1
Current measurement Calculate estimated voltage at k + 1
Measured voltage value at k + 1 k + 1 time estimation error
Calculate filter gain Correction bias
Modify the estimated S at k + 1
Get the estimated S value at k + 1
Filter output
Figure 2-4: Flowchart process of the Kalman Filtering algorithm
When using Kalman filters to estimate the battery status, an acceptable equivalent battery
model must be created. The battery is considered a power system, and the state of charge is
the system [40, 41]. The battery current charge-discharge rate is assumed to be the input of the
device, and the terminal voltage may be considered the output. by comparing the observed
terminal voltage value and the predicted state error, the device state is continuously modified
to achieve the lowest estimated variance [42]. It is an optimal self-regressive data processing
13
algorithm that is suitable for both stationary and non-stationary processes. It has got a decent
real-time output that's quick to realize. However, this algorithm has a heavy dependency on
the model and can only be extended to the linear system. The state equation and the
observation equation are defined as shown in the formula (2-4).
 X k = Ak −1 X k −1 + Bk −1Z k −1 + Wk −1
 (2-4)
Yk = Ck −1 X k + Dk Z k + Vk
Above the expression, the state variable at the time point of k is described by the parameter Xk,
and Yk is the systematic observation variable at the time point of k. Zk is the system input, as
the control variable. Ak is the transfer matrix of state x from the time point of k-1 to the time
point of k. Bk is the input matrix [43]. Ck is the measurement matrix. D is the feed-forward
matrix. Wk is the noise of the system state equation as processing noise, and its variance is Qk.
Vk is the noise of the measurement equation as the observation noise, whose variance is Rk. Its
crucial step can be realized to establish a reasonable state equation and observation equation
[44]. However, its effect applied to the nonlinear system is not ideal due to the strong non-
linearity of the battery system. Therefore, it is proposed to linearize the nonlinear system
through the Taylor series expansion. After linearization, the Kalman filtering algorithm can be
used to estimate the state variables.
2.4.2 Extended Kalman Filtering Expansion Procedure
The extended Kalman filtering (EKF) algorithm is a linearization process of the nonlinear
system. It estimates the value of the next time point according to the previous time point. The
state variables are continuously updated with the input and output observations of the system
to achieve the optimal estimation [45, 46]. When it is used to estimate the battery state, the
processing and observation noises are required to be white noise with approximate Gaussian
distribution. It is also a limitation of all Kalman filtering methods. In this case, the covariance
of processing and observation noises can be easily controlled. In the process of estimation, the
Taylor series expansion algorithm is implemented to expand the system model of the battery.
Then a first-order linear model is left after removing the high-order terms. The classical
algorithm is suitable for a linear system, while the battery system is nonlinear. It uses the
Taylor series expansion to get the approximate linear space equation and then uses the Kalman
filtering algorithm to estimate the present state, which is suitable for the discrete nonlinear
system [47]. The expression and observation equations of the discrete nonlinear system are
expressed, as shown in formula (2-5).
 X k +1 = f ( X k , k ) + Wk

 (2-5)
 Z k = h ( X k , k ) + Vk

The first part of the formula (2-5) represents the equation of state and the second part
represents the observation equation. k is a discrete-time point. Xk+1 is the n-dimensional state
vector. Zk is the m-dimensional observation vector. Wk and Vk are independent Gaussian white
14
noise. To apply the Kalman filter, the first-order Taylor series expansion of nonlinear functions
f (*) and h (*) is carried out around the estimated value. To apply the Kalman filter, the
nonlinear functions are extended to the first-order Taylor series, and the results are shown in
formula (2-6).
 f ( X k , k )
(
 f ( Xk ,k )  f X k ,k +

)
X k X k = xk
Xk − X k ( )
 (2-6)
h ( X , k )  f X k , k + h ( X k , k )


k ( )
X k X k = xk
Xk − X k ( )
The new state transformation matrix and observation driving matrix can be accomplished by
using the Taylor series expansion and avoiding the high-order term to update the linear space
formula. The values of Ak, Bk, Ck, and Dk can be calculated in a formula (2-7).
 f ( X k , k )
 Ak =
 X k Xk =X k ( )
, Bx = f X k , k − Ak , X k
 (2-7)
C = h ( X k , k )
 k
 X k Xk =X k ( )
, Dx = h X k , k − Ck , X k
This approach is based on the state of the system and the equation of measurement. The
prediction state equation involves the ampere-hour integral measurement process, and the
observation equation describes the analogous model of the Li-ion batteries. As for (2-5), the
coefficient can be linearized in formula (2-8) as follows.
 X k +1 = Ak X k + Z k Bk + k
 (2-8)
Yk = Ck X k + Dk + uk
The recursive method is accomplished by applying the basic equation to the discreet model.
The Kalman filters the basic equation for the linearized model. However, the recursive process
of the extended Kalman filtering is obtained by applying the equations to the linearized
formula (2-8) which is described in formula (2-9).
X = f X k
 k +1 ( )
 T
 P k +1 = Ak P k Ak + Qk +1

( )
T T −1
 K k +1 = P k +1 C k +1 C k +1 P k +1 C k +1 Rk +1 (2-9)

 X k +1 = X k +1 + K  Z k +1 − h X 
 k +1 
( k +1 
)
 P k +1 = I − K C
  k +1 k +1  Pk +1
Within the formula (2-9) Wherein, P is the mean square error, and K is the Kalman gain. The
parameter I should be initialed as n × m unit. Q and R are the variances of W and V, respectively,
15
which generally do not change along with the system. The initial state value is X (0) = E[X
(0)], and its variance is P(0) = Var[X (0)]. The calculation steps for the time point of k+1 is
described as follows. Firstly, the state and mean-square error of the present time can be
estimated by the last time point to obtain the initial state and its mean square error. And then,
the Kalman gain of this time point is calculated. Finally, Kk+1 is used to modify the initial state
to get the present state. The prior mean square error is modified to get the current mean square
error. In Figure 2-5 show the working process of the extended Kalman filtering algorithm.
Extended Kalman filter
State calculation process Variance calculation process
Filtering update Time update
Kalman gain matrix update
State estimation
measurement update
Figure 2-5: The processing and updating processes of EKF
The calculated value is equal since there is an absolute error between the statistical models
linearization error plus the original expected value and the estimated measured value [48, 49].
There is an individual error between all the calculated values derived by the statistical model
and the measured values displayed in Figure 2-6.
Unscented Multiply by the Unscented
transform State Equation corresponding weight transform again
Ssigma Ssigmapre Ssigma1

S(k-1) Spre
P(k-1)
Corresponding to 3 weights Wc and Wm Observation

equation
Iteration Kalman gain Kk Multiply by the
corresponding weight
ULsigmapre
P(k) S(k) ULpre
Pzz Pxz
Figure 2-6: Double unscented transform Extended Kalman Filtering process
The Kalman Extended Filter (EKF) approach is improved based on the standard measurement
technique and linearization technology. The approximate parameters are linearized, and a
linear approximation is carried out [50]. Thus, the battery state can be measured under which
the device and transmission sounds are usually approximated as white noise, consistent with
the Gaussian distribution.
2.4.3 Estimation Model Construction
The battery state estimation expression is shown in the first part of the formula (2-10). In
the measurement process, the current parameters are easy to be detected directly, so the
calculation process of the battery state can be expressed in the second part of the formula
(2-10).
16
 S = Zt Z n 100%
 (2-10)
 S1 = Z I t Z n 100%
In the above expression, S1 is the state of charge (SOC) value when the current condition is
set to be I. Zt is the remaining power, and Zn is the rated capacity. After discretization [51], the
technical implementation of the battery discrete-time state estimation can be expressed, as
shown in formula (2-11).
 t  I (t )
 S ( t ) = S ( 0 ) − 0 d
i
 Qn
 (2-11)
 S = S − i t I
 n +1 n
Qn
n
Wherein, it is the present state of charge (SOC) calculation method of superficial calculation
characteristics. For the continuous-time SOC estimation realization process, it can be
expressed by the mathematical expression of the first part of the formula [52]. In practical
application, the discrete state estimation model should be applied. The discrete state estimation
model is convenient for the control of a digital system. The complete composite pulse-power
experiment is conducted to test the battery model parameters at different state points and
establish an accurate equivalent battery model, as shown in Figure 2-7.
Start
Initialization factor setting
UT transform generates Ssigma point set
State equation prediction S sigmapre

S (k ) = S (k | k − 1) + K (k )[U L (k ) − U L (k | k − 1)]
Calculate the state prediction matrix Spre
UT transformation generates new S sigma1
Observation equation prediction UL sigmapre

U L( i ) = h[ S ( i ) ]
Calculate the observation prediction matrix

ULpre
Pxz Pzz
Calculate Kalman gain K(k)

K (k ) = PS ( k )U L ( k ) ( PU L ( k )U L ( k ) ) −1
Update status S(k) and covariance P(k)

P(k ) = P(k | k − 1) − K (k ) PU L ( K )U L ( K ) K T (k )
End
Figure 2-7: Updates iteration calculation of EKF
The reliable battery model defines the relationship between the SOC and the different battery
parameters. In the meantime, attention should also be extended to the effect of temperature,
17
aging, and other variables. After acquiring the critical parameters in the model, the state-space
equation is defined by the relationship between voltage and current.
2.5 Summary of this chapter
Cell modeling and state of charge (SOC) estimation are fundamental aspects that can
improve the electric vehicles reliability for the battery management system (BMS). This
chapter provided an overview of Li-ion batteries operation and mechanism strategies. After
then, several SOC estimation process and a detailed description of those techniques were
presented. This chapter also discussed the different mathematical analyses of the identified
SOC parameters to ensure accuracy, parameterization, and reliability. In Figure 2-8 represents
the entire structure of the chapter, and it is easy to understand.
Description of
battery critical
indication with SOC
estimation Iterative calculation
Briefly explanation
of operating and expansion of
characteristics extended Kalman
filter
Different (SOC)
estimation strategies
explanation
Figure 2-8: Overview of mathematical analysis and algorithm
The accurate parameter Identification to determine the SOC is the focus of this thesis.
Furthermore, the extended Kalman filtering algorithm was comprehensively discussed, and
the mathematical expression and model construction were also presented.
18
CHAPTER 3 MODEL BUILDING AND REALIZATION
3.1 The Battery Modeling Method Approaches
Complex chemical reactions arise in the Li-ion batteries application phase, and the
laboratory environment readily influences the reaction process. In the battery state calculation,
the offline experimental data fitting function is usually used to evaluate the values of different
battery model parameters. However, using this approach leads to a significant error in the
prediction performance. To improve the state estimation accuracy, online detection of model
parameters and their real-time correction should be crucial. The simulation of the batteries has
been widely used in the field of the estimate of SOC. However, more clarification is also
required for the relationship between RC and the accuracy of the equivalent circuit model
(ECM) [53]. Several publications also attempted to clarify the correlation between RC network
numbers and ECM exactness. Besides, relations between four-modeling methods and each for
and against methods, are discussed [54].
3.1.1 Comparison of The Batteries Modeling Systems
In a practical model-based estimation approach, the selection of the appropriate battery

models is essential. Several electrical models have been proposed to accurately simulation the
batteries characteristics, such as the Empirical models [55], the equivalent circuit models
(ECM), and the electrochemical model [56], and data-driven models. Within those models,
ECM is widely used to simulate the batteries dynamic behavior, including the voltage response
to the specific current state. The ECMs include the partnership for a new generalization of a
vehicle (PNGV) model [57], The Internal resistance (Rint) model [58], The General of non-
linear (GNL) model [59], The Thevenin model [60], and The higher-order model. These
models consist of essential circuit components, including sources of voltage, resistors, and
capacitors. The classification of battery modeling methods is presented in Figure 3-1.
Battery Modeling Methods
Empirical Equivalent Electrochemic Data-Driven

Model Circuit Model al Model Model
Neural Network
Shepherd Model Rint Model Pseud 2D Model
Model
Single Partial
Unnewehr Model RC Model ELM Model
Model
Nernst Model PNGV Model SVM Model
GNL Model
Figure 3-1: The classifications of battery modeling methods
19
As discussed in Figure 3-1, there is appeared various advantages and disadvantage in the
following battery modeling for electric vehicles. However, that advantage and disadvantage
bring suitability and complexity to use the modeling methods. Based on the batteries validation
and characterization process inside reaction, some advantages and disadvantage models are
mentioned in Table 3-1 in this literature.
Table 3-1: Advantages and disadvantages of the modeling methods
Modeling Method Advantages Disadvantages
Limited capability of
Simple expression and
Empirical Model describing the terminal
computational efficiency
voltage
Easily understand and

Equivalent Circuit The complex parameter
widely used in SOC
Model (ECM) identification process
estimation
High accuracy of voltage Required prior knowledge of

Electrochemical Model
calculation the batteries
High accuracy of voltage Laborious training dataset

Data-Driven Model
calculation, collection process
Different batteries modeling methods have some essential connections and describing the
specific reactions inside the Li-ion batteries. In comparison to statistical, electrochemical and,
electrical models are straightforward to use. The equivalent circuits consisting of passive
components including resistors and capacitors, likely inductors, and a voltage source are
typical to all-electric versions. Therefore, particularly well-suited for use in circuit simulators.
For many applications, the precision achievable with these models in terms of friction, current,
and charge conditions is appropriate. The connections between the four modeling methods are
discussed in Figure 3-2.
Empirical
Model
Electrochemical
Electrical
Model
Modeling
Data Driven
Model
Figure 3-2: Connection procedure of the modeling system
Battery model equivalent circuit was directly examined for applications such as the production
of the Battery Management System (BMS) and power control for vehicles [61]. A good battery
model should forecast the battery capacity details and the voltage response to the load. These
characteristics of Li-ion batteries can be represented by an analogous circuit model (ECM). In
20
certain situations, side reactions need to be modeled for battery losses [62] that can also be
carried out with ECM.
3.1.2 Electrochemical Battery Model
The electrochemical model aims to build the battery structure by analyzing macroscopic
battery data and internal microscopic partial operation. The batteries internal state is obtained
and coupled to the energy generation phase of the battery [63]. The model is primarily the
chemical reaction state of the internal battery. Electrochemical models shall analyze the
characteristics of internal batteries from a microscopic point of view to express specific
characteristics of the battery [64]. Electrochemical models usually involve many sets of
complex time-variable partial differential equations. Generally, these equations take a long
time to solve. The models include a wide range of cell parameters and sophisticated numerical
computational methods. Irrespective of the precision that can be obtained with these models,
they are not acceptable in a simulation environment. The electrical terminal behavior of the
battery and the charging condition is to be calculated with realistic computation times. In
Faraday’s law of electrolysis, the pore-wall flux Ji is related to the divergence of current flow
in the electrolyte phase. The pore-wall flux is calculated in the formula (3-1).
I (t ) I (t )
J n (t ) = , J p (t ) = (3-1)
F . Sn F . S p
Where n denotes is the negative electrodes, and p denotes the positive. Ohm law shows the
distribution of potential between the electrolyte and the active substance. The distribution law
of Fick shows the relationship between concentration and diffusion, which is capable of
defining the diffusion of both the electrolyte and electrode. The Butler Volmer equation shows
the effect of the electrode potential on the current of the electrode.
3.1.3 Equivalent Circuit Battery Model
The equivalent circuit model uses a circuit part to shape a particular circuit network to
describe the circuits operating characteristics. This model establishes a relationship between
the working batteries external features and the internal state of the battery [65]. The equivalent
circuit model is more intuitive, easy to process, and mild in computing. The model parameters
are easy to define and are ideal for simulation studies with circuits [54]. The analogous circuit
model is also commonly used in functional engineering applications.
The design of the battery equivalent model consists of two processes, namely, theoretical
and experimental analysis. The theoretical study is based on the interpretation of internal law's
research purpose, which deduces the complex equation of the law of change to the entity. The
experimental research has to capture the input and output signals of the object [66]. The benefit
is that the model has high accuracy and can show the battery characteristics evolution in detail.
21
The downside is that the model has a high degree of sophistication, which is not beneficial to
its engineering use.
The logical structure of the circuit model shall be defined as follows. The battery models
different configuration process can be split into a basic electrochemical model, an intelligent
mathematical model, an analogous circuit model. The equivalent circuit model is currently
commonly used because of its exact physical meaning and necessary mathematical expression
[67, 68]. Since the batteries characteristics, such as current, power, state of charge, and
temperature, are non-linear. The computational complexity is increased if the battery modeling
has to take full account of these variables. The General Controller cannot comply with the
specifications. The modeling establishment critical points can be extracted from its
mathematical description and its conceptual structure.
3.2 Evaluation of Equivalent Circuit Model
To deal with the critical behavior of the Li-ion batteries, three main battery models are
briefly discussed in the section. The Internal resistance model, Thevenin model, and the
partnership for new generation vehicles (PNGV) model presented a benefit in measuring state
space. However, these models have a low statistical numerical analysis that allows gaining
parameters.
3.2.1 Internal Resistance Modeling Description
The internal resistance model is a more common battery model, which consists of an ideal
voltage source UOC in series with an equivalent internal resistance. The UOC represents the
electromotive force of the battery, which can be obtained by measuring the terminal voltage
of the battery after the battery is in an ideal state, and after sufficient shelved [66]. The
resistance Rint characterizes the direct-current battery internal resistance, and the battery can
be tested when it is connected with the load. In Figure 3-3 illustrates the simple Rint battery
model.
Rint
Uoc UL
IL -
Figure 3-3: Schematic diagram of Rint model
The internal resistance Rint reflects the energy losses that allow the batteries to heat up. The
UL terminal voltage can only be triggered with the UOC when the circuit is the open-circuit
voltage. However, when the charge is connected and the batteries terminal voltage in formula
(3-2).
22
U L = UOC − ( Rint * I L ) (3-2)
In formula (3-2), UOC and 𝑅int are considered constant, and the battery capacity is considered
infinite. Parameter IL characterizes the current flowing through the model. When this model is
applied, the accuracy of the model is generally insufficient due to the consideration of the
shelved characteristics of the battery, and the dynamic characteristics exhibited inside the
battery are neglected, so it is not suitable for the battery modeling
3.2.2 Thevenin Electrical Battery Modeling Description
The Rint model should not take into account the transient behavior of the battery. The
addition of the parallel resistor-capacitor (RC) branch, as seen in Figure 3-4 allows
consideration of the short-term transient due to electrolyte polarization. Similar to the Rint
model, UOC and Rint is the open-circuit voltage and electrode resistance batteries, RP is the
polarization resistance, and CP is the polarization capacitance [69]. To improve model
precision and consider-transient phenomena with various time constants, other RC division
can be used in the Thevenin model sequence.
Rp
Rint
Cp
+ Ui _ +
+ U1 _
UOC UL
IL _
Figure 3-4: Schematic diagram of the Thevenin model
The parameterization process of the model becomes even more complicated. Suppose the
model is employed to simulate the battery behavior in one operating condition at a given SOC.
In that case, the models parameters can be counted as constants. Otherwise, if wide SOC
operating range has to be simulated, the parameters can be considered dependent on
temperature and SOC. The terminal voltage is shown in formula (3-3).
U L = UOC − U1 − ( I L * Rint ) (3-3)
However, the above four parameters are required to parametrize the model completely UOC,
Rint, RP, CP, and all of them are SOC dependent. The manufacturer's requirements do not
contain adequate detail for this model to continue with the parametrization process. Any
preliminary experiments must also be carried out to the particular pulse discharge test (PDT).
The PDT consists of unloading a fully charged cell with a current pulse of defined amplitude
and length. At the end of the pulse, the cell is left to rest in the open circuit. At the end of the
23
rest time, another current pulse is added, and the process is repeated before the cell hits the
cut-off voltage.
3.2.3 PNGV Electrical Battery Modeling Description
The partnership for new generation vehicles (PNGV) concept reflects a relationship of
two condensers and two resistors for a next-generation vehicle comparable cycle models for
Li-ion battery. The Li-ion batteries OCV model is distinguished by a more accurate model, as
the PNGV similar circuit model requires a battery polarization and internal ohms resistance
[70]. According to Figure 3-5 shows the PNGV electrical battery modeling circuit.
Rp
Rint
Cp
+ _ +
+ Ui
Ub
-
+ U1 _
UL
UOC
IL _
Figure 3-5: Schematic diagram of the PNGV model
The PNGV circuit equivalent model consists of a Uoc, which is an OCV, Rp polarization
resistor, Cp polarization capacitor, the ohmic internal resistor is R0, and Ub is the bulk capacity.
the battery terminal voltage is shown in formula (3-4).
U L = UOC − Ub − U1 − ( I L * Rint ) (3-4)
When the battery is drained, the accumulation of current and time induces a difference in the
state of charge. The open-circuit voltage of the battery also varies, which is expressed in the
capacitor voltage shift model. The size of the capacitor characterizes both the battery power
and the direct-current response of the battery to compensate for the limitations of the Thevenin
model. To balance the curve of the model with the calculated voltage curve, the updated model
incorporates a series of resistor-capacity circuits to achieve a better degree of alignment in the
curve fitting process.
3.3 Establishment of High-Order Battery Model
The high-order battery model is a standard lithium-ion battery control circuit model with
high precision for the simulation of transient reaction. It is sufficient for high current, phase
sort, and complicated load-unload conditions. The internal resistance modeling and the
estimation of the status space provided in this section to obtain a zero-input status equation.
24
3.3.1 High-Order Electrical Modeling
The change in the high-order model versus the Rint model is to add two RC network
circuits to describe the Li-ion batteries operations polarization effects. The high-order model
will describe the batteries complex response. During and after charge and discharge, the RC
circuits will be reflected. The incremental shift in the battery voltage as the high-order
equivalent circuit model is simple and capable of satisfying the simulation needs. The model
is also used for practical applications. A correct battery model is essential to ensure safe and
effective operation [71]. To prevent unsuitable operations such as overloading, overcharging,
and high temperatures, predict battery behavior under various operating conditions. Battery
architecture will also lead to onboard control and continuous maintenance design techniques.
Another crucial component, such as charging status and protection, are leading battery
operating states, but cannot be calculated directly. Nevertheless, the model-based estimation
approach is the other vital use of the model battery seen in Figure 3-6.
1. H-Infinity Based
2. Recursive Least Square [RLS] Based
3. Kalman Filtering Based [EKF,UKF,DEKF]
Effective Model Model Based Adaptive Filter Imposed Demonstration

Selection Selection Based Selection Algorithms Results
1. Thevenin Model (2-RC, 3-RC)

2. PNGV Model
3. GNL Model
4. Rint Model
Figure 3-6: Essential steps for selecting appropriate models
However, the high-order model is made by the essential circuit components such as voltage
sources, resistors, and capacitors [72]. In general, an improved resistance capacitor (RC) is the
high-order equivalent circuit, which increases precision and structural models. A similar
circuit model is used in this study to extend two RC networks, which consist of a resistor R0,
two RC networks, and Uoc voltage source. The schematic diagram of the high-order model is
presented in Figure 3-7.
RP1 RP2
R0
Ohmic Int ernal

resistance
+ 1 time constant 2 time constant
fast and IL current fast slow Up
EM(OCV) dependent (Hys teresis)
UL/Load
- Up1 Up2
IL
Figure 3-7: Schematic diagram of the high-order modeling
25
The configuration of the corresponding circuit model is adequate for two-state. Within the
model, R0 is the internal ohmic resistance of the Li-ion battery. The Rp1 and Cp1 have indicated
the electrochemical polarization resistance and capacitance respectively. Another parameter is
Rp2, and Cp2 denotes the resistance and capacitance of the concentration polarization,
respectively. UOC denotes the voltage sources output voltage, which features a monotonous
relationship with the SOC, and IL is that the charge current and UL are the terminal voltage.
The battery terminal voltage is as shown in formula (3-5).
U L = Em (OCV ) − U p − U p1 − U p 2 (3-5)
The measuring equipment used in the hybrid pulse-power characteristic system is not difficult
to use the cost of measurement is not high, and high accuracy can be obtained. However,
dynamic integrals are challenging to apply in functional circumstances, and it is difficult to
use the approach mentioned above to achieve a posterior probability filtering density.
Therefore, this approach has been limited to available promotion and needs to be paired with
other approaches to solve the underlying probability density. The required posterior
probability density is obtained to measure the required parameters.
3.3.2 Ohmic Internal Resistance Methods
The internal resistance approach theory is similar to the open-circuit voltage system.
There is an individual practical interaction between SOC and internal resistance. The internal
resistance is measured and converted to a SOC value [73, 74]. The functional relationship
between SOC and internal resistance is relatively consistent because the internal resistance
method precision depends solely on the internal resistance calculation precision. However,
other factors also conflict with the internal resistance method SOC accuracy when calculating
internal resistance. However, the other polarization voltage decrease terminates the fast phase
exponentially before the voltage returns to the steady-state voltage value [23, 60]. Assuming
that the voltage before the current changes is U1 and that the voltage after the current changes
is U2, the rule for measuring the ohm resistance is as follows in formula (3-6).
(U1 − U 2 )
R0 ( S ) = (3-6)
IL
According to formula (2-5), batteries terminal voltage will suddenly change at the moment of
discharge and the moment of stop, both caused by internal ohmic resistance [75]. Therefore,
the overview of internal resistance and voltage curved presented in Figure 3-8
26
0.001505
R0
0.001462
0.001419
R0()
0.001376
0.001333
0.001290
0.0 0.2 0.4 0.6 0.8 1.0

S1(%)
Figure 3-8: Relation between ohmic resistance with SOC
The method of resistance may also be called a conductivity method. The relationship between
conductivity or internal resistance of the Li-ion batteries and the SOC can be studied from
several experimental results from a long-term conductivity of the Li-ion batteries internal
resistance test to estimate the SOC estimation.
3.3.3 State Space Iteration Process
The study performed in develops a model considering three aspects such as electrical
model, thermal model, and degradation model for Li-ion batteries installed in EVs [76].
However, U1, U2, and SOC are taken as state variables, and UL is the vector observation. The
equivalent circuit model equation can be derived as formula (3-7).
 dU p1 U p1 I U p1
C p1 + = I L  U p1 = L −
 dt R p1 C p1 C p1R p1
 (3-7)
C dU p 2 + U p 2 = I  U = I L − U p 2
 p 2 dt Rp 2
L p2
C p 2 C p 2 Rp 2

The state of charge (SOC) vector can express the open-circuit voltage, and a non-linear
function relationship can be formed. The analogous circuit model can be discretized using the
experience of current control theory. For the selected chosen high-order equivalent circuits,
[SOC Up1 Up2] is selected as the state variable. Following absolute formula (3-5) with approx.
Formula (3-7) and the description of SOC, its state-space equation, as seen in formula (3-8).

  SOCk+1  1 0 0   SOCk   − t /  Q0 

 U    
  p ,k+1  = 0 e

(-t/ p )
 (
0  U p ,k  +  R p 1-e p
(-t/ )
)  I (k )
 (3-8)
  0 e (-t/ c )  U c ,k   R 1-e(-t/ c )
 U c ,k+1  0
  c ( ) 

U k +1 = U oc ( SOCk +1 ) − U p ,k+1 − U c ,k+1 −iRo
In formula (3-8) can be substituted for the recursive least-square method, and formula (3-9) is
taken as the directly identified parameters. The circuit modeling parameters of R0, Rp1, Cp1,
Rp2, and Cp2 are derived from the identification results of these parameters. The values of each
parameter can be obtained from it was based on the terminal voltage that can be expressed as
shown by formula (3-9).
27
 − et
U L = U OC − be − ce
− ft
 −t 1 −t  2
(3-9)
 
U L = U OC − ( I L R p1e ) − ( I L R p 2e ) 

In the equivalent circuit model, these equations are the values of any parameter. During the
data analysis and processing process, the thriving data section is derived from the initial
experimental data, and then the extracted data segment is analyzed. The parameters of 1 and
2 where the two-time constant is calculated, as shown in the formula (3-10).
 1 =  Rp1 ( S )  C p1 ( S ) 
 (3-10)
 2 =  Rp 2 ( S )  C p 2 ( S ) 
Assuming that the voltage before the current changes is SP1 and the voltage after the moment
the current changes in SP2. The ohmic internal resistance can be considered as a unique factor
that causes the voltage drop at the moment. An efficient method of processing is used to create
a relationship between the internal parameters of the equivalent circuit model and the state of
charge. Also, the values of the remaining parameters can be obtained by integrating. However,
compared with the co-efficient, the values of the parameters can be obtained in formula (3-11).
 R p1 ( S ) = b I L , C p1 = 1  R p1 ( S )  e 

 (3-11)
 R p 2 ( S ) = c I L , C p 2 = 1  R p 2 ( S )  f 

The step-by-step variation of the terminal voltage is consistent with the pure resistance
characteristic for the instantaneous disappearance of the current. The ohmic internal resistance
can be considered as a unique factor that causes the voltage drop at the moment. The working
procedure for the identification of a high-order model is shown in Figure 3-9.
The step of The zero input The zero input
voltage climb response response
1 Rp1 Cp1
 = RC
The step of Least square 2 Least square
Rp2 Cp2
discharge algorithm algorithm
The transient with Least square R0
or without current algorithm
Figure 3-9: Working steps for identification parameter
The expression of the step of the voltage rise was a zero-input response. The discharge step is
to express the least square algorithm. Finally, a combination of concentration polarization
resistance and capacitance is presented. The curve fitting approach utilizes a method in which
a standard feature of parameters matches the entire pulse discharge curve to obtain model
parameters. The selection of the equivalent circuit model can reasonably reflect the
characteristics of the battery and the CPU processing power of the BMS. To investigate the
28
dynamic behavior following terms are the superiority of high-order models for electric
vehicles (EVs).
1. This model does not require matrix calculation, so the computation cost is
significantly low.
2. Added an extra RC network which in the improvement of the first-order RC

model.
3. Six parameters of high-order modeling that ensure that reliably investigate the
insight state of the battery.
4. The model exhibits high precision for determining the dynamic voltage
characteristic of the battery.
5. The presented method has a higher estimation accuracy and faster convergence.
6. Comprehensively independent to understand and trusted for simulation approach.
3.4 Model Parameter Description
The battery model is essential in deciding an exact mathematical relation that can produce
significant inputs and outputs with the development of an actual battery model, analyzing the
feasibility and reliability. The model battery can be obtained using various methods.
Ultimately, the various characteristics of the battery function can be contrasted with the actual
model.
3.4.1 Ampere-Hour Counting (Ah) Method
The integral method of ampere-hour, also known as the method of coulomb counting,
measures the cumulative charge and discharge charge over a period when the SOC basis for a
given time is known in advance. The actual state of the batteries charge can be obtained by
superimposing the two techniques. The error is obtained by adding the expected value to the
real value. The error grows with the period and does not ensure the stability of the test results.
And this procedure relies on the initial value, so it does not take the right solution approach to
the issue of self-discharge.
The most frequently used method of SOC calculation is Ampere hour (AH). The intention
is to concentrate on the exterior properties of the battery device without taking into
consideration the battery's electrochemical reaction and the dynamic interaction between the
parameters [57]. The principle of the ampere-hour integral method is shown in formula (3-12).
1 t
C 0
SOCt =SOC0 −  Idt (3-12)
29
Where SOC0 is the batteries original electrical quantity, SOCt is the batteries electrical quantity
at time t, C is the batteries rating power, I is the charging and discharging current, and the
discharging direction is the positive direction  is the coulomb performance coefficient, which
reflects the internal electrical quantity dissipation throughout the charging and discharging
process. The working principle of the improved ampere-hour integral is shown in formula
(3-13).
1 t
 C 0
SOCt =  SOC0 −  Idt (3-13)
Where  is the aging factor and self-discharge correction factor  is the correction factor of
battery capacity C. The reason for adding the correction factor is that the battery will age after
a long time of use. The functional relationship between the correction factor  of the total
battery capacity C and the cycle times of the lithium battery is determined through experiments
to improve the accuracy of SOC estimation [77]. However, the approximation method of
ampere-hour integral that the accuracy of SOC0 is very important to the real-time estimation
results. The popular method is to measures the initial battery power using the open-circuit
voltage method.
By defining a correction factor and defining a re-calibration point, ampere-hour counting

errors can be reduced. Ah, counting (known as Coulomb counting) has better precision than
other SOC estimation methods [11, 78]. It is simple and effective if the current measurement
is correct, and the re-calibration point is usable.
3.4.2 Exponential Curve Fitting Technique
The curve fitting method is a technique in which the formal equation with the parameters
matches the whole pulse discharge curve to obtain the model parameters. Its rate of data
acceptance is far higher than that of the point-to-point data acquisition method. In the process,
just need to give a suitable formula to the function containing the symbols of the parameter to
be obtained for curve fitting. The curve fitting method has been designed for relatively
complex calculation parameters such as Rp1, Rp2, and Cp1, Cp2 for the high-order battery
modeling systems. However, the high-order model expression following in the formula (3-14).
t
−
U L = U OC − IRO − IRp (1 − e  ) (3-14)
The equation is abstracted to obtain a parameterized expression, as shown in the formula

(3-15).
y=a-b*(1-exp(-x/d))-c*(1-exp(-x/e)) (3-15)
Above the following, in formula (3-15), y represents the coordinate vector such as terminal
voltage UL, and x represents the abscess variable, i.e., time t, where a, b, c, d, e are the
30
parameters. Depending on the conditions of the OCV, current, and ohmic internal resistance.
The fitted curve will help express the real shift in voltage and the fitting curve of the phase
pulse zero state reaction with SOC of 1.0%, The actual view in Matlab can be obtained in
Figure 3-10, which appeared a real view of the process and not required for the internal ohmic
resistance.
Curve fitting Equation
d is expressed 1
e is expressed 2
Figure 3-10: Curve fitting method for identification parameters
The Identification of parameters has made by each step of the SOC point using a curve-fitting
tool. The polarization resistance and capacitance are done by the method but the internal ohmic
resistance has done by the experimental data which obtained between 0.1% to 1.00%.
However, the experimental effects of the hybrid pulse-power test, the curve fitting is done
using the least square method.
To accurately the state of charge (SOC) estimation, and a significant deal with Li-ion
battery, different types of battery modeling are discussed elaborately in this chapter. Many
battery modeling strategies such as Rint, PNGV, Thevenin model are presented in this section.
Expansion
evaluation of
equivalent circuit
modeling Model parameter
Model comparison
among other are identified by
modeling system adapted modeling
system
High-order
modeling adaption
and state-space
equation
Figure 3-11: Overview of battery modeling system and techniques
31
In Figure 3-11 represent the whole structure of the chapter, and it is easy to understand.
However, the mathematical description briefly explains also in this portion. Significantly, four
types of standard battery modeling are studied. Also, their pros and cons are provided under
different udder connection. Furthermore, the high-order Li-ion modeling with its mathematical
analysis is introduced briefly.
32
CHAPTER 4 EXPERIMENTAL VERIFICATION
4.1 Build Experiment Platform and Data Acquisition
According to the freedom car battery experiment manual, the standard hybrid pulse-power
experimental test has been carried out on Li-ion batteries. The battery capacity has been
extracted by using the same way also carried out.
4.1.1 Real-Time Platform Implementation
The battery test platform arrangement was introduced by the secondary leading test
equipment "CT-4616-5V100A-NTFA" produced by Shenzhen Xinwei New Energy
Technology Corporation. Expose the platform in Figure 4-1 capable of charging and
discharging in a ternary battery module with a maximum voltage of 380V at the frequency of
50/60Hz and the maximum current is 100A. Major parameters such as voltage, current, and
temperature have been calculated by equipment to be promising. The host computer with the
BTS-7.6 software installed the program the experimental procedure and handled data
acquisition in real-time. The measuring data has been transmitted via TCP/ IP ports to the host
computer. The incubator has three independent layers manufactured by Dongguan Bell
Experimental limited company qualified to monitor the high or low-temperature chamber
(DGBELL-BTT-331C). This experiment took place at a constant temperature of 25˚C.
I/V/SOC/T
TCP/IP
Power line I/V/T
Temperature test
chamber
Figure 4-1: The schematics of the battery test platform
The battery will age due to recycling and other reasons, and the actual battery capacity will be
significantly different from the calibration capacity. The actual discharge capacity of the
battery is essential for the SOC estimation of the Li-ion battery. Therefore, the capacity
calibration of the Li-ion battery must be performed first. In this study, the identification of the
33
online parameter will increase the complexity of the algorithm, and the accuracy is not
improved. Therefore, an offline method of identification is selected. The battery is subjected
to the pulse discharge experiment, and the parameters of the battery model are obtained by
analyzing the operating characteristics of the battery during operation. Therefore, an offline
method of identification is selected. The battery has been subjected to the pulse discharge
experiment, and the parameters of the battery model have been obtained by analyzing the
operating characteristics of the battery during operation. The Li-ion ternary battery is picked
as the experimental object in this experiment. The primary technical parameters of the battery
are shown in Table 4-1.
Table 4-1: Key function of battery parameters
Criteria Parameters
Size: length*width*height/mm 200*80*180
Rated Voltage /V 3.75
Maximum load current /A 1.5C
The charge cut off voltage /V 4.2
Discharge cut-off voltage /V 3.0
Working temperature /C° 25
Rated capacity /Ah 4.0
The experiment has carried out using specified requirements with parameters corresponding.
The experiment was done in the laboratory in a day according to parameters. Therefore, an
offline method of identification is selected. The battery has been subjected to the pulse
discharge experiment, and the parameters of the battery model have been obtained by
analyzing the operating characteristics of the battery during operation.
4.1.2 Hybrid Pulse Power Characteristic Test
The hybrid pulse-power characteristic test is a battery performance test method that is
described in the freedom car battery test manual for power-assist hybrid electric vehicles.
Through hybrid pulse-power characteristic tests, sufficient identification can be performed for
the model parameters. The hybrid pulse-power characteristic experimental process has been
described. Firstly, the Li-ion battery is discharged at 1C for 10s, which is then charged with
0.75C for 10s after shelved for the 40s. In the cyclic test, the battery is subjected to a pulse-
power characteristic test of an equally spaced state point. The SOC varies from 0% to 1%. The
battery needs to be left for a more extended period between adjacent pulse test to restore the
battery of electrochemical and thermal equilibrium at equally spaced points. The entire
34
experiment consists of a single repeated charge-discharge pulse test primarily, according to

the HPPC experiment procedure is designed, as shown in Figure 4-2.
1C
1C charge Set aside
discharge
10s for 40s
10s
Constant
current and N Y
Y Set 1C Set N
constant Count Count
Start I<3.5A？ aside discharge aside Count<9 End
voltage =0 ++
for 1h 6min for 1h
charging
4.2V/70A
Figure 4-2: Hybrid pulse characteristic test procedure
All the model parameters can be obtained through the hybrid pulse-power experimental test.
However, it takes a long time of about 12 hours. It generates excellent consumption of the
experimental equipment and operators. Most time of the hybrid pulse-power characteristic test
is spent on the repeated shelved stage. A hybrid pulse test was conducted consisting of a series
of HPPC profiles, constant current discharge pulses, and rests. The HPPC test will also be used
to drive parameters of the circuit module equivalent to the battery. In Figure 4-3(a) and Figure
4-3(c) represented a single cycle HPPC test voltage and current HPPC curves, respectively.
Figure 4-3(b) and Figure 4-3(d) show the full cycle current and voltage. The batteries of similar
capacity are selected to conduct the hybrid pulse-power characteristic test simultaneously,
making batteries of similar capacity for the same batch. It is used to ensure the consistency of
the quantities that affect the experimental results except for the experimental quantities.
4.4 U 90 I
70A 70A
4.0 60
30
U(V)
I(A)
3.6
0
3.2 -30
2.8 -60
2.4 -90
0 5000 10000 15000 20000 25000 30000 0 5000 10000 15000 20000 25000 30000
t (s) t (s)
(a): Closed cycle voltage response curve (b): Closed cycle current response curve
4.30 U
80 I
U5(t5) U6(t5) 60
4.25 40
U7(t7)
I (A)
U1(t1)
U (V)
4.20 20
0
4.15 U4(t4)
U7(t7)
-20
4.10 -40
U2(t2) U3(t4) -60
4.05 -80 0
0 20 40 60 80 20 40 60 80
t (s) t (s)
(c): Single cycle voltage tracking curve (d): Single cycle voltage response curve
Figure 4-3: Complete current and voltage discharge profile during the HPPC test
According to Figure 4-3(c) reflects the Li-ion batteries transitory characteristics and steady-
state characteristics. Each voltage point should satisfy the following criteria.
35
1. The discharge starts on time t1. The Li-ion terminal voltage drops abruptly from U1 to
U2. It is mainly because of the voltage changes caused by the ohmic internal resistance.
2. The Li-ion terminal voltage drops gradually from U2 to U3 during the period from t2
and t3. The effect of battery polarization. The current discharge charges the polarization
capacitor, which is the two RC networks circuit zero-state response.
3. The Li-ion batteries terminal voltage abruptly increases from U3 to U4 during the
period from t3 to t4, which is also due to the voltage shift caused by ohmic resistance.
4. The terminal voltage rises slowly from U4 to U5 during the period from t4 to t5, which
is the process that discharges the polarization capacitance to the polarization resistance
and the two RC networks circuit zero-state response.
Overall, the HPPC procedure consists of 10 voltage and current profile repetitions, as shown
in Figure 4-3 is divided by a constant current discharge of 10 percent depth of discharge (DOD)
at 1C. A continuous current discharge follows an hour rest time. The rest period should allow
the battery to achieve thermal and electrochemical equilibrium. The procedure continues until
DOD reaches 90 percent, after which another 1C rate discharge is performed until the battery
is 100 percent DOD.
4.1.3 Capacity Tracking Experiments
In charging the Li-ion battery, it is first essential to release all the remaining Li-ion battery
power that reaches the S = 0 set at the outset. Generally, as the Li-ion battery voltage decreases
to 2.5V, the Li-ion battery discharge is assumed to be complete. After the continuous current
method of charging the various currents for a certain amount of time. The Li-ion battery is
steadily charged up to around 1 hour of charging as the device cannot identify the Li-ion
battery's charge status at any time. The fixed stop state is adjusted to stop the power supply
until the battery voltage exceeds 3.65V. The waveform voltage is obtained, as shown in Figure
4-4.
4.4 4.1796 4.1809
U 40 I
4.0 20
U (V)
I (A)
3.6 0
3.2 -20
2.8 -40
0 10000 20000 30000 40000 0 10000 20000 30000 40000
t (s) t (s)
(a): Voltage transform waveform (b): Current transform waveform
Figure 4-4: Voltage and Current transition waveform
As its primary function is to charge the Li-ion battery, the waveform of power and energy
fluctuates in the process. Constant current charging and discharging cycles are carried out.
36
Capacity usually only takes into account the strain of material reactions. However, energy and
power are different since the voltage component is still used. The higher the voltage platform,
the higher the output. The change in time and the changing direction of the two variables over
time are approximately the same as seen in Figure 4-5.
50 C 190000 E
39.303 39.403
40 152000
E (mWh)
C (Ah)
30 114000
20 76000
10 38000
0 0
-10 -38000
0 10000 20000 30000 40000 0 10000 20000 30000 40000
t (s) t (s)
(a): Capacity transform waveform (b): Energy transform waveform
Figure 4-5: Capacity and Energy transition waveform
The error waveform analysis reveals that the modified algorithm is adequate. The error is
significantly decreased relative to the error. Since the given initial value varies significantly
from the initial value, the batteries internal polarization effect is improved at the initial
discharge point. The error in the state calculation of two algorithms at the beginning of the
discharge is greater than the later level.
4.2 Model Parameter Identification
Equivalent models can be used to explain the troublesome electrochemical reaction

mechanism of the battery by experiments. In short, the equivalent is to refer to the same current
in the battery system. The developed equivalent model will effectively track the battery
voltage. The model's voltage error is minimal, with a high-precision effect. Identification of
the model parameter shall be carried out with the state of charge and other factors concerning
the battery power, the current rate of charge-discharge, the degree of cell aging, and the
ambient temperature.
4.2.1 Parameters Extraction Results
In this analysis, identifying the online parameter will increase the complexity of the
algorithm, and the precision will not improve. For this cause, the method of offline recognition
is chosen for the experiment at 25°C. The battery was subjected to the HPPC experiment. The
parameters of the battery model were obtained by analyzing the batteries operational
characteristics during operation. Related practical measures have been taken below in the
HPPC experiments.
1. The battery is subjected to continuous current constant voltage charging at a

rate of 1C. The cutoff voltage charge is set to 4.2V and then transformed to a
37
constant voltage charge, with a current cutoff charge of 0.5C. To guarantee

that the battery is fully charged.
2. To reach a stable battery voltage, the battery must be fully assembled after
charging, and the preferred standing time must be 30 minutes.
3. The Li-ion battery has steadily been discharged at a rate of 1C for a period of
10s. The battery should be permitted to stand for the 40s after the discharge
has ceased.
4. At a time 10s, the lithium Li-ion is subject to continuous current charging at

a rate of 1C. The battery should be allowed to stand for the 40s after the
discharge has ceased.
5. The practical measures (3)~(4) are a complete HPPC test. To investigate the
characteristics of the reaction of the battery under various SOC values. The
battery has been the existing discharge for s with a content rate of 1C of 6
minutes, to reduce the SOC battery by 0.1 and obtain a new experimental
design for SOC. After 40 minutes, the HPPC experiment continued until the
battery power was 0, and the experiment was over.
The fitted parameter is then shown with a polynomial of the 6th order, and the following results
are obtained. The fitting results of the curve are R0, RP1, RP2, CP1, and CP2, respectively,
reflecting the parameters for the equivalent circuit of high order modeling. The fitted results
are shown in the following formula (4-1).
 R0 ( S ) = .00377 * u - 0.01272* u + 0.01776* u

6 5 4

-0.01371* u + 0.006676* u - 0.002097 * u + 0.001634
3 2

 R p1 ( S ) = 0.2705* u − 0.9849* u + 1.417 * u
6 5 4

−1.018* u + 0.3783* u − 0.06759* u + 0.004558
3 2

C p1 ( S ) = 1.136*10 * u − 3.991*10 * u + 5.509*10 * u
7 6 7 5 7 4
 (4-1)
−3.754*10 * u + 1.294*10 * u − 2.031*10 * u + 1.241*10
7 3 7 2 6 5

 R p 2 ( S ) = −0.1694* u + 0.6867 * u − 1.08* u + 0.8263* u
6 5 4 3

−0.3155* u + 0.05511* u − 0.002991
2
C ( S ) = -1.09*106 * u 6 + 4.437 *106 * u 5 - 7.464*106 * u 4


p2

+6.696*10 * u - 3.353*10 * u + 8.561*10 * u - 5.348*10
6 3 6 2 5 4
According to the schematic diagram of the HPPC terminal voltage curve shown in the part of
the HPPC experiment carried out. The battery was shelved after 6 minutes of discharge was
selected for double exponential fitting. The parameters identified results are shown in Table 4-2
according to each point's HPPC experimental data. The values of the high-order equivalent circuit
model parameters corresponding to the SOC values can be obtained.
38
Table 4-2: Discharge-profile criteria under specific SOC levels
S1(%) OCV/V R0 /Ω Rp1/Ω Rp2/Ω Cp1/μF Cp2/μF IL
10% 3.4545 0.00148 0.00069 0.0001 17529 4495 70
20% 3.5367 0.0014 0.00003 0.00045 13469 27034 70
30% 3.5900 0.00135 0.00003 0.00038 17552 31568 70
40% 3.6163 0.00132 0.00034 0.00003 33144 29758 70
50% 3.6511 0.00131 0.00003 0.00034 20421 31815 70
60% 3.7366 0.00131 0.00004 0.00057 20176 23233 70
70% 3.8309 0.0013 0.00004 0.00064 18860 25931 70
80% 3.9360 0.0013 0.00004 0.00056 23504 25388 70
90% 4.0513 0.0013 0.00003 0.00048 19710 25491 70
100% 4.1840 0.00131 0.00003 0.00045 26053 28427 70
The parameters are identified by the hybrid power pulse characterization (HPPC) test under
25°C temperature. Each parameter reflects each condition searchable of lithium-ion batteries.
However, the effect of electrochemical polarization and concentration resistance is carried
out by curve fitting results in Figure 4-6.
Rp1 Rp2
0.00064 0.00060
0.00048 0.00045
Rp2 ()
Rp1 ()
0.00032 0.00030
0.00016 0.00015
0.00000 0.00000
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
S1(%) S1(%)
(a): Polarization resistance fitted waveform (b): Concentration resistance fitted waveform
Figure 4-6: High-order resistance fitting waveform
The hybrid pulse-power curve of the typical experimental voltage variance is obtained. The
curve fitting results showed that is very closed to the real experiment result. The variation
characteristics can be derived to obtain the model parameters. The polarization and
concentration capacitance are fitted by the experiment data in Figure 4-7.
39
33200 Cp1
35000 Cp2
24900 28000
Cp1 ()
Cp2 ()
21000
16600
14000
8300
7000
0
0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
S1(%) S1(%)
(a): Polarization capacitance fitted waveform (b): Concentration capacitance fitted waveform
Figure 4-7: High-order resistance fitting waveform
The trends of R0, Rp1, Cp1, and Cp2 are coupled with state combinations, respectively. As can
be shown from the diagram, the model's internal parameters fluctuate within the particular
spectrum of the SOC. These parameters may then be replaced by their average values or by
a look-up table if the models precision is not good. One way to use model parameters is to
provide a limited amount of processing.
4.2.2 Open-Circuit Voltage Characterization
The open-circuit voltage (OCV) is the terminal voltage of the lithium-ion (Li-ion) battery
that has been left for a long time. In the perception of this, the Li-ion batteries OCV and SOC
have an excellent mapping relationship. The SOC-OCV step by step is to study the OCV
characteristic of the Li-ion battery voltage. The SOC-OCV calibration experiment for Li-ion
was performed using a battery test-time battery test device [79]. The OCV battery has a strong
nonlinear relationship with the SOC. Identify the relationship by performing battery discharge
and charging. For the charging test, charge the battery from 0 percent SOC at the
recommended C/2 rate at the factor for 12 minutes, followed by a rest period of 12 minutes to
allow the cell to return to a load balance condition before the next cycle is applied. Then the
OCV is measured [80]. The steps for designing the open-circuit voltage flow chart are shown
in Figure 4-8.
Start
Constant current constant voltage
charging 4.2V / 1.5A (1C)
N Y
I < 0.075A
Shelving for one hour
Count = 0
Constant current discharge 1.5A
Shelving for 40 minutes Count ++
Y
Count < 10 & UL > 3V
End
Figure 4-8: Open circuit voltage and state of charge calibration
40
In Figure 4-8, this process is repeated until the battery reached 0% from 100% of the SOC. As
shown in below Figure 4-9, these discrete points are extracted from the laboratory data to
obtain an OCV-SOC relationship scatter plot. The least-squares approach is a mathematical
optimization technique that determines the best function matching of the data by reducing the
error square. It is suitable for curve fittings, which use this method to fit the OCV-SOC
relationship curve and relational polynomial.
U (V) 3.984 SOC=0.7

4.08 SOC=0.5
SOC=0.3
3.901 SOC=0.1
3.91
U (V)
U(V)
3.818
3.74
3.735
3.57
3.40
3.652
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 10 20 30 40 50 60 70 80
S1(%) S1(%)
(a): Time-varying relationship between SOC and OCV (b): Different state of charge at a specific point
Figure 4-9: Specific open-circuit voltage at different SOC
Through curve fitting, it has been found that the batteries OCV of the batteries as a function
of the SOC function can be expressed as a polynomial equation sixth-order. The experiment
shows that the voltage stability of the battery after 40 minutes is equal to the open-circuit
voltage of the battery. The following formula (4-2) showed the curve fitting procedures.
U oc,k = f ( Sk ) = P1* Sk6 + P 2* Sk5 + P3* Sk4

(4-2)
+ P 4* Sk3 + P5* Sk2 + P6* Sk + P7
Given the above, after repeated experiments to compare the fitting effect, it is observed that
the six-order polynomial has a strong fitting effect and is moderately reasonable to the
processor. Thus, the functional relationship between the OCV and the SOC is accomplished
by fitting to the fifth-order polynomial. In the above equation, the P1 to P7 is the co-efficient
(with 95% conviction limits) obtained using the least square method by the MATLAB. In the
following formula (4-3).
U oc,k = f ( Sk ) = 20.08 * Sk6 − 61.22 * Sk5 + 68.12 * Sk4

(4-3)
−32.07 * Sk3 + 5.234Sk2 + 0.6794 * Sk + 3.36
In formula (4-3), Sk is the battery state-of-charge value at the time point of k, and UOC,k is the
corresponding open-circuit voltage value. To obtain the battery open-circuit voltage
characteristics in the charge-discharge experiment. The OCV is 3.36V when the SOC of the
Li-ion battery used in this experiment is zero. The SOC-OCV relationships provide a basis for
the measure of the OCV in the revised mathematical help of the following parameters. The
relation between the SOC and the OCV of the battery can be used to estimate the actual battery
power. The approximate static battery capacity, the battery capacity if left for an extended
period.
41
4.3 Detection of High-Order Model Verification
As an approached model-based battery system, it is necessary to establish a modeling-

based simulation direction in the MATLAB. The simulation model can compare experimental
data and identified parameters. To build the simulation model in MATLAB, a variety of
working processes has been done.
4.3.1 High-order Model Validation Strategies
An ordinary differential equation is usable in the time domain of the high-order equivalent
circuit. The equation for the resulting voltage response has been fixed. In Simulink, the process
of discretization is required before the model. To obtain a discreet state-space equation for the
high-order model. The largest block in the diagram is the corresponding circuit model. The
input contains current I, internal ohmic resistance R0, electrochemical polarization resistance
Rp1, Concentration polarization resistance Rp2, Electrochemical polarization capacitor Cp1,
Concentration polarization capacitor Cp2, open-circuit voltage UOC, terminal voltage UL, and
load current IL. Besides, to the current in the input parameters, the other parameters are the
internal parameters of the model. They are SOC functions as an independent variable. If the
status of the charge varies, the beneficial relationship is taken from the identification of the
previous segment. The state of charge is the result of the latest changes at every moment. A
simulation model has been developed in Figure 4-10.
Figure 4-10: Top view structure of terminal voltage verification model
The real-time state of charge update can be accessed from a built-in ampere-time integration
module. It is related to the end of the parameter described above. And then the parameters of
the model are obtained corresponding to the constant shift of the current input. To get the SOC
42
with discharge current, an ampere-hour (Ah) integral model is approached in Figure 4-11.
𝟏
𝐒 -K- + 1
+
Figure 4-11: Inner structure of ampere-hour (Ah) integral model
In Figure 4-11, the central component of the whole module is the high-order internal circuit.
The device configuration is used directly to construct the module, containing the internal
ohmic resistance, two parallel resistance systems, a controllable voltage source, a controllable
source of energy, voltage, and current sensors, and the input and output interface. The internal
structure of the high-order equivalent circuit model is shown in Figure 4-12, where each circuit
component is a controllable parameter that changes over time. The high-order model has seven
inputs and one output, which indicates the inputs are IL, R0, RP1, RP2, CP1, CP2, and UOC. The
one output is the terminal voltage UL.
Figure 4-12: The central part of inside the high-order model in Simulink
The controllable source of the voltage and the controllable source of the current are signal
interfaces that can transform the signal into a material port. Convert external input voltage
inputs to the voltage and current supply that can be attached to the circuit. Voltage sensors and
current sensors are both signal transducers, which transform physical interfaces to signal
interfaces.
4.3.2 Corroboration of Model Voltage Parameters
To validate the high-order equivalent circuit model and the characterization of the actual
battery voltage. Actual voltage and current data under the cyclic discharge keeping condition
are imported into a high-order equivalent model developed in MATLAB/Simulink. The model
43
is verified by integrating the effects of the identification of the identified parameters. The
present state of SOC values of the battery is measured and combined with the nonlinear
function relationship between the state of charge (SOC) and Uoc using the ampere-time
necessary procedure. The corresponding OCV value of the battery can also be accessed.
Substituting Uoc for an equivalent model will measure the value of the battery terminal voltage.
The calculated value is compared to the real terminal voltage value. The result of the
comparison and the resulting error is seen in Figure 4-13.
X1
4.32
X2
4.14
UL(V)
3.96
3.78
3.60
3.42
0 6300 12600 18900 25200
t (s)
Figure 4-13: Waveform of simulation comparison
In Figure 4-13 indicates a relation between the approximate value of the battery terminal
voltage and the real value of the discharge of the battery under the experimental conditions of
the cyclic discharge. The solid blue line X1 is the simulation voltage that compere with the
actual voltage. The solid red line X2 is the real battery terminal voltage value. However, the
comparison with two simulations results from the simulation error is presented in Figure 4-14.
0.075
0.051V Err
0.050 0.031V
Err (V)
0.025
0.000
-0.025
-0.050
0 6300 12600 18900 25200
t (s)
Figure 4-14: Error indication of simulation result
In Figure 4-14 shows the variation of the simulation error. The figure indicates that the
projected value has a strong monitoring impact on real value. The average estimated variance
is roughly 0.051V, which may describe the batteries voltage value at the working end. A study
of the voltage contrast error between the voltage calculation deviation increases at the end of
the batteries discharge. In this respect, the battery voltage varies significantly before the final
discharge. A calculation error accompanies the simulation result. The model accuracy is
99.14 %, while the highest lithium battery voltage is 4.2V.
44
4.4 SOC Estimation Effect Based on EKF
To verify the model accuracy and robustness SOC estimation in Mathlab. To reliably
approximate the SOC, a model-based method is proposed to create a relationship between
experimental accuracy and model accuracy.
4.4.1 Simulation Interfacing Process
The simulation system model was implemented in MATLAB/Simulink to verify the SOC
estimation effect of the algorithm. According to the above extended Kalman iteration method,
a simulation structure of SOC estimation for Li-ion battery using the EKF algorithm was built
in Simulink. The current and voltage data obtained from the HPPC experiment were imported
into the model. The simulation's crucial part is to write the S-function (System function)
module of the EKF program. The input module is the current, voltage, and six polynomial
functions created by the model. The parameters are including R0(SOC), Rp1(SOC), Cp1(SOC),
Rp2(SOC), Cp2(SOC), and OCV-SOC, and the output is the SOC value. The simulation model
is shown in Figure 4-15.
Figure 4-15: Structure of high-order model with EKF based simulation approach
In Figure 4-15, the actual current IL, the available capacity Z, and the original SOC of the
device at a given working condition can be obtained employing an ampere-time integration
process. The exact value of the SOC has been taken as the standard for the effects of the EKF
filtering. The UL voltage calculation data in response to the IL state was used to determine the
residual. The short sample time of the experimental instruments is 0.01s. The SOC value
45
obtained by amp-time integration of the current data by setting the exact initial SOC value has
a high precision, which can be assumed to be the real value according to the improved EKF
algorithm. The value obtained by the improved EKF algorithm is compared to the actual value,
and the observation algorithm is used to estimate the effect. However, the estimation of the
EKF simulation MATLAB code is below.
Function [sys,x0,str,ts] = ekf_2RCmodel(t,x,u,flag)
switch flag
case 0
[sys,x0,str,ts]=mdlInitializeSizes(t,x,u);
case 1
sys=mdlDerivatives(t,x,u);
case 2
sys=mdlUpdate(t,x,u);
case 3
sys=mdlOutputs(t,x,u);
case 4
sys=mdlGetTimeOfNextVarHit(t,x,u);
case 9
sys=mdlTerminate(t,x,u);
otherwise
DAStudio.error('Simulink:blocks:unhandledFlag', num2str(flag));
end
function [sys,x0,str,ts]=mdlInitializeSizes(~,~,~)
sizes = simsizes;
sizes.NumContStates = 0;
sizes.NumDiscStates = 12;
sizes.NumOutputs = 1;
sizes.NumInputs = 8;
sizes.DirFeedthrough = 0;
sizes.NumSampleTimes = 1;
sys = simsizes(sizes);
x0 = [0.5;0;0;0.01;0;0;0;0.01;0;0;0;0.01];
str = [];
46
ts = [0.1 0];
function sys=mdlDerivatives
sys = [];
function sys=mdlUpdate(~,x,u)
X1=[x(1);x(2);x(3)];
P=[x(4),x(5),x(6);x(7),x(8),x(9);x(10),x(11),x(12)];
q=0.000001;Q=[q,0,0;0,q,0;0,0,q];
R=0.00002;
A=[1,0,0;0,exp(-0.1/(u(4)*u(5))),0;0,0,exp(-0.1/(u(6)*u(7)))];
B=[-0.1/250580;u(4)*(1-exp(-0.1/(u(4)*u(5))));u(6)*(1-exp(-0.1/(u(6)*u(7))))];
C=[-12.032*X1(1)^3+21.978*X1(1)^2-10.82*X1(1)+1.966,1,1];
I=[1 0 0;0 1 0;0 0 1];
X1=A*X1+B*u(1);
P=A*P*A'+Q;
K=P*C'*(inv(C*P*C'+R));
X1=X1+K*(u(2)-[0,1,1]*X1-u(1)*u(3)-u(8));
P=(I-K*C)*P;
x(4)=P(1,1);x(5)=P(1,2);x(6)=P(1,3);x(7)=P(2,1);x(8)=P(2,2);x(9)=P(2,3);x(10)=P(3,1);x(11)=P(3,2);x(12)=P(3,3);
x=[X1;x(4);x(5);x(6);x(7);x(8);x(9);x(10);x(11);x(12)];
sys=x;
function sys=mdlOutputs(~,x,~)
sys =x(1);
function sys=mdlGetTimeOfNextVarHit
sampleTime = 1;
sys = t + sampleTime;
function sys=mdlTerminate(~,~,~)
sys = [];
According to the EKF iteration program and simulation model, an SOC online reliable
estimation model of the ternary Li-ion battery is constructed in the S-function module.
According to the extended Kalman iteration process mentioned above, the simulation
framework of the SOC calculation of the Li-ion battery using the EKF algorithm has
developed into Simulink.
47
4.4.2 The approach of Algorithm Verification Strategy
The high-order equivalent circuit model shall be calculated by parameter identification

with the experimental hybrid pulse. Model data and real data were compared and evaluated
with other battery state data to check the model's validity. The continuous power discharge at
a given time was used to model the batteries different operating conditions. The current value
in the experimental data collected from the test instruments shall be taken as the input state.
Simulation terminal voltage is obtained from a simulation model. The experimental terminal
voltage is compared to the effects as seen in Figure 4-16.
1.0 SOC1
SOC2
0.8
SOC (%)
0.6
0.4 0.903
0.860
0.817
0.2 0.774
8820 9450 10080 10710
0.0
0 6500 13000 19500 26000 32500
t (s)
Figure 4-16: EKF estimation results with SOC
In Figure 4-16, SOC1 is marked by the red curve of real terminal voltage data obtained by the
test equipment. The SOC2 is marked by the solid blue curve output terminal voltage curve
obtained by the simulation model under the input current condition. The deviation of the
simulation curve is close to that of the real test curve, which may help stimulate the batteries
discharge. The initial value calculates the error. The actual value has arbitrarily seen in Figure
4-17, which is the error curve produced by subtracting two SOC value curves.
0.26 Err
0.13
0.027V
Err (V)
0.029V
0.00
-0.13 0.0000
-0.0087
-0.0174
-0.26 -0.0261
-0.0348
6660 7030 7400 7770 8140
-0.39
0 6500 13000 19500 26000 32500
t (s)
Figure 4-17: EKF estimation error with SOC
In Figure 4-17, the red curve is the difference between the EKF estimation of the extended
Kalman filter model value and the actual SOC value. The experimental findings suggest that
the initial SOC error is minimal. This algorithm can be quickly addressed and has strong
robustness. Under complex conditions, the SOC prediction error is less than 0.029V.
48
The experiment verification derived parameters, validated graphs, and algorithm results
have illustrated in this chapter. A variety of tests, including the capacity test and the HPPC
test, were carried out to determine the state of charge (SOC) reliably, including the Capacity
test, HPPC test.
Parameters
identification based
on High-order
Arranged test Proposed algorithm
platform and off- and SOC estimation
line data acquisition based EKF
Real-time made a
Simulink block and
model validation
Figure 4-18: Overview of experimental verification and verification
According to Figure 4-18, The model parameter was identified using the curve fitting method
and the relationship between SOC-OCV presented briefly. After then, Simulation verification
using the parameters established the Mathlab then compare the real data versus model data has
verified. Finally established Mathlab code and built EKF with high-order modeling to satisfied
the accurate state of charge estimation, which error less than 0.029V.
49
CHAPTER 5 CONCLUSION AND RESULT
CHAPTER 5 CONCLUSION AND RESULT
An accurate state of charge (SOC) estimation approaches numerous experiments and

methodologies were presented in the thesis. However, more research needs to examine the
subjects of the battery management system (BMS). The main focus of this study is to make
the best BMS for the reliable and effective outcome for SOC estimation for EVs. However, to
understand Li-ion behavior and realistic validation simulation is a beneficial way to
understand the behavior of dynamic systems under several conditions.
The 0introduce the concept of the need for this research. The research motivation, thesis
objective, and problem formulation Li-ion batteries technology are also presented. The
research objectives and structure of this thesis are briefly summarized. In the Chapter 2 is
established based on the mathematical analysis. Also, the extended Kalman filter (EFK)
algorithm mathematical analysis was briefly presented, which was used in the research study.
In Chapter 3 introduced a comparison of battery modeling with different battery modeling
with the state-space explanation. Also, high-order modeling is established, which is the major
part of this thesis as an equivalent circuit. In rest Chapter 4 gives the comprehensive results
of all the experiments, parameters extraction of high-order model is briefly discussed.
Furthermore, the simulation platform was conducted concerning high-order modeling and
verification the reliability. Also, coding helps to estimate the SOC-EKF algorithm briefly.
In conclusion, A thermal dependency electrical high-order equivalent circuit of a

rechargeable lithium-ion battery cell, and an experimental identification of the SOC estimation
is carried out. The high-order model is also verified in MATLAB/Simulation, and the
simulation error comparatively low is 0.051V. Furthermore, An Ah integration and an EKF
method for SOC estimates were developed in MATLAB, and the algorithm was written by the
S-function. The results from the simulation show that EKF precision is significantly better
than the method of integration Ah is 0.029V. The result is comprehensively robust and benefits
other researchers to take a reference. However, the major highlights parts of the thesis are
presented in the below table.
➢ Preciously model parameters identified based on high-order

modeling
➢ The model verification has made that error is remarkably less than
0.051V which made a strong effect on the state of charge
Research estimation.
Highlights ➢ The algorithm has established through an extended Kalman filter
(EKF) algorithm, which is imposed adequate impact on the state of
charge estimation approach. The rate of error is less than is 0.029V.
➢ Strongly monitoring the Li-ion batteries internal situation such as
charging, discharging temperature.
50
ACKNOWLEDGMENT
ACKNOWLEDGMENT
At the very beginning, I would like to express my deepest gratitude to Almighty “Allah”
for allowing me with strength and pertinence to carry out this study.
First and foremost, I would like to express my sincere gratitude to my supervisor Dr. Wang
Shunli for my postgraduate thesis and continuous support, patience, inspiration, and
enthusiasm. His encouragement kept me going in my studies and the writing process of this
thesis.
I am also extremely grateful to all my colleagues for their assistance in various ways and
I appreciate their suggestions and their time. The support from my colleagues helped me
achieving this great milestone. Special thanks to Ratul, Xiong Xin, and Xu Wenhua, and
Emmanuel Appiah, Islam Md Monirul who were closed with me throughout my study.
I would like to thank the Chinese government for assisting me during the studies. And I
also would like to acknowledge the School of Information Engineering at the Southwest
University of Science and Technology, Mianyang, Sichuan province, China, for their
assistance.
Finally, I must express my sincere appreciation to my parents for providing me with

unwavering guidance and continuous encouragement during my studies. Throughout the
process of studying and writing this thesis. Without them, this feat would not have been
practicable.
/Thank you.
51
REFERENCES
REFERENCES
1. Zhang, J.N., et al., An online heat generation estimation method for lithium-ion batteries using dual-
temperature measurements. Applied Energy, 2020. 272.
2. Yuan, H., et al., Model-based observers for internal states estimation and control of proton exchange
membrane fuel cell system: A review. Journal of Power Sources, 2020. 468.
3. Peng, J.K., et al., An improved state of charge estimation method based on cubature Kalman filter for
lithium-ion batteries. Applied Energy, 2019. 253.
4. Zhang, M., K. Wang, and Y.T. Zhou, Online State of Charge Estimation of Lithium-Ion Cells Using Particle
Filter-Based Hybrid Filtering Approach. Complexity, 2020. 2020: p. 10.
5. Lin, C.P., et al., SOH Estimation and SOC Recalibration of Lithium-Ion Battery with Incremental Capacity
Analysis & Cubic Smoothing Spline. Journal of the Electrochemical Society, 2020. 167(9): p. 15.
6. Yan, Y.D., K.Y. Hu, and C.H. Tsai, Digital Battery Management Design for Point-of-Load Applications
With Cell Balancing. Ieee Transactions on Industrial Electronics, 2020. 67(8): p. 6365-6375.
7. Tang, S.X., et al., State-of-Charge estimation from a thermal-electrochemical model of lithium-ion batteries.
Automatica, 2017. 83: p. 206-219.
8. Wang, S.L., et al., Open circuit voltage and state of charge relationship functional optimization for the
working state monitoring of the aerial lithium-ion battery pack. Journal of Cleaner Production, 2018. 198:
p. 1090-1104.
9. Ko, S.T., J.N. Ahn, and B.K. Lee, Enhanced Equivalent Circuit Modeling for Li-ion Battery Using
Recursive Parameter Correction. Journal of Electrical Engineering & Technology, 2018. 13(3): p. 1147-
1155.
10. Xuan, D.J., et al., Real-time estimation of state-of-charge in lithium-ion batteries using improved central
difference transform method. Journal of Cleaner Production, 2020. 252.
11. Wang, S.L., et al., An improved packing equivalent circuit modeling method with the cell-to-cell consistency
state evaluation of the internal connected lithium-ion batteries. Energy Science & Engineering, 2019. 7(2):
p. 546-556.
12. He, Y.L., et al., Modeling of Dynamic Hysteresis Characters for the Lithium-Ion Battery. Journal of the
Electrochemical Society, 2020. 167(9).
13. Mayur, M., et al., Identification of stoichiometric and microstructural parameters of a lithium-ion cell with
blend electrode. Physical Chemistry Chemical Physics, 2019. 21(42): p. 23672-23684.
14. Meng, J.H., et al., An optimized ensemble learning framework for lithium-ion Battery State of Health
estimation in energy storage system. Energy, 2020. 206.
15. Zhang, Y.Z., et al., Lithium-Ion Battery Pack State of Charge and State of Energy Estimation Algorithms
Using a Hardware-in-the-Loop Validation. Ieee Transactions on Power Electronics, 2017. 32(6): p. 4421-
4431.
16. Ai, S.J., et al., Effect of Derivatives of Glycerol Sulfite as Electrolyte Additive on Electrochemical
Performance of Lithium Ion Battery. Chemical Journal of Chinese Universities-Chinese, 2018. 39(11): p.
2520-2528.
52
REFERENCES
17. Appiah, W.A., et al., A coupled chemo-mechanical model to study the effects of adhesive strength on the
electrochemical performance of silicon electrodes for advanced lithium ion batteries. Journal of Power
Sources, 2018. 407: p. 153-161.
18. Wang, C.W., et al., Synthesis and electrochemical properties of Ca-doped LiNi0.8Co0.2O2 cathode
material for lithium ion battery. Journal of Solid State Electrochemistry, 2007. 11(3): p. 361-364.
19. Li, X.Y., C.G. Yuan, and Z.P. Wang, State of health estimation for Li-ion battery via partial incremental
capacity analysis based on support vector regression. Energy, 2020. 203.
20. Tian, J.P., R. Xiong, and W.X. Shen, State-of-Health Estimation Based on Differential Temperature for
Lithium Ion Batteries. Ieee Transactions on Power Electronics, 2020. 35(10): p. 10363-10373.
21. Liu, Q., et al., Li-B Alloy as an Anode Material for Stable and Long Life Lithium Metal Batteries. Energies,
2018. 11(10).
22. Ding, X.F., et al., An improved Thevenin model of lithium-ion battery with high accuracy for electric
vehicles. Applied Energy, 2019. 254: p. 8.
23. Yang, Y., et al., A review on structuralized current collectors for high-performance lithium-ion battery
anodes. Applied Energy, 2020. 276.
24. Wang, Y.J., Z.D. Sun, and Z.H. Chen, Development of energy management system based on a rule-based
power distribution strategy for hybrid power sources. Energy, 2019. 175: p. 1055-1066.
25. Fan, Y.C., et al., Theoretical investigation of lithium ions' nucleation performance on metal-doped Cu
surfaces. Journal of Energy Chemistry, 2019. 39: p. 160-169.
26. Pavkovic, D., et al., Dual EKF-Based State and Parameter Estimator for a LiFePO4 Battery Cell. Journal
of Power Electronics, 2017. 17(2): p. 398-410.
27. Qiu, Y., et al., State of charge estimation of vanadium redox battery based on improved extended Kalman
filter. Isa Transactions, 2019. 94: p. 326-337.
28. Dai, K.W., J. Wang, and H.W. He, An Improved SOC Estimator Using Time-Varying Discrete Sliding Mode
Observer. Ieee Access, 2019. 7: p. 115463-115472.
29. Rosewater, D., et al., , Battery Energy Storage State-of-Charge Forecasting: Models, Optimization, and
Accuracy. IEEE Transactions on Smart Grid, 2018. 10(3): p. 2453 - 2462.
30. Abe, Y., N. Hori, and S. Kumagai, Electrochemical Impedance Spectroscopy on the Performance
Degradation of LiFePO4/Graphite Lithium-Ion Battery Due to Charge-Discharge Cycling under Different
C-Rates. Energies, 2019. 12(23): p. 4507.
31. Xu, W., J.L. Xu, and X.F. Yan, Lithium-ion battery state of charge and parameters joint estimation using
cubature Kalman filter and particle filter. Journal of Power Electronics, 2020. 20(1): p. 292-307.
32. Shen, P., et al., Error Analysis of the Model-Based State-of-Charge Observer for Lithium-Ion Batteries.
Ieee Transactions on Vehicular Technology, 2018. 67(9): p. 8055-8064.
33. Liu, D.T., et al., Hybrid state of charge estimation for lithium-ion battery under dynamic operating
conditions. International Journal of Electrical Power & Energy Systems, 2019. 110: p. 48-61.
34. Bhayo, B.A., et al., Power management optimization of hybrid solar photovoltaic-battery integrated with
pumped-hydro-storage system for standalone electricity generation. Energy Conversion and Management,
2020. 215.
53
REFERENCES
35. Bhowmik, P., S. Chandak, and P.K. Rout, State of charge and state of power management among the energy
storage systems by the fuzzy tuned dynamic exponent and the dynamic PI controller. Journal of Energy
Storage, 2018. 19: p. 348-363.
36. Nejad, S. and D.T. Gladwin, Online Battery State of Power Prediction Using PRBS and Extended Kalman
Filter. Ieee Transactions on Industrial Electronics, 2020. 67(5): p. 3747-3755.
37. Wang, L.M., et al., State of charge estimation for LiFePO4 battery via dual extended kalman filter and
charging voltage curve. Electrochimica Acta, 2019. 296: p. 1009-1017.
38. Wang, S.L., et al., The parameter identification method study of the splice equivalent circuit model for the
aerial lithium-ion battery pack. Measurement & Control, 2018. 51(5-6): p. 125-137.
39. Wang, S.L., et al., A novel weight coefficient calculation method for the real-time state monitoring of the
lithium-ion battery packs under the complex current variation working conditions. Energy Science &
Engineering, 2019. 7(6): p. 3038-3057.
40. Linghu, J.Q., et al., Estimation for state-of-charge of lithium-ion battery based on an adaptive high-degree
cubature Kalman filter. Energy, 2019. 189.
41. Tang, X.P., et al., Model migration based battery power capability evaluation considering uncertainties of
temperature and aging. Journal of Power Sources, 2019. 440.
42. Singh, K.V., H.O. Bansal, and D. Singh, Hardware-in-the-loop Implementation of ANFIS based Adaptive
SoC Estimation of Lithium-ion Battery for Hybrid Vehicle Applications. Journal of Energy Storage, 2020.
27.
43. Xu, X.M., et al., State Estimation of Lithium Batteries for Energy Storage Based on Dual Extended Kalman
Filter. Mathematical Problems in Engineering, 2020. 2020.
44. Hou, J., et al., Adaptive Dual Extended Kalman Filter Based on Variational Bayesian Approximation for
Joint Estimation of Lithium-Ion Battery State of Charge and Model Parameters. Applied Sciences-Basel,
2019. 9(9): p. 23.
45. Shu, X., et al., An adaptive fusion estimation algorithm for state of charge of lithium-ion batteries
considering wide operating temperature and degradation. Journal of Power Sources, 2020. 462.
46. Wang, S.L., et al., A novel charged state prediction method of the lithium ion battery packs based on the
composite equivalent modeling and improved splice Kalman filtering algorithm. Journal of Power Sources,
2020. 471.
47. Xia, B.Z., et al., A State of Charge Estimation Method Based on Adaptive Extended Kalman-Particle
Filtering for Lithium-ion Batteries. Energies, 2018. 11(10).
48. Yang, F.F., et al., A coulombic efficiency-based model for prognostics and health estimation of lithium-ion
batteries. Energy, 2019. 171: p. 1173-1182.
49. Yang, N.X., et al., An improved semi-empirical model for thermal analysis of lithium-ion batteries.
Electrochimica Acta, 2019. 311: p. 8-20.
50. Yin, Y.L. and S.Y. Choe, Actively temperature controlled health-aware fast charging method for lithium-
ion battery using nonlinear model predictive control. Applied Energy, 2020. 271.
51. Zhang, L., et al., An Accurate Time Constant Parameter Determination Method for the Varying Condition
Equivalent Circuit Model of Lithium Batteries. Energies, 2020. 13(8).
52. Zhang, N., et al., A Novel Method for Estimating State-of-Charge in Power Batteries for Electric Vehicles.
International Journal of Precision Engineering and Manufacturing, 2019. 20(5): p. 845-852.
54
REFERENCES
53. Wang, S.L., et al., Adaptive State-of-Charge Estimation Method for an Aeronautical Lithium-ion Battery
Pack Based on a Reduced Particle-unscented Kalman Filter. Journal of Power Electronics, 2018. 18(4): p.
1127-1139.
54. Zheng, W.H., et al., State of Charge Estimation for Power Lithium-Ion Battery Using a Fuzzy Logic Sliding
Mode Observer. Energies, 2019. 12(13).
55. Tran, N.T., et al., A Computationally Efficient Coupled Electrochemical-Thermal Model for Large Format
Cylindrical Lithium Ion Batteries. Journal of the Electrochemical Society, 2019. 166(13): p. A3059-A3071.
56. Chen, L., et al., Electrochemical Model Parameter Identification of Lithium-Ion Battery with Temperature
and Current Dependence. International Journal of Electrochemical Science, 2019. 14(5): p. 4124-4143.
57. Liu, X.Y., W.L. Li, and A.G. Zhou, PNGV Equivalent Circuit Model and SOC Estimation Algorithm for
Lithium Battery Pack Adopted in AGV Vehicle. Ieee Access, 2018. 6: p. 23639-23647.
58. Zheng, Y.J., et al., State-of-charge inconsistency estimation of lithium-ion battery pack using mean-
difference model and extended Kalman filter. Journal of Power Sources, 2018. 383: p. 50-58.
59. Yan, X.-W., Guo, Y.-W., Cui, Y., Wang, Y.-W., & Deng, H.-R, Electric Vehicle Battery SOC Estimation
based on GNL Model Adaptive Kalman Filter. Journal of Physics, 2018.
60. Yao, S.G., et al., Study on Thevenin Equivalent Circuit Modeling of Zinc-Nickel Single-Flow Battery.
International Journal of Electrochemical Science, 2018. 13(5): p. 4455-4465.
61. Zhao, Y., et al., A review on modeling of electro-chemo-mechanics in lithium-ion batteries. Journal of Power
Sources, 2019. 413: p. 259-283.
62. Zheng, Q.F., et al., A cyclic phosphate-based battery electrolyte for high voltage and safe operation. Nature
Energy, 2020. 5(4): p. 291-298.
63. Wang, S.L., et al., Analyses and optimization of electrolyte concentration on the electrochemical
performance of iron-chromium flow battery. Applied Energy, 2020. 271.
64. Tian, Y., et al., A combined method for state-of-charge estimation for lithium-ion batteries using a long
short-term memory network and an adaptive cubature Kalman filter. Applied Energy, 2020. 265.
65. Xu, Y.D., et al., State of Charge Estimation for Lithium-Ion Batteries Based on Temperature-Dependent
Second-Order RC Model. Electronics, 2019. 8(9).
66. Zhao, R.X., et al., A Compact Methodology Via a Recurrent Neural Network for Accurate Equivalent Circuit
Type Modeling of Lithium-Ion Batteries. IEEE Transactions on Industry Applications, 2019. 55(2): p. 1922-
1931.
67. Yang, Q.X., et al., State-of-health estimation of lithium-ion battery based on fractional impedance model
and interval capacity. International Journal of Electrical Power & Energy Systems, 2020. 119.
68. Yang, L., et al., Supervisory long-term prediction of state of available power for lithium-ion batteries in
electric vehicles. Applied Energy, 2020. 257.
69. Lyu, Z.Q. and R.J. Gao, Li-ion battery state of health estimation through Gaussian process regression with
Thevenin model. International Journal of Energy Research, 2020.
70. Pai, K.-J., A Reformatory Model Incorporating PNGV Battery and Three-Terminal-Switch Models to
Design and Implement Feedback Compensations of LiFePO4 Battery Chargers. Electronics, 2019. 8(2).
71. Wang, H., et al., An Energy Balanced and Lifetime Extended Routing Protocol for Underwater Sensor
Networks. Sensors, 2018. 18(5).
55
REFERENCES
72. Huo, Y.T., et al., Research on parameter identification and state of charge estimation of improved
equivalent circuit model of Li-ion battery based on temperature effects for battery thermal management.
International Journal of Energy Research, 2020.
73. Xie, G., et al., Remaining useful life prediction of lithium-ion battery based on an improved particle filter
algorithm. Canadian Journal of Chemical Engineering, 2019.
74. Xiong, R., et al., Lithium-Ion Battery Health Prognosis Based on a Real Battery Management System Used
in Electric Vehicles. Ieee Transactions on Vehicular Technology, 2019. 68(5): p. 4110-4121.
75. Campestrini, C., S. Kosch, and A. Jossen, Influence of change in open circuit voltage on the state of charge
estimation with an extended Kalman filter. Journal of Energy Storage, 2017. 12: p. 149-156.
76. Wang, M.S., et al., State Space Model of Aggregated Electric Vehicles for Frequency Regulation. Ieee
Transactions on Smart Grid, 2020. 11(2): p. 981-994.
77. Shrivastava, P., et al., Overview of model-based online state-of-charge estimation using Kalman filter family
for lithium-ion batteries. Renewable & Sustainable Energy Reviews, 2019. 113.
78. Wang, T.P., et al., Model-based unscented Kalman filter observer design for lithium-ion battery state of
charge estimation. International Journal of Energy Research, 2018. 42(4): p. 1603-1614.
79. Zheng, Y.J., et al., An accurate parameters extraction method for a novel on-board battery model
considering electrochemical properties. Journal of Energy Storage, 2019. 24.
80. Vijayaraghavan, V., et al., Crash analysis of lithium-ion batteries using finite element based neural search
analytical models. Engineering with Computers, 2019. 35(1): p. 115-125.
56
View publication stats

SchoolofInformationEngineering 7420190042 WahhabMdAbdul ShunliWang

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SchoolofInformationEngineering 7420190042 WahhabMdAbdul ShunliWang

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Master's Degree Final Defense PPT

Presentation · January 2020

Md Abdul Wahhab Shunli Wang

SEE PROFILE SEE PROFILE

001_JCYJS_BMS View project

The user has requested enhancement of the downloaded file.

论 文 题 目 Thesis Title Accurate power state

作者姓名 Name Wahhab Md Abdul

2021 年year 01 月month

UDC: Security classification:

ACADEMIC MASTER DISSERTATION

Thesis Title: Accurate power state estimation situated by extended

Date of Dissertation Defense

ACADEMIC MASTER DISSERTATION

ACADEMIC MASTER DISSERTATION

Candidate 申请人: Wahhab Md Abdul

Supervisor 导师: Wang Shunli, Associate Professor

5、将在 Simulink 中构建的模型与高阶模型进行比较，已验证模型精确性。

6、通过 S-function 实现算法并验证模型参数辨识精度。

1. The high-order battery model is picked.

KEYWORDS: Li-ion batteries; high-order modeling; open-circuit voltage; parameter

TYPE OF THESIS: Application Research

Thesis originality statement

Graduate Signature: Date：year / month / day

Statement on the use of the License

Graduate Signature： Date: year / month / day

Supervisor signature： Date: year / month / day

3.4.2 Exponential Curve Fitting Technique ................................................................ 30

1.1 Thesis Motivation

Moreover, Equivalent circuit modeling (ECM) batteries have been investigated

1.2 Thesis Objective

1. An accurate state of charge (SOC) estimation.

2. Understand the dynamic characteristics of battery behavior.

Ampere Hour Method

Test Platform Design and

Simulation Model Design

Figure 1-1: Essential research contents of the thesis

“CT-4616- TCP/IP port

Selected ternary Input expected

Extended State-space model

Figure 1-2: Major working procedure for SOC estimation approach

1.3 Problem Formulation

maintain permanent energy supply—various types of rechargeable industrial batteries, such as

1.4 Thesis Organization

The thesis includes the chapters listed below,

Chapter 1 : INTRODUCTION – The research motivation and objective of Li-ion battery

Chapter 2 : MATHEMATICAL ANALYSIS – The chapter broadly discussed mathematical

Chapter 3 : MODEL BUILDING AND REALIZATION – The various types of battery

Chapter 4 : EXPERIMENTAL VERIFICATION – The experimental platform design, data

CHAPTER 2 MATHEMATICAL ANALYSIS

2.1 Operating Characteristic of Li-ion battery

2.1.1 Operating principle Li-ion batteries

Figure 2-1: Schematic structure of a Li-ion battery

positive：LiCoO ⎯⎯⎯ ⎯⎯⎯⎯ch arg e

2.1.2 Prolong the life cycles

Depth of Discharge Discharge Cycles

2.1.3 Critical Parameters Description

Table 2-2: The major parameters of Li-ion batteries

Parameters Name Parameter definition

The amount of active material determines the batteries capacity,

The possible differentiation throughout the batteries between

When the batteries are fully discharged, the end-of-discharge

The amount of the batteries overall power expended when it is not

论文题目 Thesis Title Accurate power state