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

net/publication/341778248

A Review on AI based Predictive Battery Management System for E-Mobility

Article in Test Engineering and Management · May 2020

CITATIONS READS

0 291

2 authors:

Satyashil Nagarale B P Patil

National Institute of Technology Tiruchirappalli Army Institute of Technology
2 PUBLICATIONS 5 CITATIONS 138 PUBLICATIONS 302 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Indoor Positioning System View project

Ion Generator and Ion Counter View project

All content following this page was uploaded by B P Patil on 31 May 2020.

The user has requested enhancement of the downloaded file.

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

A Review on AI based Predictive Battery

Management System for E-Mobility
Satyashil D. Nagarale, B. P. Patil
Asst. Professor, Dept of ETC, PCCOE, Pune
Principal, Army Institute of Technology, Pune
Article Info Abstract:
Volume 83
Page Number: 15053 – 15064 Abstract: The advancement in digitalization and availability of reliable sources of
Publication Issue: information that provide credible data, Artificial Intelligence (AI) has emerged to
May - June 2020 solve complex computational real life problems which was challenging earlier. The
Artificial Neural Networks (ANNs) play a very effective role in digital signal
processing. However, ANNs need rigorous main processors and high memory
bandwidth, and hence cannot provide expected levels of performance. As a result,
hardware accelerators such as Graphic Processing Units (GPUs), Field
Programmable Gate Arrays (FPGAs), and Application Specific Integrated Circuits
(ASICs) have been used for improving overall performance of AI based
applications. FPGAs are widely used for AI implementation as FPGAs have
features like high-speed acceleration, low power consumption which cannot be
done using central processors and GPUs. FPGAs are also a reprogrammable unlike
central processors, GPU and ASIC. In Electric-powered vehicles (E-Mobility),
Battery Management Systems (BMS) perform different operations for better use of
energy stored in lithium-ion batteries (LiBs). The LiBs are a non-linear
electrochemical system which is very complex and time-variant in nature. Because
of this nature, estimation of States like State of Charge (SoC), State of Health
(SoH) and Remaining Useful Life (RUL) is very difficult. This has motivated
researchers to design and develop different algorithms which will address the
challenges of LiBs states estimations. This paper intends to review AI based data-
driven approaches and hardware accelerators to predict the SoC, SoH and RUL of
the LiBs. The goal is to choose an appropriate algorithm to develop an advanced AI
based BMS that can precisely indicate the LiBs states which will be useful in E-
Mobility.
Article History
Article Received: 1May 2020
Revised: 11 May 2020 Keywords: Artificial Intelligence, Battery Management Systems, Electric-powered
Accepted: 20 May 2020 automobiles, State of Charge, Remaining Useful Life, State of Health, Field
Publication: 24May 2020 Programmable Gate Arrays, Graphic Processing Units, Application Specific
Integrated Circuits, and Li-ion batteries.

I. Introduction industry due to their safety, low maintenance,

In recent decades, E-Mobility has gained longer life span, high efficiencies, nominal
importance due to environmental concerns and voltage, reasonable cost, and ability to operate
the depletion of fossil fuels. Renewable energy in a large temperature range [1].
is preferred over fossil fuels due to The availability of extensive credible data and
characteristics like renewability, non- advancement in digital electronics that have
depletion, and it does not give rise to high computing power, have revived AI [2].
environmental issues. LiBs have achieved The accurate determination of the LiB states is
considerable attention in the E-Mobility extremely challenging because of its high non-

Published by: The Mattingley Publishing Co., Inc. 15053

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

linear, time-variant and complex The past decades have witnessed an increase in
electrochemical system. AI based algorithms E-Mobility. Over 125 million electric vehicles
can be used to estimate battery states will be in use by 2030 according to the
accurately independent of granular information projections [5]. BMS ensures a reliable,
of electrochemical processes of batteries. AI efficient, and safe operation of the LiBs. BMS
algorithms or Data driven approaches will be is an electronic system that estimates the SoC,
useful for predicting battery states like SoC, SoH, and RUL; controls over-charge and
SoH and RUL. The SoC, SoH & RUL protect under-charge; controls voltage and
the LiBs by providing visual signals or temperature, and ensures safety and fault
warnings for any problems in the BMS. This assessment of the cells. Due to the battery
facilitates monitoring of the voltage, internal being an electrochemical product, it poses an
resistance, current, temperatures, ageing and uncertainty of performance in various
life cycles of the LiBs [3, 4]. This paper environmental and operational conditions.
discusses a general working of BMS and BMS, therefore, controls and monitors the
energy storage systems. This paper presents a states of the batteries at various levels like
relative analysis of AI based data driven module, cell and battery pack levels making it
estimation approaches of SoC, SoH and RUL a critical component in E-Mobility for
and evaluates various AI based approaches for optimum operation of LiBs in varying driving
enhanced battery management. Finally, as part conditions. Relays, Sensors, and breakers are
of future approach, this review paper incorporated in the BMS of E-Mobility to
recommends use of AI based BMS for E- ensure protection from under voltage, over
Mobility which will be beneficial for voltage, over-charge, under-charge, under-
automobile engineers, experts, and researchers. current, over current, etc. [6]. The table below
showcases BMS components and their
II. Battery management system (BMS) for operational details [7].
E- Mobility
Table 1 – BMS Components and their Operational Details
BMS Components Operational Details of Component
Measurement or Measures physical or electrical parameters such as current, voltage,
Data acquisition temperature, pressure. A computer converts them into digital values.
Measurement can be done with the help of sensors, hardware and software.
Battery Algorithm Estimate SOC, SOH and RUL with reference to current, voltage, ageing, and
temperature.
Capability After battery states estimation, BMS has to find the maximum available charge
Estimation and discharge current at any instant in accordance by using an algorithm. The
output of this component is passes to the Electronic Control Unit (ECU) to
prevent over charge and discharge of battery.
Cell Balancing Cell Balancing equals or balances voltage level of each cell.
Thermal Any divergence from an operational temperature range is addressed by
Management activating cooling or heating system, etc. which facilitates thermal stability.
Vehicle Communication with all battery components.
Communications
(CAN and
Ethernet)
Safety Safety and Reliability of LiB relates to the construction and installation of the

Published by: The Mattingley Publishing Co., Inc. 15054

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

cells. Protects the pack from damaging events such as over-charge and under-
charge and proactively detects potentially dangerous conditions within the
battery pack.

The literature reports that BMS can be developers use suitable structure like a
designed using various approaches based on distributed one for scalability and portability
the suitable operation for a particular [11]. Microcontroller Unit (MCU) or FPGAs
application, but most approaches focus on a are used as a hardware component in BMS.
specific operation, such as SoC estimation, LiBs are extensively equipped in both high-
SoH estimation, RUL estimation or Cell power applications and low-power electronics
Balancing [8-10]. products, such as hybrid-motor engines,
BMS has three different types of design electric cars, smart phones, tablet, laptops, etc.
structures viz. centralized, modular, and The table below showcases Commercial BMS
distributed. Based on the objectives to be met, used in E-Mobility [12-17].
Table 2 - Commercial BMS used in E-Mobility
Sr. BMS Description Key Features
No. Manufactures

1 SPIKE SPIKE’s developed Each module comes with an advanced Battery

advanced BMS with Management System and is optionally
highly efficient thermal equipped with liquid cooling plates to
management. significantly improve lifetime, safety and
performance.

2 AMP AMP provides the best- Estimate SoC & SoH, Power Prediction,
in-class BMS for E- temperature & Humidity Sensing, Lower
Mobility. system cost, Highest battery utilization and
etc.

3 ION ENERGY ION Energy’s battery Configurable single board compact

designs and the architecture.It supports multiple battery
ecosystem use a chemistries. SoC & SoH estimations based on
customizable and advanced algorithms. Easily Customizable
modular approach. features like additional temperature
measurements, current sensor, higher
balancing current, heater, etc.

4 RENESAS Renesas' provide BMS Cell equalization and safety.

ELECTRONICS solutions are specifically Designed for next generation E-Mobility
designed to meet the stiff
safety, performance
requirements and
reliability

5 GREENECO Green Eco products are Charge – Discharge Protection for each cell
used in renewable and Over Current Protection, Impendence and
energy efficient products. Temperature measurements.

Published by: The Mattingley Publishing Co., Inc. 15055

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

6 ORION The Orion BMS, BMS settings include: Over Charge and
specifically designed to Under Charge voltage, Over current limits.
meet the difficult SoC Estimation, Thermal Management along
requirements of with current sensor, communication unit and
protecting and managing etc.
packs of battery for E-
Mobility with automotive
grade quality.

BMS for E-Mobility require efficient hardware Currently, LiB technology is growing very fast
and software along with advanced battery state as the most reliable electrical power source for
estimation algorithms. The manufacturers have E- Mobility. Various ESDs like LiB, VRFB,
not disclosed the advanced algorithms they NiCd, NaNICl, ZBFB, NaS and lead acid are
have used in BMS. available. The table below lists the Different
types of ESD for E- Mobility with their
III. Energy Storage Device (ESD) for E- properties [18-19].
Mobility
Table 3 - Types and Properties of ESD for E- Mobility
Energy Specific Power Cell Life Nominal Round trip Estimated
storage Energy density Voltage Cycle Capacity Efficiency cost,
device Density (W/L) (V) (%) (%) (USD/kWh)
(Wh/L)
LiB 200–400 1500- 4.3 10,000 95 96 200–1260
10,000
VRFB 25–33 1–2 1.4 13,000 100 70 315–1050
NiCd 60–150 80–600 1.3 2500 85 83 -
NaNiCl 160–275 150–270 - 3000 100 84 315–488
ZBFB 55–65 1–25 1.8 10,000 100 70 525–1680
NaS 140–300 140–180 2.08 5000 100 80 263–735
Lead- 50–80 10–400 2.0 1500 50 82 105–475
acid

Table 3 shows that VRFB has the highest life its power density, specific energy density, cell
cycle and LiB has the highest power densities voltage, life cycle, round trip efficiency and
and specific energy as compared to the other cost. Although LiB is the best choice for E-
storage devices. The nominal voltage is Mobility, the challenges of reduction in capital
important point as it determines the quantity of cost and high life cycle need to be addressed.
single cell and number of cells required in a BMS is an essential component of all LiB
pack of battery for reliable and safe operations. packs. These battery packs can be classified
Round Trip efficiency is defined as the ratio of into Low Voltage (LV) or High Voltage (HV)
energy recovered and energy input to ESS. (UNECE 2013).
Compare to other Energy storage device LiB
has better round trip efficiency which plays an
important role in BMS to improve overall
performance. LiB is better choice because of

Published by: The Mattingley Publishing Co., Inc. 15056

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

Table 4: Classifications of LiB packs

Classifications of Range Usability
LiB packs

Low Voltage (LV) 30 – 1000 VAC Light electric and hybrid vehicles, two and three wheelers.

High Voltage (HV) 60 – 1500 VDC Heavy electric vehicles such as elevators, conveyors, trolleybuses and
etc.

2. SoC, SoH and RUL Estimation Methods SoH = (maximum charge capacity /rated
As pointed out earlier, accurate battery states capacity*) X 100
estimation is a very critical function of the RUL = Total life cycles - Actual life cycles
BMS design. Therefore, different literature (*Rated capacity is given by manufactures
presents various methods to calculate the SoC, which is constant through the battery lifetime)
SoH and RUL precisely. The SoC estimation Table 5 shows the estimation methods of
helps to optimize the battery performance and different batteries.
extend battery lifespan.
SoC = (available capacity/rated capacity*) X
100
Table 5 - Estimation methods of different batteries [19-23]
Classification
Sr. Working
of Estimation Methods Key Limitations
No. Principle
Methods
1 Direct Physical Internal Unable to measure initial states.
Measurement properties such Resistance(IR), Open loop, minute errors will build
as impedance, Open-circuit voltage up with time due to the integration
and (OCV), term.
charge/discharge Electromotive Requires high precision sensors.
current. force(EMF) &
Impedance
Spectroscopy (IS)
2 Book- Input will be Coulomb Counting Open Loop, Sensitive to current
keeping Battery charge (CC) sensor precision.
and discharge Accuracy depends on quality of
current sensors and initial LiB states.
Accuracy depends on battery
history, temperature, discharge
current, and life cycle of battery.
Not suggested for online BMS as
the battery needs long time resting.
3 Model-based By developing Adaptive methods: Accuracy highly depends on quality
battery models The Kalman filter; of the model. The algorithms have
battery ageing is Particle filter; & high complexity, high
evaluated but Least square computational cost, and need
these methods experimental tests to form different
involve datasets. The performance of the
complicated KF variants is required on previous
mathematical knowledge and measurement noise

Published by: The Mattingley Publishing Co., Inc. 15057

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

operations. covariance which is causing

performance to degrade and leads
to poor convergence or slow
adaptation.
4 Data Driven Used Advanced FL, ANN, SVM, Require high computational time
or AI or Algorithms. RNN and storage size. Accuracy depends
Computer on quantity of data and quality
Intelligence training NN.

As compared to other methods, the data driven Another algorithm which is based on nature-
approaches can successfully capture the inspired approach using the social behavior of
nonlinear relationship of battery parameters different species, such as birds or fishes,
and are more suitable to estimate SoC, SoH interacting with each other or with the
and RUL. These methods work independent of surroundings is PSO. Main objective of this
battery operating condition and complex method is data sharing in the group, where
electrochemical process of LiBs. For battery each bird in the flock does not know the exact
state estimation by the data driven FL, SVM, location of the food, but they can track the
ANN are accepted by few of researchers food site very easily through information
accepted [24-27]. Based on the study of sharing method [30].
various methods to estimate battery states, the iv. iv. Fuzzy Logic (FL)
AI based data-driven methods generate more FL is another method like LiB which recognize
accurate results and are simpler as compared to the unknown parameters of a highly complex
the direct, model-based, and adaptive filter and nonlinear system. The working of FL has
methods. four stages: fuzzification, fuzzy rule base,
inference engine, and defuzzification. To
IV. AI based Estimation Methods evaluate the parameters of the SOC with
i. Genetic Algorithm (GA) improved accuracy FL was also used [31].
GA is an optimization method and used to However, FL requires large memory for
estimate the model parameters of LiB system. storage and time consuming.
Zheng Chen used GA to measure SoH for v. v. Neural Network (NN)
electric and hybrid electric vehicle applications The NN is composed of nodes or neurons
and also developed formula to calculate SoH which is similar to human brains.
[28]. Fundamentally, NN uses a three-layer, input,
ii. Bacterial Foraging Algorithm (BFA) output and hidden layers containing neurons
This is a nature inspired optimization with system specifications. The NN has self-
algorithm. Which is based on Escherichia coli adaptability and self learning abilities. The
bacteria’s social foraging behavior. This is BSA and back propagation neural network
used to find the solution of engineering and (BPNN) can be used to improve the
mathematical problems because of its high performance of NN [32]. The NN can be use to
efficiency and simplicity. To estimate the estimate battery states efficiently.
unknown parameters of the LiB, BFA is also vi. vi. Adaptive Neuro Fuzzy Inference System
utilized. [29]. (ANFIS)
iii. Particle Swarm Optimization (PSO) An ANFIS is better form of the artificial NN,
which is based on the Takagi–Sugeno fuzzy

Published by: The Mattingley Publishing Co., Inc. 15058

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

inference system. The ANFIS has the for LiB state estimation because battery states
advantages of FL and NN in a single not only depend on present state also depend
framework. For modeling, optimization, and previous state. Also, number of inputs to FNN
nonlinear mapping ANFIS is used as an is fixed and cannot be randomly changed,
extraordinary tool. The comparative study which is not easy for battery states estimation
showed the better performance of ANFIS over at any given time moment. This enables the
NN. [33]. RNN to deal with sequential or time series
vii. vii. Support Vector Machine (SVM) problem by remembering, storing, and
The SVM is machine learning algorithm that processing past complex signals. RNNs have
uses both classification or regression been mostly used in time series forecasting,
challenges to change to a linear model from language or speech recognition, machine
nonlinear model. However, the complexity of translation, and system modeling. RNNs has
the SVM system is higher due to its complex short-term memory problem. If a sequence is
quadratic programming. SVM can be used to long enough, they will have a hard time to
estimate states of LiB [34]. capture long-term sequential dependencies
viii. Multivariate Adaptive Regression Splines [36]. To overcome these problems there are
(MARS) commonly-used RNN such as Gated Recurrent
Novel type of flexible regression analysis for Unit (GRU) and Long-Short Term Memory
high dimensional data called as MARS was (LSTM) networks. GRU–RNN and LSTM
suggested by Friedman in 1991. Similar with Networks both are popular ANN or self
data driven approach, no prior knowledge on learning networks have given excellent
the form of the numerical operation is required solution to the problems in various domain
for MARS. The key advantages of MARS lie such as speech recognition, language
in its capacity to capture the intrinsic correction, robotic assistance, etc. GRU-RNN
complicated data mapping in high-dimensional and LSTM networks both are a type of RNN
data patterns and produce simpler, easier-to- which uses separate units in addition to RNN
interpret models, and its ability to perform structure. LSTM units consist of a 'memory
analysis on parameter relative importance. The cell' that can maintain data in memory for long
important advantage of their suggested method period. A set of logical gates are used to
is that it can be implemented using low cost control while information enters the memory,
microcontroller [35]. when it's output, and it's forgotten. This
ix. ix. Recurrent neural network (RNN) architecture helps them to learn longer-term
Current input decides the output of FNN. It is dependencies. GRU-RNNs are similar to
inaccurate to use the FNN process for time LSTMs, but uses a simple structure. Due to use
sequence problems such as battery states of set of gates for control of flow of
estimation. The RNN is a type of an ANN that information they don’t require separate
add additional weights to the neurons to create memory which reduces the count of gates.
feedback loop in the NNs to maintain an GRU-RNN and can measure SoC, SoH and
internal neuron. In traditional feed forward RUL more accurately by experimenting battery
neural network (FNN), all test cases are historical data factors such as voltage, current
considered to be independent whereas in RNN, and at different temperature levels, hence does
present state output depend on input as well as not require information about battery complex
output from previous state which more suitable chemical reactions, internal chemistry, and

Published by: The Mattingley Publishing Co., Inc. 15059

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

model parameters. FPGAs are best suitable to

implement on hardware compared to CPU,
ASIC and GPU. Further betterment can be
done on the GRU-RNN architecture’s by
experimenting or implementing various
activation function used in GRU to improve
accuracy [37].
Table 6 - Summary of the AI Algorithms available for LiBs states estimation.
AI Advantages Limitations Applicable
Algorithms in E-Mobility
GA Robust to noisy environment Time consuming and more
and accuracy level is high. computational complexity. Yes

BFA Gives moderate accuracy for High Latency.

No
dynamic current.
PSO Moderate accuracy in High Latency.
Yes
measuring the SOC online
FL Moderate accuracy in different Requires large data storage,
conditions computational complexity and Yes
Expensive data processing units.
NN Most suitable methods for state Requires large data storage and
estimation in E-Mobility and depend on training of NN.
Yes
works in any condition.

ANFIS Most suitable methods for state Requires large data storage and
estimation in E-Mobility and depend on training of NN.
Yes
works in any condition.

SVM Suitable for non linear model High Latency. Yes

MARS Moderate accuracy Number of experiments are
needed to validate this technique
No
are more. Allowed only for
certain input variables
RNN No prior knowledge of Requires large data storage and
battery’s internal parameters or depend on training of NN.
Yes
chemical process is required. Gradient Vanishing Problem.
Suitable for non linear model.
GRU-RNN/ Prior information of battery’s Reason of gradient decay over
internal parameters or chemical layers is use of sigmoid or tangent
LSTM process is not required and is as an activation function. Yes
suitable for nonlinear model. Accuracy depends on activation
functions.

Based on the above summary, GRU-RNN can To extend this work, battery lifetime
give much better accuracy. The problem of experiments must be conducted to obtain
long term dependencies in the RNN is battery aging data for evaluating accuracy.
overcome and is suitable for nonlinear model. Future work will include prototype hardware

Published by: The Mattingley Publishing Co., Inc. 15060

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

implementation for the BMS by using FPGA supports general purpose programming
based accelerators. languages like C or C++ along with hardware
description languages (HDLs) which resolves
V. FPGA Accelerators the programming concerns and shortens the
FPGAs offer high computing capabilities, re- development time [38-43].
configurability, low power consumption,
concurrency, and customization of architecture VI. Motivation of the research
for particular applications which make it a very The government of India intends a 30%
suitable platform for hardware acceleration. penetration in E-Mobility market and is
FPGAs are being used for implementing AI or encouraging production and use of (Hybrid &)
machine learning algorithms to improve Electric Vehicles. In some industries, the
overall performance. Hardware accelerators government had compelled adoption of E-
such as ASIC, FPGA, and GPU have been Mobility. Use of E-Mobility provides health
used to betterment of performance of AI based benefits by reducing air pollution, noise
applications. AI based applications generally pollution, and water pollution caused by
use GPUs as hardware accelerator for gasoline or oil spills from petrol/diesel
enhancing training and classification process. vehicles. E-mobility is also much safer in a
This is because GPUs support complex possible collision and is less like to roll over.
mathematical operations and provide high E-mobility is powered by renewable energy
memory bandwidth. The problem with GPU sources and reduces a country’s fuel exports.
accelerators is that these are power hungry. [44-45] International Data Corporation
Hence, GPU used in AI applications on cloud predicts spends on AI and Machine Learning
service, large servers, or in E-Mobility will (ML) will grow from to $57.6 billion by 2021.
consume more power. Whereas, FPGA have Revenue from AI-based software will reach
comparatively limited on-chip memory, I/O $105.8 billion by 2025 is predicted by Tractica
bandwidths and computing resources, FPGA [46].
offers better results in terms of low power
consumption, faster data transfer and Tested and reliable datasets played vital role in
flexibility than GPU. The throughput of AI data driven approach for accurate prediction of
application can be enhanced by using ASIC as battery states estimation. Below are a few
hardware accelerator. But, ASICs are one time public testing datasets, refer [47-49] for more
programmable, requires long development details on data sets.
time and high design cost. Present EDA tools
Public Testing Datasets
Prognostic Center of Excellence (PCoE) of NASA
Panasonic 18650PF Samsung 18650-20R
Ames Research Center
Department of Center for Advanced Data repository focuses exclusively on data
Mechanical Engineering Life Cycle Engineering sets that can be used for development of
at McMaster University, at University of prognostic algorithms (prognostic data sets)
Ontario, Canada. Maryland

3. Challenge in BMS and Possible solutions AI based BMSs are in a developing stage. The
hardware required for BMS are very costly and

Published by: The Mattingley Publishing Co., Inc. 15061

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

impractical. Whereas the testing of BMS processing of data in real-time is a limitation.

should be ideally performed in real operating The concern related to real-time processing
conditions of the roads, most of the research is can be minimized with the advancement of the
theoretical and conducted in laboratories. The computational tools for implementing AI based
limitation is that AI based BMS do not take predictive algorithms in BMS for E-mobility.
into account effect of factors like vibration, The outcomes of this review will contribute to
temperature variation, snow, rain, etc. To build the future of E-Mobility as engineers and
end-user trust, the difference between real experts develop an efficient SoC, SoH and
operating conditions and laboratory tests RUL estimation method/algorithm. The AI
should be viewed by next research. based data driven proposed approach should be
Researchers must direct efforts to design an chosen over other existing BMS because of
efficient predictive hardware prototype low cost, power consumption, high speed,
algorithm to improve speed, power, and reprogrammable logic, and high on-chip
accuracy of the BMS. memory storage. The battery states
determination algorithms, developed with the
VII. Conclusion help of data driven approach, are highly
The key take-away from this review is data accurate and are a strong basis for practical
driven estimation techniques suitable for implementation. It will be fascinating to
different battery types, and accurate estimation experience the impact of this technology
of SoC, SoH and RUL. On the other hand, the applied in automobile, healthcare,
availability of testing datasets and robust transportation, and robotics industry.

VIII. References classifier observer. Energy Procedia 2017, 105,

[1] Hannan MA, Lipu MSH, Hussain A, Mohamed 2071–2076.
A. A review of lithium-ion battery state of [7] Cheng, K.W.E.; Divakar, B.P.; Wu, H.; Ding,
charge estimation and management system in K.; Ho, H.F. Battery-management system
electric vehicle applications: challenges and (BMS) and SOC development for electric
recommendations. Renew Sust. Energ Rev vehicles. IEEE Trans. Veh. Technol. 2011, 60,
2017; 78:834–54. 76–88.
[2] Chen, J.; Ouyang, Q.; Xu, C.; Su, H. “Neural [8] Cai, Y.;Wang, Q.; Qi,W. D-ukf based state of
network-based state of charge observer design health estimation for 18650 type lithium
for lithium-ion batteries”, IEEE Trans. Control battery. In Proceedings of the 2016 IEEE
Syst. Technol. 2017, PP, 1–9. International Conference onMechatronics and
[3] Neeta Khare, Rekha Govil, and Surendra Automation, Harbin, China, 7–10 August
Kumar Mittal, “A Process of determining State 2016, pp. 754–758.
of Charge and State of Health of a Battery”, [9] Chaoui, H.; Gualous, H. Adaptive state of
Indian Patent, 813/KOL/2005. charge estimation of lithium-ion batteries with
[4] Waag, W.; Fleischer, C.; Sauer, D.U. Critical parameter and thermal uncertainties. IEEE
review of the methods for monitoring of Trans. Control Syst. Technol. 2017, 25, 752–
lithium-ion batteries in electric and hybrid 759.
vehicles. J. Power Sources 2014, 258, 321– [10] Beir, M.D.; Calado, M.d.R.A.; Pombo, J.A.N.;
339. Mariano, S.J.P.S. Balancing management
[5] https://www.cnbc.com/2018/05/30/electric- system for improving li-ion batteries capacity
vehicles-will-grow-from-3-million-to-125- usage and lifespan. In Proceedings of the 2016
million-by-2030-iea.html [accessed on IEEE 16th International Conference on
22/03/2019] Environment and Electrical Engineering
[6] Tang, X.; Liu, B.; Gao, F. State of charge (EEEIC), Florence, Italy, 7–10 June 2016; pp.
estimation of lifepo4 battery based on a gain- 1–6.

Published by: The Mattingley Publishing Co., Inc. 15062

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

[11] Feng, Y.; Cao, Z.; Shen,W.; Yu, X.; Han, F.; [23] Huang, M.; Kumar, M. Electrochemical
Chen, R.;Wu, J. Intelligent battery Model-Based Aging Characterization of
management for electric and hybrid electric Lithium-Ion Battery Cell in Electrified
vehicles: A survey. In Proceedings of the 2016 Vehicles. In Proceedings of the ASME 2018
IEEE International Conference on Industrial Dynamic Systems and Control Conference
Technology (ICIT), Taipei, Taiwan, 14–17 (DSCC 2018), Atlanta, GA, USA, 30
March 2016; pp. 1436–1441. September–3 October 2018; Volume 3, pp. 1–
[12] https://spike.global/battery- 10.
module/14s35p/[Last accessed on 22/03/2019] [24] Bijender Kumar, Neeta Khare ,P.K.
[13] https://www.automotivepower.com/products/b Chaturvedi, “FPGA Design Scheme for Battery
attery-management SoC & SoH Algorithms for Advanced BMS”,
systems.html?gclid=Cj0KCQjwmpb0BRCBA [25] https://www.sciencedirect.com/topics/engineeri
RIsAG7y4za5peLhh6QPI4DECn_v5WAnQ8S ng/battery-management-system [Last accessed
XvZh71L6AdPlre2KZjyt4Chqm9lAaAnF3EA on 22/03/2019]
Lw_wcB[Last accessed on 22/03/2019] [26] https://www.maximintegrated.com/en/design/te
[14] https://www.ionenergy.co/platform/battery- chnicaldocuments/appnotes/6/6638.html?s_kw
management-systems/fs-ct/[Last accessed on cid=AL!8732!3!422714005632!b!!g!!&gclid=
22/03/2019] Cj0KCQjw9tbzBRDVARIsAMBplx9aLoxNnt
[15] https://www.renesas.com/in/en/[Last accessed xvYXLCBnUsksm4IMuls98z8ptrV9Xdm5IX
on 22/03/2019] XYmRGdBhuPIaAuo0EALw_wcB [Last
[16] http://greenecoindia.com/datasheet.html[Last accessed on 22/03/2019]
accessed on 22/03/2019] [27] https://www.grandviewresearch.com/industry-
[17] https://www.orionbms.com/[Last accessed on analysis/battery-management-system-bms-
22/03/2019] market [Last accessed on 22/03/2019]
[18] Wang, Q.; Jiang, B.; Li, B.; Yan, Y. A critical [28] Chen, Z.; Mi, C.C.; Fu, Y.; Xu, J.; Gong, X.
review of thermal management models and Online battery state of health estimation based
solutions of lithium-ion batteries for the on Genetic Algorithm for electric and hybrid
development of pure electric vehicles. Renew. vehicle applications. J. Power Sources 2013,
Sustain. Energy Rev. 2016, 64, 106–128 240, 184–192.
[19] Muhammad Umair Ali , Amad Zafar , Sarvar [29] Ma, Y.; Ru, J.; Yin, M.; Chen, H.; Zheng, W.,
Hussain Nengroo , Sadam Hussain “Electrochemical modeling and parameter
,Muhammad Junaid Alvi and Hee-Je Kim , identification based on bacterial foraging
“Towards a Smarter Battery Management optimization algorithm for lithium-ion
System for Electric Vehicle Applications: A batteries”. J. Appl. Electrochem. 2016, 46,
Critical Review of Lithium-Ion Battery State of 1119–1131.
Charge Estimation”, Energies 2019, 12, 446; [30] Ali, M.U.; Habib, B.; Iqbal, M. Fixed head
doi:10.3390/en12030446 short term hydro thermal scheduling using
[20] P. Shen, M. Ouyang, L. Lu, J. Li, and X. Feng, improved particle swarm optimization.
``The co-estimation of state of charge, state of Nucleus (Islamabad) 2015, 52, 107–114.
health, and state of function for lithium-ion [31] Singh, P.; Vinjamuri, R.; Wang, X.; Reisner,
batteries in electric vehicles,'' IEEE Trans. D. Design and implementation of a fuzzy
Veh. Technol., vol. 67, no. 1, pp. 92-103, Jan. logic-based state of charge meter for li-ion
2018. batteries used in portable defibrillators. J.
[21] Lipu, M.H.; Hannan, M.; Hussain, A.; Hoque, Power Sources 2006, 162, 829–836.
M.; Ker, P.J.; Saad, M.; Ayob, A. A review of [32] Hannan, M.A.; Lipu, M.S.H.; Hussain, A.;
state of health and remaining useful life Saad, M.H.; Ayob, A. Neural network
estimation methods for lithium-ion battery in approach for estimating state of charge of
electric vehicles: Challenges and lithium-ion battery using backtracking search
recommendations. J. Clean. Prod. 2018, 205, algorithm. IEEE Access 2018, 6, 10069–
115–133. 10079.
[22] Juan Pablo Rivera-Barrera,Nicolás Muñoz- [33] Dai, H.; Guo, P.; Wei, X.; Sun, Z.; Wang, J.
Galeano and Henry Omar Sarmiento- Anfis (adaptive neuro-fuzzy inference system)
Maldonado,“SoC Estimation for Lithium-ion based online soc (state of charge) correction
Batteries: Review and Future Challenges” considering cell divergence for the EV (electric

Published by: The Mattingley Publishing Co., Inc. 15063

 May – June 2020
ISSN: 0193-4120 Page No. 15053 – 15064

vehicle) traction batteries. Energy 2015, 80, [44] https://www.weforum.org/agenda/2019/10/how

350–360. -can-india-transition-to-electric-vehicles-heres-
[34] Anton, J.A.; Nieto, P.G.; Viejo, C.B.; Vilan, a-roadmap [accessed on 9th April, 2020]
J.V. Support vector machines used to estimate [45] https://www.ergon.com.au/network/smarter-
the battery state of charge. IEEE Trans. Power energy/electric-vehicles/benefits-of-electric-
Electron. 2013, 28, 5919–5926. vehicles [accessed on 9th April, 2020]
[35] Antón, J.C.Á.; Nieto, P.J.G.; de Cos Juez, F.J.; [46] https://www.capgemini.com/consultingde/wpc
Lasheras, F.S.; Viejo, C.B.; Gutiérrez, N.R. ontent/uploads/sites/32/2017/09/artificialintelli
Battery state-of charge estimator using the gence-report.pdf [accessed on 20th Feb 2020]
mars technique. IEEE Trans. Power Electron. [47] https://www2.deloitte.com/content/dam/Deloitt
2013, 28, 3798–3805. e/global/Images/infographics/technologymedia
[36] Esteva, A.; Kuprel, B.; Novoa, R.A.; Ko, J.; telecommunications/gxdeloitte-tmt-2018-
Swetter, S.M.; Blau, H.M.; Thrun, S. predictions-full-report.pdf [accessed on 20th
Dermatologist-level classification of skin Feb 2020]
cancer with deep neural networks. Nature [48] https://semiengineering.com/using-fpgas-for-
2017, 542, 115–118. ai/ [accessed on 20th Feb 2020]
[37] Zhang, Y.; Xiong, R.; He, H.; Pecht, M.G. [49] https://www.apriorit.com/dev-blog/586-fpgas-
Long short-term memory recurrent neural for-ai [accessed on 20th Feb 2020]
network for remaining useful life prediction of
lithium-ion batteries. IEEE Trans. Veh.
Technol. 2018, 67, 5695–5705.
[38] A. Shawahna, S. M. Sait & A. El-Maleh,
“FPGA-Based Accelerators of Deep Learning
Networks for Learning and Classification: A
Review”, IEEE Access, vol. 7, pp. 7823-7859,
2019
[39] A. Vasudevan, A. Anderson, and D. Gregg,
“Parallel multi channel convolution using
general matrix multiplication,” in Application-
specific Systems, Architectures and Processors
(ASAP), 2017 IEEE 28th International
Conference on. IEEE, 2017, pp. 19–24.
[40] J. Misra and I. Saha, “Artificial neural
networks in hardware: A survey of two decades
of progress,” Neurocomputing, vol. 74, no. 1-3,
pp. 239–255, 2010.
[41] L. Du, Y. Du, Y. Li, J. Su, Y.-C. Kuan, C.-C.
Liu, and M.-C. F. Chang, “A reconfigurable
streaming deep convolutional neural network
accelerator for internet of things,” IEEE
Transactions on Circuits and Systems Regular
Papers, vol. 65, no. 1, pp. 198–208, 2018.
[42] A. Putnam, A. M. Caulfield, E. S. Chung, D.
Chiou, K. Constantinides, J. Demme, H.
Esmaeilzadeh, J. Fowers, G. P. Gopal, J. Gray
et al., “A reconfigurable fabric for accelerating
large-scale datacenter services,” ACM
SIGARCH Computer Architecture News, vol.
42, no. 3, pp. 13–24, 2014. 3, 13
[43] Y. Liang, K. Rupnow, Y. Li, D. Min, M. N.
Do, and D. Chen, “High-level synthesis:
productivity, performance, and software
constraints,” Journal of Electrical and
Computer Engineering, vol. 2012, p. 1, 2012.

Published by: The Mattingley Publishing Co., Inc. 15064

View publication stats