2022 International Conference on Smart Generation Computing, Communication and Networking (SMART GENCON)

Karnataka, India. Dec 23-25, 2022

Artificial Intelligence and IoT based Smart Battery

2022 International Conference on Smart Generation Computing, Communication and Networking (SMART GENCON) | 978-1-6654-5499-5/22/$31.00 ©2022 IEEE | DOI: 10.1109/SMARTGENCON56628.2022.10083768

Management System for Electric Vehicle

Shradha Umathe Rutuja Hiware
EE, Assistant Professor EE, Assistant Professor
G H Raisoni College of Engg G H Raisoni College of Engg
Nagpur, India Nagpur, India
[email protected] [email protected]

Abstract— The battery of an electric vehicle is critically significant advantage. Remember that Li-Ion batteries do not
essential. As a result, thorough battery maintenance is essential have a memory effect, which is vital to keep in mind.
to ensure that it continues to operate properly. The lead-acid Therefore, consumers will not have to completely discharge
battery is taken for this research, which is the most common their batteries before recharging them. The vehicle is
type of battery used in an electric vehicle. The battery must be powered by battery packs that provide the necessary energy.
thoroughly inspected to ensure that they work optimally under The battery pack is made up of battery cells that are
all circumstances. Controlling the operation of battery systems connected in both a serial and parallel fashion, depending on
is critical to ensuring their overall safety, dependability, and the configuration [3]. Using battery management systems
performance. It is necessary to implement a more organized
(BMS), you can verify that batteries are utilized safely and
battery management system to keep track of the battery's
performance. The Internet of Things (IoT) is used to collect
within the upper limit values of their respective battery
and track battery statistics in the cloud, which are then packs. Some BMSs are capable of performing functions
analyzed. The ThingSpeak makes it feasible to collect and An electric vehicle's battery pack is made up of Lithium-
analyze battery-related data in the cloud. When it comes to Ion cells that are connected in series to give the requisite
batteries, the most important thing to consider is their present voltage output. These cells are typically packaged together in
State of Charging (SoC). To predict the SoC, current, voltage, modules to save space. When it comes to lithium-ion
and temperature must all be measured. It is possible to
batteries, they are delicate electrical components that must
anticipate the SoC of the battery by utilizing a deep learning
operate securely within a certain voltage and temperature
model called a Recurrent Neural Network (RNN). The
difference between the anticipated SoC and the actual SoC is
range. The cell will break down if the voltage or temperature
used to determine how well the RNN model performs. of the cell falls below the required limits. The cell may burst
into flames if such values exceed the specified maximum
Keywords— Battery, SoC, Recurrent Neural Network, threshold. -The voltage and temperature specifications for
ThingSpeak, Accuracy. most Lithium-Ion cells are 2.5 volts-4.2 volts and -13 to 60
degrees Celsius. The safety window of a cell is defined as the
I. INTRODUCTION range of values that fall within that range. It is vital for the
As the price of fuel continues to rise, electric cars (EVs) safety of the battery pack that the battery cells remain inside
are becoming popular. Many automotive manufacturers are this safety window. To accomplish this, the Battery
looking for alternative fuel sources to gas as a result of these Management System (BMS) is in charge of doing so.
circumstances. Because electrical energy sources emit less Consequently, it can direct the equipment that charges or
pollution than other forms of energy, they may be beneficial draws current from battery to limit or stop its flow of power,
to the environment. Because they consume less energy, or it can instruct the equipment to stop working entirely. If
electric vehicles can save money and improve the an impending failure is detected on the high voltage power
environment. Lithium-ion batteries are often found in electric rail, the primary relay on the high voltage power rail can also
vehicles, and they have a long lifespan (EVs). When be opened. The technical specifications for a single Li-Ion
compared to lead-acid, it's a drop in the bucket. It provides battery cell and battery pack are listed in Table 1.
more steady power and has a six-to-ten-times longer life
cycle than a lead-acid battery, according to the manufacturer.
Overcharging and performing deep discharges on a lithium- TABLE I. TECHNICAL SPECIFICATION OF LI-ION BATTERY
ion battery can both shorten the battery's life expectancy
Parameter Single Battery Unit
significantly. While EVs have a greater travel range than Cell (Combination of a
conventional vehicles, the size and architecture of their single cell)
batteries restrict their travel range. Now, the safety of
existing battery technology is a critical concern in limiting Rated Capacity 3.2 25.6 Ah
the usage of electric vehicles [1]. Li-Ion batteries, which
have an organic electrolyte and are more stable and Voltage Nominal 3.6 100.8 V
lightweight than NiMH batteries, should be used for practical
Charging 4.2 117.6 V
cell voltages of more than 4 V. Consequently, they can
provide large energy densities while also making use of Cut-off 2.5 70 V
several cells in tandem [2]. The need for portable electricity
has increased in recent years as a result of the reduction in Charging current 1.625 13 A
the number of applications. In other, the battery is half as
heavy and bulky as the gadget it is powering, which is a

978-1-6654-5499-5/22/$31.00 ©2022 IEEE 1

Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
 The battery management system (BMS) also estimates traveling backward, to smooth the data. For the estimation of
the battery's State of Charge (SoC) [11]. The SoC of the the combined SO2 and SO3 concentrations, a multi-scale
battery is a measurement of how much charge it can hold, or, hybrid Kalman filter is used. This comprises the BS-SRCKF
in other words, how much charge it has at any one time. and extended Kalman filters, among other things. It is
integrated into the normal PF to improve its performance by
A mathematical representation of SOC is shown below transferring particles from the prior distribution region to the
ܳ௡ highest likelihood area. The improved cuckoo search method
ܱܵ‫ܥ‬Ψ ൌ ‫ͲͲͲͳ כ‬ is used to do this. The research [7] presents an approach to
ܳ௠
SOC estimate that is based on machine learning-based
optimization. A well-known subclass of machine learning
ܳ௡ Æ Available Battery capacity
algorithms, the exogenous nonlinear autoregressive neural
ܳ௠ Æ Outstanding battery capacity network algorithm (RNARX), is widely used in the building
of time-series and dynamic systems, among other
Several different ways can be used to determine the applications. With the use of the lightning search algorithm
battery's SoC parameter. All of these alternatives have (LSA), it is possible to improve the accuracy of SOC
disadvantages. It will not be necessary to worry about estimates while simultaneously increasing RNARX's
overcharging the battery if the percentage is determined processing capability. Its capacity to accurately estimate
correctly. As a result, the battery may become overcharged SOC under a variety of operating conditions, as
as a result of the restrictions of each device. Alternators, like demonstrated by the results, suggests that the suggested
vehicle batteries, are equipped with a built-in voltage method is both accurate and resilient. A cloud battery
regulator that ensures a constant flow of electricity. If management system has been developed based on the
something goes wrong, there could be a slew of bad Internet of Things and cloud computing technologies.
consequences. Through the use of prototypes, the capabilities and stability
II. LITERATURE SURVEY of the cloud battery management system's hardware and
software may be validated and improved. Based on an
Model filtering and machine learning algorithms were improved Thevenin model, new estimation algorithms for
applied to forecast the SOH and RUL of lithium-ion batteries SOC and SOH were developed and validated. Both
[4]. A combination of the LSTM and the NN architecture is stationary and mobile applications are being used in the field
combined with the Unscented Kalman (UKF) algorithm to and being tested in an experimental setting [10]. The Internet
produce a data model fusion strategy that effectively reduces of Things was used to develop a wireless battery monitoring
noise. Comparisons with other recognized methods system for battery-operated cars, which was described in full
employing four cell cycle tests have indicated that this in this study. It is possible to keep track of the state of your
method has a higher estimation ability than the others. To battery with the assistance of a middleware application
provide a more accurate estimation of lithium-ion battery server. In this project, I designed and developed a hardware
RUL, the DGWO-ELM algorithm has been suggested [5] device for battery monitoring as well as a web-based user
which integrates the ELM, Gray Wolf Optimization (GWO), interface for it. By incorporating an MQTT server-based
and Differential Evolution (DE) algorithms]. The method system, the system may provide the user with real-time
makes use of three weaknesses in the GWO algorithm to information on the state of the battery SoC, including a deep
improve the optimization of ELM's weight and threshold drain level indicator, battery terminal voltage, and estimated
values. The square inaccuracy of this algorithm is 0.43 mileage [13]. This paper describes the various Wireless
percent, as represented by the minimum root of the equation. Battery Management Systems that have been implemented
A forgotten online sequential extreme learning machine was so far, as well as some of the most significant findings that
used in building a new method, and the Hybrid Gray Wolf have been discovered. Bluetooth and Zigbee are rapidly
Optimizer and attention mechanism were combined to replacing CAN communication as the preferred
predict SOC and forecast RUL by Fan et al [6]. The communication technology in the automotive industry,
algorithm was designed to detect SOC. By removing noise owing to their low power consumption [14]. The Battery
from the original data, Variant Mode Decomposition (VMD) Management System (BMS) of the future will be built on
can help to make it more understandable. To train a load cloud and Internet of Things (IoT) technology, which will be
classification neural network, requires 12 US06 driving used in electric vehicles.
cycles; however, different neural networks are used for
idling, charging, and discharging activities. When extra Electric vehicles can now be equipped with the Internet
filtering is conducted using this strategy, there is a 3.8 of Things (IoT), which allows them to keep track of all of
percent average estimation error or a 2.6 percent average their essential parameters, including the amount of electricity
estimation error, respectively [9]. Because the validation is consumed during the charging process. As a result of this
based on pulse discharge testing, it is currently impossible to concern, a smart charging management system has been
determine how well the method will function in real-world installed in the home to assist in reducing energy
situations. Furthermore, machine learning algorithms, consumption [15]. During this case, the Internet of Things
particularly deep neural networks (DNN), have been widely (IoT) plays an important role by sending an alarm message to
used and proposed for the assessment of the state of charge all connected users when the battery is about to run out of
of electric vehicle batteries in addition to the previously charge or is fully charged. Inter-vehicular users can also
described techniques [12]. Researchers have collected battery access a set of wireless networks in a different way than they
parameters using Kalman filters and similar circuit battery can with traditional networks. The client-server architecture
models, and they have used DNNs to predict the state of will not function if the network quality is inadequate or if no
charge (SOC) of the batteries. It was proposed to utilize the precise measurements of node energy consumption have
BS-SRCKF, which is a filter that smooth square root data been made. Because the Internet of Things is an automation

2
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
process, a swapping technique in charging stations is are currently being employed in the most adaptive client
presented by providing all essential statistics via the internet equipment, such as cells and personal computers, because of
only to certain approved users [16]. In this situation, their high energy density per mass. When hearing the phrase
blockchain technology is essential, and it will be bolstered by "Internet of Things," you're referring to the interconnection
the Internet of Things and the Internet of Things devices. The of common things through the use of a computer network.
researcher have created a unified communication system for When commonplace objects are integrated into a self-
processing large amounts of data, but they have not organizing wireless linkage of gadgets, the resulting system
addressed the issue of encrypting the information so that only is referred to as a "smart home." The goods are equipped
authorized users may decrypt it. It is also recommended that with electronic IDs, sensors, and two-dimensional codes,
the engine be switched off rather than swapped out to ensure which allow them to communicate with the wireless network
safe functioning in electric vehicles. A novel state of charge through the interface. Machines and humans can connect
counting approach using coulombs has been developed for through the Internet of Things (IoT), which is a piece of
the Android platform to address this issue. The Raspberry-pi technology developed to facilitate this communication. The
module was used to create a little computer with a massive Internet of Things is distinguished by three characteristics:
amount of storage space. It also makes use of an Internet of its size, its intelligence, and its connectivity to the internet.
Things platform known as node-red to keep track of battery We're connecting the Arduino to the internet with the use of
life and deliver notifications to a specific user at regular an ESP8266 WiFi module. Then check whether the WiFi
intervals, among other things [17]. In addition to making the board was in proper working order before moving forward.
construction of this high-processing device substantially Performing this procedure is necessary to explain how the
simpler, it enables the monitoring of an even greater number ESP8266 programming technique works as well as to check
of parameters. However, this effective technique has only that the chip is functioning properly. The use of an external
been used for the monitoring of battery charging, with no power supply is essential because the ESP8266 WiFi
consideration given to the other advantages of sensor operations consume a large amount of electricity. According
technology. to the datasheet, the ESP8266EX IC requires 140mA to
transfer data, which the Arduino Nano, to everyone's
Following the guidelines, the journal is processed. A surprise, is unable to give. As a result, to ensure stable
comprehensive discussion of electric vehicles and batteries ESP8266 module programming and operation, it is highly
are found in Section 1. Section 2 goes into greater detail suggested that an external breadboard power source be
about the previous work that's been done on the domain. attached to meet the power requirements of the WiFi module.
Section 3 presents a block diagram to show how the concept A sensor network is utilized to collect data on current,
would be implemented. In Section 4, the Artificial voltage, and temperature, and a neural network is used to
Intelligence algorithm employed in this study is briefly anticipate the outcome utilizing cloud data visualization. The
explained, along with the results of the training. Section 5 block diagram of the proposed design is shown in figure 1.
focuses on putting the plan into action. Section 6 details the
results of using data visualization in ThingSpeak,
suggestions from the AI model, and predictions made by the
AI model. As a final note, section 7 summarises the
importance of this work in electric vehicle research.
III. PROPOSED IDEA
According to data from the National Renewable Energy
Laboratory, the mass manufacture of electric vehicles has
only a minor impact on the need for electricity generation.
By 2050, half of all vehicles in the United States will be
electric, necessitating just an 8% increase in electricity
generation and a 4% increase in generating capacity, while
conventional vehicle emissions are cut greatly and fuel
consumption in the transportation sector is decreased.
Transformers' operating temperatures rise as a result of the
added load imposed by charging electric vehicles. Using less Fig. 1. Block diagram of the proposed system
energy means that you will end up spending more money
over time. Carbon footprints of nuclear and renewable IV. RNN
energy sources, such as solar and wind power, are among the The recurrent neural networks have a thread of memory
lowest available. The supply and demand curves must, information flowing through them, it is possible to refer back
however, be brought into alignment, which is extremely to previous memories. Figure 2 depicts the general
challenging. If the charging profile for electric vehicles construction of an RNN. The RNN's input is the previous
(EVs) accumulates during peak demand periods, there will sequence's memory unit, denoted by ht-1, and W is the
be a scarcity of electric power available. concatenation of the weight matrix for that sequence's
memory unit and the weight matrix for the current sequence's
The lithium-ion batteries would be used in electric input, which is defined as (1). The weight matrix, W, of the
vehicles. As a result of their high energy thickness and current sequence's memory unit, as well as the activation
increased force per mass battery unit, Li-Ion batteries are function, are denoted by the numbers (1) and (2),
currently the most widely used invention in electric vehicles, respectively. It is possible to obtain the memory information
and their use has allowed battery producers to create more ht (3) as well as the network output yt at time t with the help
efficient batteries with reduced weight and measures at lower of these.
costs than they ever could before. Lithium-particle batteries

3
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
 sequence will continue to function before causing the
aforementioned concerns to manifest themselves. Since the
capacity of RNNs to estimate SOC is what distinguishes
them as being well-suited for SOC estimation.
The RNN is trained with the data like the current,
voltage, and temperature acquired. The accuracy of the RNN
model in the training phase is depicted in figure 4 with the x-
axis having epochs and the y-axis having training accuracy.
At the 0th epoch, the accuracy is 33%. In the epochs of 10th,
20th, 30th, 40th, and 50th epochs – the percentage of accuracy
is 78%, 85%, 90%, 93%, and 97% respectively. Since the
accuracy is achieving nearly 97%. The RNN model is
finalized because the accuracy obtained in training is really
good. If the accuracy is not up to the requirement, it can be
Fig. 2. Architecture of RNN
solved by changing the hyperparameters like the number of
hidden layers, neurons, epochs, learning rate, etc.
The forward propagation procedure, as illustrated in
RNN (1)–(4), demonstrates how the RNN transfers memory
information to the network. RNNs save a large amount of
time when compared to feed-forward neural networks, owing
to their basic structure and weight-sharing mechanism, which
makes them more efficient. According to Benigo, when an
RNN examines a long series of inputs, the information
forgetting, and gradient explosion problems are typically
seen. Bengio concluded that serial multiplication of shared
resources is to blame for the problems he's been
experiencing.

ܹ ൌ ሾܹ௫௛ ǡ ܹ௛௛ ሿ (1)

݁ ௫ െ ݁ ି௫ (2)
‫ ݄݊ܽݐ‬ൌ
݁ ௫ ൅ ݁ ି௫

݄௧ ൌ –ƒŠሾܹǤ ሾ‫ݔ‬௧ ǡ ݄௧ିଵ ሿ ൅ ܾሿ (3)

Fig. 4. Accuracy of RNN model in training phase Implementation
‫ݕ‬௧ ൌ ܹ௛௬ Ǥ ݄௧ ൅ ܾ (4)
V. IMPLEMENTATION OF THE RESEARCH WORK
The working flow of the proposed system is depicted and
As input data, an RNN is utilized to estimate battery explained in figure 5. In this proposed system, three
SOC, and its performance is testes using the live data from parameters like current, voltage, and temperature are been
battery of electric vehicle. Voltage, current, temperature, and taken. To detect those three parameters, respective three
the state of charge (SOC) of the battery are all examples of sensors such as LTS25-NP Current sensor, LV25P Voltage
network inputs. The weight Whh is increased in the repeated sensor, and LM35 Temperature sensor is been incorporated
sequence back-propagation method which is shown in figure with the lithium battery. The working of the used sensor is
3. In actuality, it's unknown how long the current input explained in table 2.
The Internet of Things (IoT) is a cutting-edge technology
that has gained significant acceptance in recent years. Having
a battery management system that is integrated with the
Internet of Things (IoT) provides for on-the-go access to
battery data through a smartphone, or for battery data to be
saved in the Cloud and accessed anytime it is needed. Each
time a battery is charged or discharged, all battery-relevant
data is transferred in real-time to the cloud, where it is
analyzed using machine learning and artificial intelligence
algorithms. When a failure or flaw in the battery is
discovered, the driver or service provider will be notified
immediately. More batteries will be salvageable as a result,
preventing them from being damaged or destroyed for good.
When it comes to monitoring and managing devices, the
Internet of Things (IoT) has become increasingly significant.
It also allows for remote data tracking for battery metrics ,
weather conditions, and a variety of other variables. With the
Fig. 3. Architecture of RNN with multiple sequences Internet of Things, which has been described as a

4
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
 TABLE II. SENSOR AND ITS WORKING network of sensors, software, and other technologies
Sensor Working embedded in physical objects (things), devices can
Current sensor The LTS25-NP is used for measuring the current communicate with one another and exchange data with other
quantity of the Li-Ion battery. The exceptional devices and systems over the internet. It is predicted that by
performance of this hall effect sensor, despite its 2020, there will be 25 billion internet-connected devices,
expensive cost, is balanced by its ease of installation.
according to the well-known research and consulting firm
On each side of the sensor, there are three pins, and
there are an additional three pins on the front of the Gartner [8]. These read three parameters are given as input
sensor. The three pins on the left and right sides of into the Arduino ESP8266, which is further sent to the cloud.
the board should be linked in series. The emission is In the cloud, it is visualized as data and this data is being
emitted from the very top. Three pins and an arrow analyzed. After analysis, one of the deep learning models
sign are prominently shown on the face of the card. It called the RNN is been employed. Using this RNN, the SOC
indicates that the current is flowing in the direction
indicated by the arrow when you see an arrow sign. % (State of Charge) will be predicted. Besides, suggestions
An external 5 volt supply was required on the front will also be provided to the customer.
side of this IC to correctly read the current readings.
LV25P Voltage To obtain an accurate reading of the voltage, use a VI. RESULT AND DISCUSSION
sensor voltage sensor namely LV25P. This sensor not only
The hardware setup is done as per the discussion made in
provides insulation but also helps to reduce the
effects of temperature changes on the surrounding section 3. Next, the physical parameter of the battery is
environment. The input voltage affects the output acquired and sent to the cloud for further process. The
current of this sensor, which is capable of monitoring temperature of the battery is monitored lively and it is shown
both alternating current and direct current voltages in figure 6. During testing, the output of the temperature
and demands symmetrical feeding to function achieved is 33.8-degree Celsius. If the battery temperature
properly. The voltage of this sensor can be monitored
by connecting any analog microcontroller input to the
level is exceeded the threshold Celsius, then the suggestion
voltage at the resistor terminals. in the output will be popping out with a warning as “Battery
LM35 When the temperature varies in the Li-Ion battery, the
overheated”. With the suggestion received, the user shall
Temperature LM35 is used to measure this quantity. The LM35 take measures accordingly and it is depicted in figure 7. This
sensor linear temperature sensor from National popping of warning messages or alerts is due to the backend
Semiconductor is a three-terminal linear temperature working of the RNN model as shown in figure 5.
sensor. Precision temperature measurement is
possible between -55 and +150 degrees Celsius. The
output voltage of the LM35 increases by 10mV as the
temperature of the circuit rises. When powered by a
5V source, the LM35 has a standby current of less
than 60uA, which is very low. In an LM35 sensor,
the voltage across a diode increases at a predictable
rate as the temperature rises above a certain
threshold. By appropriately magnifying the voltage
change, it is straightforward to generate an analog
signal that is precisely proportional to temperature in
real-time.

Fig. 6. Live monitoring of battery temperature

Fig. 5. Workflow of the proposed system

Fig. 7. Warning message by RNN model

5
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
 Next, the other two important parameters are visualized
in the ThingSpeak. The voltage and current of the battery are
shown in Figures 8 and 9. In both graphs, the time is taken in
the x-axis and the parameter is taken in the y-axis. The
voltage is measured in volt and the current is measured in
ampere. In figure 8, the voltage increases gradually and
reaches the constant after the time 20:24:00. From figure 9 it
is clear that the current increases gradually and maintains the
constant after 20:24:00 seconds. This graph helps the user to
visualize the battery parameter and aids them to understand
the condition of the battery.

Fig. 10. SOC prediction in percentage using RNN

VII. CONCLUSION
This article detailed the design and implementation of
Artificial Intelligence and IoT-based smart battery
monitoring systems for an electric vehicle. The battery
parameters like voltage, current, and temperature are
collected using the appropriate sensor fixed at the battery.
With the advancement in technology, the above-mentioned
parameters are made available on the internet. This helps the
user to analyze the voltage and current used by their vehicle
over one week or month. The SoC is much important for
maintaining battery health. For predicting the SoC the RNN
model is employed. The model is finalized when the
accuracy reaches 97% in training. Then the RNN model is
used to predict the SoC of the battery based on the live
voltage, current, and temperature in the cloud. The proposed
Fig. 8. Live monitoring of battery voltage system also gives an alert to the user if the temperature
becomes high than the threshold. The proposed idea helps
the electric vehicle owner to analyze their battery
performance and also aids them to take precautionary steps
to maintain the battery health and increase the life cycle of
the battery.
REFERENCE
[1]. S. Yonghua, Y. Yuexi, H. Zechun, “Present Status and
Development Trend of Batteries for Electric Vehicles,”
Power System Technology, Vol. 35, No. 4, pp. 1-7, 2011.
[2]. Chang, W.Y, “State of charge estimation for LiFePO4
battery using artificial neural network”. Int. Rev. Electr. Eng.
7, 5874–5800, 2012.
[3]. Yong Tian, Dong Li, Jindong Tian, “An optimal nonlinear
observer for state-of-charge estimation of lithium-ion
batteries”, Industrial Electronics and Applications (ICIEA),
12th IEEE Conference. 2017.
[4]. Cui, X., and Hu, T, “State of Health Diagnosis and
Remaining Useful Life Prediction for Lithium-Ion Battery
Based on Data Model Fusion Method”. IEEE Access 8,
207298–207307, 2020.
Fig. 9. Live monitoring of battery current [5]. Zhu, J., Tan, T., Wu, L., and Yuan, H. “RUL Prediction of
Lithium-Ion Battery Based on Improved DGWO-ELM
Method in a Random Discharge Rates Environment”. IEEE
Eventually, the SoC% (State of Charge) is predicted Access 7, 125176–125187, 2019.
using RNN. The predicted SoC is compared with the actual [6]. Fan, J., Fan, J., Liu, F., Qu, J., and Li, R, “A Novel Machine
SoC to analyze how the model is performed and it is shown Learning Method Based Approach for Li-Ion Battery
in figure 10. The graph is plotted by keeping the x-axis as Prognostic and Health Management”. IEEE Access 7,
time in terms of seconds. While the y- axis is about SOC in 160043–160061, 2019.
percentage. From the figure, it is clear that the designed [7]. M. A. Hannan et.al, “Toward Enhanced State of Charge
RNN model gives the SoC% which is nearly equal to the Estimation of Lithium-ion Batteries Using Optimized
original value of the Li-Ion battery. Machine Learning Techniques” Scientific
Reports volume 10, Article number: 4687. 2020
[8]. W. Iqbal, H. Abbas, M. Daneshmand, B. Rauf and Y. A.
Bangash, "An In-Depth Analysis of IoT Security
Requirements, Challenges, and Their Countermeasures via

6
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.
 Software-Defined Security," in IEEE Internet of Things Supercapacitor Energy Storage System”, International
Journal, vol. 7, no. 10, pp. 10250-10276, Oct. 2020. Journal of Recent Innovation in Engineering and Research,
[9]. S. Tong, J.H. Lacap, J.W. Park, “Battery state of charge e- ISSN: 2456 – 2084,2017.
estimation using a load classifying neural network”, Journal [21]. N. R. Dhawas, S. K. Umathе,” Performance of STATCOM
of Energy Storage, 236–243, 2016. Incorporating Supеr capacitor”, International Journal for
[10]. Divya D Shende, Prof.Prashant Jagtap , Prof.Rutuja Hiware, Research in Applied Science & Engineering Technology
“Review On Power Quality Improvement Using Unified (IJRASET), Volume 5 Issue IV, April 2017, IC Value: 45.98
Power Flow Controller System In Electrical Network,” ISSN: 2321-9653.
International Journal for Innovative Engineering and [22]. S. Haldar, S. Mondal, A. Mondal, and R. Banerjee, "Battery
Management Research, Vol 10 Issue02, Feb2021 ,p.no36- Management System Using State of Charge Estimation: An
41,10.48047/IJIEMR/V10/I02/09. IOT Based Approach," National Conference on Emerging
[11]. Weihan Li, Monika Rentemeister, Julia Badeda, Dominik Trends on Sustainable Technology and Engineering
Jöst, Dominik Schulte, Dirk Uwe Sauer, “Digital twin for Applications, pp. 1-5, 2020.
battery systems: Cloud battery management system with [23]. V. K, R. K. Nema and A. Ojha, "Various Types of Wireless
online state-of-charge and state-ofhealth estimation,Journal Battery Management System in Ev," IEEE International
of Energy Storage”,Volume 30,2020. Students' Conference on Electrical, Electronics and
[12]. Shridhika Malode, Rutuja Hiware, “Application of Inverter Computer Science (SCEECS), pp. 1-5, 2020.
less Flywheel for Energy Storage System,” - International [24]. Shradha Umathe, Rutuja Hiware, Ankur Gupta, “A study of
Journal for Scientific Research & Development| Vol. 4, Issue automatic mains failure control panel (AMFC), “AIP
02, 2016 | ISSN (online): 2321-0613. Conference Proceedings,vol no.2417,Issue 1,2021,pages
[13]. Ankit Ghate , Rutuja Hiware, “Solar Tracking Development 030001.
using Arduino along with an Efficient Boost Converter and [25]. Sousa, R.A.; Melendez, A.A.N.; Monteiro, V.; Afonso, J.L.;
Charge Controller,” International Journal of Electrical Ferreira, J.C.; Afonso, J.A, “Development of an IoT system
Engineering,ISSN 0974-2158 Volume 8, Number 1 (2015), with smart charging current control for electric vehicles”. In
pp. 57-63. Proceedings of the IECON 2018-44th Annual Conference of
[14]. Shridhika Malode ,R.S.Hiware, “Application Of Flywheel the IEEE Industrial Electronics Society, Washington, DC,
For Energy Storage System,” International Journal Of USA, 21–23, pp. 4662–4667, October 2018.
Engineering Research In Electrical And Electronic [26]. Divya Shende, Prashant Jagtap, Rutuja Hiware, “Enhanced
Engineering,Vol No.03,Issue.03,March 2016. power quality using unified power flow controller systems,”
[15]. R S. Hiware, “Speed Control of DC Motor Using Chopper”, Journal of Physics: Conference Series,Nov 2021, Vol no.
International Journal of Science and Research (IJSR), 2017, 2089 (2021),Issue 1,pages: 012035, doi:10.1088/1742-
vol 6, page no.1751-53. 6596/2089/1/012035.
[16]. Bergveld, H. J., Kruijt, W. S., & Notten, P.H.L, “Battery [27]. Gao, D.; Zhang, Y.; Li, X, “The internet of things for electric
management systems”, Netherlands: Springer, pp. 9–30, vehicles: Wide area charging-swap information perception,
2002. transmission, and application”. Adv. Mater. Res, 608, 1560–
1565, 2013.
[17]. E. Chemali, P.J. Kollmeyer, M. Preindl, A. Emadi, “State-of-
charge estimation of Li-Ion batteries using deep neural [28]. Asaad, M.; Ahmad, F.; Alam, M.S.; Rafat, Y., “IoT enabled
networks: a machine learning approach”. J. Power Electric Vehicle’s Battery Monitoring System”. EAI SGIOT,
Sources, 400, 242–255, 2018. 8, 2017.
[18]. Meenakshi S. Bawankar, S.K.Umathe, Dr.S.G. Tarnekar,” [29]. Rahul A. Jagtap, Dr. P.M. Daigavane , Dr.S.G. Tarnekar, “Design of
Ultra-capacitor Application in Electric Vehicles for Intelligent Charger for Electric Vehicles,” International Research
Regenerative Braking of DC Motor”, IJSRD - International Journal of Engineering and Technology (IRJET), Volume: 03 Issue:
Journal for Scientific Research & Development| Vol. 4, Issue 03 | Mar-2016.
03, 2016 | ISSN (online): 2321-0613. [30]. S Raut, P Daigavane, MB Shaikh, “Test Bench of
[19]. Priyanka Kubade, S. K. Umathe, Dr. D. R .Tutakne,” Automotive Component of an Electric Vehicle for Electrical
Regenerative Braking in an Elevator Using Super capacitor”, Parameter Measurement,” Innovations in Electrical and
International Research Journal of Advanced Engineering and Electronic Engineering, 479-488
Science ISSN (Online): 2455-9024,2017.
[20]. Namrata R. Dhawas and Shradha .K. Umathе,” Improving
Power System Performsance With Integrated Statcom And

7
Authorized licensed use limited to: AMITY University. Downloaded on March 04,2024 at 18:56:33 UTC from IEEE Xplore. Restrictions apply.