Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The automobile industry is currently undergoing a paradigm change from conventional, diesel, and
Received 7 January 2023 gasoline-powered vehicles to hybrid and electric vehicles of the second generation. Lithium-ion (Li-
Revised 14 February 2023 ion) batteries have sparked the automotive industry’s interest for quite some time. One of the most cru-
Accepted 23 February 2023
cial components of an electric car is the battery management system (BMS). Since the battery pack is an
Available online xxxx
electric vehicle’s most significant and expensive component, it must be carefully monitored and con-
trolled. The precise measurement and calculation of the many states of a Li-ion battery’s cells, such as
Keywords:
the State of Health (SOH) and State of Charge (SOC) is a difficult procedure as they cannot be monitored
Artificial Intelligence
Battery Management System
directly. This paper examines various methodologies and approaches for estimating the SOC and SOH of
Lithium-ion Batteries Li-ion batteries using Artificial Intelligent methods. Six machine learning algorithms are intensively uti-
Neural Network lized to investigate the Li-ion battery state estimation. The employed methods are linear, random forest,
State of Charge gradient boost, light gradient boosting (light-GBM), extreme gradient boosting (XGB), and support vector
State of Health machine (SVM) regressors. In comparison to all other models employed in this study, the discharge pre-
diction made using random forest exhibits significantly greater performance at a low loss of accuracy. For
instance, with the highest R2-score of 0.999, the random forest regressor achieves only 0.0035, 0.0013,
and 0.0097 for mean and median absolute error, and root means squared error (RMSE), respectively.
We showed that the state estimation of Li-ion batteries can be precisely predicted using AI methods,
which can be combined with a battery management system to improve electric vehicle performance.
2023 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams Uni-
versity. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/
by-nc-nd/4.0/).
https://doi.org/10.1016/j.asej.2023.102213
2090-4479/ 2023 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Please cite this article as: Y. Khawaja, N. Shankar, I. Qiqieh et al., Battery management solutions for li-ion batteries based on artificial intelligence, Ain
Shams Engineering Journal, https://doi.org/10.1016/j.asej.2023.102213
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
BMS and their merits and demerits. Section 4 gives a detailed anal- v ðsÞ is the battery voltage in the Laplace domain [V].
ysis of the different AI approaches used with their merits and iðsÞis the battery current in the Laplace domain [A].
demerits followed by Section 5 which details the results and anal-
ysis for a sample battery dataset. Section 6 gives a comparative
analysis of the different AI techniques used in Section 5. 3. AI strategies in battery management systems
2. Lithium-Ion battery model A Battery Management System (BMS) is a software and hard-
ware system that regulates the battery for effective functioning
Currently, battery modeling for SOC determination is routinely [23]. A BMS is made up of various functional units, such as a cell
created using a variety of equivalent circuit (RC network) models, voltage balance, fuel gauge monitor, cut-off field effect transistor,
each with its own set of material properties and accuracy criteria. a cell voltage monitor, a state machine, temperature monitors,
The generic model, on the other hand, is based on the assumption and a real-time clock [24]. There are several varieties of BMS-
that the internal resistance remains constant during charge and integrated chips on the market. The functional pieces are organized
discharge cycles. Therefore, the correctness of this model is debat- differently for different systems; they might range from a simple
able. The battery deterioration model based on capacity fading was analog front end with a microcontroller capable of balancing and
simulated and created while taking the SOH estimate into account. monitoring to a stand-alone fully integrated solution capable of
These model parameters were mostly determined by the physical running autonomously.
properties of the individual anode and cathode. However, in a The BMS in EVs may incorporate a variety of actuators, controls,
dynamic setting, external factors such as ambient temperature and sensors. BMSs are responsible for safeguarding batteries, oper-
and discharge current load will cause these stationary models to ating batteries within acceptable parameters of voltage, current,
be erroneous. and temperature, and accurately monitoring battery parameters
In this investigation, a fourth-order electrochemical model was [25]. In terms of hardware structure, three basic types of topologies
used due to its ability to correctly record the battery’s complicated have been used: modular architectures, centralized, and dis-
terminal voltage behavior. As indicated in Fig. 2, ten characteristics tributed. Richter and Meissner presented a layer structure for mon-
must be determined, including the open circuit voltage source itoring and managing the status of a battery [26]. According to
(Eoc), series resistance (R0), and the resistance and capacitance of Gold, BMSs can be classified based on their different functionalities
each RC network (Rm, Cm), where m = 1:4. [27]. These concepts could be used to create a broad framework
At time t, the terminal voltage (Vterm) may be computed as with basic functionality. Various sensors located within the battery
follows: pack collect data at the monitoring layer [28]. The BMS is linked to
all of the battery-pack components as well as the vehicle control
V term ðnÞ ¼ Eoc ðSOC n Þ iðt ÞR0 i1 ðtÞR1 i2 ðtÞR2 i3 ðtÞR3 i4 ðtÞR4 processor BMSs have a long history of using safety [28]. However,
ð1Þ the suggested BMS uses more sensors than the existing safety cir-
cuitry designs, allowing for advancements like the addition of pre-
where iðt Þ denotes the applied current and im ðtÞ (where m = 1:4) cise warnings and controls to avoid overcharge, over-discharge,
denotes the currents flowing through the mth RC branches, respec- and overheating. To monitor and measure battery characteristics
tively. The voltage varies as a function of current change at the such as cell voltage, current, and temperature of a battery, a system
SOC-defined operating point, and the voltage Eoc for the electro- of sensors is required. However, the viability of these measures
chemical model in Fig. 2 is determined by a fourth-order transfer outside of laboratory contexts is hampered by both space limita-
function. tions and device expense. Therefore, it is important to measure
current, temperature, and voltage to enhance the capabilities of
b0 þ b1 s1 þ b2 s2 þ b3 s3 þ b4 s4
BðsÞ ¼ K ; ð2Þ state tracking in practical applications. Modeling, data collection
1 þ þa1 s1 þ a2 s2 þ a3 s3 þ a4 s4
(via data acquisition system DAQ), and data storage are essential
where: components for obtaining the desired data.
The charge–discharge procedure is controlled by a subsystem
K is the gain of the transfer function called the charge controller. For balancing cells or monitoring
bi " i = 0,. . .4 – numerator parameters of the transfer function internal resistance, a variable resistor could be required. Cell bal-
ai " i = 0,. . .4 – denominator parameters of the transfer function ancing management is still an important design aspect that can
BðsÞ is the battery transfer function. be improved to equalize the cells of the battery pack and accu-
rately determine the battery’s state. Since most BMS subsystems
In the Laplace domain, the battery voltage is given by the are standalone modules, internal data transmission is necessary.
expression: The main method of data transfer inside the BMS is communica-
tion over a Controller Area Network (CAN) bus. More information
v ðsÞ ¼ Eoc þ BðsÞiðsÞ; ð3Þ can be gathered by building intelligent batteries such that micro-
where: chips embedded in the battery can communicate with the users
and the chargers. Furthermore, in charging systems, radio, and
communication technologies are gradually being implemented
that facilitate communication between the battery and charger.
As temperature variations affect cell imbalance, reliability, and
C1 C2 C3 C4 performance, a thermal management module is essential. It is
R0 therefore crucial to reduce the temperature disparity between
cells, which must be regulated and run under the right tempera-
Eoc Vt ture circumstances.
R1 R2 R3 R4
The BMS software is the brains of the entire system since it
manages all equipment functions and evaluates sensor data for
decision-making and status predictions. The software of a BMS
Fig. 2. Equivalent Circuit Model of the battery. should be able to handle control switching, sample rate tracking
3
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
in the sensor module, cell balance management, and even the con- protecting the battery. The estimates of SOH and SOC, particularly
struction of dynamic safety circuits. In addition, for continuous in Li-ion batteries, influence the instructions provided to the bat-
updates and control of battery functions, web-based data analysis tery [32]. Fig. 4. Below shows the different AI techniques that can
and processing are required. Robust and reliable automated data be used for state estimation.
analysis is vital for success since it determines state estimation As a result, these parameters are in control of preventing over-
and problem detection. The user will see this information via a charge or over-discharge scenarios that might cause the batteries
user-friendly interface with the necessary suggestions. The calcula- to explode. The usage of neural networks and fuzzy logic is the
tion of SOH and SOC will be a part of a capability evaluation that solution to these challenges [33]. This section investigates the
also displays the battery’s life status and establishes operational machine learning algorithms in BMS applications and compares
restrictions using cutting-edge techniques like fuzzy systems, arti- them. Moreover, a robust literature review of the most recent
ficial neural networks, state-space-based models, etc. research in the literature regarding state estimating is illustrated
Cell balancing is performed to enhance the performance of the in Table 2. Table 3.
battery without overcharging and over-discharging, its purpose is
to bring cells with similar SOC levels together. The controller will
manage the charging procedure using a detailed program based on 3.1. Artificial neural network (ANN)
the level of charge retained by each cell. Accordingly, a key compo-
nent of enhancing the balancing is an accurate SOC calculation of The Artificial Neural Network (ANN) is made up of nodes or
every cell. To deliver battery fault warnings and identify out-of- neurons, which are analogous to human brains. Fundamentally,
tolerance conditions, smart data evaluation is required. Historical neural network (NN) employs a multiple-layer structure with
data will be kept and will show the pre-alarm situation before input, output, and hidden layers comprising neurons with system
potential issues. The interface must provide users with access to requirements [34]. The NN is capable of self-adaptation and self-
the key BMS data. Depending on the SOC of the battery, the remain- learning. A backtracking search algorithm (BSA) and back propaga-
ing range should be displayed on the dashboard. The consumers tion neural network (BPNN) can be utilized to increase NN perfor-
must also be informed about the prediction and estimation of the mance [35]. To train the network, a back-propagation technique
said battery through alarms and replacement recommendations. based on experimental data is utilized. Several inputs are taken
According to Fig. 3, a BMS’s functionality may be categorized as from previously recorded data and applied to the input layer, each
follows [29]: of which is multiplied by a weight before the final product is
added. The summation is then passed via a sigmoid function, com-
1. Protection: against extreme temperatures, over-discharge, monly known as the activation function. The algorithm changes
overcurrent, overcharge, and short circuits. the network weights to minimize the MSE in the network’s out-
2. High-voltage Control and Sensing: measure temperature, volt- come. The NN can be used to effectively estimate battery states.
age, current, temperature, thermal management, control, con- The hybrid pulse power characterization (HPPC) test is used to
tactor, pre-charge, and ground-fault detection. identify the battery parameters of the first-order equivalent circuit
3. Diagnostics: state-of-life (SOL) estimation, SOH estimation, and model. Then, it can be used to train the multiple-layer back prop-
abuse detection. agation neural network, allowing the structure and parameters of
4. Performance management: power-limit computation, balance/ the network including weights and thresholds to be determined.
equalize cells, and SOC estimation. SOH is also estimated using this well-trained neural network as
5. Interface: data recording, reporting, communications, and range well [36]. Choi et al. [37] implement a multilayer perceptron
estimation. (MLP) model that employs a number of nonlinear parameters.
For model design, 300Li-ion battery data are utilized, and the
There are several traditional ways, but AI-based strategies are model is validated by categorizing the data into training and test
gaining popularity for improving dynamic responsiveness and data. The information on the capacity predicted by the MLP model
dependability. BMSs are currently installed with a fixed control was utilized for SOC estimation, and the accuracy of the SOC esti-
algorithm, without adjustments or modifications during the sys- mate was evaluated by comparing it to the SOC measurement
tem’s lifetime. [30]. However, the batteries have a relatively long value.
life span, with many warranties ranging from 5 to 10 years. The
operational environment may vary during the system’s lifespan
due to such a long period. These changes may need adjustments
to the control technique to improve performance characteristics. Artificial Neural
The significance of SOH and SOC for the BMS extends beyond the Network
Artificial
Intelligence
Techniques
Evolutionary
Expert Systems
Computing
Support Vector
Machine
4
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Table 2
Extensive Review of the Diverse AI Technologies Employed in Battery Management Systems.
The functionality of the AI technique AI Outcomes from the Technique Employed Challenges Ref.
Technique
employed
Advanced battery management system HIS Cheap non-recurring engineering costs. Parameters selection and data [29]
ANN Low power consumption and improved collection.
FLS performance.
High operation speed and huge data storage
capacity.
Large reconfigurable logic.
Estimation Model for Lithium-Ion Batteries EA Increased precision and efficiency. It is difficult to track the life of a [31]
ANN Dynamic cycle.
Cheap Can only be used on new cells
Global estimation performance that is It is dependent on the accuracy of
optimal the current sensor and can
accumulate errors.
Electric vehicle Li-ion battery SOC Estimations ANN Non-linearity is exploited. Temperature variations affect [39]
Close results are obtained. training data.
System-based simulation environment for designing and HIS Optimization of the BMS algorithms in The cells in a battery pack do not [64]
optimizing the BMS, which includes a lithium-ion various test conditions. have the same capacitance or
battery model and CAN communication interface. System simulations run quickly. nominal voltage.
Using ANFIS MATLAB for SOC Estimation AI Easier to implement. For a similar battery, training data [65]
ANN Accurate SOC value. is required.
FLS
Enhancing the Performance of Electric Vehicle AI Lowering emissions. Environment of packaging. [66]
The cost is negligible. Choosing a Battery Technology.
Infrastructure recharging and safety.
State of charge estimation ANN ANN can estimate the state of charge The initial cost of using battery [67]
accurately. systems.
ANN can self-learn battery dynamics. Expensive elements.
SOC and SOH estimation HIS Improved health. A large number of cycles must be [68]
ANN Higher accuracy. trained to estimate battery
FLS Non-linearity exploitation. availability and energy.
Data globalization.
Analysis of the cost of utilizing battery electric vehicle HIS Advantageous to battery life. V2G is reliant on driving [69]
(BEV) batteries as energy storage in power grids Most of the cost of V2G battery wear would efficiency.
be attributable to DoD and cycle.
Real-Time Fault Diagnosis Capability for Electric Vehicles HIS The users’ safety and enhanced lifespan of Cell aging and recycling. [70]
the battery cells.
State of charge estimation ANN Online real-time implementation. One must compromise between [71]
Accurate SOC values. accuracy and computing time to
Elimination of the relative SOC prediction suit each specific application.
error and voltage measurement noise.
Surface Charge Estimation Using an Electrochemical HIS The bulk and interface solid concentrations A relatively narrow range of [72]
Model-Based Extended Kalman Filter are estimated, allowing the critical surface charge states.
charge and SOC to be calculated.
High efficacy.
Modeling and estimation of SOC of Li-Ion batteries using ANN Rapid convergence and good accuracy. The current and voltage of each [73]
neural networks and the extended Kalman filter (EKF) Could be applied to a wide range of cell must be measured.
rechargeable batteries. When the battery ages, the
trained NN may no longer produce
acceptable results.
Design of a Nonlinear Autoregressive with External Input ANN Mean squared error of less than 1e-6 Restricted range The vehicle (and [74]
(NARX) artificial neural network to estimate the SOC of The proposed model’s precision is very its traction battery)
an EV using real data extracted from the car significant. is not in a controlled environment.
Employs sophisticated machine-learning
tools.
A probabilistic neural network (PNN) is used to estimate ANN Simple to use. When it comes to classifying new [75]
the SOH of Li-ion batteries. Extremely fast for medium-sized databases. cases, multilayer perceptron
This method can improve the estimation of networks outperform them.
the battery’s SOC. PNNs necessitate more memory
space to store the model.
SOC estimation using Back propagation Neural Network HIS Real-time implementation [76]
with search-based optimization algorithm ANN Less Complexity. necessitates the use of a high-end
Model-free strategy. processor.
More precise. The implementation of the coding
Temperature sensitivity is included. requires the use of skilled
engineers.
3.2. Neuro-Fuzzy logic systems (FLS) adaptive battery control has relied solely on neural network
models or fuzzy logic. However, in recent years, research has
Fuzzy logic controllers (FLSs) in general are insufficient for begun to tilt toward combining these two domains. The ability
estimating the SOC and SOH of a battery pack. Only one research to counteract the limits of one with the constraints of the other
team used a fuzzy approach to determine the SOC and SOH of has caused a change in study and application toward employing
the battery pack [38]. Previously, the majority of research on neuro-fuzzy logic [39].
5
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Table 3
Comparative Analysis of Regression of various AI techniques.
The usage of neural networks and fuzzy logic reduces the false data while also offering more accurate estimates for all trans-
demand for experimental data. As an alternative to static solutions, mission network quantities.
neural networks, and fuzzy logic provide dynamic solutions to the
majority of battery management system challenges [40]. The lack 3.4. Support vector Machine (SVM)
of a data set to train the system regarding the operating circum-
stances at the time of system deployment limits the use of deep Support vector machine (SVM)-based SOC estimate techniques
neural networks in BMSs [41]. This data, on the other hand, maybe were introduced in [47–52]. For instance, in [47], the SOC of the
acquired continually and in real-time. The neuro-fuzzy approach’s battery was calculated as a function of voltage and current. Tem-
goal is to use a learning mechanism with input–output training perature, voltage, current, and measurements, on the other hand,
data to alter the fuzzy system’s parameters. Sugeno fuzzy rules were utilized as inputs in [48] and [49] to train the SVM model
are used in the framework of ANFIS. It has five layers. Layer 1 is and estimate the SOC. This theory has been demonstrated to be
made up of input variables. Layer 2 is a membership layer that reg- superior to the common theory of empirical risk reduction used
ulates the membership function’s weights. This layer accepts input by conventional neural networks. A quadratic programming prob-
values from the first layer and specifies membership values to rep- lem, which has the global optimal solution, is solved by the learn-
resent fuzzy collections of related input variables. Layer 3 is known ing method of SVMs, although neural networks may have local
as the rule layer, and it contains the items from the previous layer. minimization issues, SVM may generalize better as a result of this
This layer determines the amount of activation for each rule. The characteristic. It is not immediately clear, yet, that SVMs are sup-
fourth layer is the defuzing layer. The output values obtained from posed to completely replace ANN. For instance, multilayer feedfor-
rule extraction are provided by the definition layer. Layer 5 is the ward networks (MLPs), compete directly with SVMs.
output layer, which gathers all of the preceding layer’s inputs A two-step approach of SVM training and virtual testing is
and turns the fuzzy classification results into a clear value [42]. applied to battery data obtained with a typical EV current profile
The efficiency of a neuro-fuzzy system was projected to outper- to calculate the SOH figures of merit, capacity, and resistance
form that of merely neural networks [43]. The performances of [53]. An SVM is trained using current, temperature, and SOC as
these two systems, however, cannot be compared since they do inputs and voltage as output in the first step. The resulting mode-
not provide quantifiable estimates of their accuracy. The research l acts as a voltage look-up table for hypothetical current/tempera-
on neuro-fuzzy systems is also insufficiently thorough. Because ture SOC input based on the actual current/temperature SOC
of the scarcity of previous implementations, future research will profile from the relevant real standard test [54]. As in genuine
be time-consuming [44]. standard performance testing, the virtual test result may then be
utilized to calculate resistance and capacity [55].
3.3. Expert systems (ES)
3.5. Evolutionary algorithm (EA)
The expert systems (ESs) are made up of a feature data repository,
a knowledge base, an explanation machine, a reasoning machine, a To address the growing concerns in BMS, Evolutionary Compu-
knowledge acquisition mechanism, and a human–machine interac- tation (EA) is recommended. To magnify on a nanoscale and
tion interface [45]. The fault symptom information after feature develop a model for lithium-ion battery capacity, multigene
extraction is necessary to construct a feature information database genetic programming (GP) is presented [56]. The second case study
to acquire the unified form of fault symptom information [46]. Most uses a Non-dominated Sorting Genetic Algorithm II (NSGAII) to
energy management systems, for example, require the existence of optimize battery enclosure design at the pack level [57]. The vali-
logic that executes a state estimator and security analysis on a regu- dation results demonstrate NSGAII’s remarkable optimization
lar clock interval, with an override if switching actions occur within a function as well as the enclosure’s increased performance. Further-
few seconds of the start. Furthermore, the operator can request the more, when paired with other artificial intelligence systems, the
execution of the state estimator and security analysis at any time. results demonstrate that NSGAII performs better [58]. The findings
Actions can be made based on the sort of switching action that has imply that EC may be implemented into an EV system to monitor
occurred with an expert system directing the state estimator and its performance and ensure its safety. The motivating factors for
security analysis. What the difference in load level is from the last using EA are as follows: i) EA is a traditional derivative or
run, and so on. For the operators to be able to use vast volumes of gradient-based optimization method, ii) EA is very good at solving
analog data, significant numeric-intensive algorithms must be used non-convex optimization problems with multiple local optima and
to analyze it. Analog measurement data collected from a transmis- finally, iii) EA is very flexible in addressing multi-objective opti-
sion system is far more helpful to an operator when it has been pro- mization problems. A typical EA is an iterative and population-
cessed by a state estimator than when it has been supplied to the based algorithm [59] that begins its search for the best answer
operator as raw data. The state estimator can identify and remove with a random population. The original population is then sub-
6
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
jected to an iterative process including numerous operations such The dataset for this particular work is acquired from GitHub
as reproduction, fitness evaluation, and selection until particular which is an open-source online library where users may work on
stopping criteria are fulfilled. or adopt open-source code projects, fork code, discuss ideas, and
even improvise existing code. Several datasets were obtained and
3.6. Hybrid intelligent systems (HIS) based on the parameter required for the research, the required
dataset was filtered out. The accuracy of the dataset was estimated
For plug-in hybrid electric and electric cars, an intelligent pre- by running it through linear regression and random forest.
dictive battery management system is presented as an ideal energy Among those models, the experiments in this section are con-
and time management solution [60]. This research proposes a sys- ducted for both the linear and the random forest regression mod-
tem to aid drivers in choosing an optimal route and driving profile els. However, a comparative analysis to evaluate the capability of
to save travel time and energy consumption. It investigated and the six AI models on SOC and SOH predictions is introduced in
proved the benefits of the predictive intelligent battery manage- the following section.
ment system for improving battery energy usage and journey The formula for the linear regressor is given as,
duration using both analysis and simulation [61]. Because of the Ci = f (Bi, Y) + E (4).
possibility of nonlinear response, this hybrid model is based on The Ci is the dependent variable which is the output and f (Bi) is
clustering; where the modeling dataset is separated into groups the independent variable. Y is the unknown parameters and E is
with comparable features. Each cluster received a unique regres- the error rate. Thus, the Ci is estimated from the independent input
sion model, to improve overall model performance. The number classes with an error rate of E.
of clusters is determined by evaluating various topologies. This pic- The random forest regression analysis takes the average of the
ture depicts a generic hybrid model with n clusters and n models, sub-sample of the decision tree and uses this averaging to improve
each with its own set of parameters. To develop a hybrid model, the accuracy by reducing the over-fitting of the results. The random
the modeling process might be separated into the following steps: forest regression model controls the maximum sample parameter
clustering phase, regression modeling phase for each cluster, selec- when the bootstrap is true and then each model will produce a data-
tion of the best local model (per cluster), and selection of the best set. Since we use both models, we can interrelate the regression
topology for the hybrid model. model with the linear aggression model. The random forest model
To establish groups with comparable traits, the k-means provides better accuracy when compared with the linear regression
method was utilized during the clustering step. The ANN method model. It is used in the proposed system to validate the regression of
was chosen for the regression modeling phase because of its ability the linear model when we map the exponential response of both
to forecast the output of nonlinear systems with a simple internal these models given in Fig. 5 and Fig. 6. The output of the models
setup. Even though this is a hybrid system, the model performs when captured as a CSV file provides similar responses of both linear
better when the regression models are intelligent systems rather and random forest models in the order of the exponents.
than standard regression methods. From the dataset further analysis was done on the charging and
discharging cycles as observed in Fig. 7 where the positive half of
the cycle depicts charging while the negative half of the cycle rep-
4. Future prospects and inference
resents discharging of the Li-ion battery in the EV. Fig. 7 is used as
the instruction signal to estimate the SOC and SOH.
Lithium-ion batteries are one of the major achievements in
Fig. 8 illustrates the SOC of the battery, while Fig. 9 shows the
electrochemistry, and their uses as ESS have grown dramatically
SEI thickness. The capacity retention of a battery is a measurement
[62]. With no major competition, Li-ion batteries have established
of its capacity to hold onto stored energy during a prolonged open-
themselves in the consumer electronics and electric automobile
circuit rest time. The length of the rest interval, the temperature of
sectors. They will soon dominate the automotive transportation
the cell during that time, and the cell’s past performance all affect
and renewable energy storage industry by utilizing clever adjust-
retained capacity. The layout of the cell also has an impact on
ments in the various functional blocks of the BMS [63].
capacity retention. Several electrode formulation techniques are
One can tap into the latent potential and opportunity of battery
used to create nickel-.
life cycle management by using Machine Learning (ML) and Neural
Networks. The functioning of the battery packs used in EVs can be
optimized by combining modern electronics with testing and
training data. Predictive consciousness derived from experimental
laboratory data can be utilized to forecast battery life, diagnose
problems and their causes, and correct errors before they occur.
7
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Fig. 6. Random Forest Model. passivation layer known as the solid electrolyte interphase (SEI).
To ensure that electrochemical reactions continue and that the
electrolyte doesn’t degrade further, the SEI permits Li + flow while
blocking electrons [77]. The inherent consumption of the anode
and electrolyte during the development of the SEI results in poor
efficiency and the SEI effectively stops Li from coming into contact
with the solvent again, rendering Li dynamically stable in some
organic solvents. The SEI can change how Li-ions are distributed
between the anode and the bulk electrolyte. We propose a novel
method for SOH assessment based on the physical interpretation
of the model parameters that combines parameters of the impe-
dance spectrum by constructing the ratio of the SEI resistance to
the sum of the SEI resistance and the charge transfer resistance.
This ratio, which describes the charge-transfer efficiency at the
electrode surfaces, should consistently decline with SOH. Apart
from the SEI thickness other factors such as overcharging and cycle
count affect the SOH of the battery.
Fig. 10. Comparative Analysis of Various Machine Learning Regression models for SoH and SOC Analysis.
7. Conclusion References
The battery management system along with its two main [1] Abdallah L, El-Shennawy T. Reducing Carbon Dioxide Emissions from
Electricity Sector Using Smart Electric Grid Applications. J Eng
parameters, the state of charge and state of health, has been com- 2013;2013:1–8. doi: https://doi.org/10.1155/2013/845051.
prehensively discussed in this proposed work. The primary goal of [2] Sharma P. Rani Chinnappa Naidu, Optimization techniques for grid-connected
this work is to identify the best AI technology for battery manage- PV with retired EV batteries in centralized charging station with challenges
and future possibilities: A review. Ain Shams Eng J 2023;14(7):. doi: https://
ment systems. Therefore an extensive literature review of the tar- doi.org/10.1016/j.asej.2022.101985101985.
geted AI technologies is conducted, highlighting the challenges [3] Khawaja Y, Qiqieh I, Alzubi J, Alzubi O, Allahham A, Giaouris D. Design of cost-
encountered. This proposed work focused on Li-ion batteries as a based sizing and energy management framework for standalone microgrid
using reinforcement learning. Sol Energy 2023;251:249–60. doi: https://doi.
preferred energy storage solution for electric vehicles due to their org/10.1016/j.solener.2023.01.027.
longer lifespan, reduced discharge rates, excellent efficiency, and [4] Aldbaiat B, Nour M, Radwan E, Awada E. Grid-Connected PV System with
power and energy densities. The key parameters and their roles Reactive Po wer Management and an Optimized SRF-PLL Using Genetic
Algorithm. Energies 2022;15:2177. doi: https://doi.org/10.3390/en15062177.
in a battery management system’s safe operation and operating
[5] Khawaja Y, Allahham A, Giaouris D, Patsios C, Walker S, Qiqieh I. An Integrated
cycle to ensure safe and reliable battery operation in electric vehi- Framework for Sizing and Energy Management of Hybrid Energy System Using
cles have also been described. Automata. Appl Energy 2019;250:257–72. doi: https://doi.org/10.1016/j.
apenergy.2019.04.185.
Six well-known AI technologies are applied for the battery man-
[6] Y. Khawaja, D. Giaouris, H. Patsios and M. Dahidah, ‘‘Optimal cost-based model
agement system state of charge and state of health estimation. for sizing grid-connected PV and battery energy system,” 2017 IEEE Jordan
Detailed results are presented for the linear regression model Conference on Applied Electrical Engineering and Computing Technologies
and random forest, showing that the random forest model outper- (AEECT), Aqaba, Jordan, 2017, pp. 1-6, DOI: 10.1109/AEECT.2017.8257779.
[7] Osama Ayadi, Reem Shadid, Abdullah Bani-Abdullah, Mohammad Alrbai,
forms the linear regression by obtaining more accurate dataset. Mohammad Abu-Mualla, NoorAldeen Balah, Experimental comparison
Additionally, a comparative analysis is conducted between all the between Monocrystalline, Polycrystalline, and Thin-film solar systems under
algorithms including linear, random forest, gradient boost, light- sunny climatic conditions, Energy Reports, Volume 8, Supplement 9, 2022,
Pages 218-230, ISSN 2352-4847, https://doi.org/10.1016/j.egyr.2022.06.121.
gradient boosting machine, and support vector machine regressors. [8] M. Bani Khalid, A. Qandil, N. Beithou, H.Sß . Aybar, Renewable hydrogen driven
Similarly, random forest has demonstrated much superior perfor- CHCP device, International Journal of Hydrogen Energy, Volume 47, Issue 4,
mance for the discharge prediction compared to all other models 2022, Pages 2208-2219, https://doi.org/10.1016/j.ijhydene.2021.10.148.
[9] Ravi SS, Aziz M. Utilization of Electric Vehicles for Vehicle-to-Grid Services:
used in this work in terms of higher R2_score (0.9999), lower mean Progress and Perspectives. Energies (Basel) 2022;15:589. doi: https://doi.org/
absolute error (0.0035), median absolute error (0.0013), means 10.3390/en15020589.
squared error ((0.0001), and root means squared error (0.0097). [10] Abdi H, Mohammadi-ivatloo B, Javadi S, Khodaei AR, Dehnavi E. Energy Storage
Systems. Distributed Generation Systems, Elsevier; 2017, p. 333–68.
Accordingly, i it is concluded from the results achieved that the
https://doi.org/10.1016/B978-0-12-804208-3.00007-8.
Li-ion battery state can be accurately estimated using random for- [11] Hannan MA, Hoque MdM, Hussain A, Yusof Y, Ker PJ. State-of-the-Art and
est regression. With the increasing size of the dataset, a similar Energy Management System of Lithium-Ion Batteries in Electric Vehicle
Applications: Issues and Recommendations. IEEE Access 2018;6:19362–78.
application with more enhanced features and samples may require
https://doi.org/10.1109/ACCESS.2018.2817655.
higher computing power and complexity, thus requiring the pres- [12] Changki Choi, Seongyun Park, Jonghoon Kim, Uniqueness of multilayer
ence of a deep neural network or federated learning, which could perceptron-based capacity prediction for contributing state-of-charge
be addressed in future implementations. estimation in a lithium primary battery, Ain Shams Engineering Journal,
Volume 14, Issue 4, 2023, 101936, https://doi.org/10.1016/j.asej.2022.101936.
[13] Hesse H, Schimpe M, Kucevic D, Jossen A. Lithium-Ion Battery Storage for the
Declaration of Competing Interest Grid—A Review of Stationary Battery Storage System Design Tailored for
Applications in Modern Power Grids. Energies (Basel) 2017;10:2107. doi:
https://doi.org/10.3390/en10122107.
The authors declare that they have no known competing finan- [15] Kumar B, Khare N, Chaturvedi PK. Advanced battery management system
cial interests or personal relationships that could have appeared using MATLAB, Simulink IEEE International Telecommunications Energy
to influence the work reported in this paper. Conference (INTELEC). IEEE 2015;2015:1–6. doi: https://doi.org/10.1109/
INTLEC.2015.7572447.
[16] Chang W-Y. The State of Charge Estimating Methods for Battery: A Review.
Acknowledgement ISRN Applied Mathematics 2013;2013:1–7. doi: https://doi.org/10.1155/2013/
953792.
[17] Arora S, Abkenar AT, Jayasinghe SG, Tammi K. Battery Management System:
The authors would like to thank Applied Science Private Univer- Charge Balancing and Temperature Control. Heavy-Duty Electric Vehicles,
sity (Jordan) and Al-Balqa Applied University (Jordan) for their Elsevier; 2021, p. 173–203. https://doi.org/10.1016/B978-0-12-818126-
3.00005-1.
funding and support.
9
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
[18] Shankar N, Chitra A, Banerjee D, Sharma V, Zhutshi K, Razia SW, et al. [43] Talpur N, Abdulkadir SJ, Alhussian H, Hasan MH, Aziz N, Bamhdi A. Deep
Innovations in Power and Advanced Computing Technologies (i-PACT). IEEE Neuro-Fuzzy System application trends, challenges, and future perspectives: a
2021;2021:1–7. doi: https://doi.org/10.1109/i-PACT52855.2021.9697046. systematic survey. Artif Intell Rev 2023;56:865–913. doi: https://doi.org/
[19] Gabbar H, Othman A, Abdussami M. Review of Battery Management Systems 10.1007/s10462-022-10188-3.
(BMS) Development and Industrial Standards. Technologies (Basel) 2021;9:28. [44] Shihabudheen KV, Pillai GN. Recent advances in neuro-fuzzy system: A survey.
doi: https://doi.org/10.3390/technologies9020028. Knowl Based Syst 2018;152:136–62. doi: https://doi.org/10.1016/
[20] Hossain Lipu MS, Miah MdS, Ansari S, Wali SB, Jamal T, Elavarasan RM, et al. j.knosys.2018.04.014.
Smart Battery Management Technology in Electric Vehicle Applications: [45] Choi SY, Kim SH. Knowledge Acquisition and Representation for High-
Analytical and Technical Assessment toward Emerging Future Directions. Performance Building Design: A Review for Defining Requirements for
Batteries 2022;8:219. https://doi.org/10.3390/batteries8110219. Developing a Design Expert System. Sustainability 2021;13:4640. doi:
[21] Cabrera J, Vega A, Tobajas F, Deniz V, Fabelo HA. Design of a reconfigurable Li- https://doi.org/10.3390/su13094640.
Ion Battery Management System (BMS). XI Tecnologias Aplicadas a la [46] Qiu W, Qiu Z. Design for symmetrical management of storage battery expert
Ensenanza de la Electronica (Technologies Applied to Electronics Teaching) system based on single battery. In: 2006 International Conference on
(TAEE). IEEE 2014;2014:1–6. doi: https://doi.org/10.1109/TAEE.2014.6900162. Mechatronics and Automation. p. 1141–6. doi: https://doi.org/10.1109/
[22] Feng Y, Cao Z, Shen W, Yu X, Han F, Chen R, et al. Intelligent battery ICMA.2006.257786.
management for electric and hybrid electric vehicles: A survey. 2016 IEEE [47] Feng Y, Xue C, Han Q-L, Han F, Du J. Robust Estimation for State-of-Charge and
International Conference on Industrial Technology (ICIT), IEEE; 2016, p. 1436– State-of-Health of Lithium-Ion Batteries Using Integral-Type Terminal Sliding-
41. https://doi.org/10.1109/ICIT.2016.7474969. Mode Observers. IEEE Trans Ind Electron 2020;67:4013–23. doi: https://doi.
[23] Lelie M, Braun T, Knips M, Nordmann H, Ringbeck F, Zappen H, et al. Battery org/10.1109/TIE.2019.2916389.
Management System Hardware Concepts: An Overview. Appl Sci 2018;8:534. [48] Aslan E, Yasa YA. In: Review on The Battery State of Charge Estimation
doi: https://doi.org/10.3390/app8040534. Methods For Electric Vehicle Battery Management Systems. p. 281–5.
[24] Uzair M, Abbas G, Hosain S. Characteristics of Battery Management Systems of [49] Liu K, Li Y, Hu X, Lucu M, Widanage WD. Gaussian Process Regression With
Electric Vehicles with Consideration of the Active and Passive Cell Balancing Automatic Relevance Determination Kernel for Calendar Aging Prediction of
Process. World Electric Vehicle Journal 2021;12:120. doi: https://doi.org/ Lithium-Ion Batteries. IEEE Trans Industr Inform 2020;16:3767–77. doi:
10.3390/wevj12030120. https://doi.org/10.1109/TII.2019.2941747.
[25] Balasingam B, Ahmed M, Pattipati K. Battery Management Systems— [50] Chen Y, Gao C, Yang W, Wang Z, Yu Z, Xia H. A Recurrent Gaussian Process
Challenges and Some Solutions. Energies (Basel) 2020;13:2825. doi: https:// Regression Model with Composite Kernel for Industrial Process Quality Prediction.
doi.org/10.3390/en13112825. In: 2019 International Conference on Networking and Network Applications
[26] Meissner E, Richter G. Battery Monitoring and Electrical Energy Management. J (NaNA), IEEE. p. 229–34. doi: https://doi.org/10.1109/NaNA.2019.00048.
Power Sources 2003;116:79–98. doi: https://doi.org/10.1016/S0378-7753(02) [51] Zhao J, Guo H, Wang L, Han M. Computer Modeling of the Eddy Current Losses of
00713-9. Metal Fasteners in Rotor Slots of a Large Nuclear Steam Turbine Generator Based
[27] Gold S. A PSPICE macromodel for lithium-ion batteries. The Twelfth Annual on Finite-Element Method and Deep Gaussian Process Regression. IEEE Trans
Battery Conference on Applications and Advances, IEEE; n.d., p. 215–22. Ind Electron 2020;67:5349–59. doi: https://doi.org/10.1109/TIE.2019.2931487.
https://doi.org/10.1109/BCAA.1997.574106. [52] Thenaisie G, Park C-H, Lee S-G. A Real-Time Entropy Estimation Algorithm for
[28] Xing Y, Ma EWM, Tsui KL, Pecht M. Battery Management Systems in Electric Lithium Batteries Based on a Combination of Kalman Filter and Nonlinear
and Hybrid Vehicles. Energies (Basel) 2011;4:1840–57. doi: https://doi.org/ Observer. IEEE Trans Ind Electron 2020;67:8034–43. doi: https://doi.org/
10.3390/en4111840. 10.1109/TIE.2019.2945283.
[29] Yin H, Zhou W, Li M, Ma C, Zhao C. An Adaptive Fuzzy Logic-Based Energy [53] Junping W, Quanshi C, Binggang C. Support vector machine based battery
Management Strategy on Battery/Ultracapacitor Hybrid Electric Vehicles. IEEE model for electric vehicles. Energy Convers Manag 2006;47:858–64. doi:
Trans Transp Electrif 2016;2:300–11. doi: https://doi.org/10.1109/ https://doi.org/10.1016/j.enconman.2005.06.013.
TTE.2016.2552721. [54] Klass V, Behm M, Lindbergh G. Capturing lithium-ion battery dynamics with
_
[30] Ozadowicz A, Grela J. Energy saving in the street lighting control system—a support vector machine-based battery model. J Power Sources
new approach based on the EN-15232 standard. Energy Effic 2017;10:563–76. 2015;298:92–101. doi: https://doi.org/10.1016/j.jpowsour.2015.08.036.
doi: https://doi.org/10.1007/s12053-016-9476-1. [55] Klass V, Behm M, Lindbergh G. A support vector machine-based state-of-
[31] Huang S-C, Tseng K-H, Liang J-W, Chang C-L, Pecht M. An Online SOC and SOH health estimation method for lithium-ion batteries under electric vehicle
Estimation Model for Lithium-Ion Batteries. Energies (Basel) 2017;10:512. doi: operation. J Power Sources 2014;270:262–72. doi: https://doi.org/10.1016/j.
https://doi.org/10.3390/en10040512. jpowsour.2014.07.116.
[32] Venugopal P, T. V. State-of-Health Estimation of Li-ion Batteries in Electric [56] Garg A, Yun L, Shaosen S, Goyal A, Niu X, Gao L, et al. A combined
Vehicle Using IndRNN under Variable Load Condition. Energies (Basel) experimental-numerical framework for residual energy determination in
2019;12:4338. https://doi.org/10.3390/en12224338. spent lithium-ion battery packs. Int J Energy Res 2019;43:4390–402. doi:
[33] Obert J, Trevizan RD, Torres-Castro L, Preger Y. Ensemble Learning, Prediction https://doi.org/10.1002/er.4564.
and Li-Ion Cell Charging Cycle Divergence. IEEE Open Access Journal of Power [57] Cui X, Panda B, Chin CMM, Sakundarini N, Wang C, Pareek K. An application of
and Energy 2021;8:303–15. doi: https://doi.org/10.1109/OAJPE.2021.3100004. evolutionary computation algorithm in multidisciplinary design optimization
[34] Alzubaidi L, Zhang J, Humaidi AJ, Al-Dujaili A, Duan Y, Al-Shamma O, et al. of battery packs for electric vehicle. Energy Storage 2020;2. doi: https://doi.
Review of deep learning: concepts, CNN architectures, challenges, org/10.1002/est2.158.
applications, future directions. J Big Data 2021;8:53. doi: https://doi.org/ [58] Morais H, Sousa T, Castro R, Vale Z. Multi-Objective Electric Vehicles
10.1186/s40537-021-00444-8. Scheduling Using Elitist Non-Dominated Sorting Genetic Algorithm. Appl Sci
[35] Scrosati B, Garche J. Lithium batteries: Status, prospects and future. J Power 2020;10:7978. doi: https://doi.org/10.3390/app10227978.
Sources 2010;195:2419–30. doi: https://doi.org/10.1016/j. [59] Ali AF, Tawhid MA. A hybrid particle swarm optimization and genetic algorithm
jpowsour.2009.11.048. with population partitioning for large scale optimization problems. Ain Shams
[36] Yang D, Wang Y, Pan R, Chen R, Chen Z. A Neural Network Based State-of- Eng J 2017;8:191–206. doi: https://doi.org/10.1016/j.asej.2016.07.008.
Health Estimation of Lithium-ion Battery in Electric Vehicles. Energy Procedia [60] Rajalakshmi M, Razia Sultana W. Intelligent Hybrid Battery Management
2017;105:2059–64. doi: https://doi.org/10.1016/j.egypro.2017.03.583. System for Electric Vehicle. Artificial Intelligent Techniques for Electric and
[37] Choi C, Park S, Kim J. Uniqueness of multilayer perceptron-based capacity Hybrid Electric Vehicles, Wiley; 2020, p. 179–206. https://doi.org/10.1002/
prediction for contributing state-of-charge estimation in a lithium primary 9781119682035.ch10.
battery. Ain Shams Eng J 2023;14:. doi: https://doi.org/10.1016/j. [61] Abdul-Hak M, Al-Holou NITS. based Predictive Intelligent Battery Management
asej.2022.101936101936. System for plug-in Hybrid and Electric vehicles. IEEE Vehicle Power and
[38] Chitnis MS, Pandit SP, Shaikh MN. Electric Vehicle Li-Ion Battery State of Propulsion Conference. IEEE 2009;2009:138–44. doi: https://doi.org/10.1109/
Charge Estimation Using Artificial Neural Network. In: 2018 International VPPC.2009.5289858.
Conference on Inventive Research in Computing Applications (ICIRCA), IEEE. p. [62] el Kharbachi A, Zavorotynska O, Latroche M, Cuevas F, Yartys V, Fichtner M.
992–5. doi: https://doi.org/10.1109/ICIRCA.2018.8597234. Exploits, advances and challenges benefiting beyond Li-ion battery
[39] Dai H, Guo P, Wei X, Sun Z, Wang J. ANFIS (adaptive neuro-fuzzy inference technologies. J Alloys Compd 2020;817:. doi: https://doi.org/10.1016/
system) based online SOC (State of Charge) correction considering cell j.jallcom.2019.153261153261.
divergence for the EV (electric vehicle) traction batteries. Energy [63] Das CK, Bass O, Kothapalli G, Mahmoud TS, Habibi D. Overview of energy
2015;80:350–60. doi: https://doi.org/10.1016/j.energy.2014.11.077. storage systems in distribution networks: Placement, sizing, operation, and
[40] Martinez DA, Poveda JD, Montenegro D. Li-Ion battery management system power quality. Renew Sustain Energy Rev 2018;91:1205–30. doi: https://doi.
based in fuzzy logic for improving electric vehicle autonomy. IEEE Workshop org/10.1016/j.rser.2018.03.068.
on Power Electronics and Power Quality Applications (PEPQA). IEEE [64] Buccolini L, Ricci A, Scavongelli C, DeMaso-Gentile G, Orcioni S, Conti M.
2017;2017:1–6. doi: https://doi.org/10.1109/PEPQA.2017.7981677. Battery Management System (BMS) simulation environment for electric
[41] Chen C, Zhang P, Zhang H, Dai J, Yi Y, Zhang H, et al. Deep Learning on vehicles. 2016 IEEE 16th International Conference on Environment and
Computational-Resource-Limited Platforms: A Survey. Mob Inf Syst Electrical Engineering (EEEIC), IEEE; 2016, p. 1–6. https://doi.org/10.1109/
2020;2020:1–19. doi: https://doi.org/10.1155/2020/8454327. EEEIC.2016.7555475.
[42] Abdalla SA, Abdullah SS, Kassem AM. Performance enhancement and power [65] Manish Mawatwal, Aayush Mohanty, Anitha G. S. State of Charge Estimation
management strategy of an autonomous hybrid fuel cell/wind power system for Rechargeable Lithium-ion Battery using ANFIS MATLAB. International
based on adaptive neuro fuzzy inference system. Ain Shams Eng J 2022;13:. Journal of Engineering Research And 2020;V9. https://doi.org/10.17577/
doi: https://doi.org/10.1016/j.asej.2021.101655101655. IJERTV9IS030565.
10
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
[66] Abdul-Hak M, Al-Holou N, Mohamm U. Predictive Intelligent Battery Issa Qiqieh is currently an assistant professor at Al-
Management System to Enhance the Performance of Electric Vehicle. Electric Balqa Applied University, Jordan. He received his Ph.D.
Vehicles - Modelling and Simulations, InTech; 2011. https://doi.org/10.5772/ degree in computer and electrical engineering from the
17099. School of Engineering, Newcastle University, Newcastle
[67] Ismail M, Dlyma R, Elrakaybi A, Ahmed R, Habibi S. Battery state of charge upon Tyne – UK in (2018), and his M.Sc. (with distinc-
estimation using an Artificial Neural Network. 2017 IEEE Transportation tion) and B.Sc. degrees from Hijjawi Faculty of Engi-
Electrification Conference and Expo (ITEC), IEEE; 2017, p. 342–9. https://doi. neering Technology, Yarmouk University – Jordan, in
org/10.1109/ITEC.2017.7993295. 2007&2005, respectively. He is working as a full-time
[68] Watrin N, Blunier B, Miraoui A. Review of adaptive systems for lithium lecturer in the Faculty of Engineering at Al-Balqa
batteries State-of-Charge and State-of-Health estimation. 2012 IEEE
Applied University from 2007 till now. He is also the
Transportation Electrification Conference and Expo (ITEC), IEEE; 2012, p. 1–
author/co-author of 15+ IEEE/ACM/Elsevier journal and
6. https://doi.org/10.1109/ITEC.2012.6243437.
[69] Zhou C, Qian K, Allan M, Zhou W. Modeling of the Cost of EV Battery Wear Due conference articles. His research interests include multi
to V2G Application in Power Systems. IEEE Trans Energy Convers and interdisciplinary environments involving approximate, power-adaptive, and
2011;26:1041–50. doi: https://doi.org/10.1109/TEC.2011.2159977. energy-efficient computing for intelligent embedded and IoT systems, as well as
[70] Gao Z, Chin C, Chiew J, Jia J, Zhang C. Design and Implementation of a Smart research solutions in machine learning and cyber security.
Lithium-Ion Battery System with Real-Time Fault Diagnosis Capability for
Electric Vehicles. Energies (Basel) 2017;10:1503. doi: https://doi.org/10.3390/
en10101503. Jafar A. Alzubi is an associate professor at Al-Balqa
[71] Eddahech A, Briat O, Vinassa JM. Adaptive voltage estimation for EV Li-ion cell Applied University, Jordan. Received his Ph.D. degree in
based on artificial neural networks state-of-charge meter. 2012 IEEE Advanced Telecommunications from Swansea Univer-
International Symposium on Industrial Electronics, IEEE; 2012, p. 1318–24. sity, Swansea – UK (2012). Master of Science degree
https://doi.org/10.1109/ISIE.2012.6237281. (Hons.) in electrical and computer engineering from
[72] di Domenico D, Stefanopoulou A, Fiengo G. Lithium-Ion Battery State of Charge New York Institute of Technology, New York - USA
and Critical Surface Charge Estimation Using an Electrochemical Model-Based
(2005). And Bachelor of Science degree (Hons.) in Elec-
Extended Kalman Filter. J Dyn Syst Meas Control 2010;132. https://doi.org/
trical Engineering, majoring in Electronics and Com-
10.1115/1.4002475.
munications, from the University of Engineering and
[73] Charkhgard M, Farrokhi M. State-of-Charge Estimation for Lithium-Ion
Batteries Using Neural Networks and EKF. IEEE Trans Ind Electron Technology, Lahore – Pakistan (2001). Jafar works and
2010;57:4178–87. doi: https://doi.org/10.1109/TIE.2010.2043035. researches in multi and interdisciplinary environment
[74] Jiménez-Bermejo D, Fraile-Ardanuy J, Castaño-Solis S, Merino J, Álvaro- involving Machine Learning, classifications and detec-
Hermana R. Using Dynamic Neural Networks for Battery State of Charge tion of Web scams, Internet of Things, Wireless Sensor Networks and its security,
Estimation in Electric Vehicles. Procedia Comput Sci 2018;130:533–40. doi: and using Algebraic– Geometric theory in channel coding for wireless networks.
https://doi.org/10.1016/j.procs.2018.04.077. Managed and directed few projects funded by the European Union. A cumulative
[75] Lin H-T, Liang T-J, Chen S-M. Estimation of Battery State of Health Using research experience for over ten years resulted in publishing more than sixty papers
Probabilistic Neural Network. IEEE Trans Industr Inform 2013;9:679–85. doi: in highly impacted journals. Currently, he is IEEE senior member and serving as an
https://doi.org/10.1109/TII.2012.2222650. editor for IEEE Access Journal, and Wireless Sensor Networks area editor for Turkish
[76] Hannan MA, Lipu MSH, Hussain A, Saad MH, Ayob A. Neural Network Journal of Electrical Engineering and Computer Sciences. In addition, he is editorial
Approach for Estimating State of Charge of Lithium-Ion Battery Using board member and reviewer in many other prestigious journals in computer
Backtracking Search Algorithm. IEEE Access 2018;6:10069–79. doi: https://
engineering and science field. He also managed several special issues in high
doi.org/10.1109/ACCESS.2018.2797976.
impacted journals.
[77] Wang, A., Kadam, S., Li, H. et al. Review on modeling of the anode solid
electrolyte interphase (SEI) for lithium-ion batteries. npj Comput Mater 4, 15
(2018). https://doi.org/10.1038/s41524-018-0064-0.
[78] Abdelbasset WK, Elkholi SM, Ismail KA, Alshehri S, Alobaida A, Huwaimel B, Dr. Omar A. Alzubi received his BSc degree in 1986
et al. Development a novel robust method to enhance the solubility of from The University of Jordan, Jordan. He obtained his
Oxaprozin as nonsteroidal anti-inflammatory drug based on machine- MSc degree with distinction in Computer and Network
learning. Sci Rep 2022;12:13138. doi: https://doi.org/10.1038/s41598-022- Security from New York Institute of Technology, New
17440-4. York, USA in 2006. In 2013, he received his Ph.D. degree
in Computer and Network Security from Swansea
University, Swansea, UK. Currently he is an associate
professor in the Computer Science Department at Al-
Yara Khawaja is currently an assistant professor at
Balqa Applied University, Jordan. Dr. Alzubi research
Applied Science Private University, Jordan. She received
interests include machine learning and cyber security.
her Ph.D. degree in Renewable Energy/Electrical Engi-
His cumulative research experience for over ten years
neering from the School of Engineering, Newcastle
resulted in publishing more than forty articles in highly
University, Newcastle upon Tyne – UK in 2019, and her
impacted journals. Dr. Alzubi is the vice dean of Prince Abdullah Bin Ghazi Faculty
M.Sc. (with distinction) in Computer Science from
of Information and Communication Technology. In addition, he is an editorial board
Prince Abdullah bin Ghazi Faculty of Information and
member and reviewer of many prestigious journals in computer science field.
Communication Technology, Al-Balqa Applied Univer-
sity–Jordan in 2013. Yara received her B.Sc. degrees in
Power Engineering from Hijjawi Faculty of Engineering
Dr. Nallakaruppan M.K is working as the Assistant
Technology, Yarmouk University – Jordan, in 2007. Her
Professor in School of Information Technology and
current research interests include sizing of Microgrids
Engineering, Vellore Institute of Technology, Vellore,
containing distributed generation, energy management and control of Microgrid
Tamilnadu, India. He completed his bachelor degree in
using new methods such as Automata, and sizing and managing energy storage
B.E.Electronics and Communication Engineering in J.J.
systems.
College of Engineering and Technology. He completed
his master’s degree in M.E.Software Engineering in Anna
University. He completed his PhD in ‘‘Decision Support
Nathan Shankar is currently pursuing his B.Tech in EEE
Systems for IoT based Infrastructures” in VIT University,
from Vellore Institute of Technology, Vellore, and will
Vellore. His area of expertise are IoT and smart net-
graduate in 2023. He has published around 2 papers in
works, Machine learning, soft computing, Decision Sci-
international journals and conferences and has received
ence and Deep learning. He has 31 International
the best paper in one of the international conferences.
publications in various International Journals under his area of interest.He has 31
Besides this, he also has 1 preprint and 2 other accepted
international publications including 16 research papers with various reputed
papers currently undergoing publication. His research
publishers like MDPI, Springer and Inderscience , 9 conferences with various
areas include Robotics, Microcontrollers, the Internet of
reputed publishers like IEEE and springer and 3 book chapters including Elseiver
Things (IoT), Power Electronics, and Solar PV. He has
and Springer.
worked on several projects that involve Robotics, IoT, AI,
and their applications in the different fields of Electrical
Engineering.
11
Y. Khawaja, N. Shankar, I. Qiqieh et al. Ain Shams Engineering Journal xxx (xxxx) xxx
Sanjeevikumar Padmanaban (Member’12–Senior tution of Engineering and Technology, U.K. He received a lifetime achievement
Member’15, IEEE) received a PhD degree in electrical award from Marquis Who’s Who - USA 2017 for contributing to power electronics
engineering from the University of Bologna, Bologna, and renewable energy research. He is listed among the world’s top 2 scientists
Italy 2012. He is a Full Professor in Electrical Power (from 2019) by Stanford University USA. He is an Editor/Associate Editor/Editorial
Engineering with the Department of Electrical Engi- Board for refereed journals, in particular the IEEE SYSTEMS JOURNAL, IEEE Trans-
neering, Information Technology, and Cybernetics, action on Industry Applications, IEEE ACCESS, IET Power Electronics, IET Electronics
University of South-Eastern Norway, Norway. S. Pad- Letters, and Wiley-International Transactions on Electrical Energy Systems, Subject
manaban has authored over 750+ scientific papers and Editorial Board Member—Energy Sources—Energies Journal, MDPI, and the Subject
received the Best Paper cum Most Excellence Research Editor for the IET Renewable Power Generation, IET Generation, Transmission and
Paper Award from IET-SEISCON’13, IET-CEAT’16, IEEE- Distribution, and FACETS Journal (Canada).
EECSI’19, IEEE-CENCON’19, and five best paper awards
from ETAEERE’16 sponsored Lecture Notes in Electrical
Engineering, Springer book. He is a Fellow of the Institution of Engineers, India, the
Institution of Electronics and Telecommunication Engineers, India, and the Insti-
12