Exploring Machine Learning and Deep Learning Approaches for Battery Management Systems in EVs: A Comprehensive Review

Abstract

Electric vehicles (EVs) are a promising zero-emission technology in the automobile industry, but they face several challenges in terms of performance, reliability, and safety. Batteries are the heart of the EV system which helps to run the vehicle with reliability. Batteries during the process of running undergo various changes that need to be addressed. On the other hand, real-time data analysis and online access to information are necessary conditions in the modern world. Machine learning and deep learning algorithms mimic humans by focusing on statistical data and algorithms on a real-time basis. Therefore, in today’s research, machine learning and deep learning algorithms are used in EV technologies to obtain a more efficient and capable system. The battery management system (BMS) is the main part that is often in need of data processing of battery parameters and diagnosis of the problem. This paper explores the comprehensive literature review on machine learning and deep learning approaches for BMS in EVs. The state of charge (SOC) estimation, charge equalization and cell balancing, fault detection and diagnosis, and thermal management systems using various combined machine learning and deep learning techniques are discussed. By synthesizing insights from various studies, this article presents improved parameters and valuable inferences. This article aims to highlight the pivotal role of artificial intelligence (AI) and deep learning in improving the functionality of the BMS, ultimately contributing to the performance and longevity of EVs.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

1. Introduction

With respect to the accountability of the growing population, there is a great demand for transportation services. The demand for transportation services is directly proportional to the amount of population in that area. It is seen that Asia-Pacific was the largest transportation services market in the year 2022. The compound annual growth rate [1] measured between 2022 and 2023 is 8.11%. This calculation is done considering rails, roads, commodities transport, and materials through pipelines. According to research in Reference [1], there will be USD 15.94 trillion in transportation services by 2032. The multiplication of transportation services is the main cause of climate change [2]. Twenty-five percent of air pollution is because of traffic and vehicle emissions. Global warming is of great concern, and it can be counteracted by reducing vehicle emissions. Therefore, to mitigate climate change, electric vehicles (EVs) with zero emission are of great importance [3]. EVs have batteries or extra-vehicular-sourced motors used for transportation. They have a wide range of applications on roads, rails, airways, and waterways. During 1900–1921, electric cars accounted for almost 28% of the total cars [4], but gasoline and steam cars overwhelmed electric cars after 1921 and decreased their usage [5]. Later, in the 20th century, because of peak oil prices, there was a shift toward EVs, which depended on the backup of power sources from fossil fuels, solar power, wind power, and nuclear power.

There are two different structures of EVs—pure EVs and hybrid EVs [6]. Hybrid EVs have three forms—series, parallel, and combination. Pure EVs contain a convertor, motor, fixed gear, and differential gear. In a series of hybrid EVs, the total energy sources are obtained from the combination of two or more energy sources. Parallel hybrid EVs have two or more drive shafts with a combination of internal combustion converter machines and electrical machines. The combination method contains both series and parallel forms. The utilization of renewable energy sources for EVs is discussed in the paper [7, 8] where the use of fossil fuels and greenhouse gas emissions are reduced. On reviewing the existing literature, the best smart charging method is chosen, which reduces the installation of transmission and distribution systems. Solar energy and wind energy charging stations are of great importance. In the paper [9], the negative impact of EVs without adequate charging management systems is discussed. The use of smart grids such as vehicle-to-grid technology is discussed as a potential method for charging EV and increasing the deployment of EVs. The survey of smart grid integration with EVs was conducted in the paper [10] where cost-aware charging, emissions-aware charging, and effective aware charging systems were discussed in detail. The growing EV technology involves wireless charging systems [11], nonlinear control, and smart charging systems. Approximately, 95% of total EVs are sold mostly in China, USA, Japan, Canada, Norway, the United Kingdom, France, Germany, The Netherlands, and Sweden. As batteries play an important role in the development of EV technology, graphene produces high power and charges in a short time period [12]. In the demanding market of EV technology, policy schemes such as financial incentives, access to high-occupancy vehicle lane, public charging, building codes, carbon pricing, supply focussed zero-emission vehicle mandate, vehicle emission standards, and low carbon fuel standards were reviewed in the paper [13]. CO₂ emissions are reduced by using shared EVs compared to self-service and ride-sourcing EV. Life cycle analysis is required to study the overall performance of the EV, which is surveyed in the paper [14] where lithium ion batteries have high benefits compared to lithium batteries. Cybersecurity challenges such as EV supply equipment authentication protocols were also discussed in the article which is part of smart cities. There are different EVs, such as battery EV (BEV), hybrid EV (HEV), fuel cell EV, and internal combustion EV (ICEV). Their overall performance was calculated in the paper [15], and the performance of BEV was the main concern. Well to wheel analysis was also studied.

2. Battery Management Systems (BMSs)

Rechargeable batteries are the heart of an EV, and they need to be maintained in a proper condition to increase the lifetime of the system. Lithium-ion batteries are mostly used in EVs because of their high life cycle analysis. The basic monitoring of lithium-ion batteries is required to enhance protection, lifetime, and energy balancing. Different modeling approaches for the electrochemical system to forecast aging were discussed in the paper [16, 17]. They are (i) the black-box approach, (ii) the electrical lumped model, and (iii) the electrochemical model. The authors in this paper proposed that the EC models are best for studying the prediction of aging compared to the statistical models. The LiFePO₄ achieves an accuracy of 1% for currents and 0.1% for voltage levels. The hierarchical platform for BMS is provided where efficient charge balancing and state of charge (SOC) estimation are performed. The software and hardware framework of the BMS in the EV is provided in the paper [18]. Parameters like general digital input, general analog output, cell voltage measurement, temperature sensors, bus voltage/current measurement, charging system, internal power supply module, thermal management module, balancing control module, high voltage safety control, communication module, general digital output, global clock module, and man-machine interface module are studied. The authors ensured the reliability and safety of the system using an adaptive control system to identify the battery parameters, estimate the data, and communicate to the vehicular network. In the paper [19], the main characteristics of Li-ion batteries such as time invariant, nonlinearity, and complex electrochemical systems are explained. The parameters of Li-ion batteries considered for evaluations are provided in Table 1.

Table 1

Parameters of Li-ion batteries.

Parameters	Values
Energy density (Wh/L)	200–400
Power density (W/L)	1500–10,000
Nominal voltage (V)	4.3
Life cycle	10,000
Depth of discharge (%)	95
Round trip efficiency (%)	96
Estimated cost (USD/kWh)	200–1260

The different SOC estimation methods such as direct measurements, bookkeeping measurements, model-based methods, and computer intelligence were discussed in the paper, and it was concluded that machine learning–based SOC shows higher accuracy under different challenging conditions. The evolution of BMSs was given in the paper [20]. The different EV standards [21–43] are provided in Table 2. Four generations were provided which are described in Figure 1.

Table 2

Different standards of EV.

Systems	Standard numbers
ISO	ISO 6469-1
	ISO 6469-2
	ISO 6469-3
	ISO 6469-4
	ISO 12405-4
IEC	IEC 62660-1
	IEC 62660-2
	IEC 62660-3
SAE	SAE J1766
	SAE J2464
	SAE J2929
EN	EN 1987-1
	EN 1987-2
	EN 1987-3
U.S	DOE/ID-11069
	UL 2580
China	GB/T31484
	GB/T31485
	GB/T31486
	GB/T 31,467.1
	GB/T 31,467.2
	GB/T 31,467.3
	GB/T 38,661

[figure(s) omitted; refer to PDF]

Similar characteristics of NiMH and Li-ions are compared and analyzed in the paper [44]. Parameters such as open circuit voltage, SOC, aging, and temperature management were studied. Various Li-ion batteries like LiCoO₂, LiMn₂O₄, LiFePO₄, LiNiMnCoO, LiNiCoAlO₂, and Li₄Ti₅Ol₂ were explained in the paper [45], with their different characteristics and applications being provided. Intelligent algorithms are used for BMSs and control schemes such as feed neural network (FNN), radial basis function neural network (RBFNN), efficient machine learning (ELM), support vector machine (SVM), random forest (RF), wavelet neural network (WNN), gaussian process regression (GPR), time delay neural network (TDNN), long short-term memory (LSTM), gated recurrent unit (GRU), genetic algorithm (GA), backtracking search optimization (BSA), gravitational search algorithm (GSA), latent semantic analysis (LSA), and firefly algorithm (FA) are explained in the paper [46]. EVs are also explored with regard to advanced technologies like cloud computing, block chain technology, multimodeling co-estimation, and artificial intelligence (AI) [47]. The Lithium-ion metal and sodium ion technologies are also used for manufacturing batteries for EVs. Advanced modeling methods such as electrochemical model, equivalent circuit model, and data-driven model are explained in the paper [47]. The future of the BMS of EV depends on the hybrid intelligent algorithms, universal BMS, efficient prototype design, BMS virtual type, etc., as discussed in the paper [48]. The real-time SOC and state of health (SOH) estimations were also discussed in this paper.

The BMS plays an important role in the proper functioning of the EV. The four important things that must be considered are battery state estimation, battery equalization, battery fault diagnosis, and thermal management systems. Battery state estimation helps to study the battery charge available for the functioning of the EV. Equalization battery is required to overcome the effects of overcharging. In the battery fault diagnosis, the residual signals are used to find the system fault operation. The thermal management system is used to protect and maintain the temperature of the system at a nominal value. Figure 2 shows the amount of researches that have taken place in various sections of BMS in EVs.

[figure(s) omitted; refer to PDF]

2.1. Machine Learning–Based SOC Estimation

Figure 3 show the various ML algorithms with SOC estimation.

[figure(s) omitted; refer to PDF]

2.1.1. Data-Driven Approach

Estimating the battery’s SOH is necessary to determine whether to replace it or extend its useful life for EV operation. The real-time basis functioning of the system is required to maintain constant working. Therefore, machine learning algorithms are used to study the SOC of the battery. SOH based on the data-driven approach [49] is the first method in real-time battery in EVs. The current, voltage, and temperature data sets are trained from the historical distribution to build the data set. Using the data-driven approach, the time-varying distribution density is done.

2.1.2. Feed Forward Neural Network Approach

The trustable estimation RUL for online applications is done using feed forward neural network using importance sampling [50]. In this paper, the authors used 40 hidden neurons for a balanced MCU of the system, the lithium-ion charge curve reconstitution is done with the help of IS using the amount of input voltage points, and the difference in voltage curves among batteries of various cycles was used for RUL definition.

For SOC estimation, deep neural FNN is utilized, in which time dependencies are encoded into network weights that are evaluated at ambient temperatures and display an MAE of 10%.

2.1.3. Extreme Learning Machine Algorithm–Based Approach

The extreme learning machine algorithm is used to indicate the SOH of batteries as proposed in the paper [55] where the correlation between existing health indicators and capacity degradation is carried out to improve the speed and accuracy of the system. The result outperforms the BP network.

2.1.4. SVM Algorithm–Based Approach

The dynamic performance of the batteries in the EV is estimated considering the battery OCV, and model parameters were estimated with a precision of 2.3% using the SVM algorithm [56]. The probability of finding the global optimal solution, simpler solving methods, and the fastest solution approach were the key advantages of the SVM used.

2.1.5. Fuzzy Algorithm–Based Approach

In Reference [57], a fuzzy genetic clustering technique is proposed for EV SOC estimation. The method utilizes subtractive clustering and a direct search algorithm to model the topology and parameters from previous studies. The recursive least-squares approach is used to extract those parameters. The optimization was also done with the back propagation algorithm.

2.1.6. RF-Based Approach

The RF machine learning algorithm is used by the authors in the paper [58] to estimate battery conditions. The RF algorithm uses two features: (i) the number of trees and (ii) the number of random features. The inputs were given directly from the recorded data, which reduces the computation complexity of the work. Using the CC curves, the peaks in the curve are directly proportional to the capacity fade of the battery. Thus, it efficiently estimates the capacity of the battery at various cycling conditions with the RMSE less than 1.3%.

2.1.7. GPR-Based Approach

The GPR machine learning algorithm is used in the paper [59] where the authors used the advanced Gaussian filter to smooth the capacity curves and used mean and covariance functions to predict the SOH of the battery. For different battery health conditions, the GPR algorithm was robust and reliable. The RMSE was calculated at approximately 3% for the GPR machine learning algorithm.

In the paper [60], the GPR-based ML algorithm used along with the electrochemical impedance spectroscopy measurement showed an error occurrence of less than 3.8% in the SOC estimation.

2.1.8. Back Propagation Neural Network (BPNN)-Based Approach

The BPNN-based backtracking search algorithm was used to estimate strategy with the help of hidden neurons and the learning rate in the article [61]. This method didn’t consider any models, parameters, and equations. The BPNN-based BSA works efficiently in all temperatures and load conditions.

2.1.9. WNN-Based Approach

It is known that the battery estimation is nonstationary, the paper [62] proposed the use of discrete wavelet transform in the WNN, and the parameters are decomposed into wavelets in the frequency domain. The estimation was carried out at different temperatures, and the results were compared with the LMWNN method where the DWT WNN outperforms. The work showed 0.59% error in terms of RMSE.

2.1.10. Discrete Neural Network–Based Approach

The authors in the paper [51] map the battery signals like voltage, current, and temperature directly to SOC and estimate the error below 1% and self-learn all the weights, which reduces the cost and time consumption. This method is tough toward the measurement offsets, gains, and noise. Table 3 recorded the ML methods for SOC estimations.

Table 3

ML methods for SOC estimations.

Reference	Method	Datasets	Contributions	RMSE% or MAE%
[49]	Data-driven approach, BPNN	Voltage, current, and temperature	Time-varying distribution data set is done	2.18%
[50]	FFNN + IS	Voltage	Online applications are done using FFNN and reconstitution of charge cure with the IS	< 5%
[55]	ELM	Voltage, current, and temperature	Speed and accuracy of the system is enhanced by health indicators monitoring	< 2.5%
[56]	PSO-LSSVR	Voltage, current, and temperature	Dynamic performance enhanced by using SVM considering the OCV parameters	2.3%
[57]	Genetic algorithm–based fuzzy C-means clustering technique and back propagation algorithm used for optimization	Voltage, current, and temperature	Topology is modeled using subtractive clustering and direct search algorithm, and optimized using the back propagation system	< 2.5%
[58]	Random forest	Voltage and current	Estimation done using various cycling conditions	< 1.3%
[59]	GPR	Voltage and current	Advanced gaussian filter to smooth the capacity curves; mean and covariance functions are used to predict the state of health of the battery	3%
[61]	BPNN	—	It shows efficient performance at different loads and temperature conditions without the modeling parameters	0.59%
[62]	DWT-based WNN	Voltage, current, and temperature	Analysis is done in the frequency domain for various temperatures of the battery	3.58% (0°C),2.23% (25°C),1.75% (45°C),0.59%1.10%
[51]	DNN	Voltage, current, and temperature	Cost and time consumption reduced by self-learning of weights	1.10%

2.1.11. Neuro Fuzzy Algorithm–Based Approach

The authors in the paper [63] used the subtractive clustering algorithm along with the neuro fuzzy system for 10 different drive cycles of charge cycles, which outperforms the conventional neural network and Elman neural networks.

2.1.12. GA-Based Approach

In the paper [57], fuzzy C means the genetic algorithm is used to reduce the SOC error by recursive least square algorithm, and back propagation is used to reduce the feedback error.

2.2. Deep Learning–Based SOC Estimation

Different approaches of the deep learning method for SOC estimation in EV is depicted in Figure 4.

[figure(s) omitted; refer to PDF]

2.2.1. LSTM-Based Approach

The authors in the paper [64] proposed LSTM, FNN, and CNN algorithms for the estimation of battery lifetime where LSTM outperforms and produces less RMSE. The authors also concluded that sparsely sampled signals are a better way to consider estimation rather than densely sampled battery signals. The authors in the paper [65] propose LSTM deep learning algorithm for predicting the states of different batteries to obtain the end-of-life and RUL of batteries. The variational inference method is used for three types of batteries: one for source and the other two for targets. It shows an MAE of 0.03% error. However, the large model size and high computational cost are the drawbacks of the system. The storing and training of internal parameters reduces the capacity of the system.

2.2.2. CNN-Based Approach

The convolution neural network for online estimation of battery estimation is done by the paper [66]. Using voltage, current, and temperature, the estimation is done where the RMSE is reduced to approximately 1.7%. The fixed-size input matrix is the limitation of the proposed work.

2.2.3. Transfer Learning and Ensemble Learning–Based Approach

Transfer learning [67] is a method that uses predetermined data for learning the current processes. The ensemble learning [68] method combines all the weak learners to form a strong learning algorithm. The authors in the paper [69] use both the ensemble and training algorithm with the DCN deep learning algorithm using the partial curve method.

2.2.4. DCN-Based Approach

The discrete convolution network–ensemble and transfer learning (DCN-ETL) showed better performance compared to DCN, discrete convolution network–transfer learning DCN-TL [70], and discrete convolution network–ensemble learning DCN-EL in terms of RMSE error percentage. The training period for DCN-ETL is high compared to the DCN-EL and DCN-TL. The amount of data sets is also comparatively high.

2.2.5. Orthogonal Least Squares Algorithm–Based Approach

The RBF neural network–based SOC estimation of lithium ion phosphate (LFP) battery along with the OLS (orthogonal least squares) algorithm is proposed in the paper [71] where it uses three input neurons for terminal voltage, discharge current, and temperature of the battery. It uses 24 hidden neurons based on the OLS algorithm. Nearly 24 centers are chosen by the OLS algorithm depending on the input voltage, current, and temperature values.

2.2.6. GRU RNN-Based Approach

The paper [72] shows the use of self-learned network parameters using the GRU of the deep learning algorithm. The authors established the nonlinear mapping relation between the observable variables and SOC. The same sets of variables are adopted for different temperatures. Table 4 indicates the deep learning methods for SOC estimations.

Table 4

Deep learning methods for SOC estimation.

Reference	Method	Datasets	Contributions	RMSE% or MAE%
[66]	CNN	Voltage, current, and temperature	Online estimation using real-time parameters	1.7%
[69]	DCN-ETL	Voltage and current	Partial curve method employed by training using ETL	1.503%
[64]	LSTM, FNN, and CNN	Voltage, current, and temperature	SOC estimation done using sparsely sampled signals instead of densely sampled signals
[71]	RBFNN + OLS	Voltage, current, and temperature	Three input were given as input selection of 24 centers taken place	0.55%
[65]	LSTM + TL	Voltage and current	SOC estimation done using sparsely sampled signals instead of densely sampled signals	0.03%
[72]	GRU RNN	Voltage, current, and temperature	Nonlinear estimation done between the SOC and input variables at different temperatures	1.05%

2.3. Machine Learning–Based Charge Equalization

Charge equalization is required to protect the lithium-ion batteries of EV from overcharging effects and to maintain the battery’s capacity and voltage at the same state (refer Figure 5). Both the shunt and series battery voltage and capacity need to be monitored. The influential factors of charge equalization are aging, ambient temperature variation, change in internal resistance and capacitance of the battery, effectiveness of the battery, formation of sulfate crystal because of overcharging, and lifespan. There are two types of equalization techniques, namely, rule-based and optimization-based algorithms. Table 5 shows the existing charge equalization techniques for EV.

[figure(s) omitted; refer to PDF]

Table 5

Existing charge equalization technique in EV.

Reference	Contributions
[73]	Rule-based algorithms	Buck-boost equalization circuit is used for equalizing the Li-ion batteries connected in a series manner with the help of voltage and SOC variables
[74]		The inconsistency in series connected battery cells using FLC-based equalization with a two-stage bidirectional equalizer circuit
[75]		An adaptive neuro-fuzzy algorithm is used to design individual cell equalizers which is executed in 0.13 μm CMOS technology
[76]		To minimize the SOC imbalance and energy dissipation of the Li-ion battery pack, the model predictive control (MPC) method is used and shows SOC of 4% and energy dissipation of 7%
[77]		Intelligent control scheme FL-NN is used for battery equalization with a two-stage DC/DC converter topology and static SOC showing good performance in terms of energy and time
[78]	Optimization algorithms	SOC-based equalization using the PSO technique for lithium-ion batteries connected in series which showed a 13.2% increase in the capacity
[79]		Ant colony optimization algorithm for series connected battery cells for equalization process which showed minimum energy loss, high balancing efficiency, and high speed
[77]		Fuzzy logic control integrated with GA optimization for series connected lithium-ion batteries provides 93% efficiency on the basis of equalization strategy when compared with the mean difference algorithm

Machine learning–based cell balancing techniques are used for the equalization of series and parallel-connected lithium-ion battery cells in EVs. Table 6 summarizes the ML algorithms proposed for cell balancing techniques in EVs.

Table 6

ML-based cell balancing techniques for BMS in EV.

Reference	Contributions	Inference
[80]	Distance-based outlier detection algorithm with online clustering technique and bleeding circuits used for equalization	Battery pack increased by 9.5%
[81]	FNN, CNN, and LSTM are used for equalization and estimation of multichannel charging profiles of the Li-ion batteries	The proposed method yields 25%–58% improvement in terms of mean absolute percentage error
[82]	Particle swarm optimization–based feed forward neural network for battery aging estimate	On average, the RMSE obtained is 0.0019
[83]	LSTM + GPR for RUL prediction for multiple Li-ion batteries	The LSTM + GPR method showed a 1.8% error, which is less compared to the LSTM and GPR methods
[84]	Model-free reinforcement learning to control hybrid battery management systems (HBMS) using super capacitors	(i) 59% reduction in SOC deviation(ii) 22% reduction in temperature variation(iii) 4.4% reduction in RMS current in most aged cells
[85]	BPNN, RBNN, LSTM are used for estimating optimum resistor values for variable resistors for cell balancing in Li-ion battery packs	(i) RMSE of BPNN is 0.176(ii) RMSE of RBNN is 0.2677(iii) RMSE of LSTM is 0.095
[86]	Redundant smart battery topology for series connected battery cells and k-nearest algorithm used for balancing SOC and temperature variations	(i) Maximum temperature spread is reduced(ii) Maximize the lifetime of the battery pack
[87]	Comparative study of cell balancing using top-balancing algorithm	Passive balancing is simpler and implementation cost is low, but thermal management system needs high cost

2.3.1. Outlier Detection Algorithm

The authors in the paper [80] proposed a distance-based outlier detection algorithm to find the unbalanced cells using voltage and SOC parameters. The online clustering technique classifies the balanced and unbalanced cells where the unbalanced cells were equalized with bleeding circuits.

2.3.2. FNN, LSTM, and CNN Based Approach

In the paper [81], the authors used three deep learning algorithms, FNN, CNN, and LSTM, for calculating the estimation of charging profiles and capacity of the system where three steps are present: (i) preprocessing where the abnormal cells are removed, (ii) training sets used for cleansing and min-max normalization, and (iii) estimating voltage, current, and temperature parameters. It shows 58%, 46%, and 25% improvement, respectively, for FNN, CNN, and LSTM methods. The aging of Li-ion batteries is estimated using NASA data sets of B0005, B0006, and B0007 with the help of the PSO technique and biased using the FNN technique. The results were compared with the FNN technique, which showed 0.0169, 0.0271, and 0.0138 of RMSE values.

2.3.3. Data-Driven Approach

The data-driven approach is proposed in the paper [83] in which LSTM and GPR together combine to predict the uncertain capacity of the batteries in EVs. LSTM helps to find the long-term degradation of the capacitor as the GPR technique predicts the uncertainties of the capacity regenerations.

2.3.4. Reinforcement Learning Approach

The authors in the paper [84] used reinforcement learning to control HBMS. A quasi-static model was developed for power convertors which improved more than 3000-fold the real-time simulation factor of the HBMS and accelerated the RL training. It reduced SOC by 59%, temperature deviations by 22%, and RMS current in most aged cells by 4.4%.

2.3.5. LSTM-Based Approach

A passive cell balancing system uses variable resistors according to the optimum resistor value obtained from the BPNN, RBNN, and LSTM applied to the Li-ion battery packs by considering the balancing time, power loss, temperature rise, and operating constraints proposed in the paper [85]. All three ML algorithms outperform the conventional system, but the LSTM shows better results compared to BPNN and RBNN.

2.3.6. Multidimensional K-Clustering Algorithm

Multidimensional K-Nearest neighboring algorithm [86] is proposed for optimal balancing to limit the temperature spread and SOC estimation in battery packs. Either SOC or temperature is selected based on the weighting factors provided. Maximum temperature spread and lifetime are achieved. A comparative study of cell-balancing algorithms is made in the paper [87] where the passive balancing of series connected battery cells shows better performance and needs more thermal stability. Table 6 cataloged the ML-based cell balancing methods for BMS in EV.

2.4. Machine Learning–Based Fault Detection and Diagnosis in BMS

The reliability of the EV mainly depends on the battery packs used. The battery used may undergo many variations during the processes where the probability of fault occurrence is a major concern. Therefore, fault detection and diagnosis is the method used to find the fault occurrence and provide a solution for the diagnosis. There are three kinds of FDD in the literature given in Figure 6 and also recorded in Table 7.

[figure(s) omitted; refer to PDF]

Table 7

ML algorithms for fault detection and diagnosis.

Reference	Proposed work	Inference
[88]	Rule-based method used to diagnose the overcharging leads to a rise in battery temperature	Real-time execution takes place, which detects faults and provides warning to the system
[89]	Two dimensional wavelet transform–based filtering is used for sensor data fault diagnosis followed by KNN algorithm for classification	Improved CCR (correct classification rate)Low FA (false alarm)Low MD (missed detection rate)
[90]	Fuzzy logic–based residual evaluation algorithm	Proposed work efficiently detects the variation in the battery using SOC, temperature, and voltage residuals
[91]	Big data statistical models using the local outlier factor (LOF) algorithm and clustering outlier detection algorithm	Fault detection accuracy is improved based on seasons
[92]	Model parameters are identified using genetic algorithm and fault detection is done using pretrained RF algorithm	Less computation with improved fault detection and also leakage is diagnosed
[93]	LSTM deep learning model for multiforward step voltage prediction	Fast online prediction of battery voltages of 3s
[94]	LSTM	(i) Accuracy for voltage sag–96.97%(ii) Accuracy for voltage swell–97.05%(iii) Accuracy for voltage sag line to ground–98.23%(iv) Accuracy for line to line–99.37%
[95]	Realistic deep learning framework for EV LIB anomaly detection	Expected fault and inspection cost reduced by 33%–50%

2.4.1. Rule-Based Algorithm

Machine learning algorithms are used for FDD in BMS of EVs for real-time processing. The rule-based algorithm is used for battery packs in the paper [88] to find the faults and give a warning to the system to reduce the overcharging and to maintain the battery temperature.

2.4.2. Wavelet Transform–Based Approach

The two-dimensional wavelet transform is applied to the data collected from the sensors [89] which uses filters to focus only on areas with a wealth of information, and the KNN algorithm was used for classification. The applied wavelet transform reduces the complexity of the classifiers and reduces the false alarm rate and missed detection rate too.

2.4.3. Fuzzy Logic Algorithm

The authors in the paper [90] propose the fuzzy logic algorithm for residual evaluation using residual voltage, temperature, and current which uses hybrid pulse power characterization for the output description.

2.4.4. Clustering Outlier Algorithm

The 3σ multilevel screening technology using big data analysis and LOF and clustering outlier detection algorithm for fault detection is proposed in the paper [91], and it is applied to the EV which was monitored for four seasons. It is noted that the winter season shows more fault detection and summer shows less and fault detection is done accordingly.

2.4.5. GA Optimization–Based Approach

The RC model is used for the ESC fault process using the GA optimization technique in the paper [92] using 15 cells and 3 different temperature conditions. The RF algorithm is employed to classify the leakage and nonleakage conditions.

2.4.6. LSTM-Based Approach

Real-world data obtained from SMC-EV in Beijing were utilized for multistep forward voltage prediction using the LSTM model proposed in the paper [93]. This model demonstrates excellent training ability and higher prediction accuracy, achieved within a reduced online time period of 3 s. The LSTM deep learning method is used to detect and diagnose the short circuit and open circuit faults using extracted features from the sensor data of EV battery packs.

2.4.7. Anomaly Detection Algorithm

It shows better performance compared to Kalman filter and fuzzy expert system, extreme ML and PSO, wavelet packet transform and GA, FFT, ANN, and SVM. The paper [95] proposes a real-time anomaly detection algorithm for Li-ion batteries using existing BMS data, which results in a 33%–50% reduction in failure rates and inspection costs.

2.4.8. Machine Learning–Based Thermal Management System

Because of overcharging and discharging, the excess electrons get deposited on the plates of Li-ion; therefore, when the minority carriers start to flow, there will be more heat dissipation. Here, both excess and less heat lead to changes in the reaction rate of the system, hence the battery temperature needs more attention. The battery temperature is directly proportional to the battery power and battery cycle life. The temperature range between 20°C and 40°C is the desired range for good battery cycle life. There are four functions [96] required for proper thermal management of the battery packs (refer Figure 7).

[figure(s) omitted; refer to PDF]

2.4.9. Mamdani-Type Fuzzy Algorithm

The paper [97] focuses on the SOC and thermal imbalances of the battery packs; it used fuzzy logic algorithm to six battery banks containing three lithium-ion batteries that are connected in series. The cell switching topology is used to switch between charging and discharging actions, and the proposed algorithm was applied. The input parameters of the fuzzy logic algorithms were SOC and temperature parameters, and a Mamdani-type fuzzification algorithm was used. The proposed work was observed during 10 Ah of 30,000 s where the temperature lies between 20°C and 40°C.

2.4.10. ROM Approach

The authors in the paper used the reduced order model (ROM) to capture both the cell core and temperature on the surface of the cell core and developed a module-based observer to estimate cell core temperature in the paper [98]. The proposed work reduces the temperature nonuniformity and temperature fluctuations over time. The amount of cooling was also reduced to 38%.

2.4.11. MPC-Based Approach

The LQ-MPC-based MLC modular battery with parametric imbalance is proposed in the paper [99] where it uses orthogonal decomposition for two components—voltage and balance control. The temperature and SOC balancing was accomplished in the first step of MPC itself with a voltage error of less than 3.2% and a slight resistance variation between 25°C and 40°C that filters to provide the concentration only on the information-rich regions. The KNN algorithm was used for classification.

2.4.12. Particle Swarm Optimization–Based Approach

In the paper [100], the particle swarm optimization algorithm is incorporated with the fuzzy logic algorithm for thermal management where the PSO parameters like N, T, D, and space range quantities are set and then MAE was calculated. The obtained error and change in error values were given as input to the fuzzy logic cooling controller. The settling time and overshoot were reduced both in heating and cooling conditions, as discussed in Table 8.

Table 8

ML-based BTMS.

Reference	Proposed work	Inference
[97]	Fuzzy logic–based battery bank model system	(i) Proposed work has self-healing capabilities which maintain SOC and thermal imbalances(ii) It maintains a temperature range between 20°C and 40°C for both charging and discharging cycles
[98]	Reduced order model and observer-based controller for the cooling system	(i) Temperature nonuniformity reduced from 4.2 to 1°C(ii) The amount of cooling consumed was reduced by 38%
[99]	Electrothermal control of a multilevel controller (MLC)-based modular battery using LQ-MPC	(i) Terminal voltage error < 3.2%(ii) Small resistance variation between 25°C and 40°C
[100]	PSO-based fuzzy logic controller	Type settling	Time	Overshoot
		Heating	32 min 30 s	0.497%
		Cooling	28 min 46 s	0.975%
[101]	Combination of GA and PSO with FLC	Easy method for multiobjective optimization and obtained at 0.9 normalization value
[102]	EVTMS-based control strategy using FLC	Battery lifetime-loss reduced by 3.11%–3.76% with unaffected temperatures
[103]	ANN supervised learning algorithm	Power consumption reduced by 48.5% at nominal temperature range
[104]	SVR model–based automatic calibration model for BTMS-based HPACS model	Cooling capacity improved by 2.1%, and system COP improved by 2.8%
[105]	GPR-based U-shaped lightweight liquid cooling method	(i) Decrease in temperature of the battery cell by 21%(ii) Weight of the cooling plate reduced by 45%
[106, 107]	Bayesian ridge, XG Boost, and SVR	It proves that knowing a single temperature helps to find other temperature variations in the system

2.4.13. GA and PSO Based Approach

Multiobjective functions like Nusselt number, friction coefficient, and maximum temperature values were optimized in Li-ion batteries using GA and PSO algorithm with fuzzy logic controller using operating parameters, namely, Reynolds number, conduction-convection parameter, aspect ratio, and cell spacing in the paper [101]. It performs well with safe operating temperature and multiobjective optimization functions with a normalized value of 0.9.

2.4.14. EVTMS Fuzzy–Based Approach

The EVTMS fuzzy–based control strategy was introduced in the paper [102], comprising three segments: cabin, heating coolant circuit, and battery coolant circuit in order to regulate the PTC heater power according to the battery current. The cabin and battery temperature are used as parameters used for constraint conditions. The proposed work reduced the loss of battery lifetime by 3.28% at −5°C ambient temperature.

2.4.15. ANN-Based Approach

The ANN model was incorporated with the TM system in the paper [105]. Here, the trained sets obtained from EV vehicles during normal conditions are used in the interconnected ANN systems. The proposed method collects the data online from various driving environments. The power consumption reduced by 48.6% using the ANN model, and overall integrated performance reduced by 6.9% at the nominal temperature range.

2.4.16. SVR-Based Approach

The paper [104] proposed liquid cooled BTMS-based HPACS using support vector regression (SVR) algorithm at different compressor speeds, ambient temperatures, and air flow rates of external heat exchanger. The end-to-end automatic calibration of this system demonstrates better performance in terms of cooling capacity, system COP, and coolant temperature of the battery inlet.

In this research, the Bayesian Ridge, XG Boost, and SVR were utilized to optimize heat management in the battery’s cooling system, and the ARIMA algorithm was used to predict Temperature 2 using the air inlet velocity at 2 m/s and at 300 k temperature [107]. It proves that understanding the characteristics of one-temperature formations is necessary to predict forthcoming scenarios.

2.4.17. GPR-Based Approach

A light weight, U-shaped cooling system is proposed in the paper [105] where the computation complexity was reduced by using GPR with Matern 5/2 kernel functions, and the temperature of the battery cell was reduced by increasing the fluid rate of the coolant, leading to larger temperature variations. The temperature of the cell was reduced by 21%, and the cooling plate weight was reduced by 45% using the proposed work.

3. Future Research Directions in AI-Driven EV BMSs

As the review enters the exciting realm of AI and deep learning for EV BMS, several avenues for future research emerge. These areas have immense potential to advance EV technology and address critical challenges:

1. Adaptive Learning Algorithms: Investigate adaptive algorithms that dynamically adjust BMS parameters based on real-time driving conditions, battery health, and user behavior. Such algorithms could enhance efficiency and extend battery life [108, 109].

2. Edge Computing and Real-Time Decision-Making: Explore edge computing solutions to enable real-time decision-making within the BMS. This involves processing data directly at the sensor level, reducing latency, and improving responsiveness.

3. Multimodal Data Fusion: Combine data from various sensors (voltage, current, temperature, etc.) to create a holistic view of battery health. Integrating visual and thermal imaging data could enhance fault detection accuracy.

4. Predictive Maintenance Models: Develop predictive models that anticipate battery failure or degradation. By analyzing historical data, these models could recommend timely maintenance actions, preventing costly breakdowns.

5. Energy Harvesting Integration: Investigate how AI can optimize energy harvesting (such as regenerative braking) and seamlessly integrate it into BMS strategies. This could enhance the overall energy efficiency.

6. Cybersecurity and Anomaly Detection: As BMS systems become more interconnected, robust cybersecurity measures are essential. Research anomaly detection techniques to identify potential cyber threats or unauthorized access.

7. Battery Swapping Optimization: Explore AI-driven algorithms for efficient battery swapping stations. Optimal battery selection based on charge cycles, health, and user preferences could revolutionize the EV infrastructure.

8. User-Centric BMS Interfaces: Design intuitive interfaces that empower EV owners with insights into battery health, charging patterns, and personalized recommendations. User-friendly BMS interfaces enhance user trust and engagement.

9. Environmental Impact Assessment: Investigate the environmental impact of BMS algorithms. Assess their energy consumption, carbon footprint, and overall sustainability.

10. Collaboration with Material Science: Foster interdisciplinary collaboration between AI researchers and material scientists. Innovations in battery materials (such as solid-state batteries) could benefit from AI-driven BMS optimization.

In conclusion, the future of EVs lies at the intersection of AI, deep learning, and sustainable mobility. By exploring these research directions, this review paves the way for greener, smarter, and more reliable electric transportation.

3.1. Implications

Figure 8 shows the comparison of machine learning algorithm and deep learning algorithm for four parameters of BMSs. From the survey done, it is seen that there are a greater number of articles of machine learning–based BMS available compared to deep learning algorithms. Figure 9 depicts the percentage of availability of research articles present in the IEEE, Elsevier, ScienceDirect, Wiley, and other journals.

[figure(s) omitted; refer to PDF]

4. Conclusion

This paper reviews various machine learning and deep learning methodologies used to improve BMS in EV. The papers were reviewed using Google Scholar, Elsevier, ScienceDirect, Wiley, IEEE, and Springer. The reviewing process focused specifically on machine learning, deep learning, and combined ML techniques. The integration of machine learning and deep learning techniques in BMS for EVs holds immense promise. By accurately estimating SOC, ensuring cell balancing, detecting faults early, and optimizing thermal management, these algorithms contribute significantly to the reliability and longevity. The collaboration between AI and EV technology will drive the entire transportation system toward a cleaner and more efficient automotive future.

Funding

No funding was received for this research work.

References

[1] "Transportation Services Market (By Purpose: Commuter Travel, Tourism and Leisure Travel, Business Travel, Cargo and Freight Travel, Shipping and Delivery Travel; By Destination: Domestic, International; By Type: Public Buses, Electric Buses, Subways, Taxis, Auto Rickshaws, Ferries, Other Public Transport Vehicles)-Global Industry Analysis, Size, Share, Growth, Trends, Regional Outlook, and Forecast 2023-2032," Precedence Research, vol. 11, 2023.

[2] P. Gireesh Kumar, P. Lekhana, M. Tejaswi, S. Chandrakala, "Effects of Vehicular Emissions on the Urban Environment-a State of the Art," Materials Today: Proceedings, vol. 45 no. Part 7, pp. 6314-6320, DOI: 10.1016/j.matpr.2020.10.739, 2021.

[3] Transport, "Chapter 10 in Climate Change 2022: Mitigation of Climate Change," 2022.

[4] M. Guarnieri, "Looking Back to Electric Cars," 2012 Third IEEE HISTory of ELectro-Technology CONference (HISTELCON),DOI: 10.1109/HISTELCON.2012.6487583, .

[5] T. Quiroga, Driving the Future, 2009.

[6] G. Maggetto, J. Van Mierlo, "Electric and Electric Hybrid Vehicle Technology: a Survey," IEE Seminar Electric, Hybrid and Fuel Cell Vehicles (Ref. No. 2000/050), pp. 1/1-111, DOI: 10.1049/ic:20000261, .

[7] K. Jorgensen, "Technologies for Electric, Hybrid and Hydrogen Vehicles: Electricity from Renewable Energy Sources in Transport," Utilities Policy, vol. 16 no. 2, pp. 72-79, DOI: 10.1016/j.jup.2007.11.005, 2008.

[8] D. B. Richardson, "Electric Vehicles and the Electric Grid: A Review of Modeling Approaches, Impacts, and Renewable Energy Integration," Renewable and Sustainable Energy Reviews, vol. 19, pp. 247-254, DOI: 10.1016/j.rser.2012.11.042, 2013.

[9] J. Y. Yong, V. K. Ramachandaramurthy, K. M. Tan, N. Mithulananthan, "A Review on the State-Of-The-Art Technologies of Electric Vehicle, its Impacts and Prospects," Renewable and Sustainable Energy Reviews, vol. 49, pp. 365-385, DOI: 10.1016/j.rser.2015.04.130, 2015.

[10] L. Liu, F. Kong, X. Liu, Y. Peng, Q. Wang, "A Review on Electric Vehicles Interacting with Renewable Energy in Smart Grid," Renewable and Sustainable Energy Reviews, vol. 51, pp. 648-661, DOI: 10.1016/j.rser.2015.06.036, 2015.

[11] F. Zhang, X. Zhang, M. Zhang, A. S. E. Edmonds, "Literature Review of Electric Vehicle Technology and its Applications," 2016 5th International Conference on Computer Science and Network Technology (ICCSNT), pp. 832-837, DOI: 10.1109/ICCSNT.2016.8070276, .

[12] J. A. Sanguesa, V. Torres-Sanz, P. Garrido, F. J. Martinez, J. M. Marquez-Barja, "A Review on Electric Vehicles: Technologies and Challenges," Smart Cities, vol. 4 no. 1, pp. 372-404, DOI: 10.3390/smartcities4010022, 2021.

[13] M. S. Hossain, L. Kumar, M. Islam, J. Selvaraj, "A Comprehensive Review on the Integration of Electric Vehicles for Sustainable Development," Journal of Advanced Transportation, vol. 2022,DOI: 10.1155/2022/3868388, 2022.

[14] M. Bharathidasan, V. Indragandhi, V. Suresh, M. Jasiński, Z. Leonowicz, "A Review on Electric Vehicle: Technologies, Energy Trading, and Cyber Security," Energy Reports, vol. 8, pp. 9662-9685, DOI: 10.1016/j.egyr.2022.07.145, 2022.

[15] M. S. Bin Ahmad, A. Pesyridis, P. Sphicas, A. Mahmoudzadeh Andwari, A. Gharehghani, B. M. Vaglieco, "Electric Vehicle Modelling for Future Technology and Market Penetration Analysis," Frontiers of Mechanical Engineering, vol. 8,DOI: 10.3389/fmech.2022.896547, 2022.

[16] A. El, B. A. El Halim, Ahmed, E. Bayoumi, W. El-Khattam, A. Ibrahim, "EVs: a Review of Their Components and Technologies," International Journal of Power Electronics and Drive Systems, vol. 13, pp. 2041-2061, DOI: 10.11591/ijpeds.v13.i4, 2022.

[17] M. Brandl, "Batteries and Battery Management Systems for EVs," 2012 Design, Automation & Test in Europe Conference & Exhibition (DATE), pp. 971-976, DOI: 10.1109/DATE.2012.6176637, .

[18] L. Lu, X. Han, L. Jianqiu, J. Hua, M. Ouyang, "A Review on the Key Issues for Lithium-Ion Battery Management in EVs," Journal of Power Sources, vol. 226, pp. 272-288, DOI: 10.1016/j.jpowsour.2012.10.060, 2013.

[19] M. U. Ali, A. Zafar, S. H. Nengroo, S. Hussain, M. Junaid Alvi, H.-J. Kim, "Towards a Smarter Battery Management System for Electric Vehicle Applications: A Critical Review of Lithium-Ion Battery State of Charge Estimation," Energies, vol. 12 no. 3,DOI: 10.3390/en12030446, 2019.

[20] H. Dai, B. Jiang, X. Hu, X. Lin, X. Wei, M. Pecht, "Advanced Battery Management Strategies for a Sustainable Energy Future: Multilayer Design Concepts and Research Trends," Renewable and Sustainable Energy Reviews, vol. 138,DOI: 10.1016/j.rser.2020.110480, 2021.

[21] "ISO 6469-1 Electrically Propelled Road Vehicles-Safety Specifications-Part 1: Rechargeable Energy Storage System (RESS)," 2019. https://www.iso.org/standard/68665.html

[22] "ISO 6469-2 Electrically Propelled Road Vehicles-Safety Specifications-Part 2: Vehicle Operational Safety," 2018. https://www.iso.org/standard/81346.html

[23] "ISO 6469-3 Electrically Propelled Road Vehicles-Safety Specifications-Part 3 Electrical Safety," 2018. https://www.iso.org/standard/81746.html

[24] "ISO 6469-4 Electrically Propelled Road Vehicles-Safety Specifications-Part 4: Post Crash Electrical Safety," 2015. https://www.iso.org/standard/60584.html

[25] "ISO 12405-4 Electrically Propelled Road Vehicles-Test Specification for Lithiumion Traction Battery Packs and Systems-Part 4: Performance Testing," 2018. https://www.iso.org/standard/71407.html

[26] "IEC 62660-1 Secondary Lithium-Ion Cells for the Propulsion of Electric Road Vehicles-Part 1: Performance Testing," 2018. https://webstore.iec.ch/en/publication/28965

[27] "IEC 62660-2 Secondary Lithium-Ion Cells for the Propulsion of Electric Road Vehicles-Part 2: Reliability and Abuse Testing," 2018. https://webstore.iec.ch/en/publication/27387

[28] "IEC 62660-3 Secondary Lithium-Ion Cells for the Propulsion of Electric Road Vehicles-Part 3: Safety Requirements," 2016. https://webstore.iec.ch/en/publication/65084

[29] "SAE J1766 Recommended Practice for Electric. Fuel Cell and Hybrid EV Crash Integrity Testing," 2014. https://www.sae.org/standards/content/j1766_201401/

[30] "SAE J2464 Electric and Hybrid EV Rechargeable Energy Storage System (RESS) Safety and Abuse Testing," 2009. https://www.sae.org/standards/content/j2464_202108/

[31] "SAE J2929 Safety Standard for Electric and Hybrid Vehicle Propulsion Battery Systems Utilizing Lithium-Based Rechargeable Cells," 2013. https://www.sae.org/standards/content/j2929/

[32] "EN 1987-1 Electrically Propelled Road Vehicles-Specific Requirements for Safety--Part 1: On-Board Energy Storage," 1997. https://batterystandards.info/standard/en-1987-1

[33] "EN 1987-2 Electrically Propelled Road Vehicles-Specific Requirements for Safety-Part 2: Functional Safety Means and Protection against Failures," 1998. https://standards.iteh.ai/catalog/standards/cen/758c6238-3fb1-4e3f-b5c5-58063b9dd406/en-1987-2-1997?srsltid=AfmBOoqeZVAO8YESdCQCQ9Pj4wdsDg3YSKTwEvntWd43mLSLOD1AAXKl

[34] "EN 1987-3 Electrically Propelled Road Vehicles-Specific Requirements for Safety-Part 3: Protection of Users against Electrical Hazards," 1998.

[35] "DOE/ID-11069 FreedomCAR Battery Test Manual for Power-Assist Hybrid EVs," 2003. https://standards.iteh.ai/catalog/standards/cen/ac8ce30c-0383-4600-a197-4cdf73a6621f/en-1987-3-1998?srsltid=AfmBOorJIjXj7_Ih5YGy0WzoVFgqvybGFzPTnWv6bThXkybGj1nuzr7q

[36] "UL 2580 Batteries for Use in EVs," 2020. https://explorestandards.ul.org/batteries-in-ev.html

[37] "GB/T 31484 Cycle Life Requirements and Test Methods for Traction Battery of EV," 2015. https://www.chinesestandard.net/PDF.aspx/GBT31484-2015

[38] "GB/T 31485 Safety Requirements and Test Methods for Traction Battery of EV," 2015. https://www.chinesestandard.net/PDF.aspx/GBT31485-2015

[39] "GB/T 31486 Electrical Performance Requirements and Test Methods for Traction Battery of EV," 2015. https://www.chinesestandard.net/PDF.aspx/GBT31486-2024

[40] "GB/T 31467.1 Lithium-Ion Traction Battery Pack and System for EVs-Part 1: Test Specification for High Power Applications," 2015. https://GB/T31467.1Lithium-iontractionbatterypackandsystemforEVs-Part1:testspecificationforhighpowerapplications

[41] "GB/T 31467.2 Lithium-Ion Traction Battery Pack and System for EVs-Part 2: Test Specification for High Energy Applications," 2015. https://www.chinesestandard.net/PDF.aspx/GBT31467.2-2015

[42] "GB/T 31467.3 Lithium-Ion Traction Battery Pack and System for EVs-Part 3: Safety Requirements and Test Methods," 2015. https://www.chinesestandard.net/PDF.aspx/GBT31467.3-2015

[43] "GB/T 38661 Technical Specifications of Battery Management System for EVs," 2020. https://www.chinesestandard.net/PDF.aspx/GBT38661-2020

[44] V. Karkuzali, P. Rangarajan, V. Tamilselvi, P. Kavitha, "Analysis of Battery Management System Issues in EVs," IOP Conference Series: Materials Science and Engineering, vol. 994,DOI: 10.1088/1757-899X/994/1/012013, 2020.

[45] M. A. Hannan, M. M. Hoque, A. Hussain, Y. Yusof, P. J. Ker, "State-of-the-Art and Energy Management System of Lithium-Ion Batteries in Electric Vehicle Applications: Issues and Recommendations," IEEE Access, vol. 6, pp. 19362-19378, DOI: 10.1109/ACCESS.2018.2817655, 2018.

[46] M. S. Hossain Lipu, M. A. Hannan, T. F. Karim, "Intelligent Algorithms and Control Strategies for Battery Management System in Electric Vehicles: Progress, Challenges and Future Outlook," Journal of Cleaner Production, vol. 292,DOI: 10.1016/j.jclepro.2021.126044, 2021.

[47] W. Liu, T. Placke, K. T. Chau, "Overview of Batteries and Battery Management for Electric Vehicles," Energy Reports, vol. 8, pp. 4058-4084, DOI: 10.1016/j.egyr.2022.03.016, 2022.

[48] A. K. M. A. Habib, M. K. Hasan, G. F. Issa, D. Singh, S. Islam, T. M. Ghazal, "Lithium-Ion Battery Management System for Electric Vehicles: Constraints, Challenges, and Recommendations," Batteries, vol. 9 no. 3,DOI: 10.3390/batteries9030152, 2023.

[49] G. You, S. Park, D. Oh, "Real-time State-Of-Health Estimation for Electric Vehicle Batteries: A Data-Driven Approach," Applied Energy, vol. 176, pp. 92-103, DOI: 10.1016/j.apenergy.2016.05.051, 2016.

[50] J. Wu, C. Zhang, Z. Chen, "An Online Method for Lithium-Ion Battery Remaining Useful Life Estimation Using Importance Sampling and Neural Networks," Applied Energy, vol. 173, pp. 134-140, DOI: 10.1016/j.apenergy.2016.04.057, 2016.

[51] E. Chemali, P. J. Kollmeyer, M. Preindl, A. Emadi, "State-of-charge Estimation of Li-Ion Batteries Using Deep Neural Networks: A Machine Learning Approach," Journal of Power Sources, vol. 400, pp. 242-255, DOI: 10.1016/j.jpowsour.2018.06.104, 2018.

[52] S. Hong, H. Hwang, D. Kim, S. Cui, I. Joe, "Real Driving Cycle-Based State of Charge Prediction for EV Batteries Using Deep Learning Methods," Applied Sciences, vol. 11 no. 23,DOI: 10.3390/app112311285, 2021.

[53] D.-W. Chung, J.-H. Ko, K.-Y. Yoon, "State-of-charge Estimation of Lithium-Ion Batteries Using LSTM Deep Learning Method," Journal of Electrical Engineering & Technology, vol. 17 no. 3, pp. 1931-1945, DOI: 10.1007/s42835-021-00954-8, 2022.

[54] W. Lang, Y. Hu, C. Gong, X. Zhang, H. Xu, J. Deng, "Artificial Intelligence-Based Technique for Fault Detection and Diagnosis of EV Motors: A Review," IEEE Transactions on Transportation Electrification, vol. 8 no. 1, pp. 384-406, DOI: 10.1109/TTE.2021.3110318, 2022.

[55] H. Pan, Z. Lü, H. Wang, H. Wei, L. Chen, "Novel Battery State-Of-Health Online Estimation Method Using Multiple Health Indicators and an Extreme Learning Machine," Energy, vol. 160, pp. 466-477, DOI: 10.1016/j.energy.2018.06.220, 2018.

[56] D. Yang, Y. Wang, R. Pan, R. Chen, Z. Chen, "State-of-health Estimation for the Lithium-Ion Battery Based on Support Vector Regression," Applied Energy, vol. 227, pp. 273-283, DOI: 10.1016/j.apenergy.2017.08.096, 2018.

[57] X. Hu, S. E. Li, Y. Yang, "Advanced Machine Learning Approach for Lithium-Ion Battery State Estimation in Electric Vehicles," IEEE Transactions on Transportation Electrification, vol. 2 no. 2, pp. 140-149, DOI: 10.1109/TTE.2015.2512237, 2016.

[58] Y. Li, C. Zou, M. Berecibar, "Random Forest Regression for Online Capacity Estimation of Lithium-Ion Batteries," Applied Energy, vol. 232, pp. 197-210, DOI: 10.1016/j.apenergy.2018.09.182, 2018.

[59] X. Li, C. Yuan, X. Li, Z. Wang, "State of Health Estimation for Li-Ion Battery Using Incremental Capacity Analysis and Gaussian Process Regression," Energy, vol. 190,DOI: 10.1016/j.energy.2019.116467, 2020.

[60] I. Babaeiyazdi, A. Rezaei-Zare, S. Shokrzadeh, "State of Charge Prediction of EV Li-Ion Batteries Using EIS: A Machine Learning Approach," Energy, vol. 223,DOI: 10.1016/j.energy.2021.120116, 2021.

[61] M. A. Hannan, M. S. H. Lipu, A. Hussain, M. H. Saad, A. Ayob, A. Ayob, "Neural Network Approach for Estimating State of Charge of Lithium-Ion Battery Using Backtracking Search Algorithm," IEEE Access, vol. 6, pp. 10069-10079, DOI: 10.1109/ACCESS.2018.2797976, 2018.

[62] D. Cui, B. Xia, R. Zhang, "A Novel Intelligent Method for the State of Charge Estimation of Lithium-Ion Batteries Using a Discrete Wavelet Transform-Based Wavelet Neural Network," Energies, vol. 11 no. 4,DOI: 10.3390/en11040995, 2018.

[63] T. Zahid, K. Xu, W. Li, C. Li, H. Li, "State of Charge Estimation for Electric Vehicle Power Battery Using Advanced Machine Learning Algorithm under Diversified Drive Cycles," Energy, vol. 162, pp. 871-882, DOI: 10.1016/j.energy.2018.08.071, 2018.

[64] K. Kaur, A. Garg, X. Cui, S. Singh, B. K. Panigrahi, "Deep Learning Networks for Capacity Estimation for Monitoring SOH of Li‐ion Batteries for Electric Vehicles," International Journal of Energy Research, vol. 45 no. 2, pp. 3113-3128, DOI: 10.1002/er.6005, 2020.

[65] S. Kim, Y. Y. Choi, K. J. Kim, J.-I. Choi, "Forecasting State-Of-Health of Lithium-Ion Batteries Using Variational Long Short-Term Memory with Transfer Learning," Journal of Energy Storage, vol. 41,DOI: 10.1016/j.est.2021.102893, 2021.

[66] S. Shen, M. Sadoughi, X. Chen, M. Hong, C. Hu, "A Deep Learning Method for Online Capacity Estimation of Lithium-Ion Batteries," Journal of Energy Storage, vol. 25,DOI: 10.1016/j.est.2019.100817, 2019.

[67] S. Shen, M. K. Sadoughi, C. Hu, "Online Estimation of Lithium-Ion Battery Capacity Using Transfer Learning," 2019 IEEE Transportation Electrification Conference and Expo (ITEC), .

[68] G. Wen, Z. Hou, H. Li, D. Li, L. Jiang, E. Xun, "Ensemble of Deep Neural Networks with Probability-Based Fusion for Facial Expression Recognition," Cognitive Computation, vol. 9 no. 5, pp. 597-610, DOI: 10.1007/s12559-017-9472-6, 2017.

[69] S. Shen, M. Sadoughi, M. Li, Z. Wang, C. Hu, "Deep Convolutional Neural Networks with Ensemble Learning and Transfer Learning for Capacity Estimation of Lithium-Ion Batteries," Applied Energy, vol. 260,DOI: 10.1016/j.apenergy.2019.114296, 2020.

[70] Y. Li, K. Li, X. Liu, Y. Wang, L. Zhang, "Lithium-ion Battery Capacity Estimation—A Pruned Convolutional Neural Network Approach Assisted with Transfer Learning," Applied Energy, vol. 285 no. 2021,DOI: 10.1016/j.apenergy.2020.116410, 2021.

[71] W.-Y. Chang, "Estimation of the State of Charge for a LFP Battery Using a Hybrid Method that Combines a RBF Neural Network, an OLS Algorithm and AGA," International Journal of Electrical Power & Energy Systems, vol. 53, pp. 603-611, DOI: 10.1016/j.ijepes.2013.05.038, 2013.

[72] C. Li, F. Xiao, Y. Fan, "An Approach to State of Charge Estimation of Lithium-Ion Batteries Based on Recurrent Neural Networks with Gated Recurrent Unit," Energies, vol. 12 no. 9,DOI: 10.3390/en12091592, 2019.

[73] B. Wang, F. Qin, X. Zhao, X. Ni, D. Xuan, "Equalization of Series Connected Lithium-Ion Batteries Based on Back Propagation Neural Network and Fuzzy Logic Control," International Journal of Energy Research, vol. 44 no. 6, pp. 4812-4826, DOI: 10.1002/er.5274, 2020.

[74] Y. Ma, P. Duan, Y. Sun, H. Chen, "Equalization of Lithium-Ion Battery Pack Based on Fuzzy Logic Control in EV," IEEE Transactions on Industrial Electronics, vol. 65, 2018.

[75] N. Nguyen, S. K. Oruganti, K. Na, F. Bien, "An Adaptive Backward Control Battery Equalization System for Serially Connected Lithium-Ion Battery Packs," IEEE Transactions on Vehicular Technology, vol. 63 no. 8, pp. 3651-3660, DOI: 10.1109/tvt.2014.2304453, 2014.

[76] J. Liu, Y. Chen, H. K. Fathy, "Nonlinear Model-Predictive Optimal Control of an Active Cell-To-Cell Lithium-Ion Battery Pack Balancing Circuit," IFAC-PapersOnLine, vol. 50 no. 1, pp. 14483-14488, DOI: 10.1016/j.ifacol.2017.08.2297, 2017.

[77] S. Zhang, L. Yang, X. Zhao, J. Qiang, "A GA Optimization for Lithium-Ion Battery Equalization Based on SOC Estimation by NN and FLC," International Journal of Electrical Power & Energy Systems, vol. 73, pp. 318-328, DOI: 10.1016/j.ijepes.2015.05.018, 2015.

[78] Z. Liu, X. Liu, J. Han, W. Yang, "An Optimization Algorithm for Equalization Scheme of Series-Connected Energy Storage Cells," Proceedings of the IEEE Transportation Electrification Conference and Expo, Asia-Pacific (ITEC Asia-Pacific), .

[79] J. Sun, C. Zhu, R. Lu, K. Song, G. Wei, "Development of an Optimized Algorithm for Bidirectional Equalization in Lithium-Ion Batteries," Journal of Power Electronics, vol. 15 no. 3, pp. 775-785, DOI: 10.6113/jpe.2015.15.3.775, 2015.

[80] C. Piao, Z. Wang, J. Cao, W. Zhang, S. Lu, "Lithium-Ion Battery Cell-Balancing Algorithm for Battery Management System Based on Real-Time Outlier Detection," Mathematical Problems in Engineering, vol. 2015,DOI: 10.1155/2015/168529, 2015.

[81] Y. Choi, S. Ryu, K. Park, H. Kim, "Machine Learning-Based Lithium-Ion Battery Capacity Estimation Exploiting Multi-Channel Charging Profiles," IEEE Access, vol. 7, pp. 75143-75152, DOI: 10.1109/ACCESS.2019.2920932, 2019.

[82] N. Junhuathon, G. Sakulphaisan, K. Chayakulkheeree, "Li-ion Battery Aging Estimation Using Particle Swarm Optimization Based Feedforward Neural Network," 2020 International Conference on Power, Energy and Innovations (ICPEI), pp. 73-76, DOI: 10.1109/ICPEI49860.2020.9431432, .

[83] K. Liu, Y. Shang, Q. Ouyang, W. D. Widanage, "A Data-Driven Approach with Uncertainty Quantification for Predicting Future Capacities and Remaining Useful Life of Lithium-Ion Battery," IEEE Transactions on Industrial Electronics, vol. 68 no. 4, pp. 3170-3180, DOI: 10.1109/TIE.2020.2973876, 2021.

[84] J. Mirwald, R. de Castro, J. Brembeck, J. Ultsch, R. E. Araujo, "Learning-Based Control for Hybrid Battery Management Systems," Springer Optimization and Its Applications, pp. 187-222, DOI: 10.1007/978-3-030-84474-5_7, 2022.

[85] T. Duraisamy, D. Kaliyaperumal, "Machine Learning-Based Optimal Cell Balancing Mechanism for Electric Vehicle Battery Management System," IEEE Access, vol. 9, pp. 132846-132861, DOI: 10.1109/ACCESS.2021.3115255, 2021.

[86] R. Fonso, X. Sui, A. Acharya, R. Teodorescu, C. Cecati, "Multidimensional Machine Learning Balancing in Smart Battery Packs," IECON 2021–47th Annual Conference of the IEEE Industrial Electronics Society, .

[87] T. Duraisamy, D. Kaliyaperumal, "Evaluation and Comparative Study of Cell Balancing Methods for Lithium-Ion Batteries Used in EVs," International Journal of Renewable Energy Development, vol. 10, pp. 471-479, DOI: 10.14710/ijred.0.34484, 2021.

[88] J. Xiong, H. Banvait, L. Li, "Failure Detection for Over-discharged Li-Ion Batteries," Proceedings of the IEEE International EV Conference, .

[89] A. A. Silva, A. M. Bazzi, S. Gupta, "Fault Diagnosis in Electric Drives Using Machine Learning Approaches," Proceedings of the 2013 IEEE International Electric Machines and Drives Conference, IEMDC 2013, pp. 722-726, DOI: 10.1109/IEMDC.2013.6556173, .

[90] V. K. S. Muddappa, S. Anwar, "Electrochemical Model Based Fault Diagnosis of Li-Ion Battery Using Fuzzy Logic," .

[91] Y. Zhao, P. Liu, Z. Wang, L. Zhang, "Fault and Defect Diagnosis of Battery for Electric Vehicles Based on Big Data Analysis Methods," Applied Energy, vol. 207, pp. 354-362, DOI: 10.1016/j.apenergy.2017.05.139, 2017.

[92] R. Yang, R. Xiong, H. He, Z. Chen, "A Fractional-Order Model-Based Battery External Short Circuit Fault Diagnosis Approach for All-Climate Electric Vehicles Application," Journal of Cleaner Production, vol. 187, pp. 950-959, DOI: 10.1016/j.jclepro.2018.03.259, 2018.

[93] J. Hong, Z. Wang, Y. Yao, "Fault Prognosis of Battery System Based on Accurate Voltage Abnormity Prognosis Using Long Short-Term Memory Neural Networks," Applied Energy, vol. 251,DOI: 10.1016/j.apenergy.2019.113381, 2019.

[94] H. Kaplan, K. Tehrani, M. Jamshidi, "A Fault Diagnosis Design Based on Deep Learning Approach for Electric Vehicle Applications," Energies, vol. 14 no. 20,DOI: 10.3390/en14206599, 2021.

[95] J. Zhang, Y. Wang, B. Jiang, "Realistic Fault Detection of Li-Ion Battery via Dynamical Deep Learning," Nature Communications, vol. 14 no. 1,DOI: 10.1038/s41467-023-41226-5, 2023.

[96] A. A. Pesaran, "Battery Thermal Management in EVs and HEVs: Issues and Solutions," Advanced Automotive Battery Conference, vol. 10, 2001.

[97] T. Ma, B. Serrano, O. Mohammed, "Fuzzy Logic Based Power and Thermal Management System Design for Multi-Cell Lithium-Ion Battery Bank Protection and Operation," 2014 Clemson University Power Systems Conference, PSC 2014,DOI: 10.1109/PSC.2014.6808111, .

[98] F. He, L. Ma, "Thermal Management of Batteries Employing Active Temperature Control and Reciprocating Cooling Flow," International Journal of Heat and Mass Transfer, vol. 83, pp. 164-172, DOI: 10.1016/j.ijheatmasstransfer.2014.11.079, 2015.

[99] F. Altaf, B. Egardt, J. Mårdh, Lars, "Load Management of Modular Battery Using Model Predictive Control: Thermal and State-Of-Charge Balancing," IEEE Transactions on Control Systems Technology, vol. 25,DOI: 10.1109/TCST.2016.2547980, 2016.

[100] M. A. Hannan, Y. Young, M. M. Hoque, P. J. Ker, M. N. Uddin, "Lithium Ion Battery Thermal Management System Using Optimized Fuzzy Controller," 2019 IEEE Industry Applications Society Annual Meeting,DOI: 10.1109/IAS.2019.8912339, .

[101] A. Afzal, M. K. Ramis, "Multi-objective Optimization of Thermal Performance in Battery System Using Genetic and Particle Swarm Algorithm Combined with Fuzzy Logics," Journal of Energy Storage, vol. 32,DOI: 10.1016/j.est.2020.101815, 2020.

[102] H. Min, Z. Zhang, W. Sun, Z. Min, Y. Yu, B. Wang, "A Thermal Management System Control Strategy for Electric Vehicles under Low-Temperature Driving Conditions Considering Battery Lifetime," Applied Thermal Engineering, vol. 181,DOI: 10.1016/j.applthermaleng.2020.115944, 2020.

[103] J. Park, Y. Kim, "Supervised-Learning-Based Optimal Thermal Management in an Electric Vehicle," IEEE Access, vol. 8, pp. 1290-1302, DOI: 10.1109/ACCESS.2019.2961791, 2020.

[104] X. Tang, Q. Guo, M. Li, C. Wei, Z. Pan, Y. Wang, "Performance Analysis on Liquid-Cooled Battery Thermal Management for Electric Vehicles Based on Machine Learning," Journal of Power Sources, vol. 494,DOI: 10.1016/j.jpowsour.2021.229727, 2021.

[105] S. A. Khan, C. Eze, K. Dong, "Design of a New Optimized U-Shaped Lightweight Liquid-Cooled Battery Thermal Management System for Electric Vehicles: A Machine Learning Approach," International Communications in Heat and Mass Transfer, vol. 136,DOI: 10.1016/j.icheatmasstransfer.2022.106209, 2022.

[106] A. Li, J. Weng, A. C. Y. Yuen, "Machine Learning Assisted Advanced Battery Thermal Management System: A State-Of-The-Art Review," Journal of Energy Storage, vol. 60,DOI: 10.1016/j.est.2023.106688, 2023.

[107] M. Boujelbene, M. Goodarzi, M. Ashraf, "Machine-learning Optimization of an Innovative Design of a Li-Ion Battery Arrangement Cooling System," Journal of Energy Storage, vol. 58,DOI: 10.1016/j.est.2022.106331, 2023.

[108] Y. Khawaja, N. Shankar, I. Qiqieh, "Battery Management Solutions for Li-Ion Batteries Based on Artificial Intelligence," Ain Shams Engineering Journal, vol. 14 no. 12,DOI: 10.1016/j.asej.2023.102213, 2023.

[109] Y. Khawaja, I. Qiqieh, J. Alzubi, O. Alzubi, A. Allahham, D. Giaouris, "Design of Cost-Based Sizing and Energy Management Framework for Standalone Microgrid Using Reinforcement Learning," Solar Energy, vol. 251, pp. 249-260, DOI: 10.1016/j.solener.2023.01.027, 2023.