The DOI matches its Crossref record; 185 of the 186 references it cites were found in a registry or an index; 21 of the 116 sources its claims cite were checked for support.
Artificial Intelligence Approaches for SOH Estimation in Lithium-Ion Batteries: A Comprehensive Review of Progress, Critical Evaluations, Challenges, and Toward Future Outlooks
Open accessAbstract
In recent years, safety issues with electric vehicles (EVs) have grown. These EVs are powered by rechargeable lithium‐ion batteries (LIBs), which are key energy storage systems. The accurate estimation of these LIBs state of health (SOH) is crucial for reliable performance and reduction in safety issues. This estimation is challenging due to the complex electrochemical dynamics and nonlinear behavior of LIBs and variability in external environmental conditions. Traditional direct and model‐based frameworks significantly produce inaccurate results, resulting in increased interest in artificial intelligence hybrid and coestimation methods that combine multiple models to enhance precision and reliability. This review explores models and AI‐based methods, highlighting methodologies, implementation strategies, and evaluation criteria for SOH estimation. Furthermore, it identifies limitations in previous studies, including unclear classifications and inadequate evaluations, while examining essential operational factors such as battery parameters, model execution, and characteristics. Hybrid methods exhibit significant accuracy across various conditions, whereas coestimation techniques indicate the potential for simultaneous examination of SOH. Challenges including capacity degradation, model limitations, data integrity, and leading toward Green AI are explored and proposed, along with prospective advancements such as advanced algorithms, cloud‐based frameworks, and second‐life applications. This review presents insights for researchers and industries, facilitating the development of effective BMS frameworks to improve EV performance, battery longevity, and sustainability.
Overview
CoChat Summary
TL;DR
This review surveys AI-based state-of-health (SOH) estimation for lithium-ion batteries, covering direct, model-based, data-driven, hybrid, and coestimation methods, with attention to second-life batteries and Green AI. It concludes that AI and hybrid methods improve accuracy and adaptability but still face data, computational, generalization, and real-time BMS integration challenges.
Key findings
- Hybrid and coestimation methods combine model-based physical insights with data-driven learning, improving accuracy, robustness, and adaptability; examples include LSTM-KF, ECM-UPF, and LSTM-PF.
- Reported accuracies from cited studies include PINN SOH error reduced to less than 1%; internal-resistance measurement paired with ML yielding 97.3% SOH estimation accuracy; CNN voltage-pattern SOH prediction accuracy of 95.6%; transfer learning with small second-life data increasing accuracy by 15%; ultrasonic-assisted AI increasing accuracy by 20%; AI digital twins increasing SOH estimation accuracy by 20% in extremely hot settings; and deep domain adaptation increasing SOH estimation accuracy by 15% across temperatures.
- Degradation-related findings include SEI diffusion coefficient decreasing by 50% after 1000 cycles; NMC electrode diffusion coefficient falling by 40% after 500 cycles; silicon anode failure after 200 cycles when hoop stress exceeded 200 MPa at high lithium concentrations (c_avg > 0.8); and after 500 cycles, Qdischarge was 92% of Qcharge.
- Direct methods are simple but slow and less accurate; model-based methods are complex and require reparameterization; data-driven methods have high precision but rely heavily on datasets and hyperparameters and consume more energy; SLB estimation is hindered by variable histories, cell inconsistency, and unregulated operating conditions.
Methods
- Design: comprehensive literature review using three search engines—Google Scholar, Elsevier, and Web of Science—and four keywords: lithium-ion SOH estimation, battery status estimation, LIB aging, and BMS; further selection was based on title, abstract, keyword, and other relevant signs.
- Sample: research statistics in this area over the last 12 years (2012–2024) are shown in Figure 3; no formal sample size or inclusion/exclusion count is stated.
- Analysis: comparative and thematic synthesis of direct, model-based, data-driven, hybrid, and coestimation SOH estimation methods, evaluating speed, complexity, accuracy, robustness, and operational factors such as battery parameters and model execution.
Limitations
- Stated in the paper: previous studies have unclear classifications and inadequate evaluations; challenges include capacity degradation, model limitations, data integrity, lack of large-scale labeled datasets, laboratory data not reflecting real-world degradation, hybrid models assuming clean structured datasets, computational constraints for embedded BMS, limited universality across battery chemistries, uncertain long-term reliability, and missing historical data for second-life batteries.
- Stated in the paper: Green AI may trade efficiency against accuracy and flexibility, relies on smaller datasets, and faces implementation complexity and the need for specialized knowledge.
- Evident from the text: the review does not report a formal systematic search count or meta-analysis, and the many accuracy figures are drawn from heterogeneous cited studies with different datasets, chemistries, and conditions, so they are not directly comparable.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
12 Valid · 53 Not valid · 100 Not checked · 142 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“For EVs and energy storage applications, this presents a signifcant issue for real-time implementation in embedded BMS with constrained computational resources [122].”
What the sources say
- Disputedcites [122] [122] M. Kopp, A. Fill, M. Str¨ obel, and K. P. Birke, 'A Novel Long Short-Term Memory Approach for Online State-Of-Health Identifcation in Lithium-Ion Battery Cells,' Batteries 10, no. 3 (2024): 77, https://doi.org/10.3390/BATTERIES10030077. - sourceExperimental — accuracy not yet measured
The Hybrid Methods section describes performance benefits but does not identify high computing demand as a drawback. More directly, the Conclusions describe the paper’s hybrid approach as requiring no expensive computational resources, while the Abstract says it is well suited to online electric-vehicle applications. That is contrary to the attributed claim that hybrid SOH techniques pose a significant computational obstacle to such implementation. (LLM, conf=0.85)
Checked by a reasoning model — may be wrong.
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Te family KF have been used for SOH estimation, such as the Extended Kalman Filter (EKF). Tese include the Invariant Extended Kalman Filter (IEKF), Square Root Variant of the Cubature Kalman Filter (SCKF), and Adaptive Extended Kalman Filter (AEKF) [54].”
What the sources say
- Partly supportedcites [54] [54] S. Jo, S. Jung, and T. Roh, 'Battery State-Of-Health Estimation Using Machine Learning and Preprocessing With Relative State-Of-Charge,' Energies 14, no. 21 (2021): 7206, https://doi.org/10.3390/EN14217206. - sourceExperimental — accuracy not yet measured
The Introduction reports that Plett estimated battery SOH using an extended Kalman filter. It does not mention IEKF, SCKF, or AEKF as SOH-estimation methods, so the cited paper supports only the EKF example. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“As a practical outcome, charge transfer was restricted after 500 cycles once the Ds fell by 40%.”
What the sources say
- Could not verifycites [93] [93] T. F. Fuller, M. Doyle, and J. Newman, 'Simulation and Optimization of the Dual Lithium Ion Insertion Cell,' Journal of the Electrochemical Society 141 (1994): 1-10, https://doi.org/10.1149/1.2054684. - sourceExperimental — accuracy not yet measured
The Abstract mentions 500–1000 cycles only as a desired battery lifetime, not as a reported outcome. The charge-transfer discussion describes model assumptions, while the Liy section’s 40% figure refers to an increase in initial electrolyte concentration. Searches for a 40% fall in Ds and restricted charge transfer after 500 cycles found no passage reporting that result. (LLM, conf=0.76)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“A link was found between performance decline and decreasing difusion coefcient (Ds) over cycles.”
What the sources say
- Could not verifycites [93] [93] T. F. Fuller, M. Doyle, and J. Newman, 'Simulation and Optimization of the Dual Lithium Ion Insertion Cell,' Journal of the Electrochemical Society 141 (1994): 1-10, https://doi.org/10.1149/1.2054684. - sourceExperimental — accuracy not yet measured
The modeling section assumes a constant diffusion coefficient, and the Summary says its simulated performance values do not account for losses due to extended cycling. A search for declining diffusion coefficients and cycling-related performance loss found no passage reporting the claimed link. The paper discusses diffusion limits during discharge, not a decrease in Ds over successive cycles. (LLM, conf=0.77)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Like this, proper orthogonal decomposition (POD) and polynomial chaos expansion (PCE) can efectively project high-dimensional battery models onto lowerdimensional representations without causing a noticeable reduction in accuracy [125].”
What the sources say
- Could not verifycites [125] [125] Y. Sun, Y. Zhang, A. Boyce, and M. Faraji Niri, 'Editorial: Lithium-Ion Batteries: Manufacturing, Modelling and Advanced Experimental Techniques,' Frontiers in Energy Research 12 (2024): 1508980, https://doi.org/10.3389/ fenrg.2024.1508980. - sourceExperimental — accuracy not yet measured
The Introduction and Perspectives discuss battery prediction methods, including UKF, GA-BPNN, and decision tree–SVR, but do not address dimensionality reduction by POD or PCE. Searches for those methods and for reduced-order models found no passage supporting the claimed preservation of accuracy. (LLM, conf=0.74)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
73 more claims · 6 with their cited sources checked
Create a free account to see all 307 claims and every source verdict.
- Probably not a substantial claim. 158
- It may state two separate claims as one. 113
- The quote only partly states this claim. 1
- Borderline: one of the tests for a substantial claim is unclear. 4
- Not yet classified. 4
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6.2.2. Data-Driven Methods
“Paliwal and Schuster [119] developed a bidirectional RNN”
§ References
“- [119] F. Xiao, W. Yang, Y. Ding, X. Li, K. Zhang, and J. Liu, 'A CNN-LSTM Method Based on Voltage Deviation for Predicting the State of Health of Lithium-Ion Batteries,'”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6.2. Indirect Methods
“Doyle et al. invented the frst LIB-based electrochemical model. Tis method is known as the Doyle-Fuller-Newman (DFN) model [92].”
§ References
“- [92] K. Khodadadi Sadabadi, P. Ramesh, P. Tulpule, Y. Guezennec, and G. Rizzoni, 'Model-Based State of Health”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 10. Scalability and Performance of SOH Estimation Methods Under Extreme Temperatures and AI-Based Adaptation Strategies
“To address these issues, Jiao et al. recommend creating smart big data analysis platforms.”
§ References
“- [164] H. Jing, J. Hu, S. S. Ou, Z. Lv, R. Lyu, and J. Zhao, 'Scalable and Generalizable Deep Learning for Battery State of Health Estimation in on-Road Electric Vehicles,'”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6.2.2. Data-Driven Methods
“Shen et al. [64] were among the frst to apply DL algorithms for online LIBs capacity estimation”
§ References
“- [64] Y. Bao, W. Dong, and D. Wang, 'Online Internal Resistance Measurement Application in Lithium Ion Battery Capacity and State of Charge Estimation,'”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 6. SOH Estimation Methodologies
“Te Wheatstone bridge's four arms are depicted in Figure 6 as WX, XY, WZ, and ZY, with corresponding resistances of Ra, Rb, Rc, and Rd.”
§ 6. SOH Estimation Methodologies
“Figure 7: Te Wheatstone bridge for R and C measurements.”
10 more findings
Create a free account to see every finding of this check.
Citations
Cites
LeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Kennedy, J., & Eberhart, R. (2002) · Proceedings of ICNN'95 - International Conference on Neural Networks · doi:10.1109/icnn.1995.488968 · Cited by 58,192 · cited 1×
“Zhou et al. used RVM to assess battery health by measuring the average voltage drop during charge and discharge cycles and applied the Box–Cox transformation to enhance the linear correlation between the health indicator (HI) and capacity [110].”
Bidirectional recurrent neural networks
Schuster, M., & Paliwal, K. K. (1997) · IEEE Transactions on Signal Processing · doi:10.1109/78.650093 · Cited by 10,172 · cited 1×
“Tese data-driven methods by using AI models are also implemented in many diferent areas of research [120].”
A review on the key issues for lithium-ion battery management in electric vehicles
Lu, L., Han, X., Li, J., Hua, J., & Ouyang, M. (2013) · Journal of Power Sources · doi:10.1016/j.jpowsour.2012.10.060 · Cited by 4,837 · cited 1×
“Lithium-ion batteries (LIBs) [10] have advantages over traditional batteries, such as lead-acid and nickel metal hydride, including low environmental friendliness, a wide operating temperature range, high energy and power density [11], self-discharge rate [12], and longer cycle life.”
Modeling of Galvanostatic Charge and Discharge of the Lithium/Polymer/Insertion Cell
Doyle, M., Fuller, T. F., & Newman, J. (1993) · Journal of The Electrochemical Society · doi:10.1149/1.2221597 · Cited by 4,328 · cited 1×
“According to their outcomes, KF output is superior to RLS [91].”
Simulation and Optimization of the Dual Lithium Ion Insertion Cell
Fuller, T. F., Doyle, M., & Newman, J. (1994) · Journal of The Electrochemical Society · doi:10.1149/1.2054684 · Cited by 2,057 · cited 1×
“Te study [93] investigated the lithium restrictions in NMC electrodes utilizing Fickian difusion models.”
Hannan, M. A., Lipu, M. S. H., Hussain, A., & Mohamed, A. (2017) · Renewable and Sustainable Energy Reviews · doi:10.1016/j.rser.2017.05.001 · Cited by 2,005 · cited 1×
“[ 31 ] 2017 R e v i e w o n b a tt e r y s t a t e e s t i m a t i o n a n d m a n a g e m e n t s o l u t i o n s f o r n e x t - g e n e r a t i o n E V F o c u s e s o n a d v a n c e m e n t s i n b a tt e r y m a n a g e m e n t s o l u t i o n s ; m a y n o t p r o v i d e d e t a i l e d m e…”
A review on the key issues of the lithium ion battery degradation among the whole life cycle
Han, X., Lu, L., Zheng, Y., Feng, X., Li, Z., Li, J., & Ouyang, M. (2019) · eTransportation · doi:10.1016/j.etran.2019.100005 · Cited by 1,745 · cited 1×
“In the process of evaluating the battery’s remaining power, capacity, and RUL [21], SOC [22], SOE [22], state of health (SOH) [23], and SOP [24], other relevant studies for precise battery health assessment are critical.”
Hu, X., Xu, L., Lin, X., & Pecht, M. (2020) · Joule · doi:10.1016/j.joule.2019.11.018 · Cited by 1,320 · cited 1×
“In the process of evaluating the battery’s remaining power, capacity, and RUL [21], SOC [22], SOE [22], state of health (SOH) [23], and SOP [24], other relevant studies for precise battery health assessment are critical.”
A critical review of battery thermal performance and liquid based battery thermal management
Wu, W., Wang, S., Wu, W., Chen, K., Hong, S., & Lai, Y. (2019) · Energy Conversion and Management · doi:10.1016/j.enconman.2018.12.051 · Cited by 1,214 · cited 1×
“Battery management systems (BMSs), prognostics and systems health management, and battery thermal management systems are important because they can quickly adjust the battery’s voltage and temperature within a precise range [20].”
Research works printed but not linked: 2
- 2025, Long Short-Term Memory Recurrent Neural Network for Remaining Useful Life Prediction of Lithium-Ion Batteries
- 10.1155/2023/2060808 · DOI · Paper page
Cited by
Battery SOH estimation via an optimized CNN–BiLSTM–Attention network using ICA-Based ageing features
Lyu, Z., Wang, H., Shi, W., Qiang, X., & Wu, L. (2026) · Ionics · doi:10.1007/s11581-025-06933-7 · Cited by 5
Errakkas, K., Kissaoui, M., Lajouad, R., Magri, A. E., Khayat, A., & Atifi, Y. (2026) · Scientific African · doi:10.1016/j.sciaf.2026.e03244 · Cited by 2
Khan, M. S., Maddipatla, S., & Pecht, M. G. (2026) · Journal of Power Sources · doi:10.1016/j.jpowsour.2026.239788 · Cited by 1
Shi, H., Zhang, Z., Yang, G., Wang, X., Huo, W., Liang, S., Zuo, B., Zhang, L., Sun, G., & Cheng, S. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5825149
Zhang, X. (2026) · Transactions on Emerging Telecommunications Technologies · doi:10.1002/ett.70462
Metadata
- Authors
- Kumar, D., Nasimov, R., Singh, A., Jha, A., Jha, A., & Rizwan, M.
- Affiliations
- Delhi Technological University (IN)
- Tashkent State University of Economics (UZ)
- Corresponding author
- Kumar, Deepak
- Nasimov, Rashid
- Rizwan, M.
- Venue
- Applied Computational Intelligence and Soft Computing
- Publisher
- Wiley
- Published
- 2025-01
- Volume
- 2025
- Issue
- 1
- ISSN
- 1687-9724 (ISSN-L)
- 1687-9732
- OpenAlex
- W4416176497
- Cited by
- 8
- References
- 183
- Citations by year
- 2026: 7 · 2025: 1
- FWCI
- 1.95
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0/ (vor, 316 d)
- doi.wiley.com/10.1002/tdm_license_1.1 (tdm)
- Locations
- Applied Computational Intelligence and Soft Computing (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Engineering › Automotive Engineering › Battery Modeling and Management
- Physical Sciences › Engineering › Electrical and Electronic Engineering › Advancements in Battery Materials
- Physical Sciences › Engineering › Automotive Engineering › Electric and Hybrid Vehicle Technologies
- Keywords
- state of health estimation, lithium-ion batteries, electric vehicles, battery management systems, artificial intelligence, hybrid estimation, capacity fade, green AI, second-life batteries
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-28, under version 3 of the claim definition (prompt extract-v3).
A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Artificial Intelligence Approaches for SOH Estimation in Lithium-Ion Batteries: A Comprehensive Review of Progress, Critical Evaluations, Challenges, and Toward Future Outlooks. https://app.cochat.ai/papers/10.1155/acis/5516151
- The paper
- Kumar, D., Nasimov, R., Singh, A., Jha, A., Jha, A., & Rizwan, M. (2025). Artificial Intelligence Approaches for SOH Estimation in Lithium-Ion Batteries: A Comprehensive Review of Progress, Critical Evaluations, Challenges, and Toward Future Outlooks. Applied Computational Intelligence and Soft Computing. https://doi.org/10.1155/acis/5516151
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.