Pith. sign in

REVIEW 4 major objections 7 minor 69 references

Physics-informed mixture of experts network for interpretable battery degradation trajectory computation amid second-life complexities

T0 review · 4 major / 7 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Without any historical cycling data, PIMOE computes a retired battery's future capacity trajectory from one partial charging curve plus relaxation voltage and an assumed future load schedule, reporting 0.88% average error over 50 cycles.

desk verdict Useful architecture and real experimental breadth, but the headline accuracy is not credible until the train/test split is disclosed. read the letter →

arxiv 2506.17755 v1 pith:N7UIPKI2 submitted 2025-06-21 cs.LG

Xinghao Huang , Shengyu Tao , Chen Liang , Jiawei Chen , Junzhe Shi , Yuqi Li , Bizhong Xia , Guangmin Zhou
show 1 more author
Xuan Zhang
This is my paper · ORCID
classification cs.LG
keywords second-lifebatterydegradationtrajectorypredictionphysics-informedlearningmixtureofexpertsstatehealthcapacityfadepartialchargingcurverelaxationvoltage
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Retired electric-vehicle batteries are hard to repurpose because their degradation history is usually unavailable and their future loads are uncertain. The paper proposes a physics-informed mixture-of-experts network (PIMOE) that claims to compute a retired battery's capacity-fade trajectory from one field-accessible cycle alone: a partial charging curve starting at a random state of charge plus a 30-minute relaxation voltage measurement, together with an assumed schedule of future charge/discharge currents and temperatures. On 207 batteries across 77 use conditions and 67,902 cycles, it reports average errors of 0.88% over a 50-cycle horizon and 1.50% over a 150-cycle horizon, with a 0.43 ms inference time. The significance, if the claim holds, is that second-life batteries could be screened and routed for reuse without offline capacity tests or historical records, and the same expert weights can indicate whether a cell is fit for demanding storage duty or should be recycled.

What carries the argument

The load-bearing object is the degradation router inside the Adaptive Multi-degradation Prediction (AMDP) module. It takes twelve physics-informed features, six from the partial charging curve and six from relaxation voltage, including charge acceptance in a 0.05 V window and voltage rise per 200 mAh, applies a noisy top-k softmax over five expert networks, and synthesizes a latent degradation-trend vector. Each expert maps the 50-point charging curve to a short-horizon trend, and the router's weights select which degradation mode dominates at that moment in life. The Future-Operation Recurrent Neural Network (FORNN) then concatenates that trend with per-cycle future load triplets $(I_{\text{charge}}, I_{\text{discharge}}, T)$ and feeds an LSTM that emits the capacity trajectory, which is what lets the model respond to changing second-life loads rather than assuming a constant profile.

What would settle it

Deploy the trained PIMOE on retired cells with logged future loads but hidden capacity history, specifying the load sequence that is actually used, and compare predicted and measured capacity at cycles 50 and 150. If the reported 0.88% and 1.50% average MAPE reproduce only when the given load sequence matches the true one, the history-free claim holds; if accuracy collapses when load assumptions deviate even slightly, the effective constraint moves from data accessibility to forecastability of future operating conditions.

Watch

Extended reading notes

Core claim

PIMOE establishes a history-free route from in-field electrical signals to future degradation: the input is a random-SOC partial charge curve (50 voltage bins) and a relaxation voltage trace; twelve statistics encode polarization and aging state; a noisy top-2-of-5 degradation router produces a latent trend embedding; and a recurrent network walks that embedding forward through the assumed load profile to output capacity per future cycle. The central claims are that this pipeline computes 50-cycle capacity trajectories at 0.88% average MAPE without historical data, extends to 150 cycles at 1.50% average MAPE, stays below 6.26% maximum MAPE, and works across three material systems and both fixed and randomly varying second-life loads. The paper further claims that the router's expert weights alone classify retired cells by state of health, with high-confidence separation between cells retired at 95% SOH ('excellent') and 75% SOH ('scrap'), so the interpretable routing doubles as a screening tool.

Load-bearing premise

For every claimed trajectory, the future charge current, discharge current, and temperature of each future cycle are assumed known and are fed into the network; if those future loads cannot be specified or forecast, the history-free trajectory computation cannot be run as validated.

Editorial extensions

If this is right

  • Battery recyclers could classify and route retired cells using one partial charge and relaxation measurement, replacing full-capacity calibration tests and historical data audits.
  • Because random initial state of charge is supported, field batteries need no pre-conditioning to a fixed SOC before measurement.
  • The same trained network can be applied across first-life and second-life regimes, since the results show it tracks capacity when loads switch abruptly.
  • Long-horizon planning is feasible: 150-cycle forecasts remain at 1.50% average MAPE, so battery-management decisions could be made against predicted future fade rather than current state of health alone.
  • Small training budgets suffice, with the model remaining usable on a pruned 5 MB training set, lowering the data barrier for deployment.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension is to treat the router's confidence as a distribution-shift alarm: when expert weights are uncertain or fall outside trained regions, the forecast should be flagged rather than trusted, since the training data do not include thermal-runaway or internal-short regimes.
  • The model's expert-to-mechanism mapping (SEI formation, SEI thickening, lithium plating) is inferred from trajectory slope and literature, not from direct cell teardown; an independent validation could check whether the expert weights align with differential-voltage signatures or post-mortem diagnostics.
  • Because the method requires an assumed future load sequence, real deployments should pair PIMOE with a load-forecasting module or scenario ensemble; otherwise the validation protocol implicitly bundles trajectory-model error with load-forecast error.
  • The input features are deliberately simple statistics, so replacing or augmenting them with incremental-capacity peaks or other mechanistic descriptors might sharpen loss-of-lithium versus loss-of-active-material separation, and this is a cheap, testable modification.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. This paper proposes PIMOE, a mixture-of-experts architecture combined with a recurrent network (FORNN) to compute future capacity-degradation trajectories of retired lithium-ion batteries from a single partial charging curve and a relaxation-voltage measurement at a random initial state of charge, together with an assumed future load profile. The method is evaluated on the Uniform-Life dataset (130 cells) and the Two-Phase Second-Life dataset (77 cells) and is compared with PatchTST and Informer, with additional ablations, noise-perturbation studies, data-scarcity experiments, and horizon sweeps. The headline results are an average MAPE of 0.88% over a 50-cycle horizon and 1.50% over 150 cycles, with an inference time of 0.43 ms per battery.

Significance. If the reported accuracy is leak-free, PIMOE would be a practically valuable history-free trajectory forecaster for second-life battery management: it uses field-accessible partial-cycle signals, incorporates future load scenarios, and is computationally very light. The authors should be credited for the breadth of the evaluation (207 cells across multiple conditions), the ablations of the AMDP and FORNN modules, the feature-perturbation and data-scarcity analyses, and the public code/data link. The expert-weight specialization shown over the lifecycle is suggestive and could be a useful interpretability tool. However, the significance is conditional: the headline metric currently lacks a documented train/test split, a trivial baseline, a precise statement of the protocol underlying the 0.88% figure, and independent validation of the claimed degradation-mode labels. These issues must be resolved before the central empirical claim can be taken at face value.

major comments (4)
  1. [Results (Dataset; Model performance and generalization capability); Supplementary Notes 1 and 10] The manuscript never specifies how training and test windows are separated. Supplementary Note 1 constructs sample pairs by sliding over every cycle t, with target horizon [t+1, t+H] (Eqs. (1)-(2)); if the split is random by cycle rather than by battery or by temporally non-overlapping blocks, a model trained on cycles t-1, t-2, ... has already seen target capacity values that overlap the test horizon at cycle t, so the reported 0.88% MAPE would not reflect generalization to a new retired battery. Supplementary Note 10 states only that the test set is "unchanged" during data reduction; the exact battery/condition/cycle partition is absent. Please provide the split protocol explicitly (battery-level or contiguous-block, with no target-cycle overlap) and report the per-condition test sizes. This is load-bearing for the history-free generalization claim.
  2. [Results, Fig. 3c-3f; Abstract] The headline 0.88% average MAPE is presented without a trivial baseline and without a precise protocol for how the number is composed. No persistence or linear-extrapolation forecaster is reported, and the initial-SOC composition of the 0.88% figure is unspecified: Fig. 3f reports MAPEs of 1.44%, 2.81%, and 3.43% at 50% SOC for the three datasets, so the aggregate number depends on the mix of initial SOCs and horizons used. Because capacity fade over 50 cycles is gradual, a "repeat last capacity" baseline may already achieve low MAPE; without such a baseline, the improvement over PatchTST and Informer does not by itself establish that the trajectory shape is being predicted. Please add persistence and linear baselines and state exactly which initial-SOC conditions and horizons enter the 0.88% average.
  3. [Supplementary Table 9 (80-cycle results) and Supplementary Table 7 (40% training data)] Supplementary Table 9 is identical to Supplementary Table 7 in every row. This duplication is a clear copy/paste error and invalidates the 80-cycle results as currently presented. The table must be regenerated from the actual experimental runs, and the claims about horizon sensitivity in the Results section and Fig. 5d should be updated if the corrected numbers differ.
  4. [Results (Rationalization of statistical model performance), Fig. 4a, 4c, 4d; Discussion] The expert networks are labeled as SEI formation, SEI thickening, and lithium plating based on the temporal evolution of router weights, but no independent measurement (post-mortem analysis, incremental-capacity or dQ/dV analysis, electrochemical impedance spectroscopy) is used to validate that each expert corresponds to the claimed degradation mode. The Discussion itself states that the employed features are "essentially statistical correlations, lacking systematic integration of deeper physicochemical mechanisms." Therefore, the interpretability claims in Fig. 4 should be reframed as a hypothesized correspondence, or supported with external mechanistic measurements; otherwise the physics-informed/mechanistic interpretation is not evidenced.
minor comments (7)
  1. [Abstract; Discussion] The number of batteries is reported as 207 in the Abstract and as 203 independent cells in the Discussion; these counts should be reconciled.
  2. [Results, Fig. 3d; Supplementary Table 4] The MAPE values for TPSL-Fixed and TPSL-Random appear to be swapped between the main text (2.96% for TPSL-Fixed, 2.81% for TPSL-Random) and Supplementary Table 4 (2.957 for TPSL-Arbitrary and 2.806 for TPSL-Fixed); please correct the inconsistency.
  3. [Supplementary Note 1] The placeholder "see Supplementary Material X" should be replaced with a real reference to the prediction-horizon selection note.
  4. [Abstract; Results, Fig. 3e] The statement that the method "reduces computational time and MAPE by 50%, respectively" is ambiguous: the inference-time reduction is about 50% relative to Informer, but the MAPE reduction relative to the baseline models is much larger; please state the exact comparison baseline for each quantity.
  5. [Supplementary Figures 13 and 14] The captions of Supplementary Figs. 13 and 14 appear mismatched with their descriptions (one is captioned TPSL-Arbitrary but describes TPSL-Fixed, and the neighboring figure has the opposite inconsistency); the figure numbering and captions should be checked.
  6. [Whole manuscript] The manuscript needs a careful language and copy-editing pass; numerous typographical errors remain (e.g., "relavant", "larbor", "Desipte", "valuse", "statisitical", "pipleline"), and several informal phrases should be revised for a journal submission.
  7. [Results, Fig. 3e] The 0.43 ms inference-time measurement should be accompanied by a specification of the hardware and software environment on which it was obtained.

Circularity Check

1 steps flagged · score 2.0 of 10

The capacity-trajectory computation is an independent supervised benchmark; the only circular element is the post hoc labeling of the model's own expert-weight clusters as physical degradation modes.

  1. self definitional [Results, 'Rationalization of statistical model performance' (pp. 9-10, Fig. 4a-4c; Supplementary Note 7)]
    "Without re quiring historical data, the degradatio n router partitions battery lifecycle degradations into three distinct phases using partial cycling data, whic h is evidenced by existing literature 8,46. ... Mechanistically, this suggests that Expert Networks 1, 3, and 5 respectively dominate initial SEI layer formation, subsequent thickening, and lithium plating processes ，consistent with prior research on primary battery degradation modes 46."

    The router weights g_j(F_i) are trained exclusively through the trajectory loss L_traj (Eq. 10) against capacity sequences S_i; no external label for SEI formation, SEI thickening, or lithium plating is used in training. The 'three distinct phases' are therefore clusters of the model's own latent weights, and the mechanistic attributions in Fig. 4a are read back from those same weights. When the paper then presents this as evidence that AMDP 'classifies degradation modes', the classification target is defined by the classifier's own latent partition: observation of the weights is both the source and the confirmation of the mode labels. The capacity-trajectory MAPE remains an independent benchmark, so this circularity is confined to the interpretability claim.

full rationale

The central derivation is a supervised regression chain: Eqs. (1) and (6) map current-cycle charging curve Q_i and physics-informed features F_i to a latent trend, Eqs. (7)-(9) concatenate that trend with assumed future load conditions, and Eq. (10) trains the whole chain against the ground-truth capacity trajectory S_i. No target capacity appears among the inputs of the forward model, and no fitted constant is renamed as a test prediction, so the claimed 0.88% MAPE is an external empirical benchmark rather than an identity. The future-load input is explicitly acknowledged in the Discussion as 'difficult to predict in reality', which is a deployment limitation, not a circular derivation. The only circular element I found is interpretive: the paper's degradation-mode classification is read off the expert weights that were trained only to reproduce capacity trajectories, and then the same weight pattern is presented as evidence that AMDP identifies SEI/lithium-plating modes; no independent mechanistic label is used. Self-citations in the manuscript are contextual and not load-bearing, and the undocumented training/test split is a data-integrity risk, not a demonstrated reduction of output to input.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The model rests on no first-principles battery equations; it rests on engineered features, a supervised loss, and the assumption that future load sequences are known. The headline performance is an empirical benchmark, not a derivation. The main 'physics' is feature selection plus an interpretive mapping from router weights to aging phases, which the authors themselves describe as statistical correlations.

free parameters (6)
  • Number of experts E = 5
    Chosen via sensitivity analysis in Results and Fig. 5c; 4 experts were worse and 6 experts were less stable.
  • TopK selection = 2
    Selected together with E=5 in the degradation router (Methods Eq. 3).
  • Loss weights alpha and beta = alpha=0.75, beta=0.25
    Set in Methods Eq. 13 to balance trajectory accuracy and expert diversity.
  • Benchmark prediction horizon L = 50 cycles (10 for NCA-25-1-1, 150 in extension)
    Chosen in Supplementary Note 3 as a compromise between accuracy and utility; performance is horizon-dependent.
  • Voltage segments N = 50
    Resolution of the partial charging curve representation (Methods, data preparation).
  • FORNN hidden dimension = 64
    Model configuration in Supplementary Table 3, fixed across datasets.
assumptions (4)
  • domain assumption Future load sequences are available as model inputs.
    Methods Eq. 7 defines C_i as charge/discharge currents and temperature for each future cycle; the Discussion notes these are difficult to predict in reality.
  • domain assumption One-cycle features capture the relevant degradation state.
    The 12 extracted features are the only link between the field measurement and the future trajectory; no internal electrochemical state is directly measured.
  • domain assumption Router weights correspond to physical degradation modes.
    Fig. 4a maps experts 1, 3, and 5 to SEI formation, SEI thickening, and lithium plating, but no ground-truth mode labels are used; the mapping is post hoc.
  • domain assumption The UL and TPSL laboratory datasets represent second-life field diversity.
    The paper validates only on these two lab datasets and notes in the Discussion that real-world thermal non-uniformity, manufacturing variability, and extreme safety events are not covered.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Physics-informed mixture of experts network for interpretable battery degradation trajectory computation amid second-life complexities." pith.science (2026). https://pith.science/paper/N7UIPKI2

@misc{pith2026250617755,
  author       = {Pith},
  title        = {Pith review of: Physics-informed mixture of experts network for interpretable battery degradation trajectory computation amid second-life complexities},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N7UIPKI2}},
  note         = {Machine review of arXiv:2506.17755}
}
read the original abstract

Retired electric vehicle batteries offer immense potential to support low-carbon energy systems, but uncertainties in their degradation behavior and data inaccessibilities under second-life use pose major barriers to safe and scalable deployment. This work proposes a Physics-Informed Mixture of Experts (PIMOE) network that computes battery degradation trajectories using partial, field-accessible signals in a single cycle. PIMOE leverages an adaptive multi-degradation prediction module to classify degradation modes using expert weight synthesis underpinned by capacity-voltage and relaxation data, producing latent degradation trend embeddings. These are input to a use-dependent recurrent network for long-term trajectory prediction. Validated on 207 batteries across 77 use conditions and 67,902 cycles, PIMOE achieves an average mean absolute percentage (MAPE) errors of 0.88% with a 0.43 ms inference time. Compared to the state-of-the-art Informer and PatchTST, it reduces computational time and MAPE by 50%, respectively. Compatible with random state of charge region sampling, PIMOE supports 150-cycle forecasts with 1.50% average and 6.26% maximum MAPE, and operates effectively even with pruned 5MB training data. Broadly, PIMOE framework offers a deployable, history-free solution for battery degradation trajectory computation, redefining how second-life energy storage systems are assessed, optimized, and integrated into the sustainable energy landscape.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

69 extracted references · 63 canonical work pages

  1. [1]

    Tao, S. et al. Collaborative and privacy-preserving retired battery sorting for profitable direct recycling via federated machine learning. Nat Commun 14, 8032 (2023)

    work page 2023

  2. [2]

    Huang, J., Boles, S. T. & Tarascon, J.-M. Sens ing as the key to battery lifetime and sustainability. Nat Sustain 5, 194–204 (2022)

    work page 2022

  3. [3]

    Hannan, M. A. et al. Battery energy-storage system: A review of technologies, optimization objectives, constraints, approaches, and outstanding issues. J. Energy Storage 42, 103023 (2021)

    work page 2021

  4. [4]

    Wu, W., Lin, B., Xie, C., Elliott, R. J. R. & Ra dcliffe, J. Does energy storage provide a profitable second life for electric vehicle batteries? Energy Economics 92, 105010 (2020)

    work page 2020

  5. [5]

    Ren, Y . et al. Hidden delays of climate mitigation benefits in the race for electric vehicle deployment. Nat Commun 14, 3164 (2023)

    work page 2023

  6. [6]

    & Passerini, S

    Innocenti, A., Beringer, S. & Passerini, S. Cost and performance analysis as a valuable tool for battery material research. Nat Rev Mater 9, 347–357 (2024)

    work page 2024

  7. [7]

    & Leker, J

    Duffner, F., Wentker, M., Greenwood, M. & Leker, J. Battery cost modeling: A review and directions for future research. Renewable and Sustainable Energy Reviews 127, 109872 (2020)

    work page 2020

  8. [8]

    Han, X. et al. A review on the key issues of the lithium ion battery degradation among the whole life cycle. eTransportation 1, 100005 (2019)

    work page 2019

Show all 69 references
  1. [9]

    & Zhao, Z

    Tian, H., Qin, P., Li, K. & Zhao, Z. A review of the state of health for lithium-ion batteries: Research status and suggestions. J. Cleaner Production 261, 120813 (2020)

    2020

  2. [10]

    Attia, P. M. et al. Closed-loop optimization of fast-charging protocols for batteries with machine learning. Nature 578, 397–402 (2020)

    2020

  3. [11]

    & Sauer, D

    Li, W., Zhang, H., Van Vlijmen, B., Dechent, P. & Sauer, D. U. Forecasting battery capacity and power degradation with multi-task learning. Energy Storage Mater. 53, 453–466 (2022)

    2022

  4. [13]

    Ma, R. et al. Pathway decisions for reuse and recycling of retired lithium-ion batteries considering economic and environmental functions. Nat Commun 15, 7641 (2024)

    2024

  5. [15]

    & Chen, R

    Fly, A. & Chen, R. Rate dependency of increm ental capacity analysis (dQ/dV) as a diagnostic tool for lithium-ion batteries. J. Energy Storage 29, 101329 (2020)

    2020

  6. [17]

    Wang, J. et al. Degradation of lithium ion batteries employing graphite negatives and nickel– cobalt–manganese oxide + spinel manganese oxide positives: Part 1, aging mechanisms and life estimation. J.Power Sources 269, 937–948 (2014)

    2014

  7. [18]

    Tao, S. et al. Battery Cross-Operation-Condition Lifetime Prediction via Interpretable Feature Engineering Assisted Adaptive Machine Learning. ACS Energy Lett. 8, 3269–3279 (2023)

    2023

  8. [19]

    Geslin, A. et al. Dynamic cycling enhances battery lifetime. Nat Energy (2024)

    2024

  9. [20]

    & Teodores cu, R

    Che, Y ., Hu, X., Lin, X., Guo, J. & Teodores cu, R. Health prognostics for lithium-ion batteries: mechanisms, methods, and prospects. Energy Environ. Sci. 16, 338–371 (2023)

    2023

  10. [21]

    Xu, L., Deng, Z., Xie, Y ., Lin, X. & Hu, X. A Novel Hybrid Physics-Based and Data-Driven Approach for Degradation Trajectory Prediction in Li-Ion Batteries. IEEE Trans. Transp. 21/71 Electrific. 9, 2628–2644 (2023)

    2023

  11. [22]

    Tao, S. et al. Rapid and sustainable battery health diagnosis for recycling pretreatment using fast pulse test and random forest machine learning. J. Power Sources 597, 234156 (2024)

    2024

  12. [23]

    Li, J. et al. Degradation Pattern Recognition and Features Extrapolation for Battery Capacity Trajectory Prediction. IEEE Trans. Transp. Electrific. 10, 7565–7579 (2024)

    2024

  13. [24]

    & Flynn, D

    Roman, D., Saxena, S., Robu, V ., Pecht, M. & Flynn, D. Machine learning pipeline for battery state-of-health estimation. Nat Mach Intell 3, 447–456 (2021)

    2021

  14. [25]

    He, K. et al. A Novel Quick Screening Method for the Second Usage of Parallel-connected Lithium-ion Cells Based on the Current Distribution. J. Electrochem. Soc. 170, 030514 (2023)

    2023

  15. [26]

    & Sun, F

    Tian, J., Xiong, R., Shen, W., Lu, J. & Sun, F. Flexible battery state of health and state of charge estimation using partial charging data and deep learning. Energy Storage Mater. 51, 372–381 (2022)

    2022

  16. [27]

    & Peng, H

    Hu, X., Li, S. & Peng, H. A comparative study of equivalent circuit models for Li-ion batteries. J. Power Sources 198, 359–367 (2012)

    2012

  17. [28]

    Xiong, R. et al. A data-driven method for extracting aging features to accurately predict the battery health. Energy Storage Mater.s 57, 460–470 (2023)

    2023

  18. [30]

    & Chen, X

    Wang, F., Zhai, Z., Zhao, Z., Di, Y . & Chen, X. Physics-informed neural network for lithium-ion battery degradation stable modeling and prognosis. Nat Commun 15, 4332 (2024)

    2024

  19. [31]

    Severson, K. A. et al. Data-driven prediction of battery cycle life before capacity degradation. Nat Energy 4, 383–391 (2019)

    2019

  20. [32]

    Fu, S. et al. Data-driven capacity estimation for lithium-ion batteries with feature matching based transfer learning method. Appl. Energy 353, 121991 (2024)

    2024

  21. [34]

    Tao, S. et al. Non-destructive degradation pattern decoupling for early battery trajectory prediction via physics-informed learning. Energy Environ. Sci. 10.1039.D4EE03839H (2025)

    2025

  22. [35]

    Lu, J. et al. Battery degradation prediction against uncertain future conditions with recurrent neural network enabled deep learning. Energy Storage Mater. 50, 139–151 (2022)

    2022

  23. [36]

    & Bian, X

    Deng, Z., Hu, X., Li, P., Lin, X. & Bian, X. Data-Driven Battery State of Health Estimation Based on Random Partial Charging Data. IEEE Trans. Power Electron. 37, 5021–5031 (2022)

    2022

  24. [37]

    Wang, Z. et al. Battery health diagnostics: Bridging the gap between academia and industry. eTransportation 19, 100309 (2024)

    2024

  25. [38]

    & Wang, H

    Zhao, M., Zhang, Y . & Wang, H. Battery degrad ation stage detection and life prediction without accessing historical operating data. Energy Storage Mater. 69, 103441 (2024)

    2024

  26. [39]

    Li, R. et al. The importance of degradation mode analysis in parameterising lifetime prediction models of lithium-ion battery degradation. Nat Commun 16, 2776 (2025)

    2025

  27. [42]

    Shazeer, N. et al. Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts 22/71 Layer. Preprint at https://doi.org/10.48550/arXiv.1701.06538 (2017)

    2017 doi arXiv

  28. [43]

    & Shazee r, N

    Fedus, W., Zoph, B. & Shazee r, N. Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity. Preprint at https://doi.org/10.48550/arXiv.2101.03961 (2022)

    2022 doi arXiv

  29. [45]

    Zhou, H. et al. Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. AAAI 35, 11106–11115 (2021)

    2021

  30. [46]

    Schuster, S. F. et al. Nonlinear aging characteristics of lithium-ion cells under different operational conditions. J. Energy Storage 1, 44–53 (2015)

    2015

  31. [47]

    Desai, T., Gallo, A. J. & Ferrari, R. M. G. Multi timescale battery modeling: Integrating physics insights to data-driven model. Appl. Energy 393, 126040 (2025)

    2025

  32. [48]

    Fan, J. et al. Wireless transmission of internal hazard signals in Li-ion batteries. Nature 641, 639–645 (2025)

    2025

  33. [49]

    Huang, J. et al. Operando decoding of chemical and thermal events in commercial Na(Li)-ion cells via optical sensors. Nat Energy 5, 674–683 (2020)

    2020

  34. [50]

    Wang, F. et al. Inherently Interpretable Physics-Informed Neural Network for Battery Modeling and Prognosis. IEEE Trans. Neural Netw. Learning Syst. 1–15 (2024)

    2024

  35. [51]

    Liu, H. et al. Multi-modal framework for battery state of health evaluation using open-source electric vehicle data. Nat Commun 16, 1137 (2025)

    2025

  36. [52]

    Both" denotes the complete model incorporating both capacity-voltage curves and relaxation-voltage curves

    Zhang, H. et al. Battery lifetime prediction across diverse ageing conditions with inter-cell deep learning. Nat Mach Intell (2025). Author contributions Xinghao Huang: Writing – review & editing, Wr iting – original draft, Visualiz ation, Validation, Methodology, Formal analy...

    2025

  37. [53]

    plug-and-play

    In contrast, our innovative approach requires only partial current cycle data as input and adapts to random initial SOC sampling strategies. This breakthrough not only significantly enhances real-world applicability but also provides technical assurance for end-user deployment...

  38. [54]

    Tao, S. et al. Generative learning assisted state-of-health estimation for sustainable battery recycling with random retirement conditions. Nat Commun 15, 10154 (2024)

    2024

  39. [55]

    Wang, Z. et al. Ba ttery health diagnostics: Bridging the gap between academia and industry. eTransportation 19, 100309 (2024)

    2024

  40. [56]

    & Passerini, S

    Innocenti, A., Beringer, S. & Passerini, S. Cost an d performance analysis as a valuable tool for battery material research. Nat Rev Mater 9, 347–357 (2024)

    2024

  41. [57]

    Zhu, J. et al. Data-driven capacity estimation of commercial lithium-ion batteries from voltage relaxation. Nat Commun 13, 2261 (2022)

    2022

  42. [58]

    Pinson, M. B. & Bazant, M. Z. Theory of SEI Formation in Rechargeable Batteries: Capacity Fade, Accelerated Aging and Lifetime Prediction. J. Electrochem. Soc. 160, A243–A250 (2013)

    2013

  43. [59]

    & Sun, F

    Xiong, R., Wang, P., Jia, Y ., Shen, W. & Sun, F. Multi-factor aging in Lithium Iron phosphate batteries: Mechanisms and insights. Appl. Energy 382, 125250 (2025)

    2025

  44. [60]

    & Teodorescu, R

    Che, Y ., Hu, X., Lin, X., G uo, J. & Teodorescu, R. Health prognostics for lithium-ion batteries: mechanisms, methods, and prospects. Energy Environ. Sci. 16, 338–371 (2023)

    2023

  45. [61]

    Wang, J. et al. Degradation of lithium ion batteries employing graphite negatives and nickel–cobalt– manganese oxide + spinel manganese oxide positives: Part 1, aging mechanisms and life estimation. J. Power Sources 269, 937–948 (2014)

    2014

  46. [62]

    R., Roberts, M

    Birkl, C. R., Roberts, M. R., Mc Turk, E., Bruce, P. G. & Howey, D. A. Degradation diagnostics for lithium ion cells. J. Power Sources 341, 373–386 (2017)

    2017

  47. [63]

    Ye, J., Xie, Q., Lin, M. & Wu, J. A method for estimating the state of health of lithium-ion batteries based on physics-informed neural network. Energy 294, 130828 (2024)

    2024

  48. [64]

    Xiong, R. et al. A data-driven met hod for extracting aging features to accurately predict the battery health. Energy Storage Mater. 57, 460–470 (2023)

    2023

  49. [65]

    & Flynn, D

    Roman, D., Saxena, S., Robu, V ., Pecht, M. & Flynn, D. Machine learning pipeline for battery state-of- health estimation. Nat Mach Intell 3, 447–456 (2021)

    2021

  50. [66]

    Tan, R. et al. Forecasting battery degradation trajectory under domain shift with domain generalization. Energy Storage Mater. 72, 103725 (2024)

    2024

  51. [67]

    & Ts ou, N.-T

    Hsu, C.-W., Xiong, R., Chen, N.-Y ., Li, J. & Ts ou, N.-T. Deep neural network battery life and voltage prediction by using data of one cycle only. Appl. Energy 306, 118134 (2022)

    2022

  52. [68]

    H., Sinthong, P

    Nie, Y ., Nguyen, N. H., Sinthong, P. & Kalagnanam, J. A Time Series is Worth 64 Words: Long-term Forecasting with Transformers. Preprint at https://doi.org/10.48550/arXiv.2211.14730 (2023)

    2023 doi arXiv

  53. [69]

    Zhou, H. et al. Informer: Beyond Efficient Transf ormer for Long Sequence Time-Series Forecasting. AAAI 35, 11106–11115 (2021)

    2021

  54. [70]

    & Dai, T

    Hu, Y ., Liu, P., Zhu, P., Cheng, D. & Dai, T. Adaptive Multi-Scale Decomposition Framework for Time Series Forecasting. Preprint at https://doi.org/10.48550/arXiv.2406.03751 (2024)

    2024 doi arXiv

  55. [71]

    & Dos Reis, G

    Strange, C. & Dos Reis, G. Prediction of future capa city and internal resistance of Li-ion cells from one cycle of input data. Energy and AI 5, 100097 (2021)

    2021

  56. [72]

    Shi, X. et al. Time-MoE: Billion- Scale Time Series Foundation Models with Mixture of Experts. Preprint at https://doi.org/10.48550/arXiv.2409.16040 (2024)

    2024 doi arXiv

  57. [73]

    Wang, Y . et al. TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables. Preprint at https://doi.org/10.48550/arXiv.2402.19072 (2024)

    2024 doi arXiv

  58. [74]

    & Wang, H

    Zhao, M., Zhang, Y . & Wang, H. Battery degradat ion stage detection and life prediction without accessing 71/71 historical operating data. Energy Storage Mater. 69, 103441 (2024)

    2024

  59. [75]

    Wu, W., Lin, B., Xie, C., Elliott, R. J. R. & Radcliffe, J. Does energy storage provide a profitable second life for electric vehicle batteries? Energy Economics 92, 105010 (2020)

    2020

  60. [76]

    & Leker, J

    Duffner, F., Wentker, M., Greenwood, M. & Leker, J. Battery cost modeling: A review and directions for future research. Renew. Sustain. Energy Rev. 127, 109872 (2020)

    2020

  61. [77]

    Lu, J. et al. Battery de gradation prediction against uncertain future conditions with recurrent neural network enabled deep learning. Energy Storage Mater. 50, 139–151 (2022)

    2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.