Physics-informed Diffusion Generative Model for Time-Series Data Synthesis in Dynamic Systems

Haiteng Wang; Jiabao Dong; Lei Ren; Tao Wang; Xiaoge Zhang; Yikang Li; YunFei Zhu

REVIEW 3 major objections 6 minor 48 references

Physics-informed Diffusion Generative Model for Time-Series Data Synthesis in Dynamic Systems

T0 review · 3 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read PhysDGM embeds physical laws into every step of a diffusion generator, producing synthetic industrial time series that outperform real-data-only training on downstream tasks.

desk verdict A broadly useful physics-regularized diffusion recipe for industrial time series, with an overclaimed 'physical consistency' story and a real circular-validation problem. read the letter →

arxiv 2608.10941 v1 pith:6HGGTYXI submitted 2026-08-11 cs.LG

Haiteng Wang , Yunfei Zhu , Tao Wang , Yikang Li , Jiabao Dong , Xiaoge Zhang , Lei Ren This is my paper · ORCID

classification cs.LG

keywords physics-informeddiffusiontime-seriessynthesisdynamicsystemsindustrialdataaugmentationphysicsconstraintspredictivemaintenancefaultdiagnosis

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

Industrial time-series data — turbine temperatures, rotation speeds, battery voltages — is expensive and often dangerous to collect, which blocks AI deployment in critical machinery. PhysDGM claims that a diffusion generative model can create useful synthetic versions of such data only if the physical laws governing the system steer every step of generation, not just the final output. To that end, the paper embeds four classes of constraints (degradation trends, coupled-sensor relations, valid value ranges, and discrete sensor states) into each reverse-diffusion step, with a training schedule that first fits the data distribution and then tightens the physics penalties. Across 34 datasets from turbofan engines, aero-engines, batteries, and chemical processes, the resulting 4.4 million synthetic samples improve downstream models beyond real-data-only training by 48% (remaining useful life), 15% (health indicator), 22% (state of health), and 20% (fault diagnosis), and 5% of real data plus synthetic samples restores full-data accuracy. If the claim holds, physics-consistent synthetic data becomes a drop-in substitute for much of the real data collection that currently limits predictive maintenance and fault diagnosis.

What carries the argument

Three mechanisms carry the claim. (1) Stepwise physics embedding: physical laws are expressed as differentiable penalty functions $f_{p_i}$ evaluated on the predicted clean trajectory $\hat{x}_0^{(t)}$ at every reverse-diffusion timestep, with total objective $L = \sum_{t=1}^T \big[L_{data}^{(t)} + \Phi(t) \sum_{i=1}^n f_{p_i}(\hat{x}_0^{(t)})\big]$; this is the mechanism that is supposed to keep intermediate violations from accumulating. (2) Dynamic Physical Constraint Training (DPCT): the constraint weight $\Phi(t) = \mathrm{Sigmoid}(\alpha \ln(x/(1-x)))$ with $x=t/T$ stays low early in training so the model first fits the data distribution and only later tightens physics compliance; $\alpha=3$ was found optimal. (3) Gradient-guided Physical Conditional Sampling (GPCS): at inference, the predicted noise is corrected by subtracting $\sqrt{1-\alpha_t}\, \sum_i \nabla_{x_t} f_{p_i}$, letting a pretrained model respect new physical conditions without retraining. The four constraint families are sliding-window ReLU monotonicity losses for degradation, slope-alignment MSE between coupled sensors (grounded in Bernoulli and isentropic-compression relations), box-range ReLU penalties for valid operating ranges, and a sine-based discreteness loss for quantized sensor states, with an added PINN-style trend-consistency loss for battery degradation.

What would settle it

Run PhysDGM against a diffusion model with equally strong but non-physical regularizers, such as generic smoothing or monotonicity penalties that enforce no actual degradation or coupling law, matched in penalty budget and schedule, and record both downstream utility on the same test sets and the physics-violation score of the predicted clean trajectory at every intermediate diffusion step. If the non-physical regularizers match the downstream gains, or if PhysDGM's intermediate-step violations are no lower than the unconstrained model's until the final step, the trajectory-level physics-embedding mechanism is not what carries the result.

Watch

Extended reading notes

Core claim

PhysDGM's central claim is that standard diffusion generators, which optimize only a noise-prediction loss, cannot keep synthetic time series on the physically valid manifold, because violations that appear in intermediate reverse-diffusion states compound over the multi-step rollout and applying physics constraints only at the final output is too late to stop them. The remedy is to treat physical laws as active feedback: at every timestep $t$, the model's predicted clean trajectory $\hat{x}_0^{(t)}$ is evaluated by physics-consistency functions $f_{p_i}$, and the total objective becomes $\sum_t \big[L_{data}^{(t)} + \Phi(t)\sum_i f_{p_i}(\hat{x}_0^{(t)})\big]$, where the constraint weight $\Phi(t)$ rises from near zero to near one over training. The paper reports that this stepwise embedding yields synthetic turbofan, aero-engine, battery, and chemical-process data with discriminative scores 19.6–69.1% better than the best baseline, that augmenting real training sets with the synthetic samples surpasses real-data-only performance by 48% for remaining useful life prediction, 15% for health indicator estimation, 22% for state-of-health assessment, and 20% for fault diagnosis, and that a training-free gradient-guidance variant (PhysDGM-grad) adapts a pretrained model to new physical conditions. The paper further claims that physical consistency is what makes scale-up safe: low-fidelity baselines degrade as augmentation grows from 1× to 20×, while PhysDGM keeps improving.

Load-bearing premise

The whole result rests on the assumption that the four hand-written penalty families — sliding-window monotonicity, sensor-to-sensor slope alignment, boxed value ranges, and the sine-based discreteness loss — genuinely capture the physics of these systems; if those penalties are too weak or mis-specified, 'physical consistency' is just adherence to a few statistical heuristics, and the downstream gains could come from generic regularization instead.

Editorial extensions

If this is right

Physically consistent synthetic samples can substitute for a large fraction of real training data: in the 5%-data experiments, augmentation restored full-data predictive accuracy, implying a 10–20× reduction in data-collection cost.
Scale-up is safe only when samples respect physical laws: low-fidelity baselines such as DiT degrade at high augmentation volumes, while PhysDGM keeps improving up to 20×.
A training-free gradient-guidance variant (PhysDGM-grad) retargets a pretrained generator to new physical conditions without retraining, which matters when operating conditions shift.
Millions of physics-consistent failure trajectories expand coverage for rare-fault modeling and early-stage fault detection, demonstrated on aero-engines.
Synthetic data that preserves temporal-physical patterns can support privacy-preserving sharing of industrial operational data while concealing sensitive details.

Reading between the lines

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

The paper does not compare PhysDGM against equally strong non-physical regularizers, such as smoothing, monotonicity-only penalties, or mixup-style augmentation at the same penalty budget; if those matched the downstream gains, generic regularization, not physics embedding, could be the operative mechanism.
The four constraint families are hand-specified per system, so a natural next step is learning constraint forms from governing equations or from the data itself, which would move toward a cross-domain foundation model rather than per-dataset penalty engineering.
The 'beyond real data' result is measured under a fixed augmentation protocol with small real training sets; the claim that synthetic data beats real-data-only training may not transfer if the downstream model is itself retrained on far larger real corpora.
The trajectory-level mechanism is directly testable: PhysDGM predicts lower physics-violation scores for intermediate denoising states than the unconstrained model, but the paper mostly reports final-output metrics, leaving that check open.

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.

Desk Editor's Note

read the letter

Read the paper. Short version: the core technique is real and the empirical effort is broad, but the central physics claim is not supported as cleanly as the abstract says.

What's new: applying soft physics-style penalties at every reverse diffusion step, not just the final output, with a sigmoid-ramped weight schedule and training-free gradient guidance at sampling, is a sensible combination I haven't seen in exactly this form. The downstream augmentation results are extensive: 34 datasets, several baselines, multiple downstream architectures, and significance tests over 1000 runs. The 5% data experiment in Fig. 4o-r does show the full-data baseline, so that particular criticism in the reader report doesn't hold up. The gains over unconstrained diffusion are believable as an inductive-bias/regularization effect.

Soft spots, in order. First and most important: physical fidelity is validated with the same losses used in training. Fig. 3m reports 'physics violation scores' computed from the same degradation, coupling, range, and discrete penalties that appear in L_phys, Eq. 8. Of course the model scores better than a variant trained without those penalties. That is not independent evidence that the generated trajectories respect real thermodynamics or electrochemistry. The constraints themselves are heuristics: sliding-window monotonicity, slope alignment with one reference sensor, box bounds, and a sine discreteness loss. The limitations section even concedes that non-smooth dynamics and complex control logic remain unresolved. So the headline claim of 'trajectory-level physical consistency' is stronger than what is actually demonstrated. External physical validation or at least held-out physics metrics are needed.

Second, hyperparameter selection is partly circular: the DPCT intensity alpha is swept on the same four evaluation tasks and alpha=3 is chosen. Other loss weights and window sizes are not disclosed in the main text. The code repo link is there, but no commit hash or exact configuration, which makes exact reproduction harder than it should be.

Third, the dataset and task counts are sloppy: 34 datasets versus 21 tasks versus 14 fault classes, and the abstract's 20% for fault diagnosis does not match the 13% reported in the results. These are fixable reporting issues.

The citation pattern and use of prior work look fair. The method is coherent on its own terms; the circularity weakness is about validation, not internal contradiction.

Who this is for: anyone working on industrial time-series augmentation, diffusion for RUL/SOH, or physics-informed generative models. It deserves a serious referee, because the method is novel enough and the empirical base is large. But I would send it back for major revision on validation, hyperparameter disclosure, and claim calibration. If the authors add independent physics metrics, compare against a properly tuned physics-constrained diffusion baseline, and fix the reporting inconsistencies, this could become a solid paper.

Referee Report

3 major / 6 minor

Summary. The paper proposes PhysDGM, a diffusion-based generative model for industrial time series that applies four families of soft physical constraints at every reverse-diffusion step. The constraints are weighted by a training-progress curriculum, and a gradient-guided sampling variant adapts pretrained models to new constraints without retraining. The method is evaluated on a large collection of industrial datasets (turbofan RUL, aero-engine HI, battery SOH, and chemical-process fault diagnosis) for fidelity, physical consistency, downstream utility, and data efficiency. The authors report that synthetic data from PhysDGM improves downstream performance beyond real-data-only training and that augmenting a 5% real-data subset can approach full-data performance.

Significance. If the central claims hold, the paper would be a substantial empirical contribution: it demonstrates a stepwise physics-in-the-loop diffusion framework, provides a 4.4-million-sample synthetic dataset, compares against six strong generative baselines, includes ablations and statistical tests, and states code and data availability. The main weakness is that the 'physical consistency' of the generated data is validated with the same penalty functions used in the training loss, so the central conceptual claim is not independently supported. The downstream utility gains are plausible and well documented, but the physics claim and the '10-20x less training data' claim need stronger evidence before publication.

major comments (3)

[Results — 'PhysDGM generates physically-consistent synthetic data' (Fig. 3; Eqs. 6–8)] The aggregate physics violation scores in Fig. 3m are computed from exactly the four penalty functions that appear in L_phys in Eq. (8): the sliding-window monotonicity ReLU, the slope MSE, the box-range ReLU, and the sine discreteness loss. Because the training objective minimizes these same quantities, the fact that PhysDGM has lower violation scores than PhysDGM_non-phys is an expected optimization outcome rather than independent evidence of physical consistency. The manuscript also does not establish that these soft penalties represent the governing equations of the C-MAPSS, battery, or Tennessee Eastman systems. Please add an external validation — for example, comparing generated health-parameter trajectories against the C-MAPSS physics simulator, checking battery voltage/capacity consistency, or checking mass/energy balances for the chemical process — or replace the 'physical consistency' claim with a claim about adherence to pre-defined domain constraints.
[Abstract and 'PhysDGM achieves full-data performance with 20× less training data'] The statement 'requires 10–20× less training data than existing approaches' is not supported by the reported experiments. The 5% small-sample experiment in Fig. 4o–r trains PhysDGM on a 5% subset and compares only 5% real, 100% real, and 5% real + PhysDGM synthetic; no baseline generator (DiT, DiffWave, Diff-TS, etc.) is trained on the 5% subset. Consequently, the paper does not demonstrate that PhysDGM needs 10–20× less training data than other generative models. Please either add head-to-head comparisons of all generators at 5% data or rephrase the claim as, e.g., 'matching full-data performance using 5% of the real data with PhysDGM-generated augmentation.'
[Method — 'Embedded Physical Constraints' (Eqs. 13–27)] The implemented constraints are heuristic regularizers — sliding-window monotonicity, reference-sensor slope alignment, box range, and sine discreteness — rather than the physical laws themselves. Equations (13), (17)–(20) are motivational, but the losses in Eqs. (15)–(16), (22), (23)–(25), and (27) would apply to any monotone, slope-correlated, bounded, or discrete signal. The Discussion's stated limitation ('our current constraint library is most effective for differentiable or smoothly approximable physical laws... non-smooth dynamics... remain a challenge') underscores this gap. As written, the abstract's claim that PhysDGM 'embeds physical laws directly into each reverse diffusion step' is an overclaim. Please either provide per-system evidence that these constraints capture the relevant physics (e.g., direct derivation from the C-MAPSS equations or battery electrochemical models), or explicitly reframe the contribution as physics-inspired regularity constraints and test whether the downstream gains are distinct from generic regularization.

minor comments (6)

[Algorithm 1] Line 10 reads 'Calculate latent variable x_{t−1} according to Eqs.;' with no equation number cited; please complete this reference.
[Eq. (8) vs Eq. (9)] The symbol t denotes the diffusion step in Eq. (8) but the training step in Eq. (9)'s definition of Φ(t); this notational clash makes the dynamic weighting mechanism ambiguous and should be clarified (e.g., use τ for training progress).
[Data overview] The abstract says 34 datasets, but the components described in the text and Fig. 1e — 4 turbofan, 4 aero-engine, 6 battery, and 21 chemical fault tasks — sum to 35; please clarify how the 34 count is obtained.
[Eq. (12)] The gradient scaling in Eq. (12) uses √(1−α_t), while the DDPM update in Eq. (10) uses √(1−\bar α_t); if this is intentional, state why; otherwise it is likely a typo.
[Fig. 3] The text refers to Fig. 3k for the aggregate physics violation scores, but Fig. 3k and l are the discrete-constraint frequency panels; please renumber the panels or the references.
[Extended Data Fig. 2] The description of the PINN baseline says it 'generates data by optimizing neural networks under physical constraints in the form of differential equations,' but the experimental details of how PINN was adapted for multivariate time-series generation are not given; please add these details.

Circularity Check

1 steps flagged · score 5.0 of 10

Physical-consistency validation is self-referential: the same loss functions used to train PhysDGM are reused as the 'physics violation' metrics, so the reported physical-consistency improvement is expected by construction; the downstream performance claims remain independently evaluated.

fitted input called prediction [Methods, Eq. (8) in 'Stepwise Embedding of Physical Laws in the Generative Process'; Results, 'PhysDGM generates physically-consistent synthetic data', Fig. 3k.]
"L(t)phys = Φ(t) Σ i f_pi(ˆx(t)0) (8) ... Beyond general structural similarity, we quantified the precise adherence to physical laws using constraint-specific violation metrics (See Method section Embedded Physical Constraints for details). The integration of physical constraints significantly mitigated physical inconsistencies, reducing violation metrics by 29.7%, 45.6%, 36.7%, and 70.0%, respectively (Fig. 3k)."

The claim that PhysDGM produces physically consistent trajectories is supported by reduction in 'physics violation scores'. Those scores are computed from the same f_pi functions that Eq. (8) explicitly minimizes as the training objective L_phys. A model trained to minimize Σ f_pi(ˆx0) will by construction report lower values of those same functions than an unconstrained baseline; this is an ablation of a regularizer, not independent evidence that the trajectories obey real physical laws. The four constraints are soft heuristics (sliding-window monotonicity, slope alignment, box ranges, sine discreteness), and the paper's own limitations concede that non-smooth dynamics and complex control logic are not handled, so 'physical consistency' reduces to adherence to the paper's own soft losses.

full rationale

The evaluation of physical consistency is internally circular: Eq. (8) defines the training regularizer L_phys from the constraint functions f_pi, and the validation in Fig. 3k reports 'physics violation scores' derived from those same functions. Thus the reduction in violation scores for PhysDGM over PhysDGM_non-phys is an expected optimization outcome, not a check against an independent physical model or external governing equations. The same issue applies to the gradient-guided sampling variant, which adjusts the noise prediction using gradients of the same f_pi functions. However, the paper's downstream performance claims are not circular: synthetic data are evaluated on held-out real test sets across RUL prediction, HI estimation, SOH assessment, and fault diagnosis, and the comparisons with real-only training and other generators are externally meaningful. The self-citations in the paper (e.g., Diff-MTS and related prior work by the same group) are used as background or baseline pointers and are not load-bearing for the derivation. Overall, the circularity is partial: it affects the physical-consistency component of the central claim, while the downstream augmentation results retain independent content.

Assumptions & free parameters 7 free parameters · 7 assumptions · 0 invented entities

The method introduces no new physical entities, forces, or conserved quantities. Its load-bearing ingredients are the hand-designed soft physics penalties and the diffusion training and sampling procedures. Most physics loss weights are undisclosed, and the schedule parameter alpha is fitted to downstream performance. The central claim therefore rests on several fitted or heuristic components rather than on a parameter-free derivation.

free parameters (7)

Constraint intensity transition factor alpha = 3, selected by sweep over 1 to 5
Eq 9 defines the DPCT schedule; alpha=3 is chosen by minimizing downstream RMSE on RUL tasks in Extended Data Fig 2e-h, so downstream performance partly depends on a fitted hyperparameter.
Monotonic loss gains alpha1 and alpha2 = Not disclosed
Eqs 15-16 weight the degradation monotonicity penalties; the values are chosen by hand and not reported.
Range loss weight alpha = Not disclosed
Eq 25 combines the lower and upper bound penalties; the weighting coefficient is not reported.
Discrete loss weight alpha = Not disclosed
Eq 27 scales the sine-based discreteness penalty; the value is not reported.
Sliding window k for degradation loss = Not disclosed
Eq 14 uses a sliding window size k to compare early and late segments; the value is chosen by hand and not reported.
Range boundaries k_UB and k_LB = Per-dataset normalized bounds
Eqs 23-24 enforce box constraints whose boundaries are read from observed training data, making them dataset-dependent inputs rather than physical constants.
Number of discrete levels k_disc = Per-sensor
Eq 26 expands and applies modulo to enforce discreteness; k_disc is taken from the original data and varies by sensor.

assumptions (7)

standard math Markov Gaussian forward and reverse diffusion structure of DDPM
Adopted from Ho et al. 2020 and Song et al. 2021 without proof, used throughout Eqs 1-4.
domain assumption Predicted clean signal xhat_0 from the noise prediction is a valid proxy for computing physics losses at every reverse step
Eq 8 applies physics losses to xhat_0 reconstructed from eps_theta; early reverse steps yield crude reconstructions, so gradients from this proxy may be unreliable.
ad hoc to paper Monotonic trend, slope coupling, box range, and discreteness penalties capture the governing physical laws
Methods Embedded Physical Constraints; Bernoulli and isentropic relations in Eqs 17-20 are used only to motivate slope alignment, not enforced as governing equations.
domain assumption Regularizing each reverse step with the same soft losses does not distort the learned data distribution
The DPCT schedule assumes likelihood fitting first and physics later, but no proof is given that the weighted objective remains a valid denoising or score objective.
domain assumption Synthetic training data and downstream test data do not overlap
The paper does not state the exact train and test split used when training PhysDGM versus evaluating downstream models; leakage would inflate the reported gains.
domain assumption Battery degradation constraint Loss_deg in Eq 28 enforces physical consistency
The ReLU product-of-increments regularizer is a heuristic trend-alignment penalty, not an electrochemical degradation model.
domain assumption The 34 selected datasets are representative and not cherry-picked
No inclusion criteria, per-dataset failure analysis, or full list of all datasets with results is provided in the main text.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Physics-informed Diffusion Generative Model for Time-Series Data Synthesis in Dynamic Systems." pith.science (2026). https://pith.science/paper/6HGGTYXI

@misc{pith2026260810941,
  author       = {Pith},
  title        = {Pith review of: Physics-informed Diffusion Generative Model for Time-Series Data Synthesis in Dynamic Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6HGGTYXI}},
  note         = {Machine review of arXiv:2608.10941}
}

read the original abstract

Industrial time-series signals, such as turbine temperature and rotational speed in aero-engines, are essential for monitoring the health and operational status of complex dynamical systems. However, collecting such data is often limited by harsh environments (e.g., high temperature and high pressure) and the high cost of experimental testing. To address this challenge, we introduce PhysDGM, a stepwise physics-embedded diffusion generative model for synthesizing time-series data that are consistent with the underlying physical laws of dynamical systems. PhysDGM embeds physical laws directly into each reverse diffusion step of the generative process, ensuring trajectory-level physical consistency, rather than enforcing constraints only at the final output. A large-scale AI-synthetic dataset (4.4 million samples, 20x scale-up) constructed by PhysDGM demonstrates strong fidelity across 34 datasets spanning turbofan engines, aero-engines, batteries, and chemical processes. After incorporating the synthetic data, the downstream task performance substantially surpassed that using real data alone by 48% for remaining useful life prediction, 15% for health indicator estimation, 22% for state-of-health assessment, and 20% for fault diagnosis. Moreover, it requires 10-20x less training data than existing approaches, substantially reducing the high cost of data collection in dynamical systems. We further demonstrate PhysDGM's potential in identifying early-stage faults in aero-engines by incorporating AI-synthesized data. In summary, PhysDGM provides a solid foundation for generating physically consistent industrial time-series, paving the way for expanding physics-guided AI into diverse data-scarce environments, including both industrial machinery and complex chemical reaction dynamics.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 46 canonical work pages

[1]

Smart manufacturing must embrace big data.Nature544, 23–25 (2017)

Kusiak, A. Smart manufacturing must embrace big data.Nature544, 23–25 (2017)

work page 2017
[2]

Li, Z.et al.Learning spatiotemporal dynamics with a pretrained generative model.Nature Machine Intelligence6, 1566–1579 (2024)

work page 2024
[3]

& Wang, J

Yin, C., Xi, J., Sun, R. & Wang, J. Location privacy protection based on differential privacy strategy for big data in industrial internet of things.IEEE Transactions on Industrial Informatics 14, 3628–3636 (2017)

work page 2017
[4]

Ng, M.-F., Zhao, J., Yan, Q., Conduit, G. J. & Seh, Z. W. Predicting the state of charge and health of batteries using data-driven machine learning.Nature Machine Intelligence2, 161–170 (2020)

work page 2020
[5]

& 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.Nature Communications15, 4332 (2024)

work page 2024
[6]

Qu, M.et al.Controlling process instability for defect lean metal additive manufacturing.Nature communications13, 1079 (2022)

work page 2022
[7]

A.et al.Active learning framework to optimize process parameters for additive- manufactured ti-6al-4v with high strength and ductility.Nature Communications16, 931 (2025)

Lee, J. A.et al.Active learning framework to optimize process parameters for additive- manufactured ti-6al-4v with high strength and ductility.Nature Communications16, 931 (2025)

work page 2025
[8]

& Huang, B

Jiang, Q., Chen, S., Yan, X., Kano, M. & Huang, B. Data-driven communication efficient distributed monitoring for multiunit industrial plant-wide processes.IEEE Transactions on Automation Science and Engineering19, 1913–1923 (2021)

work page 2021

Show all 48 references

[9]

Bitharas, I.et al.The interplay between vapour, liquid, and solid phases in laser powder bed fusion.Nature Communications13, 2959 (2022)

2022
[10]

& Zeng, P

Liu, S., Song, C., Wu, T. & Zeng, P. A lightweight fault diagnosis method of beam pumping units based on dynamic warping matching and parallel deep network.IEEE Transactions on Systems, Man, and Cybernetics: Systems(2023)

2023
[11]

& Zhao, C

Zheng, J., Chen, X. & Zhao, C. Interval-aware probabilistic slow feature analysis for irregu- lar dynamic process monitoring with missing data.IEEE Transactions on Systems, Man, and Cybernetics: Systems(2023)

2023
[12]

V., Bowyer, K

Chawla, N. V., Bowyer, K. W., Hall, L. O. & Kegelmeyer, W. P. Smote: synthetic minority over-sampling technique.Journal of artificial intelligence research16, 321–357 (2002)

2002
[13]

& Viktor, H

Guo, H. & Viktor, H. L. Learning from imbalanced data sets with boosting and data generation: the databoost-im approach.ACM Sigkdd Explorations Newsletter6, 30–39 (2004)

2004
[14]

Pourhabib, A., Mallick, B. K. & Ding, Y. Absent data generating classifier for imbalanced class sizes.Journal of Machine Learning Research16, 2695–2724 (2015)

2015
[15]

Nature Machine Intelligence1–14 (2025)

Chen, Z.et al.Versatile cardiovascular signal generation with a unified diffusion transformer. Nature Machine Intelligence1–14 (2025)

2025
[16]

& Yang, C

Ren, L., Wang, H., Li, J., Tang, Y. & Yang, C. Aigc for industrial time series: From deep- generative models to large-generative models.IEEE Transactions on Systems, Man, and Cybernetics: Systems(2025)

2025
[17]

& Gui, W

Pan, Z., Wang, Y., Cao, Y. & Gui, W. Vae-based interpretable latent variable model for process monitoring.IEEE Transactions on Neural Networks and Learning Systems35, 6075–6088 (2023)

2023
[18]

Lin, S.et al.Anomaly detection for time series using vae-lstm hybrid model.In proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing(Ieee, 2020)

2020
[19]

& He, Y.-L

Chen, L., Xu, Y., Zhu, Q.-X. & He, Y.-L. Adaptive Multi-Head Self-Attention Based supervised V AE for Industrial Soft Sensing With Missing Data.IEEE Transactions on Automation Science 21 and Engineering1–12
[20]

He, Y.-L., Li, X.-Y., Ma, J.-H., Zhu, Q.-X. & Lu, S. Attribute-relevant distributed variational autoencoder integrated with LSTM for dynamic industrial soft sensing.Engineering Applications of Artificial Intelligence119, 105737
[21]

& Laili, Y

Ren, L., Wang, H. & Laili, Y. Diff-mts: Temporal-augmented conditional diffusion-based aigc for industrial time series toward the large model era.IEEE Transactions on Cybernetics54, 7187–7197 (2024)

2024
[22]

& Yan, X

Yang, C., Wang, T. & Yan, X. Ddmt: Denoising diffusion mask transformer models for multivariate time series anomaly detection.arXiv preprint arXiv:2310.08800(2023)

2023 arXiv
[23]

& Catanzaro, B

Kong, Z., Ping, W., Huang, J., Zhao, K. & Catanzaro, B. Diffwave: A versatile diffusion model for audio synthesis.In proceedings of International Conference on Learning Representations (OpenReview.net, 2021)

2021
[24]

Alcaraz, J. L. & Strodthoff, N. Diffusion-based time series imputation and forecasting with structured state space models.In proceedings of Transactions on Machine Learning Research (OpenReview.net, 2022)

2022
[25]

Fei, N.et al.Towards artificial general intelligence via a multimodal foundation model.Nature Communications13, 3094 (2022)

2022
[26]

Qiao, Z., Nie, W., Vahdat, A., Miller III, T. F. & Anandkumar, A. State-specific protein– ligand complex structure prediction with a multiscale deep generative model.Nature Machine Intelligence6, 195–208 (2024)

2024
[27]

International Conference on Machine Learning(PMLR, 2024)

Esser, P.et al.Scaling rectified flow transformers for high-resolution image synthesis. International Conference on Machine Learning(PMLR, 2024)

2024
[28]

https://cdn

Betker, J.et al.Improving image generation with better captions.Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf2, 8 (2023)

2023
[29]

& Wang, Y

Wang, H., Ren, L., Li, Y. & Wang, Y. Metaindux-ts: Frequency-aware aigc foundation model for industrial time series.IEEE transactions on neural networks and learning systems(2025)

2025
[30]

Eivazi, H.et al.Diffbatt: A diffusion model for battery degradation prediction and synthesis.In proceedings of Neurips Workshop Foundation Models for Science: Progress, Opportunities, and Challenges(Neurips, 2024)

2024
[31]

Luo, C., Zhang, Z., Zhu, S. & Li, Y. State-of-health prediction of lithium-ion batteries based on diffusion model with transfer learning.Energies16, 3815 (2023)

2023
[32]

& Babenko, A

Kotelnikov, A., Baranchuk, D., Rubachev, I. & Babenko, A. Tabddpm: Modelling tabular data with diffusion models.International Conference on Machine Learning17564–17579 (2023)

2023
[33]

E.et al.Physics-informed machine learning.Nature Reviews Physics3, 422–440 (2021)

Karniadakis, G. E.et al.Physics-informed machine learning.Nature Reviews Physics3, 422–440 (2021)

2021
[34]

& Karniadakis, G

Lu, L., Jin, P., Pang, G., Zhang, Z. & Karniadakis, G. E. Learning nonlinear operators via deep- onet based on the universal approximation theorem of operators.Nature machine intelligence 3, 218–229 (2021)

2021
[35]

& Fink, O

Wei, A. & Fink, O. Integrating physics and topology in neural networks for learning rigid body dynamics.Nature Communications16, 6867 (2025). URL https://doi.org/10.1038/ s41467-025-62250-7

2025
[36]

Rao, C.et al.Encoding physics to learn reaction–diffusion processes.Nature Machine Intelligence 5, 765–779 (2023)

2023
[37]

& Veeramachaneni, K

Patki, N., Wedge, R. & Veeramachaneni, K. The synthetic data vault.In proceedings of IEEE international conference on data science and advanced analytics(IEEE, 2016). 22

2016
[38]

Seo, S.et al.Controlling neural networks with rule representations (Neurips, 2021)

2021
[39]

Xiong, J., Fink, O., Zhou, J. & Ma, Y. Controlled physics-informed data generation for deep learning-based remaining useful life prediction under unseen operation conditions.Mechanical Systems and Signal Processing197, 110359 (2023)

2023
[40]

Nature Reviews Physics6, 320–328 (2024)

Azizzadenesheli, K.et al.Neural operators for accelerating scientific simulations and design. Nature Reviews Physics6, 320–328 (2024)

2024
[41]

ACM/IMS Journal of Data Science1, 1–27 (2024)

Li, Z.et al.Physics-informed neural operator for learning partial differential equations. ACM/IMS Journal of Data Science1, 1–27 (2024)

2024
[42]

Song, Y.et al.Score-based generative modeling through stochastic differential equations.In proceedings of International Conference on Learning Representations(OpenReview.net, 2021)

2021
[43]

& Abbeel, P

Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models.Advances in neural information processing systems33, 6840–6851 (Neurips, 2020)

2020
[44]

& Xie, S

Peebles, W. & Xie, S. Scalable diffusion models with transformers.In proceedings of the IEEE/CVF International Conference on Computer Vision4195–4205 (Ieee, 2023)

2023
[45]

& Qiao, Y

Yuan, X. & Qiao, Y. Diffusion-ts: Interpretable diffusion for general time series generation.In proceedings of International Conference on Learning Representations(OpenReview.net, 2024)

2024
[46]

J.AerodynamicsPitman aeronautical engineering series (Wiley, New York, 1975)

Clancy, L. J.AerodynamicsPitman aeronautical engineering series (Wiley, New York, 1975). Section: XVIII, 610 pages

1975
[47]

M., Hassen, M

Fakhari, S. M., Hassen, M. B. & Mrad, H. Optimizing the operation safety and performance of an axial compressor using fluid-structure coupling and high-performance computing.Results in Engineering18, 101061 (2023)

2023
[48]

Downs, J. J. & Vogel, E. F. A plant-wide industrial process control problem.Computers & chemical engineering17, 245–255 (1993). 23 Aero-engineUnit 2 Aero-engineUnit 5 Aero-engineUnit 10 Aero-engineUnit 16 ChemicalD1 ChemicalD2 ChemicalD3 ChemicalD4 ChemicalD5 0.0 0.2 0.4 0.6 0...

1993

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Smart manufacturing must embrace big data.Nature544, 23–25 (2017)

Kusiak, A. Smart manufacturing must embrace big data.Nature544, 23–25 (2017)

work page 2017

[2] [2]

Li, Z.et al.Learning spatiotemporal dynamics with a pretrained generative model.Nature Machine Intelligence6, 1566–1579 (2024)

work page 2024

[3] [3]

& Wang, J

Yin, C., Xi, J., Sun, R. & Wang, J. Location privacy protection based on differential privacy strategy for big data in industrial internet of things.IEEE Transactions on Industrial Informatics 14, 3628–3636 (2017)

work page 2017

[4] [4]

Ng, M.-F., Zhao, J., Yan, Q., Conduit, G. J. & Seh, Z. W. Predicting the state of charge and health of batteries using data-driven machine learning.Nature Machine Intelligence2, 161–170 (2020)

work page 2020

[5] [5]

& 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.Nature Communications15, 4332 (2024)

work page 2024

[6] [6]

Qu, M.et al.Controlling process instability for defect lean metal additive manufacturing.Nature communications13, 1079 (2022)

work page 2022

[7] [7]

A.et al.Active learning framework to optimize process parameters for additive- manufactured ti-6al-4v with high strength and ductility.Nature Communications16, 931 (2025)

Lee, J. A.et al.Active learning framework to optimize process parameters for additive- manufactured ti-6al-4v with high strength and ductility.Nature Communications16, 931 (2025)

work page 2025

[8] [8]

& Huang, B

Jiang, Q., Chen, S., Yan, X., Kano, M. & Huang, B. Data-driven communication efficient distributed monitoring for multiunit industrial plant-wide processes.IEEE Transactions on Automation Science and Engineering19, 1913–1923 (2021)

work page 2021