Pith. sign in

REVIEW 2 major objections 1 minor 40 references

UniWind separates a site-calibrated physical prior from latent operating states via routing to forecast day-ahead wind power across farms.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.3

2026-07-03 17:57 UTC pith:2A7267JM

load-bearing objection UniWind's hybrid of site-warped physical prior plus state routing is a reasonable attempt to disentangle meteo effects from operational states in wind forecasting, but the zero-shot transfer story rests on assumptions about the prior that need checking. the 2 major comments →

arxiv 2607.01670 v1 pith:2A7267JM submitted 2026-07-02 cs.LG

UniWind: Toward Unified Day-Ahead Wind Power Forecasting via Physics-Informed State Routing

Ronghui Xu , Tongxin Wu , Guozhen Zhang , Yihan Li , Chenjuan Guo , Bin Yang , Yong Li This is my paper
classification cs.LG
keywords wind power forecastingphysics-informed machine learningstate routingzero-shot transferday-ahead forecastingrenewable energylatent state modeling
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Day-ahead wind power forecasts must disentangle meteorological availability from local effects such as shutdowns and curtailment. Physical models supply constraints that fail to adapt across sites, while data-driven models mix the two sources of variation. UniWind first builds a physical prior by warping a shared power curve monotonically per site and enforcing an upper-bound envelope, then routes the prior through a latent state encoder and state-aware corrector that applies supervised, bounded adjustments for each operating state. Full-shot and cross-farm zero-shot tests on more than twenty real datasets support the claim that this separation yields accurate, transferable predictions. The result would allow one model to serve many farms without per-site retraining, improving grid integration of variable wind resources.

Core claim

The central claim is that a Physical Prior Estimator using site-conditioned monotonic warping of a shared physical power curve, shaped by an upper-bound constraint, produces a reliable soft envelope of available power; this prior is then transformed by a Latent State Encoder and a State-aware Power Corrector that performs knowledge-guided supervised state routing together with bounded state-specific expert corrections, enabling unified day-ahead forecasts that remain accurate under both full-shot training and zero-shot transfer to new wind farms.

What carries the argument

Physics-informed state routing, implemented through the Physical Prior Estimator that creates the site-calibrated upper-envelope prior and the State-aware Power Corrector that applies latent-state adjustments.

Load-bearing premise

A shared physical power curve combined with site-conditioned monotonic warping can produce a reliable physical prior that acts as a soft upper envelope on available wind power and remains valid across different wind farms and operating conditions.

What would settle it

Zero-shot cross-farm experiments in which UniWind shows no accuracy improvement over standard data-driven baselines, or cases where the estimated physical prior exceeds observed power on new farms more often than expected under the upper-bound constraint.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • A single trained model can be applied to new wind farms without retraining while preserving physical consistency.
  • State-specific corrections focus only on deviations from the physical envelope, reducing conflation of meteorological and operational effects.
  • Knowledge-guided supervised routing produces interpretable latent states corresponding to real operating conditions such as curtailment.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same separation of prior and state correction could be tested on solar or hydro generation where physical curves also exist.
  • If the monotonic warping remains stable, the method may lower the data volume needed to commission forecasts at new sites.
  • Extending the shared curve to include additional meteorological variables could further tighten the physical envelope.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The paper proposes UniWind, a physics-informed model for day-ahead wind power forecasting. It introduces a Physical Prior Estimator that combines a shared physical power curve with site-conditioned monotonic warping and a physical upper-bound constraint to form a soft envelope of available power. A Latent State Encoder models operating-state embeddings, and a State-aware Power Corrector applies knowledge-guided supervised state routing with bounded expert corrections to produce final forecasts. The central claim is that this architecture achieves superior accuracy and robustness in both full-shot and cross-farm zero-shot settings across more than 20 real-world datasets.

Significance. If the zero-shot generalization claims hold, the work would represent a meaningful step toward unified forecasting models that transfer physical priors across wind farms without site-specific retraining. The explicit separation of meteorological effects from latent operational states (curtailment, shutdowns) via routing could reduce confounding in data-driven forecasts and improve reliability for power-system operations.

major comments (2)
  1. [Physical Prior Estimator and zero-shot experiments] The manuscript's central claim for cross-farm zero-shot transfer rests on the Physical Prior Estimator producing a reliable, transferable soft upper envelope. However, the description of site-conditioned monotonic warping applied to a single shared power curve does not address how this prior remains valid when target farms differ in turbine models, terrain, or curtailment regimes; without target observations the warping parameters cannot be fitted, raising the risk that downstream state routing receives a systematically biased envelope.
  2. [Methods (Physical Prior Estimator, Latent State Encoder, State-aware Power Corrector)] The abstract and methods summary supply no equations for the monotonic warping function, the physical upper-bound constraint, the state routing mechanism, or the expert correction bounds. This absence prevents evaluation of whether the reported gains are independent of the fitted components or largely restate the training objective, directly undermining assessment of the soundness of the performance claims on the 20+ datasets.
minor comments (1)
  1. [Abstract] The abstract states that experiments demonstrate 'accuracy and robustness' but provides no quantitative metrics, baseline comparisons, or error analysis; these should be summarized with specific numbers and statistical significance even in the abstract.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful review and constructive feedback on UniWind. We address each major comment below and outline revisions that will clarify the technical details and strengthen the zero-shot claims.

read point-by-point responses

  1. Referee: [Physical Prior Estimator and zero-shot experiments] The manuscript's central claim for cross-farm zero-shot transfer rests on the Physical Prior Estimator producing a reliable, transferable soft upper envelope. However, the description of site-conditioned monotonic warping applied to a single shared power curve does not address how this prior remains valid when target farms differ in turbine models, terrain, or curtailment regimes; without target observations the warping parameters cannot be fitted, raising the risk that downstream state routing receives a systematically biased envelope.

    Authors: The site-conditioned monotonic warping is implemented via a metadata-conditioned neural parameterization that maps observable farm attributes (turbine specifications, terrain descriptors, and nominal capacity) to warping parameters. These attributes are available at inference time for any target farm, even in the complete absence of power observations, and the mapping is learned across source farms during training to promote transferability. The shared physical power curve is fixed, while the warping adjusts the shape in a manner that respects physical monotonicity. We will add an explicit subsection in Methods 3.1 detailing the conditioning inputs, the zero-shot inference procedure, and an ablation confirming that performance does not rely on target power data for warping. revision: yes

  2. Referee: [Methods (Physical Prior Estimator, Latent State Encoder, State-aware Power Corrector)] The abstract and methods summary supply no equations for the monotonic warping function, the physical upper-bound constraint, the state routing mechanism, or the expert correction bounds. This absence prevents evaluation of whether the reported gains are independent of the fitted components or largely restate the training objective, directly undermining assessment of the soundness of the performance claims on the 20+ datasets.

    Authors: We agree that the high-level overview in the abstract and initial methods paragraph omits the governing equations, which appear in the detailed subsections. To improve readability, we will insert the key equations directly into the methods summary: the monotonic warping (Eq. 3), physical upper-bound constraint (Eq. 5), supervised state routing (Eq. 8), and bounded expert correction (Eq. 10). This revision will allow immediate assessment of the architectural components without requiring readers to locate later equations. revision: yes

Circularity Check

0 steps flagged

No circularity: model combines external physical prior with learned corrections; claims rest on dataset experiments

full rationale

The derivation chain constructs a physical prior from a shared domain-knowledge power curve plus site-conditioned monotonic warping, then applies learned latent state embeddings and expert corrections to produce forecasts. No equation or component reduces to its own fitted inputs by construction, no self-citation is invoked as a uniqueness theorem, and no prediction is statistically forced from the training objective itself. The central claims are evaluated via full-shot and zero-shot experiments on >20 external real-world datasets, which constitute independent benchmarks outside the model's fitted parameters.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on the existence of a transferable physical power curve, the validity of monotonic warping for site calibration, and the learnability of discrete operating-state embeddings that can be routed without violating physical bounds. These are domain assumptions rather than derived quantities.

free parameters (2)
  • site-conditioned monotonic warping parameters
    Used to adapt the shared physical power curve to each wind farm; values are not stated in the abstract.
  • latent state embeddings
    Learned representations of operating states such as shutdowns and curtailment; dimension and training procedure unspecified.
axioms (2)
  • domain assumption A shared physical power curve provides a useful and transferable prior for available wind power across farms.
    Invoked when constructing the Physical Prior Estimator.
  • domain assumption Operating states can be represented as discrete embeddings that allow bounded expert corrections without violating the physical envelope.
    Required for the State-aware Power Corrector to function as described.

pith-pipeline@v0.9.1-grok · 5759 in / 1574 out tokens · 30201 ms · 2026-07-03T17:57:45.926193+00:00 · methodology

0 comments
read the original abstract

Day-ahead wind power forecasting is essential for cost-effective power-system operation. It is primarily driven by future meteorological conditions while retaining temporal dependencies in power generation. In practice, observed wind-farm power often entangles physically available power with local environmental effects and latent operational states, such as shutdowns and curtailment. Existing physical models provide useful constraints but adapt poorly across wind farms, whereas data-driven models can capture rich correlations but often conflate meteorological effects with state-induced deviations. In this study, we propose UniWind, a wind power forecasting model based on physics-informed state routing. UniWind first employs a Physical Prior Estimator to construct a site-calibrated physical prior by combining site-conditioned monotonic warping with a shared physical power curve. It further applies a physical upper-bound constraint to shape this prior as a soft envelope of available wind power generation. UniWind then proposes a Latent State Encoder to model operating-state embeddings and transforms the physical prior into final power forecasts through a State-aware Power Corrector, which uses knowledge-guided supervised state routing and bounded, state-specific expert correction. Full-shot and cross-farm zero-shot experiments on more than 20 real-world datasets demonstrate the accuracy and robustness of UniWind.

Figures

Figures reproduced from arXiv: 2607.01670 by Bin Yang, Chenjuan Guo, Guozhen Zhang, Ronghui Xu, Tongxin Wu, Yihan Li, Yong Li.

Figure 1
Figure 1. Figure 1: Comparison of UniWind with other wind power forecasting models. Day-ahead wind power forecasting plays a central role in cost-effective power-system operation [36], as electricity-market decisions must be made before power delivery [11, 37]. Compared with conventional time series forecasting [30, 19, 32, 22, 8], day-ahead wind power forecasting exhibits a stronger dependence on meteorological conditions at… view at source ↗
Figure 2
Figure 2. Figure 2: Operational states in a wind power sequence. Existing wind power forecasting methods can be categorized into phys￾ically driven and data-driven approaches. Physical models [9, 16, 34], detailed in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overall framework of UniWind. where T1 and T2 denote the lengths of the historical and forecasting horizons, respectively. Nw is the number of meteorological features, and Ns is the number of static site features, such as longitude, latitude, and rated power. 3.2 Overall Framework As shown in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Performance comparison under varied conditions on the SD_A dataset. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Ablation study on the SD_A dataset. To evaluate the contribution of each component, we construct four variants of UniWind. w/o phy. removes the physical prior and replaces it with a learnable parameter. w/o cor. directly uses the physical prior as the final prediction. w/o upper removes the physical upper-bound constraint from the Physical Prior Estimator. w/o state removes the supervised state routing los… view at source ↗
Figure 6
Figure 6. Figure 6: Ablation Studies. D.3 State Statistics [PITH_FULL_IMAGE:figures/full_fig_p017_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Sensitivity analysis on the SD_A dataset. [PITH_FULL_IMAGE:figures/full_fig_p018_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Case study on the SD_A dataset. The physical prior follows the wind-speed-driven available [PITH_FULL_IMAGE:figures/full_fig_p018_8.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages · 1 internal anchor

  1. [1]

    Chronos-2: From Univariate to Universal Forecasting

    Abdul Fatir Ansari, Oleksandr Shchur, Jaris Küken, Andreas Auer, Boran Han, Pedro Mercado, Syama Sundar Rangapuram, Huibin Shen, Lorenzo Stella, Xiyuan Zhang, et al. Chronos-2: From univariate to universal forecasting.arXiv preprint arXiv:2510.15821, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  2. [2]

    Improving day-ahead solar irradiance time series forecasting by leveraging spatio-temporal context.NeurIPS, 36:2342–2367, 2023

    Oussama Boussif, Ghait Boukachab, Dan Assouline, Stefano Massaroli, Tianle Yuan, Loubna Benabbou, and Yoshua Bengio. Improving day-ahead solar irradiance time series forecasting by leveraging spatio-temporal context.NeurIPS, 36:2342–2367, 2023

    work page 2023

  3. [3]

    Multistage wind-electric power forecast by using a combination of advanced statistical methods.IEEE Trans

    Serkan Buhan and I¸ sık Çadırcı. Multistage wind-electric power forecast by using a combination of advanced statistical methods.IEEE Trans. Ind. Inform., 11(5):1231–1242, 2015

    work page 2015

  4. [4]

    Multivariate wind power time series forecasting with noise-filtering neural odes

    Tianyu Chang, Dongming Chen, and Dongqi Wang. Multivariate wind power time series forecasting with noise-filtering neural odes. InCIKM, pages 221–230, 2025

    work page 2025

  5. [5]

    Windfm: An open-source foundation model for zero-shot wind power forecasting.arXiv preprint arXiv:2509.06311, 2025

    Hang Fan, Yu Shi, Zongliang Fu, Shuo Chen, Wei Wei, Wei Xu, and Jian Li. Windfm: An open-source foundation model for zero-shot wind power forecasting.arXiv preprint arXiv:2509.06311, 2025

    work page arXiv 2025

  6. [6]

    M2gsnet: Multi- modal multi-task graph spatiotemporal network for ultra-short-term wind farm cluster power prediction.Applied Sciences, 10(21):7915, 2020

    Hang Fan, Xuemin Zhang, Shengwei Mei, Kunjin Chen, and Xinyang Chen. M2gsnet: Multi- modal multi-task graph spatiotemporal network for ultra-short-term wind farm cluster power prediction.Applied Sciences, 10(21):7915, 2020

    work page 2020

  7. [7]

    Unraveling spatio-temporal foundation models via the pipeline lens: A comprehensive review.IEEE Trans

    Yuchen Fang, Hao Miao, Yuxuan Liang, Liwei Deng, Yue Cui, Ximu Zeng, Yuyang Xia, Yan Zhao, Torben Bach Pedersen, Christian S Jensen, et al. Unraveling spatio-temporal foundation models via the pipeline lens: A comprehensive review.IEEE Trans. Knowl. Data Eng., 2026

    work page 2026

  8. [8]

    Efficient high-dimensional time series forecasting with transformers: A channel reordering perspective

    Yuchen Fang, Shiyu Wang, Yuxuan Liang, Zhou Ye, Yang Xiang, Yan Zhao, and Kai Zheng. Efficient high-dimensional time series forecasting with transformers: A channel reordering perspective. InWWW, pages 7223–7234, 2026

    work page 2026

  9. [9]

    Previento-a wind power prediction system with an innovative upscaling algorithm

    Ulrich Focken, Matthias Lange, and Hans-Peter Waldl. Previento-a wind power prediction system with an innovative upscaling algorithm. InEWEC, volume 276, 2001

    work page 2001

  10. [10]

    wind-python/windpowerlib: Update release, 2024

    Sabine Haas, Uwe Krien, Birgit Schachler, et al. wind-python/windpowerlib: Update release, 2024

    work page 2024

  11. [11]

    A critical review of wind power forecasting methods—past, present and future.Energies, 13(15):3764, 2020

    Shahram Hanifi, Xiaolei Liu, Zi Lin, and Saeid Lotfian. A critical review of wind power forecasting methods—past, present and future.Energies, 13(15):3764, 2020

    work page 2020

  12. [12]

    A hybrid deep learning-based neural network for 24-h ahead wind power forecasting.Applied Energy, 250:530–539, 2019

    Ying-Yi Hong and Christian Lian Paulo P Rioflorido. A hybrid deep learning-based neural network for 24-h ahead wind power forecasting.Applied Energy, 250:530–539, 2019

    work page 2019

  13. [13]

    A model combining convolutional neural network and lightgbm algorithm for ultra-short-term wind power forecasting.IEEE Access, 7:28309–28318, 2019

    Yun Ju, Guangyu Sun, Quanhe Chen, Min Zhang, Huixian Zhu, and Mujeeb Ur Rehman. A model combining convolutional neural network and lightgbm algorithm for ultra-short-term wind power forecasting.IEEE Access, 7:28309–28318, 2019

    work page 2019

  14. [14]

    Wind power forecasting based on hybrid ceemdan-ewt deep learning method.Renewable Energy, 218:119357, 2023

    Irene Karijadi, Shuo-Yan Chou, and Anindhita Dewabharata. Wind power forecasting based on hybrid ceemdan-ewt deep learning method.Renewable Energy, 218:119357, 2023

    work page 2023

  15. [15]

    Tackling Climate Change with Machine Learning

    Julie Keisler and Etienne Le Naour. Winddragon: Enhancing wind power forecasting with automated deep learning. InWorkshop-"Tackling Climate Change with Machine Learning", ICLR, 2024

    work page 2024

  16. [16]

    A physical approach of the short-term wind power prediction based on cfd pre-calculated flow fields.Journal of Hydrodynamics, Ser

    Li Li, HAN Shuang, WANG Yi-mei, et al. A physical approach of the short-term wind power prediction based on cfd pre-calculated flow fields.Journal of Hydrodynamics, Ser. B, 25(1):56–61, 2013

    work page 2013

  17. [17]

    Solarcube: an integrative benchmark dataset harnessing satellite and in-situ observations for large-scale solar energy forecasting.NeurIPS, 37:3499–3513, 2024

    Ruohan Li, Yiqun Xie, Xiaowei Jia, Dongdong Wang, Yanhua Li, Yingxue Zhang, Zhihao Wang, and Zhili Li. Solarcube: an integrative benchmark dataset harnessing satellite and in-situ observations for large-scale solar energy forecasting.NeurIPS, 37:3499–3513, 2024. 10

    work page 2024

  18. [18]

    Moirai-moe: Empowering time series foundation models with sparse mixture of experts

    Xu Liu, Juncheng Liu, Gerald Woo, Taha Aksu, Yuxuan Liang, Roger Zimmermann, Chenghao Liu, Junnan Li, Silvio Savarese, Caiming Xiong, et al. Moirai-moe: Empowering time series foundation models with sparse mixture of experts. InICML, 2025

    work page 2025

  19. [19]

    itransformer: Inverted transformers are effective for time series forecasting

    Yong Liu, Tengge Hu, Haoran Zhang, Haixu Wu, Shiyu Wang, Lintao Ma, and Mingsheng Long. itransformer: Inverted transformers are effective for time series forecasting. InICLR, 2024

    work page 2024

  20. [20]

    Improv- ing time series forecasting via instance-aware post-hoc revision

    Zhiding Liu, Mingyue Cheng, GuanHao Zhao, Jiqian Yang, Qi Liu, and Enhong Chen. Improv- ing time series forecasting via instance-aware post-hoc revision. InNeurIPS, 2025

    work page 2025

  21. [21]

    Fusionsf: Fuse heterogeneous modalities in a vector quantized framework for robust solar power forecasting

    Ziqing Ma, Wenwei Wang, Tian Zhou, Chao Chen, Bingqing Peng, Liang Sun, and Rong Jin. Fusionsf: Fuse heterogeneous modalities in a vector quantized framework for robust solar power forecasting. InSIGKDD, pages 5532–5543, 2024

    work page 2024

  22. [22]

    A time series is worth 64 words: Long-term forecasting with transformers

    Yuqi Nie, Nam H Nguyen, Phanwadee Sinthong, and Jayant Kalagnanam. A time series is worth 64 words: Long-term forecasting with transformers. InICLR, 2023

    work page 2023

  23. [23]

    A novel wind power forecast model: Statistical hybrid wind power forecast technique (shwip).IEEE Trans

    Mehmet Baris Ozkan and Pinar Karagoz. A novel wind power forecast model: Statistical hybrid wind power forecast technique (shwip).IEEE Trans. Ind. Inform., 11(2):375–387, 2015

    work page 2015

  24. [24]

    Gaussian process power curve models incorporating wind turbine operational variables.Energy Reports, 6:1658–1669, 2020

    Ravi Kumar Pandit, David Infield, and Athanasios Kolios. Gaussian process power curve models incorporating wind turbine operational variables.Energy Reports, 6:1658–1669, 2020

    work page 2020

  25. [25]

    A hybrid wind power forecast- ing model with xgboost, data preprocessing considering different nwps.Applied Sciences, 11(3):1100, 2021

    Quoc Thang Phan, Yuan Kang Wu, and Quoc Dung Phan. A hybrid wind power forecast- ing model with xgboost, data preprocessing considering different nwps.Applied Sciences, 11(3):1100, 2021

    work page 2021

  26. [26]

    Kelmarsh wind farm data, 2022

    Charlie Plumley. Kelmarsh wind farm data, 2022

    work page 2022

  27. [27]

    Penmanshiel wind farm data, 2022

    Charlie Plumley. Penmanshiel wind farm data, 2022

    work page 2022

  28. [28]

    Wind power forecasting for a real onshore wind farm on complex terrain using wrf high resolution simulations.Renewable Energy, 135:674–686, 2019

    Miguel A Prósper, Carlos Otero-Casal, Felipe Canoura Fernández, and Gonzalo Miguez-Macho. Wind power forecasting for a real onshore wind farm on complex terrain using wrf high resolution simulations.Renewable Energy, 135:674–686, 2019

    work page 2019

  29. [29]

    An advanced statistical method for wind power forecasting.IEEE Trans

    George Sideratos and Nikos D Hatziargyriou. An advanced statistical method for wind power forecasting.IEEE Trans. Power Syst., 22(1):258–265, 2007

    work page 2007

  30. [30]

    xpatch: Dual-stream time series forecasting with exponential seasonal-trend decomposition

    Artyom Stitsyuk and Jaesik Choi. xpatch: Dual-stream time series forecasting with exponential seasonal-trend decomposition. InAAAI, volume 39, pages 20601–20609, 2025

    work page 2025

  31. [31]

    Solarmae: A unified framework for regional centralized and distributed solar power forecasting with weather pre-training

    Jin Wang, Bingqing Peng, Wenwei Wang, Yuanjie Hu, Yuejiang Chen, Peisong Niu, and Liang Sun. Solarmae: A unified framework for regional centralized and distributed solar power forecasting with weather pre-training. InCIKM, pages 6093–6101, 2025

    work page 2025

  32. [32]

    Timemixer: Decomposable multiscale mixing for time series forecasting

    Shiyu Wang, Haixu Wu, Xiaoming Shi, Tengge Hu, Huakun Luo, Lintao Ma, James Y Zhang, and JUN ZHOU. Timemixer: Decomposable multiscale mixing for time series forecasting. In ICLR, 2024

    work page 2024

  33. [33]

    Lightgts: A lightweight general time series forecasting model

    Yihang Wang, Yuying Qiu, Peng Chen, Yang Shu, Zhongwen Rao, Lujia Pan, Bin Yang, and Chenjuan Guo. Lightgts: A lightweight general time series forecasting model. InICML, pages 64109–64126. PMLR, 2025

    work page 2025

  34. [34]

    Short-term wind power forecasting based on clustering pre-calculated cfd method.Energies, 11(4):854, 2018

    Yimei Wang, Yongqian Liu, Li Li, David Infield, and Shuang Han. Short-term wind power forecasting based on clustering pre-calculated cfd method.Energies, 11(4):854, 2018

    work page 2018

  35. [35]

    Approaches to wind power curve modeling: A review and discussion.Renew

    Yun Wang, Qinghua Hu, Linhao Li, Aoife M Foley, and Dipti Srinivasan. Approaches to wind power curve modeling: A review and discussion.Renew. Sustain. Energy Rev., 116:109422, 2019

    work page 2019

  36. [36]

    Electric-carbon market coupling and price transmission mechanism in china: An empirical analysis and development barriers study.J

    Jiajun Wu, Yanjun Shen, Ruitong Yang, Hang Fan, and Yunjie Duan. Electric-carbon market coupling and price transmission mechanism in china: An empirical analysis and development barriers study.J. Clean. Prod., 544:147692, 2026. 11

    work page 2026

  37. [37]

    A cross- dataset benchmark for neural network-based wind power forecasting.Renewable Energy, 254:123463, 2025

    Xin Xu, Qinglong Cao, Ruizhe Deng, Zhiling Guo, Yuntian Chen, and Jinyue Yan. A cross- dataset benchmark for neural network-based wind power forecasting.Renewable Energy, 254:123463, 2025

    work page 2025

  38. [38]

    2dxformer: Dual transform- ers for wind power forecasting with dual exogenous variables.arXiv preprint arXiv:2505.01286, 2025

    Yajuan Zhang, Jiahai Jiang, Yule Yan, Liang Yang, and Ping Zhang. 2dxformer: Dual transform- ers for wind power forecasting with dual exogenous variables.arXiv preprint arXiv:2505.01286, 2025

    work page arXiv 2025

  39. [39]

    A novel bidirectional mechanism based on time series model for wind power forecasting.Applied Energy, 177:793– 803, 2016

    Yongning Zhao, Lin Ye, Zhi Li, Xuri Song, Yansheng Lang, and Jian Su. A novel bidirectional mechanism based on time series model for wind power forecasting.Applied Energy, 177:793– 803, 2016

    work page 2016

  40. [40]

    Deep latent state space models for time-series generation

    Linqi Zhou, Michael Poli, Winnie Xu, Stefano Massaroli, and Stefano Ermon. Deep latent state space models for time-series generation. InICML, pages 42625–42643. PMLR, 2023. 12 A Datasets We provide detailed information on the datasets used in this study. The wind power data were collected from 24 wind farms across four geographically distinct regions: Anh...

    work page 2023