Three-dimensional spatiotemporal wind field reconstruction based on physics-informed deep learning

Zhang, Jincheng; Zhao, Xiaowei

doi:10.1016/j.apenergy.2021.117390

Cited by 42 publications

(11 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We demonstrate that 2D and 3D TC vortices can be constructed using realistic and minimal obersvations, and we also show that a combination of artificially sampled model output combined with real-time observations can be used by a PINN to reconstruct the full 3D structure of a real TC. In all of these cases, PINN RMSE were less than 2.3 m s −1 , which is much less than 10% of the maximum wind speeds observed by the target storm (better than the results found in Zhang and Zhao 44,45 ).…”

Section: Main Textcontrasting

confidence: 51%

“…They are especially useful for inverse problems, which makes them aptly suited for DA 12 . PINNs have been applied to reconstruct wind fields in an idealized setting 44 -Zhang and Zao (2021) reconstruct the 2D 44 and 3D 45 wind field in front of a wind turbine using the Navier-Stokes equations. They found very promising results and root mean square erros (RMSE) that were 10% of the range in wind speeds observed.…”

Section: Main Textmentioning

confidence: 99%

See 1 more Smart Citation

Physics-Informed Neural Networks for Hurricane Reconstruction and Data Assimilation

Eusebi

Vecchi

Lai

et al. 2023

Preprint

View full text Add to dashboard Cite

Tropical cyclones (TCs) are responsible for large-scale loss of life and property, and billions of dollars worth of damage annually. Forecasting is typically accomplished through the use of dynamical models which directly integrate the governing equations of motion. These forecasts, and especially intensity forecasts, rely on accurate initial conditions obtained by assimilating real-time observational data. However, current methods used to generate these initial conditions are computationally expensive and limited, and TC intensity forecasts have much room for improvement. Here, we show that a Physics-Informed Neural Network can provide a promising alternative for data assimilation for TC initial conditions. Using synthetic training data sparsely sampled from hurricanes simulated in a forecast model, a PINN is able to accurately reconstruct full 2- and 3-dimensional wind and pressure fields of a TC. We also demonstrate how a set of sparse, real-time observations, can be used to accurately reconstruct Hurricane Ida. Our results demonstrate how recent advances in computational capabilities and deep learning can be applied to TC data assimilation, offering an accurate and efficient complement to existing data assimilation methods.

show abstract

Section: Main Textcontrasting

confidence: 51%

Section: Main Textmentioning

confidence: 99%

Physics-Informed Neural Networks for Hurricane Reconstruction and Data Assimilation

Eusebi

Vecchi

Lai

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Later, the authors extended the study to consider 3D spatiotemporal inflow wind fields [16]. A neural network with nine hidden layers was used, with four neurons in the first and last hidden layers and 128 neurons in the other layers.…”

Section: Related Workmentioning

confidence: 99%

“…PINNs can be applied for wind field modeling, providing accurate velocity field predictions, while also being computationally efficient as they can solve high dimensional forward and inverse problems in a single shot [15,16]. As well as other soft computing methods, PINNs have specific advantages such as being robust and able to handle incomplete or noisy data and also providing a continuous approximation for solutions [10].…”

Section: Introductionmentioning

confidence: 99%

Physics informed neural networks for wind field modeling in wind farms

Cobelli

Shukla

Nesmachnow

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

This article presents an approach for applying Physics-Informed Neural Networks (PINNs) for modeling wind fields in wind farms. The addressed problem is reconstructing the inflow velocity field for a wind turbine. Several PINN variants are implemented and validated over a real-world case study, trained with sparse numerically simulated velocity data. Results demonstrate that the proposed PINNs can accurately assimilate the numerical simulation data, and compute accurate solutions. The proposed approach is a viable alternative for modeling wind fields in wind farms, requiring significantly lower execution times than standard numerical/simulation methods.

show abstract

“…Note that some recent works use a PGDDM approach to predict the flow field before reaching the wind farm (see, e.g., Refs. 89,90 ). These works have the same approach and methodology as the papers covered in this review but with a different objective; therefore, they are not covered here.…”

Section: Data-driven Wind-farm Flow Modelingmentioning

confidence: 99%

Data-driven fluid mechanics of wind farms: A review

Zehtabiyan-Rezaie,

Iosifidis,

Abkar

2022

Preprint

View full text Add to dashboard Cite

With the growing number of wind farms over the last decades and the availability of large datasets, research in wind-farm flow modeling -one of the key components in optimizing the design and operation of wind farms -is shifting towards data-driven techniques. However, given that most current data-driven algorithms have been developed for canonical problems, the enormous complexity of fluid flows in real wind farms poses unique challenges for data-driven flow modeling. These include the high-dimensional multiscale nature of turbulence at high Reynolds numbers, geophysical and atmospheric effects, wake-flow development, and incorporating wind-turbine characteristics and wind-farm layouts, among others. In addition, data-driven wind-farm flow models should ideally be interpretable and have some degree of generalizability. The former is important to avoid a lack of trust in the models with end-users, while the most popular strategy for the latter is to incorporate known physics into the models. This article reviews a collection of recent studies on wind-farm flow modeling covering both purely data-driven and physics-guided approaches. We provide a thorough analysis of their modeling approach, objective, and methodology, and specifically focus on the data utilized in the reviewed works.

show abstract

Three-dimensional spatiotemporal wind field reconstruction based on physics-informed deep learning

Cited by 42 publications

References 35 publications

Physics-Informed Neural Networks for Hurricane Reconstruction and Data Assimilation

Physics-Informed Neural Networks for Hurricane Reconstruction and Data Assimilation

Physics informed neural networks for wind field modeling in wind farms

Data-driven fluid mechanics of wind farms: A review

Contact Info

Product

Resources

About