Evaluation of three potential machine learning algorithms for predicting the velocity and turbulence intensity of a wind turbine wake

Purohit, Shantanu; Ng, E. Y. K.; Kabir, Ijaz Fazil Syed Ahmed

doi:10.1016/j.renene.2021.11.097

Cited by 29 publications

(13 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For more details, see, e.g., Refs. 58,61 . In this study, we have developed and utilized an in-house ML code using Python version 3.8.3 to implement the ML algorithm utilized in the data-driven CFD simulations.…”

Section: Xgboost Algorithmmentioning

confidence: 99%

Data-driven quantification of model-form uncertainty in Reynolds-averaged simulations of wind farms

Eidi,

Zehtabiyan-Rezaie,

Ghiassi

et al. 2022

Preprint

View full text Add to dashboard Cite

Computational fluid dynamics using the Reynolds-averaged Navier-Stokes (RANS) remains the most costeffective approach to study wake flows and power losses in wind farms. The underlying assumptions associated with turbulence closures is one of the biggest sources of errors and uncertainties in the model predictions. This work aims to quantify model-form uncertainties in RANS simulations of wind farms by perturbing the Reynolds stress tensor through a data-driven machine-learning technique. To this end, the extreme gradient boosting algorithm is validated and employed to predict the perturbation amount and direction of the modeled Reynolds stress toward the limiting states of turbulence on the Barycentric map such that the perturbed Reynolds stress returns a more accurate representation of the Reynolds stress. The model is trained on high-fidelity data obtained from large-eddy simulation of a specific wind farm, and it is tested on two other (unseen) wind farms with distinct layouts. The results indicate that, unlike the data-free approach in which a uniform and constant perturbation amount is applied to the entire computational domain, the proposed framework yields an optimal estimation of the uncertainty bounds for the RANS-predicted quantities of interest, including the wake velocity, turbulence intensity, and power losses in wind farms.

show abstract

Section: Xgboost Algorithmmentioning

confidence: 99%

Data-driven quantification of model-form uncertainty in Reynolds-averaged simulations of wind farms

Eidi,

Zehtabiyan-Rezaie,

Ghiassi

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…In Ref. [4] three machine learning techniques are assessed for predicting the wind speed and turbulence intensity in the wake of a stand-alone wind turbine subject to a uniform non-turbulent inflow condition under different inlet velocities.…”

Section: Introductionmentioning

confidence: 99%

Estimation of wind turbine wakes with generative-adversarial networks

Bove

López

Toutouh

et al. 2023

J. Phys.: Conf. Ser.

View full text Add to dashboard Cite

The interaction of the atmospheric boundary layer with wind turbines and among wind turbines in a wind farm is a complex phenomenon. Its analysis is fundamental for wind energy development. In this work, a generative adversarial network has been developed to predict the mean streamwise velocity component at hub height in the wake of a wind turbine based on the mean streamwise velocity two rotor diameters upstream. The dataset used to train and test the model is obtained from Large Eddy Simulations (LES) of a wind farm comprising 15 wind turbines under different inlet conditions. The method is able to predict accurately the mean streamwise velocity in the wake of one wind turbine. When the model is used to predict the mean streamwise velocity at hub height of the whole wind farm based on the inlet condition, the difference between the model predictions and LES results are larger as looking downstream.

show abstract

“…With the rapid development of machine learning, its powerful nonlinear modeling ability and fast prediction ability enable it to have great potential in wind turbine wake prediction, which has been the research focus in literature. [24][25][26] This paper adopts an artificial neural network (ANN) wake model incorporating the wind turbine control parameters of tip speed ratio λ and pitch angle γ, which allows to predict the wake velocity field of a single wind turbine with high efficiency and great accuracy. 27 Combined with the best wake superposition model selected, the combined wake model can effectively predict the wake velocity field of multiple wind turbines and hence rapidly calculate the power of each wind turbine based on incoming wind scenario and control parameters.…”

Section: Introductionmentioning

confidence: 99%

Study of cooperative wake control for multiple wind turbines under variable wind speeds/directions

Zhang,

Xu,

Luo

et al. 2023

Proceedings of the Institution of Mechanical Engineers, Part A:

View full text Add to dashboard Cite

In the wind farm control field, wind turbines are normally manipulated to maximize the individual power production which is named the greedy control. However, this greedy control method can lead to massive losses of total wind farm power production, mainly caused by the wake interference between multiple wind turbines. To this end, the cooperative wake control, which seeks the maximum total power production by coordinating each individual wind turbine at the global optimum operation point, can greatly improve the wind farm output performance. In this paper, we investigate the effectiveness of two different cooperative wake control strategies, i.e., instantaneous control and wind-interval based (WIB) control under variable wind speeds/directions scenario. These two cooperative control strategies are achieved based on the power de-rating operation to the upstream wind turbines. Taking three in-line wind turbines as an example, the control parameters of the two upstream wind turbines are cooperatively optimized while the downstream third wind turbine operates at the maximum power coefficient. To account for the multiple wind turbines wake interference, an artificial neural network (ANN) wake model characterized by the fast computational efficiency and great accuracy, in combination with the best wake superposition model chosen to quantify multiple wake effect, is proposed for the control optimization. By comparing to the baseline greedy control, it shows that both cooperative control strategies are effective in improving the power production of the wind farm. More specifically, the WIB control can maintain the power production at the same level of instantaneous control with a maximum difference less than 3%, while it reduces the operating difficulty to a large extent which greatly facilitates its application under realistic more complex wind scenarios.

show abstract

Evaluation of three potential machine learning algorithms for predicting the velocity and turbulence intensity of a wind turbine wake

Cited by 29 publications

References 41 publications

Data-driven quantification of model-form uncertainty in Reynolds-averaged simulations of wind farms

Data-driven quantification of model-form uncertainty in Reynolds-averaged simulations of wind farms

Estimation of wind turbine wakes with generative-adversarial networks

Study of cooperative wake control for multiple wind turbines under variable wind speeds/directions

Contact Info

Product

Resources

About