Parameterization of wind evolution using lidar

Abstract

Abstract. Wind evolution, i.e., the evolution of turbulence structures over time, has become an increasingly interesting topic in recent years, mainly due to the development of lidar-assisted wind turbine control, which requires accurate prediction of wind evolution to avoid unnecessary or even harmful control actions. Moreover, 4D stochastic wind field simulations can be made possible by integrating wind evolution into standard 3D simulations to provide a more realistic simulation environment for this control concept. Motivated by these factors, this research aims to investigate the potential of Gaussian process regression in the parameterization of wind evolution. Wind evolution is commonly quantified using magnitude-squared coherence of wind speed and is estimated with lidar data measured by two nacelle-mounted lidars in this research. A two-parameter wind evolution model modified from a previous study is used to model the estimated coherence. A statistical analysis is done for the wind evolution model parameters determined from the estimated coherence to provide some insights into the characteristics of wind evolution. Gaussian process regression models are trained with the wind evolution model parameters and different combinations of wind-field-related variables acquired from the lidars and a meteorological mast. The automatic relevance determination squared exponential kernel function is applied to select suitable variables for the models. The performance of the Gaussian process regression models is analyzed with respect to different variable combinations, and the selected variables are discussed to shed light on the correlation between wind evolution and these variables.

The COTUR project: remote sensing of offshore turbulence for wind energy application

“…3.4. 2020; Chen et al, 2021). In the present study, such an investigation can be conducted when the elevation angle is 2 • , such that d z d x .…”

“…Schlez and Infield (1998) suggested that for a given turbulence intensity, the decay coefficient of the lateral co-coherence is independent of the mean wind speed. In the surface layer, the dependence of the decay coefficients on the spatial separation and measurement height has been highlighted for both lateral and vertical separations (Kanda and Royles, 1978;Perry et al, 1978;Shiotani et al, 1978;Kristensen et al, 1981;Cheynet et al, 2017b;Bowen et al, 1983;Cheynet, 2018), reflecting the increase in the size of the eddies further away from the ground. Equation ( 7) is a two-parameter function where C x and C y need both to be determined from measurements.…”

“…3.4. Besides, the co-coherence can increase with height as the surface blocks the flow and distorts eddies (Kanda and Royles, 1978;Bowen et al, 1983;Cheynet, 2018). A decrease in the co-coherence with the scanning distance is also possible because the CNR reduces as r increases, which may be related to the presence of uncorrelated noise in the velocity records.…”

4D wind field generation for the aeroelastic simulation of wind turbines with lidars

“…The first two models are empirical models following a similar simple exponential form. They are all based on the same assumption that turbulent eddies decay exponentially but consider different model parameters (Simley and Pao, 2015;Chen et al, 2021). The last model is a physical-deduced model which assumes that the coherence can be modelled with the square of the probability that an eddy observed at the first location can also be observed at the second location.…”

“…In recent years, Simley and Pao (2015) modified the exponential wind evolution model by including a second parameter to adjust the coherence at very low frequency, taking a similar model form as the coherence model for transverse and vertical separations proposed by Thresher et al (1981), and suggested a model to determine both model parameters based on LES simulations. On the basis of Simley and Pao's (2015) model, Chen et al (2021) suggested a concept to build parameterization models using supervised machine learning (ML) algorithms and presented the results of Gaussian process regression models. In a following work, the performance of different ML algorithms was compared considering their computational efficiencies (Chen and Cheng, 2020).…”