Abstract
Wind turbines are often placed in complex terrains, where benefits from orography-related speed up can be capitalized. However, accurately modeling the wind resource over the extended areas covered by a typical wind farm is still challenging over a flat terrain, and over a complex terrain, the challenge can be even be greater. Here, a novel approach for wind resource modeling is proposed, where a linearized flow model is combined with a machine learning approach based on the k-nearest neighbor (k-NN) method. Model predictors include combinations of distance, vertical shear exponent, a measure of the terrain complexity and speedup. The method was tested by performing cross-validations on a complex site using the measurements of five tall meteorological towers. All versions of the k-NN approach yield significant improvements over the predictions obtained using the linearized model alone; they also outperform the predictions of non-linear flow models. The new method improves the capabilities of current wind resource modeling approaches, and it is easily implemented.
Highlights
We choose the best results from Wind Atlas Analysis and Application Program (WAsP) and WindSim, which are summarized in Figure 6, where the cross-validation errors for the prediction of the average wind speed obtained with WAsP (a) and WindSim (b) are shown
Average WAsP results have an error of about 5%, the k-nearest neighbor (k-neural networks (NN)) error is about 1.5%, with the best-ranking method yielding an error of only 1.29%
The ensemble approach consists of a linear superposition of models kNNa and kNNb, where kNNa is driven by the observed wind speed data at each location, and kNNb works on the transferred microclimates from each predictor tower to the target location
Summary
Academic Editor: Davide Astolfi and Eugen Rusu. In the case of wind energy, the accurate assessment of the wind resource at a prospective wind farm site continues to be key to successful development and competitive pricing. While traditional modeling approaches [5,6] for wind project development [7] are often sufficiently accurate in flat or mildly rolling landscapes, wind power predictions in complex terrain are considerably more challenging [6,8,9,10,11]. A 5% mean error in predicted wind speed can be the threshold for the approval or rejection of a wind project. Traditional modeling approaches based on linearized flow solvers are nowadays often complemented with
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have