With the growing demand for renewable energy, optimizing wind turbine performance requires accurate understanding of upstream wind flow. This study introduces a model for characterizing upwind flow using two years of raw radial wind speed (RWS) data from a fixed 4-beam nacelle-mounted LiDAR system. The model incorporates axial and lateral wind speed components, wind shear exponent (WSE), veer, and wind turbine induction factor. These parameters are optimized by minimizing the mean absolute error (MAE) using stochastic gradient descent and the Adam optimization algorithm. Hypothesis testing for linear veer profiles, power law shear, and induction zone models demonstrated statistical significance at the 95 % confidence level in 99 %, 90 %, and 92 % of cases, respectively. The model achieved an MAE of 0.08 m/s (1.3 %) for reconstructing horizontal RWS. Notable diurnal variations were observed in model parameters; at night, when the atmospheric boundary layer (ABL) is stable, WSE, veer, and axial wind speed components increase, while during the day, ABL instability leads to higher turbulence intensity. The model’s short optimization duration (0.25 s) makes it suitable for realtime applications in wind turbine control, such as alignment and yaw control for wake steering strategies using nacelle-mounted LiDAR.

Keywords