A LiDAR-Assist Yaw Misalignment Compensation Method Based on Machine Learning Algorithms

Yaw misalignment is common in the yaw system control of wind turbines, resulting in reduced power generation efficiency and increased load. The advantage of LiDAR is that the accuracy of wind measurement is greatly improved, while its disadvantage is that the cost remains high. In this paper, an efficient machine learning method for estimating LiDAR measurement, XGBoost, is used to establish the data-driven wind turbine yaw misalignment compensation framework. To choose a suitable algorithm, four machine learning algorithms are studied, namely Linear Regression, Random Forest, Gradient Boosting and XGBoost. Moreover, to verify the effectiveness of the proposed method, a 2MW WT was taken as the research object for filed testing, and the power generation was quantitatively evaluated under the three strategies of wind vane control, LiDAR control and ML algorithm control. The results show that the proposed yaw misalignment compensation method is feasible for field application. In this way, the yaw-to-wind error can be mitigated even after the LiDAR is removed.