A Refined Prediction Method of Short Wind Velocity Along High-Speed Rail Based on Transformer

To solve the problem of wind forecasting and early warning along high-speed railway, we proposed a short-term refined gale prediction method based on Transformer. Firstly, in order to distinguish the topography of different monitoring sites along the railway, we use K-means clustering algorithm to classify the data of different monitoring sites. Secondly, the Inverse Distance Weight (IDW) interpolation method is used to reduce the spatial scaling of meteorological forecast data. Finally, the trained Transformer model makes short-term refined gale prediction. Through the comparative experiment between prediction and observation data, the results show that the proposed method can achieve 67% wind speed prediction accuracy. The proposed algorithm is compared with LSTM and XGBoost, and evaluated by root-mean-square error, which shows that the proposed algorithm is obviously superior to other algorithms.