Study on Wind Forecasting for Beijing 2022 Winter Olympic Games Based on Unsupervised Machine Learning

The accurate prediction of wind speed is an important part of the weather forecasting service for the Winter Olympic Games. The Zhangjiakou zone of the Beijing 2022 Winter Olympic Games is located in the complex terrain of the Dama Mountains, and the current numerical forecasts have large deviations in forecasting wind speed in small-scale complex terrain. Based on the unsupervised machine learning method, this paper performs historical modeling on the basis of five element forecasts of two numerical forecasts for nine stations in the Zhangjiakou Zone of the Beijing 2022 Winter Olympic Games, thus achieving objective correction to the numerical forecasts of average winds and extreme winds. The results show that the unsupervised machine learning-based winds forecasts during the Beijing 2022 Winter Olympics Test Games are better, with significant improvements over the original numerical forecasts in terms of mean absolute error and mean squared error.