基于XGBoost-SVR组合模型的气象能见度预测OA

With the acceleration of urbanization and the intensification of global climate change,accurate visibility prediction under the coupling of atmospheric pollution and complex meteorological conditions has become a key technical challenge for environmental management.In order to improve the accuracy of dy-namic urban meteorological visibility prediction,this study takes Beijing as an empirical object,screens 10 key meteorological parameters such as daily average temperature,daily humidity,daily wind speed,sea level pressure,and establishes a combined XGBoost-SVR machine learning prediction model by the stacked generalization method,the dynamic weighted fusion method,and the residual learning method,respectively.The prediction results of these three combined models are systematically evaluated and com-pared with the single machine learning model and among the combined models.The results show that the XGBoost-SVR combined model based on stacked generalization method achieves the highest prediction accuracy,with an accuracy of 84.02%,which is3.70%higher than the second best model.Additional-ly,the XGBoost-SVR combined model significantly outperforms the single machine learning model and other combined model algorithms.