基于BP神经网络的南通机场风速预报模型

Wind Speed Prediction Model of Nantong Airport Based on BP Neural Network

本文利用2016—2021年南通兴东机场整点报文数据,采用BP人工神经网络建立南通机场风速预报模型,将南通机场风速进行分级,并进行6 h和24 h模拟预报,以期对南通机场风速分级预报数值进行修订并讨论风速分级预报的准确率。结果如下:当风为1级、2级风时,风速无须修订;3级风时,6 h预报风速可修订增加0.6 m/s,24 h可修订增加3.8 m/s;而4级风时,6 h预报风速和24 h预报风速则分别修订增加1.0 m/s和4.8 m/s。利用2020年样本资料进行检验,在预报风速数值分级修订后,准确率结果如下:对民航安全运行有影响的4级风6 h准确率为66%,24 h为64%。再利用模型对2021年12月进行试预报可得:4级风6 h准确率为75%,24 h为73%。上述结果表明,该神经网络模型可以为南通机场风速预报提供一定的依据。

The forecasting model of wind velocity was built based on BP neural network by using data from the aviation routine weather report of meteorological terminal at Nantong Airport during 2016 to 2021.The correction to the model predicted wind velocity was proposed and prognosing accuracy was improved for the forecast of 6 h and 24 h.The results can be summarized as follows:The model predicted wind velocity was needed to be revised when the wind force scale at 3 or 4.When the scale at 3,the wind velocity should be increased 0.6 m/s and 3.8 m/s for the predictions of 6 h and 24 h,respectively;while for the scale 4,the wind velocity should be increased 1.0 m/s and 4.8 m/s for the predictions of 6 h and 24 h,respectively.The revised predictions were validated against the data from the entire year of 2020 and the whole month in December of 2021.For the wind at scale 4,which largely affects the security of civil aviation,the accuracy was 66%and 64%for the corrected predictions of 6 h and 24 h,respectively,in the former case,but increased significantly to 75%and 73%in the later case.This suggests that the BP model can be used to forecast the wind velocity.