基于神经网络的跑道结冰预测

Prediction of Runway Icing Based on Neural Network

针对机场道面结冰严重影响机场运营及飞行安全的现状,提出了一种基于BP神经网络的积冰预测模型。通过建立室外模拟机场跑道实验平台,控制风速和降水量的变化并进行相应的实验,得到该模型的误差精度在15%左右。相比已有的对道面结冰问题初步研究的模型,引入了输出域,有效避免了输出域内小概率预测结果的随机性与不确定性,提高了结冰预测的可靠性和准确性。

In view of the severe impact of airport icing on airport operation and flight safety,a prediction model based on BP neural network is proposed in this paper.The changes of wind speed and precipitation are controlled and the corresponding experiments are carried out.The conclusion is obtained that the accuracy of the model is about 15%.Compared with the existing model,the output domain is introduced in this paper,which effectively avoids the randomness and uncertainty of the small probability prediction results.