离港航班延误时间预测方法

Forecasting Method of Departure Flight Delay Time

在如今的民航运行体系里,航班延误已经成了机场和航空公司为了提高效率与控制成本的主要研究目标。为了构建更准确的离港航班延误时间预测模型,首先分析了导致离港航班延误发生的主要因素,并利用皮尔逊相关度系数对各因素进行相关性分析。其次基于基本BP(back propagation)神经网络算法,构建离港航班延误时间预测模型,并进行优化;然后采用自动编码器(AutoEncoder)对BP算法进行改进;接着构建了基于支持向量机(support vector machine,SVM)的预测模型并与优化后的BP模型进行对比;最后基于上海浦东机场实际历史航班数据进行仿真检验,验证了本文优化模型的准确性和高效性。

In today's civil aviation operation system,flight delays have become the main research goal of airports and airlines in order to improve efficiency and control costs.In order to build a more accurate prediction model for the delay of departure flights,the main factors leading to the occurrence of departure flight delays were firstly analysed and correlation analysis of each factor was carried out using Pearson's correlation coefficient.Secondly,the back propagation(BP)neural network algorithm was used to build a delay prediction model and optimise it,support vector machine(SVM)-based prediction model was then constructed and compared with the optimised BP model.Finally,the accuracy and efficiency of the optimised model are verified through simulation tests based on actual historical flight data from Shanghai Pudong Airport.