基于多重筛选的深度神经网络应用于航班延误预测的方法研究  

Research on Methodologies and Applications of Flight Delay Prediction Models Based on Deep Learning

作　　者：黄雍杰 HUANG Yongjie(Nanjing Aerosino Tech Co.,Ltd.,Nanjing 211899,Jiangsu,China)

机构地区：[1]南京中科航港技术有限公司,江苏南京211899

出　　处：《民航学报》2024年第3期17-28,共12页Journal of Civil Aviation

摘　　要：从全球民航信息化发展态势看,以新一代数据驱动和AI技术融合应用为主要特征的智慧民航建设正全方位重塑民航业的形态。本文将论述如何使用机器学习和深度神经网络建模来预测宏观航班延误态势,并把预测结果作为次年航延保费系数调整参考。研究将从数据处理、数据分析、预测建模、预测结果应用全流程进行阐述,总结出可适用于民航其他领域预测能力的方法论。文中也将以国内机场“航班地面保障时间预测”为例,讨论该方法论在机场运行资源优化等场景的应用,为大数据+深度学习赋能运行智慧化给予启发,充分发挥中国民航的数据资源优势,助力实现数据资产化,实现数尽其用,用其所长。The global development trends in civil aviation informatization show that,this industry has been experiencing a transformation driven by integration of big data and AI technologies.This article discusses how to conduct machine learning and deep neural network modeling to predict macro trend of nationwide flight delays,and then take the predicted results as a reference for adjusting the flight delay insurance premium coefficients for the following year.This research elaborates the entire process from data processing,data analysis,prediction modeling to the application of prediction results,and summarizes a methodology applicable to other fields in the industry.Meanwhile,the article takes the"ground handling node-time prediction at domestic airports"as an example to discuss the application of this methodology in scenarios such as optimizing airport operational resources.It aims to inspire the empowerment of intelligent operations by integrating big data and deep learning leverage data resource advantages of Chinese civil aviation industry,and contribute to the progress of data assetization.

关 键 词：智慧民航 机器学习 深度学习 大数据 航班延误预测 地面保障节点预测 

分 类 号：U8[交通运输工程]