摘要
近年来,随着航空业的蓬勃发展,机场数量大、国内外航线逐渐密集导致的空域拥堵、航班准点率低等问题也随之浮现,因此对航线进行智能分析并及时进行有效预测以改善空域拥堵等问题受到了广泛关注。对基于深度学习的4D航迹预测方法进行了综述。首先,对4D航迹预测问题进行概述,包括基本概念、基本类型、优缺点、评价方式。其次,提出将现有基于深度学习的4D航迹预测方法分为基于单一模型预测和基于聚合模型预测并详细阐述了各类模型的建模思想、基本原理及优缺点。最后,阐述了三种基于聚合模型的预测方法:模型聚合方法、聚类优化方法、自生成网络方法并总结全文。
In recent years,with the booming development of the aviation industry,airspace congestion and low on-time flight rate caused by the large number of airports and the gradual density of domestic and foreign routes have also emerged.Therefore,in⁃telligent analysis of routes and timely effective prediction to improve airspace congestion have attracted wide attention.The 4D track prediction method based on deep learning is reviewed.Firstly,the 4D flight path prediction problem is summarized,including basic concepts,basic types,advantages and disadvantages,evaluation methods.Secondly,the existing 4D flight path prediction methods based on deep learning are divided into single model prediction and aggregation model prediction,and the modeling ideas,basic principles,advantages and disadvantages of each model are elaborated.Finally,three kinds of prediction methods based on aggrega⁃tion model are described,which are model aggregation method,cluster optimization method and self-generated network method.
作者
宋岩
杜冬
侯晓雪
SONG Yan;DU Dong;HOU Xiaoxue(Civil Aviation Flight University of China,Guanghan 618300)
出处
《舰船电子工程》
2024年第2期62-66,75,共6页
Ship Electronic Engineering
基金
国家自然科学基金中国民用航空局民航联合研究基金项目“大数据驱动的飞行训练智能评估理论与方法”(编号:U2033213)资助。
关键词
4D航迹
航迹预测
深度学习
聚合模型
聚类优化
4D track
track prediction
deep learning
aggregation model
cluster optimization
