基于机器学习的空域受限制单元识别方法

Recognition Method of Restricted Spatial Units Based on Machine Learning

根据航班延误研究现状,提出一种基于机器学习的空域受限单元识别方法,旨在识别航班运行中受到限制空域单元及受限起止时间。研究基于现行空域内大量航班历史飞行数据,筛选可利用数据并对数据归类排列,采取升采样的数据处理方法,依据时间序列的分析方法,将数据按不同的时间粒度进行划分;而后计算目标机场之间的累计延误架次和平均延误时间,基于K means++聚类算法结合空域结构数据构建空域单元受限识别模型。选择2017年8月1日北京首都国际机场、上海浦东国际机场、广州白云国际机场和成都双流国际机场之间的往返航班数据进行实例分析,识别结果表明,成都双流国际机场是当天典型的受限空域单元。

This paper proposes a machine learning based identification method of restricted airspace units,which aims to identify restricted airspace units and restricted start and end time in flight operation.Based on the historical flight data of the target airspace,we divide the data into different time intervals according to the analysis method of time series.Then we calculate the cumulative delay flights and the average delay time between the target airports.After that we construct the restricted airspace unit identification model based on K means++clustering algorithm and airspace structure data.Finally,we select the round trip flight data between Beijing capital international airport,Shanghai Pudong international airport,Guangzhou Baiyun international airport and Chengdu Shuangliu international airport on August 1,2017 for case analysis.The identification results show that Chengdu Shuangliu international airport is a typical restricted airspace unit on that day.