基于DTW的机场延误特性分析  

Analysis of Airport Delay Time Series Data Using Dynamic Time Warping

作　　者：李雨吟 胡明华[1] 陈海燕 谢华[1] LI Yu-yin;HU Ming-hua;CHEN Hai-yan;XIE Hua(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)

机构地区：[1]南京航空航天大学民航学院,江苏南京210016

出　　处：《航空计算技术》2018年第3期54-59,共6页Aeronautical Computing Technique

基　　金：国家自然科学基金项目资助(61501229)

摘　　要：针对国内航班延误现状,尝试从机场延误时间序列数据本身分析,挖掘机场延误时间序列特性,找出与延误时间相关的影响因素,重点考察机场离港航班整体延误情况。基于动态时间规整(DTW)方法,分析机场延误时间序列的整体相似特性;在时序相似性度量的基础上,结合皮尔逊相关系数探究在延误情况相似情况下与机场延误时间序列相关的其他时序因素。结果表明,天气因素对于机场延误时间影响较大,且延误总时间相近的月份之间,机场时间序列也更为相似;实际进离港架次和前序延误航班对于机场延误时间序列影响较大,而发生机上延误及过站冗余时间的关联性较小。For analyzing the the nature of airport delay time series itself,finding factors related to the departure delay,Dynamic Time Warping（ DTW） distance is considered as a desirable choice in exhibiting similar delay patterns occurring at different time periods. In this paper,Pearson correlation coefficient is used on the basis of the time series similarity measure. As a result,the potentially underlying factors connected to the departure delay might be identified. The results show that weather factors have a great influence on the departure delay and the DTW distance of airport delay time series sharing similar delay situations should be more close. Besides,the actual airport throughout and propagated delays greatly affect departure delays,but the delays on board and the extra transit time have little impact.

关 键 词：机场延误 时间序列相似性 动态时间规整 皮尔逊相关系数 

分 类 号：V355[航空宇航科学与技术—人机与环境工程]