Abstract:

Flight delay often occurs in air transportation in recent years, which severely affects benefits of airports, airlines and passengers. Flight delay cannot be entirety eliminated because of the effects of weather, air control and human factors. However, we are able to analyze the internal reasons and underlying laws to reduce till avoid unnecessary loss. Based on statistics during a week at the Capital Airport in Beijing, China, the paper analyzes the delay length, delay rate, delay fluctuation of different airlines and different durations. It develops a kernel density evaluation function of a day delay duration distribution, and then proves that the delay duration distribution looked on as normal distribution is not accurate enough. The method to fit optimal bandwidth is presented. The result shows that the method meets the accuracy request to analyze flight delay, which provides reference for analyzing operation and recovery for abnormal flights.

Key words: air transportation, flight delay, delay duration distribution, kernel density evaluation function

摘要：

近年来航班延误现象愈演愈烈，严重影响旅客、机场和航空公司的利益.受天气、管制、人为因素的影响，难以彻底消除航班延误，但可以探究航班延误的内在原因及其发展规律，减少或避免不必要的损失.通过对首都机场的追踪统计，分析不同航空公司、不同时段的延误时长、延误率和延误波动；建立一天延误时长分布的核概率密度函数，证明在以往机场仿真时将延误时长分布看作是正态分布并不合理；提出直接利用延误总时长拟合最佳窗宽的方法.结果表明，该方法精度达到航班延误规律分析要求，可以为不正常航班情形下的运行调度和恢复提供参考.

关键词: 航空运输, 航班延误, 延误时长分布, 核概率密度函数 