基于随机森林的进离港航班延误预测
Prediction of Inbound and Outbound Flight Delays Based on Random Forest
牟奇锋 1衣超群 1吕晨辉1
作者信息
- 1. 中国民用航空飞行学院机场学院 广汉 618307
- 折叠
摘要
为科学、有效地预测进离港航班延误时间,论文选取了机场繁忙程度、飞行距离、天气以及地面服务保障时间等特征因素,分析了各特征因素与航班延误之间的关系,并进行了特征因素的量化.采用机场单位小时内的飞机起降架次比来量化机场繁忙度;考虑飞机在空中航行受管制的影响,将航行距离实际值量化;采用ATMAP算法为METAR报文打分量化;并使用A-CDM中的地面保障服务时间将其数字化处理.将以上特征因素编入到预测模型中,采用随机森林算法建立模型,分别对某机场内每架进港航班、离港航班做了短期延误预测.实验结果表明,85.7%的进港航班延误预测值与实际值差值小于15 min;83.7%的离港航班延误预测值与实际值差值小于15 min.这为机场、航司、空管提供决策依据.
Abstract
In order to predict the inbound and outbound flight delays scientifically and effectively,this paper selects the char-acteristic factors such as airport busyness,flight distance,weather and ground service guarantee time,analyzes the relationship be-tween each characteristic factor and flight delays,and quantifies the characteristic factors.The ratio of aircraft takeoffs and landings per unit hour at the airport is used to quantify the airport busyness.The actual value of navigation distance is quantified by consider-ing the influence of aircraft navigation in the air by control.The ATMAP algorithm is used to score and quantify the METAR messag-es,and the ground service guarantee time in A-CDM is used to digitize them.The above characteristic factors are coded into the pre-diction model,and the random forest algorithm is used to build the model,and the short-term delay prediction is made for each in-coming flight and departing flight in an airport respectively.The experimental results show that 85.7%of inbound flight delay predic-tions have a difference of less than 15 minutes from the actual value,83.7%of departing flight delay predictions have a difference of less than 15 minutes from the actual value.This provides a decision basis for airports,airlines and ATC.
关键词
随机森林回归/延误预测/ATMAP算法/机场繁忙程度Key words
random forest regression/delay prediction/ATMAP algorithm/airport busyness引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据