摘要
航班离港始于停机坪推出,经由滑行道滑行并止于跑道离地,因此机坪空间及其与跑滑系统构成的相对位置关系对航班离港滑行具有较大影响,滑行时间的预测精度对ACDM机制下航班推出时刻优化和跑滑系统效率提升具有重要作用。首先提出机位组概念表征机坪空间构型,据此设计场面实时动态航班流量、机位组无阻碍滑行时间和机位组空间影响指数等新特征变量,然后基于分类回归树构建预测模型,验证新特征引入对离港滑行时间的预测效果。以首都机场实际运行数据为例,预测结果表明:在保持较高拟合优度的同时,新引入的表征机坪构型及其与跑滑系统相对位置关系的特征变量提高了离港滑行时间预测模型精度,预测与实际误差值在3 min和5 min内的航班数量分别提高了4.88%和6.46%,每高峰小时可为首都机场减少约2~3个起飞时隙的浪费;新特征变量对离港滑行时间预测精度的整体贡献较大,且场面整体流量特征变量的贡献超过单一跑道流量,进港航班相关特征变量的贡献超过离港航班。
Flight departure starts from apron pushback, taxiway taxi until runway takeoff. Therefore, the apron space and the consequent spatial relation with runway-taxiway system have significant impacts on flight departure taxi. The prediction accuracy of flight departure taxi time plays an important role in the optimization of flight pushback time and the improvement of runway-taxiway system efficiency under Airport Collaborative Decision Making(ACDM). First,the concept of stand group is proposed to represent the spatial configuration of apron. The new feature variables are introduced, such as real-time and dynamic surface flight flow, unimpeded taxiing time and spatial index of stand groups. Then the prediction model is developed based on the classification and regression tree method to verify the impact of the new features on departure taxi time. The actual operation data of Capital Airport was used for the case study. The prediction results show that: while maintaining a high degree of goodness-of-fit, the new features characterizing apron configuration and spatial relation with runway-taxiway system improve the accuracy of the prediction model for departure taxi time. The number of flights respectively increases by 4.88% and 6.46% when the error between the predicted and actual values are within 3 minutes and 5 minutes. The method also reduces the waste of2 to 3 takeoff slots per peak hour for Capital Airport. The new features closely relate to the prediction accuracy of departure taxi time. The overall surface flow has larger impact on the prediction accuracy than the surface flow of single runway, and the arrival flight features contribute more on the prediction accuracy than the departure flight features.
作者
唐小卫
陈祯
张生润
丁叶
TANG Xiao-wei;CHEN Zhen;ZHANG Sheng-run;DING Ye(College of Civil Aviation,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China)
出处
《交通运输系统工程与信息》
EI
CSCD
北大核心
2022年第5期309-317,共9页
Journal of Transportation Systems Engineering and Information Technology
基金
国家自然科学基金(61603178)。
关键词
航空运输
离港滑行时间预测
分类回归树
机坪构型
特征构建
air transportation
departure taxi time prediction
classification and regression tree
apron configuration
feature construction