基于相似场景的机场场面离场滑行时间预测  

Similar Scenario Based Prediction of Airport Surface Taxi-out Time

作　　者：吴维 陈雨桐 乔沛然 何雨轩 尹嘉男[1] WU Wei;CHEN Yu-tong;QIAO Pei-ran;HE Yu-xuan;YIN Jia-nan(Nanjing University of Aeronautics and Astronautics,Nanjing 211000,China;The University of Melbourne,Grattan Street,Parkville,VIC 3010,Australia)

机构地区：[1]南京航空航天大学,江苏南京211000 [2]墨尔本大学,维多利亚墨尔本3010

出　　处：《航空计算技术》2025年第1期12-16,共5页Aeronautical Computing Technique

基　　金：国家自然科学基金项目资助(52002178);江苏省自然科学基金项目资助(BK20190416);南京航空航天大学校创新计划项目资助(xcxjh20230747)。

摘　　要：为准确感知机场场面运行态势,科学支撑场面交通运行管控,提出了基于相似场景的机场场面离场滑行时间预测。面向高度动态变化的复杂场面运行场景,考虑场面进离场航空器对滑行时间的影响,构建了场面场景量化指标体系,选取模糊C均值聚类算法对机场场面相似场景进行了分类;结合分类结果,采用决策树对场面运行场景进行了识别;最后,基于识别出的相似场景,采用多层感知器对离场滑行时间进行了预测。深圳宝安国际机场实例验证表明,FCM算法有效识别出4种场面运行场景,提出的基于相似场景的滑行时间预测方法可以更好地捕捉场景间的相似性,并提高了场面活动的预测精度。In order to accurately perceive the airport surface operation situation and scientifically support the surface traffic operation control,the airport surface departure taxi time prediction based on similar scenario is proposed.Facing the highly dynamic change of the complex surface operation scenario,and considering the influence of the surface arrival and departure aircraft on the taxi time,a quantitative index system of the surface scenario is constructed,and the Fuzzy C-Means Clustering(FCM)algorithm is selected to classify the similar scenario of the airport surface;combined with the classification results,the Decision Tree(DT)is used to identify the surface operation scenario;Finally,based on the identified similar scenario,Multilayer Perceptron(MLP)was used to predict the departure taxi time.Example validation at Shenzhen Bao′an International Airport shows that the FCM algorithm effectively recognizes four surface operation scenarios,and the proposed taxi time prediction method based on similar scenarios can better capture the similarity between scenarios and improve the prediction accuracy of surface activities.

关 键 词：机场场面 滑行时间 预测 聚类 场景识别 

分 类 号：V35[航空宇航科学与技术—人机与环境工程]