文档介绍：第52卷第1期 航空计算技术 Vol. 52
2022年1月t, make its can be esti­
mated in the proposed architecture perform supervised fine- tuning,In addition ,some key influencing fac ­
tors in the traffic management measures ( TMI) are added into the model as the hidden layer of Gauss
Bernoulli (GBRBM) and the next visible layer of the model,which helps to reduce the overall delay. For
the unbalanced high- dimensional data set of the test set,this study will use accuracy and sensitivity to e-
valuate the relationship between dependent variables and explanatory variables. The final data show that
the accuracy of delay prediction of DBN - SVM model reaches 89. 39% , which can provide a theoretical
basis for automatic calculation of flow management.
Key words: air traffic management ; delay prediction; deep belief learning support vector machine ; data
mining
引言 长⑴。2015—2019年间全均正常率为
航班延误是衡量航空运输系统效率的重要指标, %，平均延误时间为18 min,其中2015年客运航
准确的延误预测有助于航空运输系统中相关部门制定 班的正常率仅为68. 33%，平均延误时间超过20 min。
决策方案,减少航班延误带来的影响。2014—2018年 2020年