哈尔滨商业大学学报（自然科学版）2025，Vol.41Issue(6)：643-650,8.
一种鲁棒的航班延误原因智能判定方法
A robust intelligent identification method for flight delay causes
周薇 1陈海燕 2周智慧 2刘莉 2袁立罡2
作者信息
- 1. 中国民用航空局运行监控中心,北京 100710
- 2. 南京航空航天大学 计算机科学与技术学院,南京 211000
- 折叠
摘要
Abstract
Flight punctuality served as a critical indicator of air transport efficiency.Identifying delay causes and responsible entities remained one of the most challenging issues confronting aviation regulatory body in establishing the Flight Punctuality Statistical Regulation.To address excessive subjectivity,incomplete data records,and poor operational feasibility in the delay attribution process,this paper proposed an intelligent method for flight delay cause determination based on Generative Adversarial Networks(GANs)and semi-supervised learning.The proposed method first employed an adversarial defense mechanism to generate additional adversarial samples for training the determination model,thereby mitigating the impact of subjective factors.A semi-supervised transductive support vector machine(TSVM)was adopted as the delay cause determination model,enabling the effective utilization of unlabeled data during training to address the problem of incomplete data annotation.Experiments on real-world operational data demonstrated that the proposed intelligent method achieved more objective and accurate determination of flight delay causes while exhibiting high robustness.关键词
航班延误原因/对抗攻击/直推式支持向量机/对抗攻防/责任归属/半监督学习Key words
flight delay causes/adversarial defense/transductive support vector machine/adversarial training/responsibility attribution/semi-supervised learning分类
信息技术与安全科学引用本文复制引用
周薇,陈海燕,周智慧,刘莉,袁立罡..一种鲁棒的航班延误原因智能判定方法[J].哈尔滨商业大学学报（自然科学版）,2025,41(6):643-650,8.
