摘要
飞行安全是民航运输业生存发展的根本保障。严重的飞行事故不仅会给航空公司带来巨大的经济损失，也会对旅客的生命安全构成极大威胁。因此，飞行安全亟待引起重视。首先，进行预分析，发现与飞机机动直接相关的量是盘量和杆量，分别对应滚转机动和俯仰机动。之后，采用聚类分析和引入注意力机制的图神经网络模型对数据进行训练，得出影响飞机飞行最重要的运行因素是杆量和姿态。根据模型追溯重着陆的原因：“是由于不正确的松杆操作，导致杆尺寸和姿态异常，使着陆G值曲线明显上凸”。最后，评估了模型的优缺点并进行了灵敏度分析。本文建立的模型可以扩展到评估社会经济运行的稳定性。
关键词： 图神经网络；航空安全决策树；回归树
Abstract
Flight safety is the fundamental guarantee for the survival and development of civil aviation transport industry. Serious flight accidents will not only bring huge economic losses to airlines, but also pose a great threat to the life safety of passengers. Therefore, we need to pay close attention to flight safety. Firstly, the pre-analysis is carried out, and it is found that the quantity directly related to the aircraft manoeuvring is the disc quantity and the rod quantity, which correspond to roll manoeuvring and pitch manoeuvring respectively. After that, cluster analysis and graph neural network model introduced attention mechanism were used to train the data, and it was concluded that the most important operational factors affecting the flight of aircraft were rod size and attitude. According to the model, the reason for the heavy landing is traced back: "It is because of the incorrect loose rod that the rod size and attitude are abnormal, which makes the G-value curve of the landing significantly convex". Finally, the advantages and disadvantages of the model are evaluated and the sensitivity is analyzed. The model established in this paper can be extended to evaluate the stability of social and economic operation.
Key words： Graph neural network; Aviation safety decision tree; Regression tree
参考文献 References
[1] Li, B., Huang, J., Wang, X., Shi, Y., Lv, M., & Hu, M. (2022, August). Overview of Application in Data Mining Techniques to QAR Data Ansys. In 2022 2nd International Conference on Big Data Engineering and Education (BDEE) (pp. 13-21). IEEE.
[2] Wang, X., Zhao, X., & Yu, L. (2020, October). Data Mining on the Flight Quality of an Airline based on QAR Big Data. In 2020 IEEE 2nd International Conference on Civil Aviation Safety and Information Technology (ICCASIT (pp. 955-958). IEEE.
[3] Zixuan, W., Weizhen, L., Cong, C., Qingchi, Y., & Hao, Z. H. A. N. G. (2020, October). Research on Influencing Factors of Fuel Flow Based on QAR Data. In 2020 IEEE 2nd International Conference on Civil Aviation Safety and Information Technology (ICCASIT (pp. 800-804). IEEE.
[4] Jiao, Y., Han, J., Xu, B., Xiao, M., Shen, B., & Sun, H. (2021, October). Research on Domain Entity Extraction in Civil Aviation Safety. In 2021 IEEE 3rd International Conference on Civil Aviation Safety and Information Technology (ICCASIT) (pp. 384-388). IEEE.
[5] Hong, W., Xiuxia, H., & Hongwei, W. (2008, December). Research and implementation of QAR data warehouse. In 2008 Second International Symposium on Intelligent Information Technology Application (Vol. 3, pp. 156-162). IEEE.
[6] Kang, Z., Shang, J., Feng, Y., Zheng, L., Liu, D., Qiang, B., & Wei, R. (2020). A deep sequence-to-sequence method for aircraft landing speed prediction based on QAR data. In Web Information Systems Engineering–WISE 2020: 21st International Conference, Amsterdam, The Netherlands, October 20-24, 2020, Proceedings, Part II 21 (pp. 516-530). Springer International Publishing.
[7] Han, J., & Jiao, Y. (2018, December). Conception of QAR Data Application Mode Based on Virtual Integration. In 2018 IEEE International Conference of Safety Produce Informatization (IICSPI) (pp. 392-395). IEEE.
[8] Li, L., Hansman, R. J., Palacios, R., & Welsch, R. (2016). Anomaly detection via a Gaussian Mixture Model for flight operation and safety monitoring. Transportation Research Part C: Emerging Technologies, 64, 45-57.