随着民用航空运输业的快速发展，航空运输已成为远程旅客和贵重货物运输的首选方式。然 而，空中交通量的持续增长使得有限的空域资源面临严峻挑战，尤其是在恶劣的对流天气条件下， 空域拥堵、流量控制和航班延误等问题频发，已成为制约民航高质量发展的瓶颈之一。为应对这 一挑战，本文提出了一种基于航班偏好的路径生成与轨迹优化方法，旨在从航班的不同运行阶段 优化飞行路径，提升在对流天气条件下的运行效率。

论文首先明确了航班运行流程、对流天气表征方法、偏好路径和轨迹选项集合概念等基础知 识。基于广州和上海机场群的历史 23578 条轨迹，识别出 9124 条偏航轨迹，使用 K-means、 DBSCAN和K-medoids聚类模型识别常用偏航轨迹，并通过CHI和DBI评估选择聚类模型，根 据常用偏航轨迹和计划路径生成19条轨迹选项集合。接着研究了航班运行前的路径选择，确定 了12类特征，包括扇区气象、流量、航路和航班信息，基于XGBoost和时空图神经网络建立航 班路径选择模型，实现航班在飞行前的路径预测。其中本研究提出的GAT-LSTM路径选择模型 在测试集表现最佳。最后，论文研究了航班起飞后遭遇对流天气的应对策略，提出DWR方法。 从空管、航司和飞行员三个角度建立改航路径目标函数，旨在提高改航路径的接受度。通过引入 对流天气和飞机性能等约束条件，采用两阶段方法实现了路径求解。

论文最后以2023年3月19日10:00-18:00期间，广州至上海的DKH1078、CSZ9515、CSZ3550 三个计划航班为案例，比较预测路径与计划路径，多角度验证路径选择模型的效果。通过对比预 测路径遭遇对流天气时应用DWR方法生成的路径与实际雷达轨迹，进一步验证了多阶段路径选 择与轨迹优化模型的应用效果。

With the rapid development of civil aviation, air transportation has become the preferred mode for long-distance passengers and valuable cargo. However, the continuous increase in air traffic has put immense pressure on limited airspace resources, especially under adverse convective weather conditions. Issues such as airspace congestion, flow control, and flight delays have become frequent, posing a significant bottleneck to the high-quality development of civil aviation. To address this challenge, this paper proposes a flight path generation and trajectory optimization method based on flight preferences, aiming to optimize flight paths at different operational stages and improve operational efficiency under convective weather conditions.

The paper first clarifies fundamental concepts such as the flight operation process, convective weather representation methods, preferred paths, and trajectory option sets. Based on 23,578 historical trajectories from Guangzhou and Shanghai airport clusters, 9,124 deviation trajectories are identified. The K-means, DBSCAN, and K-medoids clustering models are used to identify common deviation trajectories, and the clustering models are selected through evaluation using CHI and DBI indices. The 19 trajectory option sets are generated based on the common deviation trajectories and planned paths. Next, the paper examines pre-flight path selection and determines 12 categories of features, including sector weather, traffic, route, and flight information. Using XGBoost and spatiotemporal graph neural networks, a flight path selection model is developed to predict the flight path before departure. The proposed GAT-LSTM path selection model shows the best performance on the test set. Finally, the paper investigates the strategy for handling convective weather encountered after flight takeoff and introduces the Dynamic Weather Route (DWR) method. The paper establishes a rerouting path objective function from the perspectives of air traffic control, airlines, and pilots, aiming to increase the acceptance of rerouted paths. By incorporating constraints such as convective weather and aircraft performance, a two stage method is used to solve the path optimization problem.

The paper concludes with a case study of three planned flights—DKH1078, CSZ9515, and CSZ3550—on March 19, 2023, from 10:00 AM to 6:00 PM, between Guangzhou and Shanghai. The predicted paths are compared with the planned paths to validate the path selection model from multiple angles. Furthermore, the DWR method is applied to reroute the flight paths encountering convective weather, and the generated paths are compared with actual radar trajectories to further validate the practical application of the multi-stage path selection and trajectory optimization model.

[1] 中国民用航空局. 2020年中国民航行业发展统计公报[R]. 北京: 中国民用航空局, 2020.

[2] Federal Aviation Administration, "The Future of the NAS," United States Department of Transporation, [online], URL:https://www.faa.gov/nextgen/media/futureOfTheNAS.pdf [retrieved Septbember 2017].

[3] Federal Aviation Administration, “Vision for Trajectory Based Operations”, Version 2.0, September 2017.

[4] Rodionova O, Arneson H, Sridhar B, et al. Efficient trajectory options allocation for the collaborative trajectory options program[C]//2017 IEEE/AIAA 36th Digital Avionics Systems Conference (DASC). IEEE, 2017: 1-10.

[5] 张宝成,胡威,刘婉纯.考虑航空公司偏好的协同航路分配优化方法[J].交通运输系统工程与信息,2024,24(04):243-252.

[6] Zhu G, Wei P. An interval-based TOS allocation model for collaborative trajectory options program (CTOP)[C]//2018 Aviation Technology, Integration, and Operations Conference. 2018: 3042.

[7] Yoo H S, Brasil C, Smith N M, et al. Impact of different trajectory option set participation levels within an air traffic management collaborative trajectory option program[C]//2018 Aviation Technology, Integration, and Operations Conference. 2018: 3040.

[8] Galindo K C G, Lee P U, Brasil C L. Modeling Relative Trajectory Costs for Airborne Trajectory Reroutes using Trajectory Option Sets[C]//2021 IEEE/AIAA 40th Digital Avionics Systems Conference (DASC). IEEE, 2021: 1-10.

[9] 饶丹,时宏伟.基于深度聚类的航空交通流识别与异常检测研究[J].计算机科学, 2023,50 (03): 121-128.

[10] Rehm F. Clustering of flight tracks[M]//AIAA Infotech@ Aerospace 2010. 2010: 3412.

[11] Liu D, Zhang Y, et al. Unsupervised flight phase recognition with flight data clustering based on GMM[C]//2020 IEEE International Instrumentation and Measurement Technology Conference (I2MTC). IEEE, 2020: 1-6.

[12] Liu G, Fan Y, Zhang J, et al. Deep flight track clustering based on spatial–temporal distance and denoising auto-encoding[J]. Expert Systems with Applications, 2022, 198: 116733.

[13] Pham V V. Optimisation Algorithms and Heuristics for Aircraft User-Preferred Routes and their Environmental Impact[D]. UNSW Sydney, 2012.

[14] Mateos Villar M, Martín I, Alcolea R, et al. Unveiling airline preferences for pre-tactical route forecast through machine learning. An innovative system for ATFCM pre-tactical planning support[C]//Proceedings of the 11th SESAR Innovation Days. 2021.

[15] J. Zhu, K. Cai, Y. Li, B. Wang and Y. Yang, "A Novel Transformer-based Trajectory Options Generator for Collaborative Air Traffic Flow Management," 2023 IEEE/AIAA 42nd Digital Avionics Systems Conference (DASC), Barcelona, Spain, 2023, pp. 1-9,

[16] Evans A, Lee P. Using machine-learning to dynamically generate operationally acceptable strategic reroute options[C]//Air Traffic Management Research and Development (ATM R&D) Seminar. 2019 (ARC-E-DAA-TN69401).

[17] Coppenbarger R. En route climb trajectory prediction enhancement using airline flight-planning information[C]//Guidance, Navigation, and Control Conference and Exhibit. 1999: 4147.

[18] 王超,郭九霞,沈志鹏.基于基本飞行模型的4D航迹预测方法[J].西南交通大学学报, 2009, 44(02): 295-300.

[19] Gong C, Chan W. Using flight manual data to derive aero-propulsive models for predicting aircraft trajectories[C]//AIAA's Aircraft Technology, Integration, and Operations (ATIO) 2002 Technical Forum. 2002: 5844.

[20] Wang C, Guo J, Shen Z. Prediction of 4D trajectory based on basic flight models[J]. Journal of southwest jiaotong university, 2009, 44(2): 295-300.

[21] Krozel J, Andrisani D. Intent inference with path prediction[J]. Journal of guidance, control, and dynamics, 2006, 29(2): 225-236.

[22] Yepes J L, Hwang I, Rotea M. New algorithms for aircraft intent inference and trajectory prediction[J]. Journal of guidance, control, and dynamics, 2007, 30(2): 370-382.

[23] Michael A K, Scott D, Robab S. Improving ground-based trajectory prediction through communication of aircraft intent[C]//AIAA Guidance, Navigation, and Control Conference, AIAA. 2009, 6080: 1-25.

[24] 生柳振.基于ADS-B数据的民航飞机轨迹预测研究[D].南京邮电大学,2022.

[25] 付茂洺,陈纪宗.基于LSTM模型的机场航班飞行轨迹预测[J].现代计算机, 2022, 28(14): 55-60.

[26] 梁海军,韩琪聪.基于机器学习的航班静态四维轨迹预测研究[J].兵器装备工程学报, 2022, 43(04): 147-151.

[27] Prats X, Puig V, Quevedo J, et al. Multi-objective optimisation for aircraft departure trajectories minimising noise annoyance[J]. Transportation Research Part C: Emerging Technologies, 2010, 18(6): 975-989.

[28] Ayhan S, Samet H. Aircraft trajectory prediction made easy with predictive analytics[C]//Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. 2016: 21-30.

[29] De Leege A, van Paassen M, Mulder M. A machine learning approach to trajectory prediction[C]//AIAA Guidance, Navigation, and Control (GNC) Conference. 2013: 4782.

[30] Zhang X, Mahadevan S. Bayesian neural networks for flight trajectory prediction and safety assessment[J]. Decision Support Systems, 2020, 131: 113246.

[31] Zhang K, Chen B. Phased flight trajectory prediction with deep learning[J]. arXiv preprint arXiv:2203.09033, 2022.

[32] 李翔宇.雷雨天气影响下基于改航的终端区短期容量预测[D].南京航空航天大学，2016.

[33] Taylor C, Wanke C, Dynamically Generating Operationally Acceptable Route Alternatives using Simulated Annealing[J]. ATC Quarterly, 2012, 20(1): 97–121.

[34] 梁洁雅,田卫萍,杨志坚.基于改进蚁群算法的飞机航路规划研究[J].火力与指挥控制，2021，46(3):67-71.

[35] 孟令航, 徐肖豪, 李善梅, 等. 不确定强对流天气下动态改航路径规划[J]. 西南交通大学学报, 2012, 47(4): 686- 691.

[36] Sridhar B, Chatterji G, Grabbe S, et al. Integration of Traffic Flow Management Decisions[C]. AIAA Guidance, Navigation, and Control Conference. Monterey, 2002.

[37] Love W, Arthur W, Heagy W, et al. Assessment of Prediction Error Impact on Resolutions for Aircraft and Severe Weather Avoidance[C]. 4th AIAA Aviation Technology, Integration and Operations (ATIO) Forum, 2004.

[38] Mukherjee A, Hansen M. A Dynamic Rerouting Model for Air Traffic Flow Management[J]. Transportation Research Part B: Methodological, 2009, 43(1): 159-171.

[39] 王莉莉, 王鑫, 位放. 移动式收缩型恶劣天气下改航模型和算法[J]. 河南科技大学学报, 2017, 38(03): 35-40.

[40] 王飞, 王红勇. 基于Maklink图和遗传算法的改航路径规划方法研究[J]. 交通运输系统工程与信息, 2014, 14(5): 154-160.

[41] 李雄, 徐肖豪, 赵嶷飞等. 散点状分布危险天气区域下的航班改航路径规划[J]. 航空学报, 2009, 30(12): 2342-2347.

[42] 张兆宁, 魏中慧. 散点状分布危险天气下的终端区动态改航方法[J]. 中国安全科学学报, 2016, 26(1): 40-44.

[43] McNally D, Sheth K, Gong C, et al. Dynamic weather routes: a weather avoidance system for near-term trajectory-based operations[C]//28th international congress of the aeronautical sciences. 2012: 23-28.

[44] McNally D, Sheth K, Gong C, et al. Operational Evaluatioin of Dynamic Weather Routes at American Airlines[C]//USA/Europe Seminar on ATM Research & Development (R&D)(ATM 2013). 2013 (ARC-E-DAA-TN7608).

[45] Ballin M, Wing D. Traffic Aware Strategic Aircrew Requests (TASAR)[C]//12th AIAA Aviation Technology, Integration, and Operations (ATIO) Conference and 14th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference. 2012: 5623.

[46] Idris H, Enea G, Burke K, et al. Assessment of Air Traffic Controller Acceptability of Aircrew Route Change Requests[C]//USA/Europe Air Traffic Management (ATM) R&D Seminar 2017. 2017 (ARC-E-DAA-TN43412).

[47] Sheth K, Clymer A, Morando A, et al. Analysis of multiple flight common route for traffic flow management[C]//16th AIAA Aviation Technology, Integration, and Operations Conference. 2016: 4207.

[48] Gong C, Mcnally D. Dynamic arrival routes: A trajectory-based weather avoidance system for merging arrivals and metering[C]//15th AIAA Aviation Technology, Integration, and Operations Conference. 2015: 3394.

[49] Zelinski S, Xue M, Bassett P. Optimized Route Capability (ORC) Intelligent Offloading of Congested Arrival Routes[C]//16th AIAA Aviation Technology, Integration, and Operations Conference. 2016: 4357.

[50] DeLaura R, Robinson M, Pawlak M, et al. Modeling convective weather avoidance in enroute airspace[C]//13th Conference on Aviation, Range, and Aerospace Meteorology, AMS, New Orleans, LA. 2008.

[51] 王世锦,林荆荆,李家豪,等.基于POT模型的航班偏差阈值确定[J].航空计算技术, 2021, 51(3):5.

[52] Liu Y, Hansen M, Lovell D J, et al. Predicting aircraft trajectory choice–a nominal route approach[C]//Proc. of the International Conference for Research in Air Transportation. 2018.

[53] Arneson H. Initial analysis of and predictive model development for weather reroute advisory use[C]//15th AIAA Aviation Technology, Integration, and Operations Conference. 2015: 3395.

[54] Wang S J, Li Y L, Yang B T, et al. Modeling sector route segment capacity under convective weather based on machine learning[J]. The Aeronautical Journal, 2024: 1-22.

[55] Wang S, Yang B, Duan R, et al. Predicting the airspace capacity of terminal area under convective weather using machine learning[J]. Aerospace, 2023, 10(3): 288.

[56] 储洁雯.对流天气下的空域可用性和改航路径研究[D]. 南京： 南京航空航天大学.

[57] 张文.危险天气下的改航路径规划方法研究[D].南京信息工程大学, 2023.

[58] 董襄宁.飞行的组织与实施[R]. 南京航空航天大学:民航学院, 2003年.

[59] 段荣荣.基于深度学习的对流天气短时预测及其在CTOP中的应用研究[D]. 南京：南京航空航天大学.