空中交通航迹预测技术的发展是新一代国家航空运输系统的一个关键目标，但终端区对流天气仍然是影响航迹预测的一个重要因素，导致实际飞行与原始飞行计划之间出现较大偏差。准确可靠地进行对流天气影响下的终端区航迹预测，既有利于管制员提前掌握航班的飞行态势，减少航迹不确定性的干扰，有效避免飞行冲突，也有利于促进空域资源科学配置和灵活使用，减少管制员工作负荷。因此，本文基于在晴好天气时航空器沿飞行计划路线飞行的思想，建立了一种基于深度学习的航迹预测模型，根据飞行计划航迹及其附近的对流天气预测航迹，以应对复杂多变的对流天气影响场景，实现高精度的终端区航迹预测。

本文首先基于空域高峰时段研究对流天气对终端区的影响，建立具有空域区域性的对流天气量化指标，采用K-means聚类实现对流天气影响场景分类，并分析不同场景下终端区航迹的特性；然后分析对流天气影响终端区航迹的特征，通过模糊c均值聚类和三次样条插值提取等间隔的计划航迹特征；通过飞行计划航迹及其附近的对流天气，提取航迹方向对流天气影响区和量化的天气危险指数等对流天气特征，并对ADS-B异常数据进行数据清洗以提高特征数据的质量；最后，为了充分捕捉天气、航迹以及天气与航迹的时空关系，建立了2D CNN和LSTM双神经网络分支的航迹预测模型，并以广州终端区为例进行验证，对比分析了1D CNN和LSTM、2D CNN和MLP双神经网络分支模型、CNN 1D和LSTM单模型在不同类别的对流天气影响场景下的航迹预测结果。结果表明，本文模型在MAE、RMSE、MAPE和R²拟合优度等方面的整体性能优于其他模型，且各类对流天气影响场景中具有最佳的预测精度，模型稳定性高。

The development of air traffic trajectory prediction technology is a key goal of the new generation of national air transportation systems, but convective weather in the terminal area is still an important factor affecting the trajectory prediction, resulting in a large deviation between the actual flight and the original flight plan. Accurate and reliable prediction of the terminal area trajectory under the influence of convective weather not only helps the controller to grasp the flight situation of the flight in advance, reduce the interference of trajectory uncertainty, and effectively avoids flight conflicts, but also helps to promote the scientific allocation and flexible use of airspace resources and reduce the workload of the controller. Therefore, in this paper, based on the idea that the aircraft will follow the flight plan route when the weather is fine, a deep learning-based trajectory prediction model is established, which predicts the trajectory according to the flight plan trajectory and the convective weather near the flight plan trajectory. In order to cope with complex and changeable convective weather scenarios to achieve high-precision terminal area track prediction.

Firstly, a quantitative index of convective weather with airspace regionality is established by studying the impact of convective weather on the terminal area based on the peak period of the airspace, establishes a regional convective weather quantitative index, and uses K-means clustering to classify the impact of convective weather and analyzes the trajectory of the terminal area under different scenarios. Then, the characteristics of the trajectory in the terminal area affected by convective weather are analyzed, and the features of the planned trajectory at equal intervals are extracted by fuzzy c-means clustering and cubic spline interpolation. Convective weather features such as the convective weather influence area in the direction of the flight trajectory and the quantified weather severity index are extracted from the flight plan trajectory and the convective weather near it. The abnormal data of ADS-B is cleaned to improve the quality of the feature data. Finally, in order to fully capture the weather, trajectory, and the temporal and spatial relationship between weather and trajectory, a trajectory prediction model of 2D CNN and LSTM dual neural network branches is established. Taking Guangzhou terminal area as an example for verification, the trajectory prediction results of 1D CNN and LSTM, 2D CNN and MLP dual neural network branch model, and CNN 1D and LSTM single model under different types of convective weather influence scenarios are compared and analyzed. The results show that the overall performance of the model in this paper is better than other models in terms of MAE, RMSE, MAPE, and R2 goodness of fit, and has the best prediction accuracy and high model stability in various convective weather scenarios.

中国民用航空局. 2019年民航行业发展统计公报[EB/OL]. http://ats.atmb.net.cn/UploadFiles/20210313153444175.pdf

罗荣龙. 美国和欧洲新一代民用航空运输系统研究综述[D].南京：南京航空航天大学, 2018.

Lvaro Rodríguez–Sanz, Puchol C C , Fernando Gómez Comendador, etal. Air traffic management based on 4D-trajectories: requirements and practical implementation[J]. MATEC Web of Conferences, 2019, 304(2):5001-5008.

杨东玲. 基于ADS-B的4D航迹预测及应用[D].天津：中国民航大学, 2017.

Nilim A, Ghaoui I, Hansen M, et al. Trajectory-based air traffic management (TB-ATM) under weather uncertainty[J]. The 4th ATM R&D Seminar, 2001,32(2):1-10.

J Prete, J Mitchell .Safe Routing of Multiple Aircraft Flows in the Presence of Time-Varying Weather Data[C]. AIAA Guidance, Navigation, & Control Conference & Exhibit, 2004:462-482.

Robinson M, DeLaura R, Underhill N. The Route Availability Planning Tool (RAPT): Evaluation of departure management decision support in New York during the 2008 convective weather season[J].Proceedings of the 8th USA/Europe Air Traffic Management Research and Development Seminar, 2009,12(34):353-362.

Matthews M, Delaura R. Assessment and Interpretation of En Route Weather Avoidance Fields from the Convective Weather Avoidance Model[C].AIAA Aviation Technology, Integration, & Operations, 2010:1-19.

McNally D, Sheth K, Gong C, Love J, Lee CH, Sahlman S, et al. Dynamic Weather Routes: A weather avoidance system for near-term trajectory-based operations[J].28th Congress of the International Council of the Aeronautical Sciences, 2012, 5(1):4200-4217.

Erzberger H , Nikoleris T , Paielli R A , et al. Algorithms for control of arrival and departure traffic in terminal airspace[J]. Proceedings of the Institution of Mechanical Engineers Part G Journal of Aerospace Engineering, 2016, 230(9):1762-1779.

González-Arribas D, Soler M, Sanjurjo-Rivo M, Kamgarpour M, Simarro J. Robust aircraft trajectory planning under uncertain convective environments with optimal control and rapidly developing thunderstorms[J]. Aerospace science and technology. 2019, 89(6):445-459.

魏彤. 强对流天气下航班改航策略研究[D].唐山：华北理工大学, 2021.

蒋昕,胡明华,张颖,田文.基于飞行受限区划设的航班改航研究[J].华东交通大学学报, 2016, 33(03):60-67.

谢春生,李雄.危险天气影响航路飞行区域的划设及评估[J].中国安全科学学报, 2010, 20(10):47-52.

仝佳璐,胡明华,张颖.恶劣天气下多条改航路径的生成[J].航空计算技术, 2018, 48(06):55-58+63.

Warren A W, Ebrahimi Y S. Vertical path trajectory prediction for next generation ATM[C]. Digital Avionics Systems Conference, IEEE, 1998:1-8.

Rodriguez J , Deniz L. A model to 4D descent trajectory guidance[J]. IEEE, 2007,12(6):1-12.

Schuster W , Porretta M , Ochieng W . High-accuracy four-dimensional trajectory prediction for

civil aircraft[J]. The Aeronautical Journal, 2012, 116(1175):45-66.

Schuster, W. Trajectory prediction for future air traffic management – complex manoeuvres and

taxiing[J]. The Aeronautical Journal, 2015, 119(1212):121-143.

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.

张军峰, 蒋海行, 武晓光, 汤新民. 基于BADA及航空器意图的四维航迹预测[J].西南交通大学学报, 2014, 49(03):553-558.

Chatterji G. Short-term trajectory prediction methods[C]. Guidance, Navigation, and Control Conference and Exhibit, 2013:1406-1506.

Avanzini, Giulio. Frenet-Based Algorithm for Trajectory Prediction[J]. Journal of Guidance Control & Dynamics, 2000, 27(1):127-135.

Inseok Hwang, Jesse Hwang, Claire Tomlin. Flight-Mode-Based Aircraft Conflict Detection using a Residual-Mean Interacting Multiple Model Algorithm[C]. AIAA Guidance, Navigation and Control Conference and Exhibit. Austin: AIAA, 2003:1-11.

Franco A, Rivas D. Optimization of Multiphase Aircraft Trajectories Using Hybrid Optimal Control [J]. Journal of Guidance Control and Dynamics, 2015, 38 (3): 452-467.

张军峰, 蒋海行, 武晓光, 汤新民. 基于BADA及航空器意图的四维航迹预测[J].西南交通大学学报，2014，49(03):553-558.

谢丽, 张军峰, 隋东, 辛正伟. 基于交互式多模型滤波算法的航迹预测[J].航空计算技术, 2012, 42(05):68-70+74.

Ayhan S, Samet H. Aircraft trajectory prediction made easy with predictive analytics[J]. Proceedings of the 22nd ACM SIGKDD International Conference on knowledge discovery and data mining, 2016,13(17):21-30.

Shi Z , Min X , Quan P , et al. LSTM-based Flight Trajectory Prediction[C]. 2018 International Joint Conference on Neural Networks (IJCNN), 2018:1-8.

Liu Y , Hansen M. Predicting Aircraft Trajectories: A Deep Generative Convolutional Recurrent Neural Networks Approach[J]. 2018,8(3):1-24.

Zhao X , Yan H , Li J , et al. Spatio-temporal Anomaly Detection, Diagnostics, and Prediction of the Air-traffic Trajectory Deviation using the Convective Weather[J]. 2019, 11(1):1-8.

Pang Y , Yao H , Hu J , et al. A Recurrent Neural Network Approach for Aircraft Trajectory Prediction with Weather Features From Sherlock[C]. AIAA Aviation 2019 Forum, 2019:1-14.

Pang Y, Xu N, Liu Y. Aircraft trajectory prediction using lstm neural network with embedded convolutional layer[J]. Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM, 2019,11(1):1-10.

Pang Y , Liu Y . Conditional Generative Adversarial Networks (CGAN) for Aircraft Trajectory Prediction considering weather effects[C]. AIAA Scitech 2020 Forum,2020:1-11.

Guan X , Lv R , Liang S, et al. A study of 4D trajectory prediction based on machine deep learning[C]. Intelligent Control & Automation. IEEE, 2016:24-27.

钱夔, 周颖, 杨柳静, 谢荣平, 何锡点. 基于BP神经网络的空中目标航迹预测模型[J].指挥信息系统与技术, 2017, 8(03):54-58.

Zhao Z , Zeng W , Quan Z , et al. Aircraft Trajectory Prediction Using Deep Long Short-Term Memory Networks[C]. 19th COTA International Conference of Transportation Professionals, 2019.

Zeng W , Quan Z , Zhao Z , et al. A Deep Learning Approach for Aircraft Trajectory Prediction in Terminal Airspace[J]. IEEE Access, 2020, 8(1):151250-151266.

田杉. 基于神经网络的4D航迹预测方法[D].天津：中国民航大学, 2020.

Lim W, Zhong Z. Re-Planning of Flight Routes Avoiding Convective Weather and the "Three Areas"[J]. IEEE transactions on intelligent transportation systems, 2018, 19(4):868-877.

Yang, Yuanchao. Practical Method for 4-Dimentional Strategic Air Traffic Management Problem With Convective Weather Uncertainty[J]. IEEE Transactions on Intelligent Transportation Systems, 2018,11(5):1-12.

Gonzalez-Arriba D , Soler M , Sanjurjo-Rivo M , et al. Robust aircraft trajectory planning under uncertain convective environments with optimal control and rapidly developing thunderstorms[J]. Aerospace Science and Technology, 2019, 89(6):445-459.

冷雨泉, 张会文, 张伟. 机器学习入门到实战[M]. 北京：清华大学出版社, 2018:225-226.

周志华. 机器学习[M]. 北京：清华大学出版社, 2015:211-214.

张和平, 李俊武. 基于模糊c均值聚类算法的控制图模式识别[J].工业工程, 2021, 24(05):108-116.

吕云翔, 卓然, 关捷雄, 等. Python深度学习[M].北京：机械工业出版社, 2020:174-178.

Zhao Jianfeng, Mao Xia, Chen Lijiang．Speech emotion recognition using deep 1D & 2D CNN LSTM networks[J].Biomedical signal processing and control, 2019, 47(2):312-323.

Lv J, Li Q, Sun Q, Wang X. T-CONV: A Convolutional Neural Network For Multi-scale Taxi Trajectory Prediction[J]. 2018 IEEE International Conference on Big Data and Smart Computing, BigComp 2018, 2018,5(4):82-89.

Minh-Thang Luong. Addressing the Rare Word Problem in Neural Machine Translation[J]. Bulletin of University of Agricultural Sciences and Veterinary Medicine Cluj-Napoca, Veterinary Medicine, 2015, 27(2):82-86.

[美]桑德罗?斯卡尼著, 杨小东译. 深入浅出深度学习:从逻辑运算到人工智能[M].北京：清华大学出版社, 2021:127-151.

Krozel J, Mitchell L J, Polishchchuk V, et al. Capacity Estimation for Airspaces with Convective Weather Constraints[C]. AIAA Guidance, Navigation and Control Conference and Exhibit, 2007: 1-15.

Mitchell J, Polishchuk V, Krozel J. Airspace Throughput Analysis Considering Stochastic Weather[C]. AIAA Guidance, Navigation and Control Conference and Exhibit, 2006: 1-19.

方旖,毕大平,潘继飞,陈秋菊.基于主成分分析的雷达行为状态聚类分析方法[J].探测与控制学报, 2020, 42(02):112-11.

Tabassum A , Allen N , Semke W . ADS-B Message Content Evaluation and Breakdown of Anomalies[C]. IEEE/AIAA 36 th Digital Avionics System Conference. IEEE, 2017:1-8.

Syd Ali B , Schuster W , Ochieng W , et al. Analysis of anomalies in ADS-B and its GPS data[J]. Gps Solutions, 2016, 20(3):429-438.

李文杰, 闫世强, 蒋莹, 张松芝, 王成良. 自适应确定DBSCAN算法参数的算法研究[J].计算机工程与应用, 2019, 55(05):1-7+148.

陈凯达, 朱永生, 闫柯, 等. 基于LSTM的船舶航迹预测[J].船海工程, 2019, 048(006):121-125.

陈肯,杨晓刚.基于改进Event模型的航路垂直方向碰撞研究[J].航空计算技术, 2021, 51(05):15-18.