隨著航空運輸的成長，飛行安全變得尤為重要，最後進場和著陸階段往往是一趟航班中風險最高的，因此本論文的研究範圍即針對了最後進場以及著陸階段進行研究。由於FOQA數據為航空公司的機密資訊且不易取得，因此作者決定使用公開的ADS-B資料中取得的基本飛行數據（速度、垂直速度、航向）以及軌跡資訊（經度、緯度、高度）再加上航空例行天氣報告（METAR/SPECI）中所獲得機場天氣資訊，進行航機進場階段的風險評估。本研究選擇了台北松山機場一年期間(2019/07/01~2020/06/30)內的ATR72-600機型降落航班資料作為主要研究對象，透過計算每個航班的軌跡偏差量及下滑角偏差量，並以這兩個偏差變量作為評估標準。將該航班當下的天氣條件視為影響因素，運用深度學習的概念建立類神經網路模型來訓練各個天氣因素之權重，經過一系列的模型調整之後，從預測準確度最高的模型中提取權重，作為各個天氣因素用於後續風險計算中的權重。將各個天氣條件乘上相對應的權重並加總，加總後的值即為該航班的風險值。最後再以整體的風險值分布來進行松山機場降落航班的風險評估。

With the increase in air transportation density, flight safety has become increasingly more important, especially in the final approach and landing phases of a flight. Since the flight operations quality assurance (FOQA) data are considered confidential for airlines, as an alternative, publicly available sources are used in this study, including the Automatic Dependent Surveillance-Broadcast (ADS–B) flight data and the airport weather information obtained from the aviation routine weather report (METAR) to carry out a risk analysis of the final approach phase. The goal of this thesis is to establish a standard by which to calculate the risk index of the flights landing at Taipei Songshan Airport (TSA). An artificial neural network with deep learning concepts is used in this study to resolve the nonlinear issues that conventional statistical regression methods cannot overcome. Trajectory deviations, as well as deviations in the glide path angle deviation of each flight, are calculated and used in the model training. These two deviation variables are taken as the evaluation standards, and the weather conditions are used as influencing factors. The weights of each weather factor are trained using the neural network as the risk coefficient for the subsequent risk calculation. A risk evaluation standard for flights bound for Taipei Songshan Airport was successfully established.

[1] T. T. S. Board, "Taiwan Aviation Occurrence Statistics 2010-2019," 2020.
[2] Boeing, "2019 Statistical Summary of Commercial Jet Airplane Accidents," 2019.
[3] K. Sheridan, T. G. Puranik, E. Mangortey, O. J. Pinon-Fischer, M. Kirby, and D. N. Mavris, "An application of dbscan clustering for flight anomaly detection during the approach phase," in AIAA Scitech 2020 Forum, 2020, p. 1851.
[4] D. ICAO, "9871: Technical Provisions for Mode S Services and Extended Squitter, AN/464," 2008.
[5] M. Schäfer, M. Strohmeier, V. Lenders, I. Martinovic, and M. Wilhelm, "Bringing up OpenSky: A large-scale ADS-B sensor network for research," in IPSN-14 Proceedings of the 13th International Symposium on Information Processing in Sensor Networks, 2014: IEEE, pp. 83-94.
[6] E. O. Eldawy and H. M. O. Mokhtar, "Clustering-Based Trajectory Outlier Detection," (in English), Int. J. Adv. Comput. Sci. Appl., Article vol. 11, no. 5, pp. 133-139, May 2020. [Online]. Available: <Go to ISI>://WOS:000540015500021.
[7] X. Olive and L. Basora, "Identifying anomalies in past en-route trajectories with clustering and anomaly detection methods," in ATM Seminar 2019, 2019.
[8] F. Gökgöz, "Anomaly detection using gans in opensky network," NATO Science and Technology Organization: Big Data and Artificial Intelligence for Military Decision Making, 2018.
[9] A. Hanifa and S. Akbar, "Detection of unstable approaches in flight track with recurrent neural network," in 2018 International Conference on Information and Communications Technology (ICOIACT), 2018: IEEE, pp. 735-740.
[10] M. Y. Pusadan, J. L. Buliali, and R. V. H. Ginardi, "Anomaly detection on flight route using similarity and grouping approach based-on automatic dependent surveillance-broadcast," International Journal of Advances in Intelligent Informatics, vol. 5, no. 3, pp. 285-296, 2019.
[11] Z. Shi, M. Xu, Q. Pan, B. Yan, and H. Zhang, "LSTM-based flight trajectory prediction," in 2018 International Joint Conference on Neural Networks (IJCNN), 2018: IEEE, pp. 1-8.
[12] Z. Zhao, W. Zeng, Z. Quan, M. Chen, and Z. Yang, "Aircraft trajectory prediction using deep long short-term memory networks," in CICTP 2019, 2019, pp. 124-135.
[13] M. C. R. Murça, R. J. Hansman, L. Li, and P. Ren, "Flight trajectory data analytics for characterization of air traffic flows: A comparative analysis of terminal area operations between New York, Hong Kong and Sao Paulo," Transportation Research Part C: Emerging Technologies, vol. 97, pp. 324-347, 2018.
[14] S. R. Proud, "Go-Around Detection Using Crowd-Sourced ADS-B Position Data," Aerospace, vol. 7, no. 2, p. 16, 2020.
[15] G. Gui, F. Liu, J. L. Sun, J. Yang, Z. Q. Zhou, and D. X. Zhao, "Flight Delay Prediction Based on Aviation Big Data and Machine Learning," IEEE Transactions on Vehicular Technology, vol. 69, no. 1, pp. 140-150, Jan 2020, doi: 10.1109/tvt.2019.2954094.
[16] M. Schultz, X. Olive, J. Rosenow, H. Fricke, and S. Alam, "Analysis of airport ground operations based on ADS-B data," in 2020 International Conference on Artificial Intelligence and Data Analytics for Air Transportation (AIDA-AT), 2020: IEEE, pp. 1-9.
[17] A. Renz-Wieland and H. Wallenburg, "Airport quality : holding and go-arounds," 2016.
[18] A. De Leege, M. Van Paassen, and M. Mulder, "Using automatic dependent surveillance-broadcast for meteorological monitoring," Journal of Aircraft, vol. 50, no. 1, pp. 249-261, 2013.
[19] J. Kopeć, K. Kwiatkowski, S. de Haan, and S. Malinowski, "Retrieving clear-air turbulence information from regular commercial aircraft using Mode-S and ADS-B broadcast," Atmospheric Measurement Techniques Discussions, vol. 8, no. 11, 2015.
[20] S.-S. Jan and Y.-T. Chen, "Development of a new airport unusual-weather detection system with aircraft surveillance information," IEEE Sensors Journal, vol. 19, no. 20, pp. 9543-9551, 2019.
[21] Y. Fan, "Analysis of Influence of Clear Air Turbulence on Aircraft," 2019.
[22] J. Mazon, J. Rojas, M. Lozano, D. Pino, X. Prats, and M. Miglietta, "Influence of meteorological phenomena on worldwide aircraft accidents, 1967–2010," Meteorological Applications, vol. 25, no. 2, pp. 236-245, 2018.
[23] Z. Wang, "A methodology for nowcasting unstable approaches," George Mason University, 2016.
[24] X. Olive and P. Bieber, "Quantitative assessments of runway excursion precursors using Mode S data," arXiv preprint arXiv:1903.11964, 2019.
[25] S. A. Sarcià, G. Cantone, and V. R. Basili, "A statistical neural network framework for risk management process-from the proposal to its preliminary validation for efficiency," in International Conference on Software and Data Technologies, 2007, vol. 2: SCITEPRESS, pp. 168-177.
[26] F. Bati and L. Withington, "Application of Machine Learning for Aviation Safety Risk Metric," in 2019 IEEE/AIAA 38th Digital Avionics Systems Conference (DASC), 2019: IEEE, pp. 1-9.
[27] G. E. Hinton and R. R. Salakhutdinov, "Reducing the dimensionality of data with neural networks," science, vol. 313, no. 5786, pp. 504-507, 2006.
[28] EUROCONTROL, "ATMAP weather algorithm " in "Algorithm to Describe Weather Conditions at European Airports," 10 May 2011 2011.
[29] R. Bogdane, O. Gorbacovs, V. Sestakovs, and I. Arandas, "Development of a model for assessing the level of flight safety in an airline using concept of risk," Procedia Computer Science, vol. 149, pp. 365-374, 2019.
[30] D. Koks, "Using Rotations to Build Aerospace Coordinate Systems," 2008.
[31] S. Karsoliya, "Approximating number of hidden layer neurons in multiple hidden layer BPNN architecture," International Journal of Engineering Trends and Technology, vol. 3, no. 6, pp. 714-717, 2012.
[32] K. G. Sheela and S. N. Deepa, "Review on methods to fix number of hidden neurons in neural networks," Mathematical Problems in Engineering, vol. 2013, 2013.
[33] T. Vujicic, T. Matijevic, J. Ljucovic, A. Balota, and Z. Sevarac, "Comparative analysis of methods for determining number of hidden neurons in artificial neural network," in Central european conference on information and intelligent systems, 2016: Faculty of Organization and Informatics Varazdin, p. 219.