本研究針對量化分析飛航事件的可行性，提出依據線性理論的相關探索，並針對FOQA系統偵測異常的航空事件情境，嘗試使用ROC分析找尋可靠的情境分類門檻。本研究首先將飛航紀錄所得的數據，以半高斯濾波的方式，轉化為代表每一瞬間飛航情境的情境矩陣，並假設每一瞬間飛航情境的變化是依據線性法則。再藉由線性代數中奇異值分解方法，將情境矩陣進行拆解。透過奇異值分解，本研究將物理空間中的飛航情境，轉換到奇異空間中，透過其線性獨立的特性，針對飛航情境的變化，進行相關之後續分析。並提出12項指標，嘗試做為區隔出飛航事件的依據，再以航空事件中的情境指標為基礎，使用ROC分析建立事件分類的初步門檻。結果顯示，利用此12項指標，平均可以篩檢出約八成的異常事件；對於正常航班的篩檢，平均更有近九成的準確率，說明本研究中用來篩檢的指標，具有相當的大可靠度。

The goal of this research is to explore the feasibility of quantifying
flight operation safety by using engineering tools. With singular value decomposition, the recorded flight data were decomposed into a mutually independent space so that the safety situation of flights is fully included.
Comparing normal flights with event flights distinguished by FOQA system, it is possible to reveal the difference of stability and divergence between normal flights and event flights.
Following using ROC analysis to find the suitable cut-off point in different flight situations, so that it is possible to define reliable situation classifiers to tell the flight is safety or not.

[1]Statistical Summary of Commercial Jet Airplane Accidents,
Worldwide Operations 1959–2010,Boeing,2011.
[2]台灣年度飛安統計,飛安會, 2012.
[3]International Civil Aviation Organization.ICAO Safety Management Manual(Doc.9859).CA,2005.
[4]Federal Aviation Administration. Introduction to Safety Management Manual for Air Operations(AC No: 120-92).Washington D.C.,2006.
[5]Heinrich,H.W.,Industrial Accident Prevention,3rd ed.,N.Y.,McGraw-Hill Book Co.,1950.
[6]Reason,J.,Human Errors,New York,Cambridge University Press,1990.
[7]Perrow C.,Normal Accident:Living with High Risk Technologies,NJ,Princeton University Press,1984.
[8]U.S General Accounting Office,Aviaion Safety:U.S.Efforts to Implement Flight Operational Quality Assurance Program,Flight Safety Digest,Vol.17,no.7-9,1998.
[9]景鴻鑫，本土化之飛安理念，飛航安全討論與提升研討會，國立成功大學，中華民國八十七年
[10]景鴻鑫，盛嘉昇，”飛航安全之幾何觀點-飛航安全裕度”，兩岸飛行研討會，財團法人中華民國民航飛行協會，2006
[11]鐘華興，”飛航安全之工程分析-線性系統觀點”，國立成功大學民航研究所，中華民國九十七年
[12]陳品妤，”民航操作品質數據之安全性分析”， 國立成功大學民航研究所，中華民國九十九年
[13]T. Fawcett, An introduction to ROC analysis, Pattern Recognition Letters 27 pp.861–874, Institute for the Study of Learning and Expertise, 2005