核能電腦安全事件肇因型態分析：以約莫理論方法

Root Cause Categories Analysis for Nuclear Computer-based Event：A Rough Set Approach

沈涵安(Han-An Shen)

指導教授 ： 周永燦

中原大學/電機資訊學院/工業與系統工程研究所/碩士(2009年)

https://doi.org/10.6840/cycu200901202

摘要

過去核電廠設備主要是以類比系統為主，而伴隨電子、資訊工業的發展，且舊有的類比系統核能級備品取得不易，可能增加維護成本，並降低整廠性能及可靠度，數位電腦系統之應用已成為趨勢。數位電腦系統具有高度資訊化與自動化之特性，然而新型電腦系統也產生了不同於以往類比系統之失效模式。
於2005年，經濟合作暨發展組織/能源署(Organization for Economic Co-operation and Development/Nuclear Energy Agency, 簡稱OECD/NEA)遂提出了Computer-Based Systems Important to Safety (COMPSIS)計畫以加強國際間使用電腦系統之失效經驗交流，COMPSIS計畫其中一個目的便是要成立一專門收集電腦安全事件之資料庫。
本研究以COMPSIS網站資料庫之事件為基礎並應用約莫理論進行核能電腦安全事件肇因型態分析。首先，本研究先對肇因進行階段、型態與影響程度分類，接著對所收集到的事件進行分析，進而透過演算法來引出規則並分析所得到之結果提出肇因型態影響強度分布圖。由研究結果顯示，COMPIS資料庫中影響電廠安全之電腦安全事件肇因發生最顯著的階段為設計（Design）、運行（Operation）與維護（Maintenance）階段，最顯著的肇因型態為硬體老化（Acts of Nature），上述結果提供相關人員未來預防方向以降低電腦安全事件所造成之傷害。
最後本研究會以約莫理論所推導出來的規則為基礎，建置出一套核能電腦安全事件分析系統（Nuclear Computer-based Safety Event Analysis System, 簡稱NCSEAS）提供診斷功能讓相關人員在事件發生時能透過本系統快速診斷出事件可能之肇因並得到建議改善方法。

關鍵字

肇因型態分析 ； 約莫理論 ； COMPSIS ； 電腦系統 ； 診斷系統

並列摘要

In the past, Nuclear Power Plants (NPPs) equipments were based on the analog-based system. Now, analog-based replacement parts and components are becoming more and more difficult to obtain. May increase the maintenance costs and decrease the whole plant performance and reliability. Computer-based systems are currently being used in operating NPPs worldwide. Computer-based systems provided with a high degree of information and automation characteristics. Moreover, the introduction of new failure modes which did not exist in traditional analog-based systems is difficult to perceive in a traditional sense.
The Computer-Based Systems Important to Safety (COMPSIS) project which under OECD/NEA was founded to enhance international cooperation and exchange operating experiences of NPPs in 2005. One of COMPSIS’s objectives was to develop a well structured and consistent computer-based event databank.
In this paper, we propose a rough set approach to analyze the COMPSIS event databank, which is intended to explore relationships between root causes and their effects on the safety of NPPs. In the first stage of the analysis, .the important impact phase, categories and effect level of root causes for event classification is identified. Then, analysis the databank’s events and the decision rules are generated by a rule induction algorithm. Based on the result, we presented the cause categories and impact strength distribution. We can observe that the most significant phases of computer-based system are design, operation and maintenance. The most significant cause category is acts of nature by hardware. The analysis result provides the prevention direction in the future.
Finally, a Nuclear Computer-based Safety Event Analysis System (NCSEAS) is presented. It also is anticipated that NCSEAS can be used to diagnosis the cause of NPPs computer-based system events.

並列關鍵字

Rough Set Theory ； Root cause categories analysis ； COMPSIS ； Computer-based system ； Diagnosis system

參考文獻

中華民國核能學會, http://chns.org/index.php

Ananthanarayana, V. S., Narasimha Murty, M., and Subramanian, D.K. (2003).Tree structure for efficient data mining using rough sets. Pattern Recognition Letters 24(6). 851-862.

Doherty, P., Lukaszewicz, W., and Szalas, A. (2007). Communication between agents with heterogeneous perceptual capabilities. Information Fusion 8(1). 56-69.

DÜNTSCH, I., and GEDIGA, G. (1997). Statistical evaluation of rough set dependency analysis. International Journal of Human-Computer Studies 46(5). 589-604.

Greco, S., Matarazzo, B., and Slowinski, R. (2001). Rough sets theory for multicriteria decision analysis. European Journal of Operational Research 129(1). 1-47.