由於數位技術的發展,以及類比控制系統逐漸淘汰,許多核能系統都已經從類比被更新至電腦基礎系統,這些新一代的電腦基礎系統將電腦應用在感應器、促動器和軟體方面,當類比系統被電腦控制設備取代以後,可以減輕操作者的負擔以及降低維修成本。雖然以電腦為基礎設計出來的系統有很多優點,但是在操作或維護階段,軟體和硬體結合系統中的人機介面或是專案管理上還是具有一些特徵可能會引起意外事件。 核能署(Nuclear Energy Agency , NEA)是國際組織經濟合作暨發展組織(Organization for Economic Cooperation and Development, OECD)下的一個機構,OECD NEA成立了一個電腦系統安全相關的計畫(Computer-based Systems Important to Safety, COMPSIS)去收集和分析核電廠安全相關意外事件的操作經驗,其目的是提供一個平台讓會員國互相分享資訊、經驗,促進國際間的合作,以期提高各國在核能上的專業技術。 在這個議題之下,我們必須發展出肇因與影響分析模型(Causal and Effect Analysis Model, CEAM)和本體論肇因分析系統(Ontology-based Cause Analysis System)去深入分析COMPSIS中的意外事件,此外我們還使用資訊網本體論語言(Web Ontology language, OWL)和推論引擎FaCT++去支援推論過程,最後顯示出統計分析結果與經驗教訓,本研究期望所提出的模型可以被應用在核電廠安全相關系統的設計、操作和維護上。
Due to the progress of digital technology and obsolescence of analog control equipment, many nuclear systems have been upgraded from analogy to computer-based systems. The newer computer-based systems utilize computers with sensors, actuators, and software. When these systems can be replaced with computer-controlled equipment, the operator’s burden and maintenance costs can often be reduced. Although there are many advantages for computer-based design, there are still some characteristics inherent in software and hardware integrated systems, human-machine interfaces and project management which may cause failure events during operation or maintenance phase. The Nuclear Energy Agency (NEA) which is a specialized agency within the Organization for Economic Cooperation and Development (OECD). OECD NEA promoted the Computer-based Systems Important to Safety (COMPSIS) project to collect and analyze operating experience on safety events. Its aim is provide a forum for sharing information and experience and promoting international co-operation; a centre of excellence which helps member countries to pool and maintain their technical expertise. In this thesis, we have developed a Causal and Effect Analysis Model (CEAM) and Ontology-based Cause Analysis System to analyze the COMPSIS events visually and in depth. In addition, we also employed OWL (Web Ontology language) and inference engine FACT++ to support reasoning process. Finally, statistics and the lessons learned are given. It is anticipated that our proposed model can apply to the design, operation and maintenance of safety system in nuclear power plant.