Submarine pipelines are the major transportation mode of marine oil and gas resources. Because of submarine pipeline damage, the leakage of oil and gas will result the serious consequences such as environmental disasters, fires and explosions, and huge economic losses. There are variously internal and external factors that initiate spill accidents of oil and gas. To prevent and mitigate such accidents, risk analysis is an efficient way. Fault tree analysis is an effective tool to identify failure causes and perform the risk assessment. In fault tree analysis, it is presumed that all basic events are statistically independent and have precise occurrence probabilities. In the case, when probability data of basic events of system fault tree are unavailable or imprecise, the concept of fuzzy set theory and expert elicitation is used to obtain qualitative data. In the quantification of qualitative data, experts‘ knowledge is used which may raise issues such as incompleteness, imprecision, and lack of consensus. In the process to minimize the uncertainty, expert’s opinions are aggregated and updated to the posterior possibilities using the prior observations. Bayesian networks have the advantages of representing the dependencies of events, updating probabilities, and dealing with uncertainties. In this research paper, a novel methodology is proposed by combining fuzzy fault tree analysis and Bayesian network to obtain updated prior possibilities of basic events and top event of system fault tree when new information are available. The main contributions of this research are: weakest t-norm based arithmetic operations on fuzzy numbers are employed for less uncertainty accumulation during the process; weakest t-norm and α-cut based similarity aggregation method is developed to evaluate the possibilities of basic events and top event in system fuzzy fault tree analysis; the obtained prior possibilities are then updated using fuzzy Bayesian network; criticality analysis is executed using the posterior possibilities of basic events and top event. Further, a case study of leakage in submarine pipeline is discussed to demonstrate the applicability and effectiveness of the proposed methodology. The obtained results are then compared with the pre-existing results which shows the validity and applicability of the proposed method.

海底管道是海洋油气资源的主要运输方式。由于海底管道损坏，油气泄漏将造成环境灾难、火灾爆炸等严重后果，造成巨大的经济损失。引发油气泄漏事故的内部和外部因素多种多样。为了预防和减轻此类事故，风险分析是一种有效的方法。故障树分析是识别故障原因和进行风险评估的有效工具。在故障树分析中，假设所有基本事件在统计上是独立的，并且具有精确的发生概率。在系统故障树的基本事件的概率数据不可用或不精确的情况下，模糊集理论的概念专家启发用于获得定性数据。在定性数据的量化中，使用了专家的知识，这可能会引发诸如不完整、不精确和缺乏共识等问题。在最小化不确定性的过程中，专家的意见被汇总并使用先前的观察更新到后验可能性。贝叶斯网络具有表示事件相关性、更新概率和处理不确定性的优点。在本研究论文中，提出了一种新的方法，将模糊故障树分析和贝叶斯网络相结合，在有新信息时获取系统故障树的基本事件和顶事件的更新先验可能性。本研究的主要贡献是：基于最弱 t 范数对模糊数进行算术运算，以减少过程中的不确定性累积；开发了基于最弱t-范数和α-割的相似性聚合方法，用于评估系统模糊故障树分析中基本事件和顶事件的可能性；然后使用模糊贝叶斯网络更新获得的先验可能性；使用基本事件和顶级事件的后验可能性执行关键性分析。此外，还讨论了海底管道泄漏的案例研究，以证明所提出方法的适用性和有效性。然后将获得的结果与预先存在的结果进行比较，表明了所提出方法的有效性和适用性。