基于动态贝叶斯网络的深水防喷器可靠性研究  

Dynamic Bayesian Networks-based Research on the Reliability of Deepwater BOP

作　　者：顾和元 侯国庆[1] 吴占伟[1] 

机构地区：[1]河北华北石油荣盛机械制造有限公司

出　　处：《石油机械》2013年第3期36-39,共4页China Petroleum Machinery

基　　金：国家863计划项目"3 000 m深水防喷器及其控制系统研制"(2007AA09A101);国家科技重大专项"深水钻井防喷器组及控制系统研制"(2011ZX05027-001-05)

摘　　要：研究深水闸板防喷器系统的可靠性具有十分重要的意义。为此,将深水闸板防喷器故障树模型转换为静态贝叶斯网络模型,再经时序扩展得到其动态贝叶斯网络模型。利用BNT和GraphViz4Matlab软件包对动态贝叶斯网络进行建模和推理分析,得出如下结论:①在400周内,深水闸板防喷器的可靠性随时间的延长逐渐下降,在第400周时趋近于0;②深水闸板防喷器的可靠性随基本事件节点退化导致失效的概率DS1和DS2的增大而下降,且DS1的影响大于DS2;③研究的4个基本事件退化导致的失效概率对系统可靠性的影响程度从大到小依次为PILFC、YPFMR、ELYCFH和LCFLYP。因此,在实际可靠性设计和评估中应根据基本事件对系统可靠性的影响大小给予不同程度的重视。The deepwater ram BOP fault tree model was converted into the static Bayesian networks model. Then, its dynamic Bayesian networks model was derived through time sequence extension. The BNT and GraphViz4Matlab software package was applied to conduct the modeling and inferential analysis of the dynamic Bayesian networks. The following conclusions have been reached. First, the reliability of the BOP gradually goes down with time within 400 weeks and it will reach zero in the 400th week. Second, the reliability will decrease with the increase of the failure probabilities DS1 and DS2 caused by the degradation of elementary event node. The effect of DS1 is greater than that of DS2. Third, the degrees of effects of the failure probability caused by the degradation of the four elementary events studied on the system reliability are PILFC, YPFMR, ELYCFH and LCFLYP respectively from big to small. Therefore, attention should be paid to them at various degrees according to the effect of elementary event on the system reliability in the practical reliability design and assessment.

关 键 词：深水防喷器 闸板 故障树 Noisy-or模型 动态贝叶斯网络 失效概率 可靠性 

分 类 号：TE931[石油与天然气工程—石油机械设备]