基于故障树和贝叶斯网络的动车组走行部运用可靠性分析方法

Analysis Method for the Operational Reliability of EMU Running Gear Based on Fault Tree and Bayesian Network

为克服故障树分析法对事件二态性假定的局限，根据故障树与贝叶斯网络的映射关系（事件-节点、逻辑门-连接强度），将动车组走行部故障树中的事件看作是贝叶斯网络中的节点，采用多维变量描述节点（事件）的多态性；将与故障树中逻辑门对应的贝叶斯网络连接强度用相应的条件概率表示，计算系统故障的概率分布。对动车组走行部基础制动系统故障概率分布的计算分析结果表明，考虑事件的多态性和逻辑关系的不确定性，综合利用故障树分析法和贝叶斯网络法能有效提高动车组走行部运用可靠性分析的质量，推算出的系统故障概率分布更为准确。

To get over the restriction that the two states of the event are supposed by Fault Tree Analysis （FTA）, the events in Fault Tree （FT） of EMU running gear can be seen as the points in Bayesian Net- works （BN） and the states can be described by multi-dimensional variables, according to the relation be- tween FT and BN, including the mapping of event to point and logical gate to connected intensity. The connected intensity in BN corresponding to the logical gate in FT will be presented as corresponding condi- tional probability to compute the probability distribution of system failure. The computing analysis of the fault probability distribution of foundation brake system in EMU running gear shows that, when different states of events and uncertain logical relationship are considered, the analysis quality for the operational re- liability of EMU running gear can be improved by comprehensively using FTA and BN, and the probability distribution of system failure can be calculated more accurately.