一种改进的可靠性分配方法及其应用验证

可靠性分配是系统设计中的重要工作，常用的可靠性分配方法如AGREE分配法、评分分配法对影响可靠性指标的因素考虑不全，并且都是针对串联系统的。利用贝叶斯网络在不确定性概率推理方面的优势，将传统的故障树转化为贝叶斯网络求解最小割集和底事件重要度。建立了改进的可靠性分配模型，首先将顶层可靠性指标在最小割集之间分配，然后根据底事件重要度和评分将最小割集对应的可靠性指标分配到底事件。以某型飞机电传操纵系统可靠性指标分配为例，仿真结果表明，贝叶斯网络相对于故障树分析计算底事件重要度更为简单、精度更高。改进的分配方法能够考虑影响可靠性指标的主要因素，实现指标的科学分配。

An improved reliability allocation method and its application verification

Reliability allocation is an important work in system design.Current reliability allocation method such as AGREE and score grading allocation cannot take all factors into consideration,and they are all fit for series system only.Bayesian Network(BN) shows great advantage on uncertainty probability reasoning.Minimum cut sets and importance of bottom events are obtained by transforming Fault Tree(FT) into BN.An improved reliability allocation model is constructed,in which reliability index of top event is allocated among minimum cut sets first,and then reliability index of minimum cut sets is allocated to bottom events based on their importance and scores.Taking reliability index allocation of Fly-by-Wire Flight Control System for aircraft as example,case study shows that importance of bottom events calculated by BN is simpler in calculation and has higher accuracy than FT Analysis.Allocation of reliability index can be realized by the improved allocation model which can take all main factors into consideration.