中文摘要：

针对航空公司安全质量管理体系（SQMS）中风险识别与可靠性改进问题,文中提出基于区间数学与贝叶斯网络的灵敏度分析方法.给出了衡量航空公司安全状况的指标;借助贝叶斯网络建立指标关联网络,结合先验知识进行参数学习以获取条件概率;通过贝叶斯双向推理机制和区间数学理论分析不同区间内重要指标与系统整体安全状况间的扰动关系;基于扰动分析结果对各指标进行灵敏度排序以识别风险并指导可靠性改进工作.实例分析验证了基于区间数学与贝叶斯网络的指标灵敏度分析方法的有效性.

英文摘要：

To meet the requirements of risk identification and reliability engineering proposed by Safety and Quality Management System（SQMS）,this paper presents a novel approach to sensitivity analysis by combining Bayesian network with interval mathematics.Firstly,the security status indicators of the airlines are given.Then,Bayesian network is utilized to construct a directed graph for these indicators,and prior knowledge is taken into account for parameters learning to obtain conditional probability table（CPT）.After that,interval mathematics and Bayesian bidirectional inference mechanisms are incorporated for the analysis of the interaction between the indicators and the overall security level.Finally,according to the obtained interaction,sensitivity analysis can be achieved to provide a guide for operators to implement risk identification and reliability engineering.A case study is presented to verify the effectiveness of the novel sensitivity analysis method in SQMS for airlines.