一种面向可靠性的改进不确定性重要度分析方法

作为识别和评估系统薄弱环节的有效工具之一的重要度理论,一直都被广泛应用于系统可靠性和安全性工程。而作为其主要环节的不确定性重要度分析更是扮演着不可或缺的角色。因此,为有效表征多态系统的可靠性,本文给出了一种变换数据处理方式的改进不确定性重要度分析方法。该模型是以两个累积分布函数之间的空间几何距离为基础,通过改变随机输入变量的不确定性范围和对应分布情况,模拟和描述输入对系统输出变量的相对影响趋势,并以两个累积分布函数之间的定积分面积表示。最后利用系统故障树对其进行不确定性重要度分析。结果表明,空间几何距离是一种用来表示输入变化对输出分布变化的相对影响的不确定性重要度度量的有效工具。

A Reliability-oriented Improved Uncertainty Importance Measure

Importance measure has been widely explored and applied in reliability and safety engineering, which is effective method to identify and evaluate the vulnerable spot of one system. The uncertainty importance measure as a primary tool has played an important role. Therefore, there is a new method to show the reliability of multi-state system efficiently based on the uncertainty importance measure. The model in the present work simulates and describes the relative impact of input changes on output uncertainty, which depends on the changes of input distributions ranges, and the uncertainty importance measure is expressed by the definite integral area between two cumulative distribution functions. At last, a fault tree analysis is used to study the proposed uncertainty importance measure. From the result, the metric distance is an effective means to describe the uncertainty importance measure of relative impact of input distributions on the change of output distributions.