利用多种信息源的可靠性评估方法

Reliability Assessment Using Multiple Information Sources

进行小样本可靠性评估的关键是充分利用由科学分析和专家经验得到的主观信息以及输入参数的基本试验信息.主观推断一般仅能得到可靠性的不完全先验信息,这些信息通常以可靠度均值或可信区间的形式存在.针对成败型及正态型试验,用最大熵原理将不完全信息转化为相应的共轭型先验信息,而输入量的试验信息通过功能函数和统计理论被转化成输出量的先验信息.先验信息和试验信息的融合则通过贝叶斯理论来实现.介绍了计算方法,通过算例分析了非试验信息对可靠度后验分布、试验次数以及可靠度评估结果的影响.

The key step for reliability assessment with small sampling is to take advantage of the subjective information and the test data of input parameters. Subjective inference could just offer the reliability some incomplete information which can be generally assumed to exist in the form of either a prior mean or a prior credibility interval. To the trials with the outcomes of either survival\|failure or normal distributed parameters, efficient approaches are developed to determine the conjugate prior distributions from the subjective information according to the principle of maximum entropy. The test data of input parameters are also transformed as the prior information of the output parameters according to statistics theory. Bayes theorem is used to synthesize the different informations and numerical examples are presented to illustrate the impact of non\|experimental information on the reliability posterior, the needed test number and the reliability assessment.