多源测试性综合评估数据等效折合模型与方法研究

Research on Multi-source Data Equivalent Methods for Testability Integrated Evaluation

由于用于开展测试性综合评估的先验信息形式多样,使得Bayes框架下能够处理的成败型数据不一致。为了解决这一问题,以测试性评估中专家数据、摸底试验数据、增长试验数据、可更换单元数据和虚拟试验数据这5种常见的先验信息为研究对象,在Bayes理论框架下分别研究提出相应的等效方法,实现了各类数据向成败型数据的折合。案例应用表明:所提方法合理有效,适用范围广,间接扩大了可用于测试性评估的数据样本量。

The form of prior data for testability integrated evaluation is too multiply to match with the bi- nomial data which is processed under the framework of Bayes theory. For this issue, five kinds of prior data, including expert data, preexposure test data, growth test data, replaceable unit data and virtual test data, are studied. Furthermore, five corresponding equivalent methods which transform the different types of data into the system-level binomial data are proposed based on Bayes theory. The proposed methods are validated via applied examples. Results show that the proposed methods have wide applicability, and also enlarge the amount of data used for testability evaluation indirectly.