针对有多个研制阶段的成败型系统,提出了一种可靠性增长的Bayes评定模型。该模型首先根据Laplace检验,确定系统的可靠性是否增长;然后根据各个阶段可靠度的点估计,运用DGM(1,1)模型对现场试验的可靠度进行预测,得到验前信息,再运用列联表分析计算验前信息与现场试验信息的相似程度;最后,结合Bayes方法对验前信息和现场信息进行融合,在给定的置信度下,得到系统的可靠度置信下限。实例分析表明了该模型的有效性。
For a binomial system that had several stages' informations,this paper introduced an Bayesian evaluation methods of reliability growth. The model first determined whether the system reliability grew ac- cording to the Laplace inspection. Then prior information could be obtained according to the reliability point estimation of each stage using DGM (1,1) model. And the similarity coefficient of prior information and field information were calculated. Finally,prior information and the field information were fused using the Bayesian method. The examt)les show the effectiveness of the proposed model.