Abstract:

Because of defects of classical statistical method and traditional Bayesian method, a reliability evaluation method of binomial products based on the multistage information was proposed. The concept of similarity coefficient was introduced based on the chi-square goodness of fit. The similarity coefficients of each stage is got test information and field test information were determined respectively. Then the confidence lower limit was determined with each stage is got test information as information prior to the field test. Finally, the confidence lower limit of the product. The simulation examples show that the calculation results are between classical statistical method and traditional Bayesian method, and the evaluation result is more reasonable.

Key words: reliability evaluation, similarity coefficient, Bayesian analysis, posteriori distribution

摘要：

针对经典统计方法和传统Beyes方法的缺陷,基于卡方拟合优度检验，引入相似系数的概念,提出了一种基于多阶段试验信息的成败型产品可靠性评价的加权Bayes方法。通过计算各阶段试验信息和现场试验信息的相似系数，求出各阶段试验信息作为验前信息的置信下限，据此计算得到了产品的可靠性置信下限。仿真算例结果表明,该方法的计算结果介于经典统计方法和传统Bayes方法之间,与根据专家意见计算的结果最接近,评价结果更加合理。

关键词: 可靠性评价, 相似系数, 贝叶斯分析, 验后分布