基于Bayes方法的小子样可靠性分析

少量数据的可靠性分析与评定一直是工程实践中的技术难题之一。Bayes方法可以融合多种验前信息,并结合现场数据,从而对少量数据作出有效的参数估计。针对多种验前信息的综合应用问题,引入支持向量机(support vectormachine,SVM)理论对不同来源的信息在验前分布中的权重分配策略进行研究,提出一种新的信息融合思路。并在此基础上,提出基于Bayes理论的威布尔分布的小样本参数估计方法。最后,运用实例对Bayes方法和传统的参数估计方法进行对比分析。

RELIABILITY ANALYSIS ON SMALL SAMPLE BASED ON BAYES

The reliability analysis and evaluation is difficult in engineering because of small sample tests.Bayes method can deal with the parameter estimation especially of small sample data,because the multi-source prior information was pretreated together with field information.First,on the basis of support vector machine(SVM) theory,the weight distribution of prior information was discussed to find a solution to manage the multi-information.Then,based on the Bayes theory,a method for parameter estimation of a little information was established to discuss the reliability assessment of Weibull distribution.Finally,availability and practicality of the methods were indicated in comparison with other conventional parameter estimation methods discussed before.