小样本条件下行波管可靠性评估方法的研究

 行波管的可靠性对许多军用装备整机系统的正常运行十分重要。但其结构复杂、价格昂贵，如何在小样本条件下提高其可靠性评估的精度是我们目前面临的问题。在此情况下，利用专家经验、历史数据等先验信息来提高其可靠性评估的精度是一个行之有效的途径。本文基于模糊隶属函数给出了一种创建行波管可靠性模糊先验分布的方法，并在此基础上利用Bayes方法实现了行波管先验与试验信息的有效融合。对某卫星用行波管进行可靠性评估的实例表明，采用正态型模糊先验分布的行波管可靠性Bayes评估可以在小样本数和试验结尾程度很高的情况下显著提高可靠性评估的精度，同时Bayes估计可随试验样本信息的增加不断得到修正;而模糊先验分布的带宽可用于调节先验信息在后验分布中所占的比重。

Reliability Evaluation for TWT with Small Samples

Reliability of Traveling Wave Tubes (TWTs) are crucial for operation of many armaments. However, due to the complexity and expensiveness of TWTs, the way to precisely evaluate their reliability under small sample size is still the problem to face. In this way, prior information, such as expert experience and historical data, can be used to enhance the estimation precisions. With a fuzzy prior distribution construction method for TWTs introduced, based on fuzzy membership functions, prior information can be effectively merged with test data using Bayes method. Reliability evaluation for some satellite TWTs shows that, the precision can be enhanced notably for data with small samples or high truncation levels by using Bayes estimation with normal fuzzy prior distributions. And the estimation can be modified by increasing test information. Furthermore, the bandwidth of the normal priors can be used to adjust the proportion of prior information within the post-Bayes distributions.