串联系统的可靠性下限评估方法研究

Estimation Methods of Low Bounds on Series System Reliability

本文针对小子样可靠性统计试验的特点，指出经典可靠性点估计和极大似然区间估计的不足。然后着重介绍了Ｂｕｅｈｌｅｒ精确置信下限评定方法以及ＭＭＬ方法、Ｌｉｎｄｓｔｒｏｍ－Ｍａｄ－ｄｅｎ方法、熵方法、似然比方法、贝叶斯方法等常用的可靠性评定方法及其特点。最后以具体数字实例将文中介绍的近似方法与精确评定方法进行比较，得出较好的适合于小子样无失效情况的核武器可靠性评估方法。本文“挑选”出来的可靠性评定方法还需作更进一步的研究。

For the case of the reliability statistical test which is a small sample, this paper presents disadvantages of point and interval estimation of classical reliability. Then, we describe common reliability estimation methods, such as Buehler's exact method and MML、L M、Entropy、Likelood、Bayes method, etc. By the comparison of those above approximation methods with Buehler's exact method using examples, better approximation methods used for the case of small sample and zero failure are obtained.[WT5”HZ〗