节点文献

寿命试验下一个失效时间的预测

Prediction of the Failure Time of Life Test

【作者】 林穗华

【导师】 程依明;

【作者基本信息】 华东师范大学 , 概率论与数理统计, 2007, 硕士

【摘要】 本文主要内容之一是依据贝叶斯统计理论与方法,分别对定时截尾寿命试验和定数截尾寿命试验,利用观察到的前r个失效时间t1,t2…,tr,求出第r+1个失效时间的后验分布,并由此得出第r+1个失效时间的点估计与置信度为1-α的区间估计,对常用的寿命分布:单参数指数分布、双参数指数分布、双参数威布尔分布分别使用本文所得的方法进行了讨论,并对预测结果进行了随即模拟。本文的另一主要内容是依据观察到的定时截尾寿命试验前r个失效时间t1,t2,…,tr,,运用非参数统计方法,求出第r+1个产品将再活过一段时间△t的概率Pr=P(tr+1-t0|t),并由此得出第r+1个失效时间的置信度为1-α的置信区间,对单参数指数分布、双参数指数分布、双参数威布尔分布分别使用三种方法估计出参数后,利用本文所得的方法进行了讨论,并利用一些现实数据对研究的结果进行了应用试验,对三种方法的预测结果进行了比较。

【Abstract】 This dissertation consists of two parts. In the first part, in reference to the Bayes Statistics theory and methods, the author draws the Posterior Distributions ofthe two failure times (r+1) based on each first failure time (r) (t1,t2,…,tr) respectivelyobserved in the Life Test of Type-I Censoring Case and the Life Test of Type-ⅡCensoring Case, and hence obtains the point estimation of each failure time (r+1) and the interval estimation of each confidence level (1-α). Whereafter, the regular life distributions (one-parameter exponential distribution, two-parameter exponential distribution, two-parameter Weibull distribution) are probed with the methods discussed above and the related predictions are verified by stochastic simulation.In the other part of this dissertation, on the thereunder of first failure time (r)(t1,t2,…,tr) observed in the Type-I Censoring Test, the author draws the probability(Pr=P(tr+1-t0|t)) of the subsequent survival time (△t) of product (r+1) bynonparametric statistical method, and hence obtains the confidence interval of the failure time (r+1) on confidence level (1-α). Then, the parameters of one-parameter exponential distribution, two-parameter exponential distribution, and two-parameter Weibull distribution are estimated respectively with three related methods, and the failure time (r+1) is discussed in the methods obtained above in this paper. Afterwards, the practical test on the results and the comparison among the predictions drew by the three methods are conducted with substantial data.

  • 【分类号】O212.8
  • 【被引频次】1
  • 【下载频次】135
节点文献中: