...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Quality Control and Applied Statistics >An application of Bayesian posterior analysis for disc drive annual failure rate (AFR) estimate
【24h】

An application of Bayesian posterior analysis for disc drive annual failure rate (AFR) estimate

机译：贝叶斯后验分析在磁盘驱动器年故障率（AFR）估计中的应用

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

In the disc drive industry annual failure rate (AFR) has important role in making business decisions. AFR can be defined as the failure probability for one year power-on operations estimated from the data generated by reliability demonstration tests (RDTs). Using AFR the manufacturer decides on the warranty cost and some other business decisions. Thus the AFR estimate has to be as accurate as possible. Due to limited sample size short duration of tests, it may be difficult to make such accurate estimations. Thus, the confidence intervals of AFR with Weibull distribution application are often carried out to see that it is too wide. Therefore, the upper bound is used as an estimate to avoid under estimation. Since the confidence intervals are too large, business decisions based on them will be erroneous. Thus there is a need for closer upper bounds for AFR so that the estimate used will be more reliable. Increasing the sample size to achieve this goal is cost prohibitive. The use of large quantity of historical data to gain the prior knowledge about actual product failure in the field couple with the test generated estimates can improve the estimation by dampening the inaccuracy in the test data. To enable this, the study proposes a Bayesian posterior analysis approach for AFR estimate. In this process, prior distributions of Weibull distribution parameters are estimated from historical data Statistical models of the AFR posterior estimate and its confidence intervals are derived. This is shown to be narrowing down the posterior confidence interval. (9 refs.)
机译：在磁盘驱动器行业中，年度故障率（AFR）在制定业务决策中具有重要作用。 AFR可以定义为根据可靠性演示测试（RDT）生成的数据估算的一年开机操作的故障概率。制造商使用AFR来决定保修成本和其他一些商业决定。因此，AFR估算必须尽可能准确。由于有限的样本量和较短的测试持续时间，可能难以进行这种准确的估计。因此，经常使用威布尔分布应用程序进行AFR的置信区间，以发现它太宽。因此，将上限用作估计以避免估计不足。由于置信区间太大，因此基于它们的业务决策将是错误的。因此，需要更接近AFR的上限，以便所使用的估计将更加可靠。增加样本数量以实现此目标的成本过高。使用大量的历史数据来获得有关现场实际产品故障的先验知识以及测试生成的估计值，可以通过减轻测试数据中的不准确性来改善估计值。为此，该研究提出了一种用于AFR估计的贝叶斯后验分析方法。在此过程中，根据历史数据估算威布尔分布参数的先验分布，然后得出AFR后验估计的统计模型及其置信区间。结果表明，这缩小了后置置信区间。 （9篇）

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号