论文部分内容阅读
可靠性参数是核电厂概率安全分析评价(PSA)的基础,参数经验贝叶斯方法(PEB)在处理少量失效数据样本时会低估待估可靠性参数的不确定性;Kass-Steffey修正方法采用泰勒展开对参数的后验方差进行修正可以解决参数低估问题。研究Kass-Steffey修正原理并推导出一阶修正公式,计算带Kass-Steffey修正的多个核电厂始发事件频率的参数后验估计方差及90%的置信区间值。计算结果表明,对于失效数据次数多的样本,Kass-Steffey修正对后验方差及估计区间影响较小;对于失效数据稀少的样本,Kass-Steffey修正值得关注,修正后的后验方差变化16%~99%,置信区间值变化4%~53%。
The reliability parameter is the basis of the probability and safety analysis and evaluation (PSA) of the NPP. The parameter empirical Bayesian method (PEB) underestimates the uncertainty of the reliability parameter to be estimated when dealing with a small number of failure data samples. The Kass-Steffey correction method uses Taylor expansion of the posterior variances of the parameters can be corrected to solve the problem of underestimation of parameters. The Kass-Steffey correction principle is studied and a first-order correction formula is derived to calculate the posteriori estimation variance and the 90% confidence interval for the frequency of events with multiple KPS-Steffey modifications. The results show that the Kass-Steffey correction has little effect on the posterior variance and the estimation interval for the samples with more invalid data. For the samples with few invalid data, the Kass-Steffey correction is of concern. The modified posterior variance is 16% ~ 99%, confidence interval value change 4% ~ 53%.