Statistical error estimation of the Feynman-α method using the bootstrap method
Statistical error estimation of the Feynman-α method using the bootstrap method
复制标题
使用 bootstrap 方法估计 Feynman-α 方法的统计误差
DOI：
10.1080/00223131.2015.1113898
复制
发表时间：
2016
期刊：
影响因子：
--
通讯作者：
C.H. Pyeon
中科院分区：
文献类型：
--
作者：
T. Endo;A. Yamamoto;T. Yagi;C.H. Pyeon
研究了基于反应堆噪声分析的次临界测量技术之一Feynman-α方法的统计误差估计的Bootstrap方法的适用性。在Feynman-α方法中，统计误差可以简单地从反应堆噪声的多次测量中估计出来，但是需要额外的测量时间来重复多次测量。利用一种称为“自助法”（bootstrap method）的重建技术，仅利用一次反应堆噪声测量，就可以估计Feynman-α方法测量结果的标准差和置信区间作为统计误差。为了验证我们提出的技术，我们在京都大学临界性大会上进行了无任何外部源的反应堆噪声被动测量，即仅使用核燃料中自发裂变和（α，n）反应的固有中子源。通过实际测量，验证了Bootstrap方法对Feynman-α方法测量结果统计误差的近似估计是可行的。
Applicability of the bootstrap method is investigated to estimate the statistical error of the Feynman-α method, which is one of the subcritical measurement techniques on the basis of reactor noise analysis. In the Feynman-α method, the statistical error can be simply estimated from multiple measurements of reactor noise, however it requires additional measurement time to repeat the multiple times of measurements. Using a resampling technique called “bootstrap method,” standard deviation and confidence interval of measurement results obtained by the Feynman-α method can be estimated as the statistical error, using only a single measurement of reactor noise. In order to validate our proposed technique, we carried out a passive measurement of reactor noise without any external source, i.e. with only inherent neutron source by spontaneous fission and (α,n) reactions in nuclear fuels at the Kyoto University Criticality Assembly. Through the actual measurement, it is confirmed that the bootstrap method is applicable to approximately estimate the statistical error of measurement results obtained by the Feynman-α method.