Research on Calculation Methods of Important Parameters in Radioactivity Measurement

Abstract [Background]: Radioactivity measurement is widely used in various fields of nuclear technology application. The measurement uncertainty, confidence interval and detection limit are important parameters of radioactivity measurement. Different calculation methods may get different results, and the calculation results directly affect some important and relevant decisions. [Purpose]: It is important to analyze whether the calculation method is used properly. [Methods]: The important parameters of α particle activity concentration measurement were studied by partial derivative method and Monte Carlo method. In this study, based on the measurement of α activity concentration, the sources of uncertainty for the measurement results were analyzed. The measurement uncertainty, confidence limits, decision threshold and detection limit of α particle activity concentration under different input modes were derived and calculated by partial derivative and MC methods. [Results]: The results show that when the input uncertainty is higher than 10%, the relative deviation between confidence interval and uncertainty results obtained by the two calculation methods is greater than 15%. When the relative uncertainty of the input is small, the detection limit is about 2 times of the decision threshold. [Conclusions]: The partial derivative method is widely used, but it does not consider the probability distribution of the input, and it is not suitable for complex and special input models. Under the circumstances, Monte Carlo method can be used to obtain more reliable calculation results. The two approaches can be applied in complementary ways.

摘要【研究背景】：放射性测量在核技术应用的诸多领域中均有着广泛应用。测量不确定度、置信区间与探测限是放射性测量的重要参数。不同的计算方法可能得到不同的结果，而计算结果将直接影响相关的重要决策。
【研究目的】：分析计算方法的使用是否恰当具有重要意义。
【研究方法】：本研究采用偏导数法（partial derivative method）与蒙特卡洛法（Monte Carlo method）对α粒子活度浓度测量的核心参数展开研究。本研究以α活度浓度测量为基础，对测量结果的不确定度来源进行了分析；并通过偏导数法与蒙特卡洛法，推导并计算了不同输入模式下α粒子活度浓度的测量不确定度、置信限、判定阈值与探测限。
【研究结果】：结果表明，当输入不确定度高于10%时，两种计算方法得到的置信区间与不确定度结果的相对偏差大于15%；当输入相对不确定度较小时，探测限约为判定阈值的2倍。
【研究结论】：偏导数法应用广泛，但未考虑输入量的概率分布，不适用于复杂及特殊输入模型。在此场景下，可采用蒙特卡洛法以获得更可靠的计算结果。两种方法可互为补充，协同应用。