蒙特卡罗方法在复杂模型测量不确定度评定中的应用

Application of Monte-Carlo method in measurement uncertainty evaluation of complicated model

《测量不确定度表示指南》(简称GUM)提供的评定方法基于不确定度传播律,主要适用于线性模型,在评定复杂模型时,由于灵敏系数、输入量间的相关系数以及有效自由度难以确定,这种方法存在很大的局限性。针对复杂模型的评定问题,《测量不确定度表示指南》补充文件1给出了用蒙特卡罗数值模拟方法通过分布传播评定测量不确定度的指南。本文参照补充文件1的要求,给出了用蒙特卡罗数值模拟方法评定复杂模型测量不确定度的原理和评定步骤,并用实例对上述两种评定方法进行对比。

The method introduced in ＂Guide to the expression of Uncertainty in Measurement＂（GUM） for evaluating measurement uncertainty is based on propagation of uncertainty, which is mainly used for linear model. Since it is difficult to determine the sensitive coefficient, correlation coefficient of input quantities and effective degrees of freedom, the GUM method has some defects for complicated model. Aiming at this problem, the propagation of distribu- tions using Monte-Carlo numerical simulation method is introduced in the GUM Supplement 1-＂Propagation of Distri- butions using a Monte Carlo method＂. According to GUM Supplement 1, the principle and procedures of Monte-Carlo method for evaluating measurement uncertainty of complicated model are outlined and discussed, and a case is shown to compare above two methods.