摘要提出一种改进蒙特卡罗方法（Modified Monte Carlo Method，简称MMCM），对非线性多变量的冲击力溯源系统进行不确定性量化评定。该方法首先建立冲击力溯源系统各不确定变量的表征模型；再根据抽样空间相邻迭代解的误差限，建立样本量的适应度函数；最后利用适应度函数使冲击力溯源模型的输入变量最佳逼近总体分布。为了验证本文方法的有效性，搭建了一套冲击力溯源系统。利用该方法对冲击力溯源系统进行不确定度评定，评定结果表明，在95%的置信水平上，冲击力溯源系统的相对扩展不确定度优于0.171%,其不确定性主要来源于落锤上表面加速度分布不均匀和横向偏摆。
Abstract：A Modified Monte Carlo Method (MMCM) is proposed to evaluate the uncertainty of a traceable measurement system of the impact force with nonlinear multi-component characteristics. First, the uncertainty components are represented by analyzing its error generation principles. Second, The relative error of adjacent solutions are applied to build the fitness function of the sample size. Third, the inputs of the traceable measurement system can approximate the population distribution by dynamic regulating the sample size of the MMCM. The MMCM is verified by applying it to a traceable measurement system of the impact force. The experimental result shows that the relative expansion uncertainty is 0.171% with 95% degree of confidence. It reveals that both the inhomogeneous distribution of acceleration and the lateral oscillations on the hammer are the main uncertainties of this system.
2020, Vol. 41 
