基于蒙特卡洛方法平尺测量直线度不确定度评估方法的研究  

Research on Uncertainty Evaluation Methods of Straightness of Straight Edge Based on Monte Carlo Method

作　　者：李元峰 孟令川[1] 黄垚 杨皓天 LI Yuan-feng;MENG Ling-chuan;HUANG Yao;YANG Hao-tian(Beijing Jiaotong University,Beijing 100044,China;National Institute of Metrology,Beijing 100029,China;University College London,London,WC1E 6BT,UK)

机构地区：[1]北京交通大学,北京100044 [2]中国计量科学研究院,北京100029 [3]伦敦大学学院,伦敦WC1E 6BT,英国

出　　处：《计量学报》2023年第4期540-548,共9页Acta Metrologica Sinica

基　　金：中央高校基本科研业务费专项资金资助(2020JBM075)。

摘　　要：采用蒙特卡洛方法(MCM)对平尺最小二乘直线度和最小条件直线度进行测量不确定度评估。通过与测量不确定度评定指南法(GUM)的评估结果进行比较发现,MCM评估出的最小二乘直线度和最小条件直线度的测量不确定度分别比GUM评估结果小0.028μm和0.026μm。在给定的0.05μm允差范围内,两种评估方法对直线度测量不确定度的评估均有效。统计检验采用了kolmogorov-smirnov检验法、jarque-bera检验法、normal probability plot图示法、偏度和峰度检验法。通过对两种不同定义直线度的测量模型进行统计检验分析发现,被测量分布函数与正态分布的峰度偏离是造成差异的主要原因。The straightness uncertainty of straight edge in least-square rule and in minimum-zone rule have been evaluated by Monte Carlo method(MCM).Comparing with the straightness uncertainty evaluated by guide to the expression of uncertainty in measurement(GUM),it s been found that the straightness uncertainty in least-square rule evaluated by MCM is 0.028μm less than that by GUM and the straightness uncertainty in minimum-zone rule evaluated by MCM is 0.026μm less than that by GUM.It has been confirmed that both the methods are valid for evaluating the straightness uncertainty of straight edge at a certain numerical tolerance of 0.05μm.Kolmogorov-smirnov test,jarque-bera test,normal probability plot,skewness and kurtosis test were employed as the statistic testing methods.By employing specific statistic testing method on measurand,it s been found that the kurtosis deviation of measurand distribution from normal distribution is responsible for the uncertainty difference.

关 键 词：计量学 平尺直线度 不确定度评估 蒙特卡洛方法 最小二乘法 最小条件法 峰度偏离 

分 类 号：TB92[一般工业技术—计量学]