测量不确定度评定中的若干问题

Some questions on evaluation of measuring uncertainty

针对误差理论及误差分析与不确定度评定的关系和异同,以及各种分布的特性,指出均匀分布是由单一因素引起的,由多因素引起的分布不是均匀分布。为此对文献[2]一些均匀分布事例提出质疑。不确定度的评定一定要安全保险且经济实用。不能随意用均匀分布代替其他分布,一般情况下扩展不确定度的概率应取99.73%。某次测量结果的不确定度与某种测量方法的不确定度是不同的,前者不能代替后者,而后者却能代替前者。

Similarity and distinction between analysis of errors and evaluation of measuring uncertainty was introduced, including characteristics of various error distributions. Evenly distribution was brought of single factor. Distribution composed of multifactor was not evenly distribution. Some instances of evenly distribution in literature [2] was interrogated. Evaluation of measuring uncertainty should be secure and practical. We shouldn＇t substitute other distribution by evenly distribution. In general probability of extending measuring uncertainty should be 99.73%. The measuring uncertainty of a certain measuring result was different with measuring uncertaint of a kind of measuring method. The latter couldn＇t be substituted by the former, but the former could be substituted by the latter.