基于评分法的测量系统能力评价

A Research on Measurement System Capability Evaluation Based on Scoring Rule

目前所用的测量系统能力评价方法是建立在正态性假设的基础上,如果测量系统误差不满足正态性,运用原来的评价指标在对测量系统能力进行评价的过程中会发生误判。将连续等级概率评分(CRPS)应用到测量系统分析评价当中,通过研究CRPS的相关特性,建立了基于CRPS评分法的测量系统能力评价指标,从而使测量系统误差非正态分布时,对测量系统能力进行评价的过程中不会发生误判。通过案例分析表明,当测量系统误差为正态分布时,运用CRPS和运用传统评价指标给出的评价结果是一致的,同时,CRPS评价方法能够有效实现非正态测量系统误差情形下的测量系统能力评价。

The existing measurement system capability evaluation methods are established on the basis of the normality assumption. However, when the measurement system error distributes non-normally, they are not appropriate. A CRPS based method is proposed. To illustrate it, examples are given. They show that when the measurement system error distributes normally, the evaluation results given by CRPS and the traditional evaluation indicators are consistent. Meanwhile, CRPS can effectively implement the measurement system capability evaluation of non-normal measurement system error.