Development of statistical methodology in robust parameter design and verification of its effectiveness

在这篇文章中，我们提出了一种统计方法，减少测量系统的变化。我们考虑具有通过原点的正值线性信号响应关系的系统。将不可观测噪声变量分为三类：噪声因子、残留噪声因子和随机误差。然后，我们提出了一个广义的信号噪声比（SN）在测量系统的分类。此外，我们讨论了两种不同的数据分析方法-响应函数建模和性能测量建模-建模和优化。两种方法的有效性进行了研究的Monte Carlo模拟研究。

In this article, we propose a statistical method which reduces the variation in measurement systems. We consider systems with positive-valued linear signal-response relationship passing through the origin. The unobservable noise variables are classified into three categories : noise factors, remained noise factors, and random errors. Then we propose a generalized signal-to-noise (SN) ratio in measurement systems with respect to the classification. Furthermore, we discuss the two different data analysis methods -response function modeling and performance measure modeling- for modeling and optimization. The effectiveness of two methods is investigated by a Monte Carlo simulation study.