製程能力指標Spk的近似不偏估計量

Abstract

製程能力指標是被廣泛地用來評估製程產出產品是否符合規格的重要工具，不僅能提供品質保證，也是提供品質改善方面的一個方針。其中，Boyles在1994年提出一個良率指標取名為Spk，該指標為常態製程提供一個精準的良率衡量，Spk可以藉由製程的良率，精確地衡量製程績效。然而，在估計Spk時，由於Spk的估計量是偏誤的，因此在呈現製程績效時並不可靠。本篇研究的目的是應用曲線模擬工具得出指標Spk的估計量的校正因子來改善估計量的準確性，我們的方法將使我們可以得到Spk的近似不偏估計量，這項研究結果對於現實工廠製程在品質保證的應用上是非常實用的。

Process capability indices (PCIs) have been extensively used to measure whether the process meets the specifications. Not only can they provide a decision on the quality assurance of the process but also they can provide a guide principal of the quality improvement on the process. Among various indices, yield index Spk can provide an exact measure on process yield of normal processes. Unfortunately, the estimator of Spk is biased. Therefore, process yield testing may be unreliable. In this paper, we propose an approximately unbiased estimator of Spk utilizing the curve fitting tool. The research is useful for the quality managers to make reliable decisions in real applications.