一些估計非常態製程能力指標的績效評估

Abstract

製程能力指標(PCIs)已廣泛地應用於衡量一製程是否具生產符合規格之產品的能力。指標 包含四個基本的指標Cp，Cpk，Cpm和Cpmk在內，這些指標在製程為常態的情況下表現的都不錯，但當製程為非常態的時候就不太適當。Pearn 和 Chen [6] 提出一般化的指標 和其估計式 ，使其在非常態的情況之下能適切地估計製程的能力。儘管他們提出這樣的一個估計方法，但並沒有任何數據來評估此一估計方法的績效。在本文中，首先將介紹此一方法，其次藉由模擬的技巧來評估其績效，模擬的過程中將包含十一種分配，並搭配不同的參數值，使其研究的成果能廣泛地被應用。

Process capability indices(PCIs) are popular used to determine whether a manufacturing process is capable of producing items within a specific tolerance. , which include the four basic indices Cp, Cpk, Cpm and Cpmk as special cases, are appropriate indices for processes with normal distribution, but have been shown to be inappropriate for processes with non-normal distributions. Pearn and Chen [6] proposed the generalizations and construct a superstructure for the estimators of , named to cover cases where the underlying distributions may not be normal. Although they proposed these estimators, the performance of these estimators didn't be shown. In this article, to be used widely, we first review the percentile method, then investigate it's performance by simulation techniques, including eleven distribution for some parameters values.