以複式模擬法構建製程良率之信賴區間

Abstract

製程能力分析（Process Capability Analysis）的目的是評估製程實際績效的優劣，並藉由量化的製程能力指標（Process Capability Index，PCI）來表現整體製程優劣的狀況，以作為製程改善或買賣雙方議價訂約之依據，目前業界常用之製程能力指標有Cp、Cpk和Cpm等。 然而製程良率（process yield）與製程能力指標之間有密不可分的關係，製程良率除了能提供品管人員判斷製程或產品優劣之資訊外，也是企業對外展現產品良好品質的依據。然而在這些製程能力指標中，只有指標Cp和Boyles（1994）所提出之新指標Spk與製程良率間有一對一的關係，而業界最常使用的指標Cpk與製程良率間只具有不等式的關係，因此要從Cpk指標直接推算製程良率非常困難。 本研究之主要目的是希望能利用Spk指標與製程良率間之一對一關係，找出合理計算製程良率的方法，並且藉由一種重複抽樣的複式模擬法（Bootstrap Simulation）來推導製程良率的信賴區間。最後並將本研究所發展之構建製程良率信賴區間的流程，寫成一套應用程序，並以Visual Basic程式語言撰寫成簡單易用的視窗軟體，提供沒有太多統計背景的工業界人士直接使用。

The process capability analysis is widely used to evaluate product quality in manufacturing industry. Consequently, process capability index (PCI) is one of the deciding criteria measuring for manufacturers to negotiate their contracts. Besides, process capability index can also be utilized as the information for process improving. Cp, Cpk and Cpm are the three commonly PCI used in the manufacturing industry. Process yield has closely relationship with PCI. Process yield not only provides quality control engineer the useful information of process or product to measure if the yield is in the prescribed range, but also exhibits the outstanding product quality to the enterprises. However, only Cp and Spk, defined by Boyles(1994), have one-by-one relationship to process yield. The objective of this study is to use a bootstrap simulation to construct the bootstrap confidence interval for the process yield based on Spk. Finally, a procedure for applying the proposed bootstrap confidence interval and a new software package based on Visual Basic language for this procedure are also developed in this study.