2001-08-012024-05-18https://scholars.lib.ntu.edu.tw/handle/123456789/703986摘要：在任何生產程序中，不管是自然變異或是特殊變異都會影響製程的良率，並且使得產品品質降低。製程業者以往對於變異的來源的分解，採用的多是所謂的經驗法則，缺乏系統性的分析工作。本研究即針對變異的來源，特別是非均勻度做系統性的研究，期望運用合理分群的方式找出影響非均勻度的自然變異以及特殊變異，並提出改善非均勻度的方法。合理分群的運用，乃是根據製程的特性，藉由統計的方法，將資料作系統性的抽樣與分群，期能將製程發生特殊變異與自然變異分離開來。這樣的變異分離提供了下列的好處（1）增加特殊變異被偵測出的敏感性，（2）針對不同變異來源，便於施予不同的消除對策。 針對以上的動機為出發點，本研究的工作項目可細分為以下幾點來進行： (1)探討非均勻度統計模型建立之文獻 (2)收集製程實驗資料 (3)運用合理分群法分解多重變異與建立多種非均勻度統計模型 (4)分析與圖解不同模型之製程意義 (5)推導非均勻度之多重變異分解數學模式 (6)提出針對不同變異來源的消除對策<br> Abstract: In a manufacturing process, variations due to natural causes and special causes affect and reduce the product quality. Process engineers usually search for the variation causes based on empirical rules. However, the search for the causes is very often empirical and ineffective. In this research, we provide a systematic methodology for controlling process non-uniformity and its variation causes. With the method of rational sub-grouping, we intend to develop methodologies for multiple variation decomposition and propose their reduction procedures. In the use of rational sub-grouping technique, process data are sampled and characterized based on the specific process characteristics. With proper rational sub-grouping, the advantages are twofold: (1) higher sensitivity to detecting special causes (2) better reduction procedures for different types of variation causes. With these goals in mind, we propose to have the following research tasks in this research: (1)Review literature on process non-uniformity variation decomposition and modeling (2)Collect process experimental data (3)Conduct variation decomposition and construct different non-uniformity models using rational sub-grouping technique (4)Analyze and explain these non-uniformity models (5)Derive mathematical relationship among these non-uniformity models (6)Propose different reduction procedures for different variation causes非均勻度變異分解合理分群Non-uniformityMultiple Variation DecompositionRational Sub-grouping