標題: 應用雙反應曲面分析法於動態系統多重品質特性最佳化之研究
Optimizing Multi-response Problems in a Dynamic System by Using Dual Response Surface Analysis
作者: 陳姣燕
Chen, Jiao-Yan
唐麗英
Tone, Lee-Ing
工業工程與管理學系
關鍵字: 田口方法;動態系統;多重品質特性;迴歸分析;雙反應曲面分析法;望想函數;Taguchi Method;Dynamic System;Multi-response Problem;Regression Analysis;Dual Response Surface Analysis;Desirability Function
公開日期: 1997
摘要: 由於田口方法的實驗次數較傳統實驗設計的次數來得少,且分析程序簡單,因此此法廣為業界所使用。然而田口方法主要是用於單一品質特性最佳化的問題,在解決多重品質同時最佳化的問題時,一般的作法均是先就個別品質分別找出其最佳因子水準組合,然後再依工程經驗由其中歸納出一最佳的因子水準設定。然而隨著產品的設計日益複雜,產品品質的良莠已非單一品質的最佳化能滿足,往往需要同時考量到多個品質特性的同時最佳化;此外,為了使產品的設計與使用更具彈性,引入信號因子的動態系統已成為近年來的研究重點,但是有關研究動態多品質問題的相關文獻卻非常罕見,因此本研究的主要目的即是發展一套能有效解決動態系統多重品質特性同時最佳化問題的演算程序。 本研究是以望想函數與雙反應曲面分析法來解決動態系統多重品質特性同時最佳化的問題。首先,利用望想函數來綜合衡量多個品質特性的靈敏度與品質變異,然後再利用雙反應曲面分析法找出可能的最佳因子水準組合,最後同時考慮系統靈敏度與品質變異以及整體的總望想值,決定出最佳的因子水準組合。本研究以一個製備S-CHBE光學活性中間體的實驗為例來說明如何應用本研究所發展的動態系統多重品質特性同時最佳化之演算程序。本研究最後並與傳統的解決動態多重品質最佳化的作法及現有之其他解決動態多品質問題的方法做比較,以說明本研究之優越性。
Tanguchi method has been widely implemented in industry for product/puocess improvement. Most of the Taguchi's applications are focused on the optimization of a single-response in a state system. Optimization of a dynamic system has received more attentions in recent years. Due to the increasing complexity of the product design, more than one quality characteristic must be considered simultaneously to improve the product's quality. Hence, optimizing multiple responses simultaneously in a dynamic system becomes an importantissue. This study proposes a procedure which utilizing the desirability function and dual response surface analysis to optimize the multi-response problems in a dynamic system. First ,the desirability function is applied to measure the overall measurements of sensitivity and quality variation. Next, the dual response surface analysis is used to obtain a set of possible optimm factor level combiantions. The optimun factor level setting is obtained according to the maximum overall desirability value. Finally, the propsed procedure is illustrated by a biological reduction of ethly acetoacetate process experiment. A comparison between the propsed method, the conventional and other ways of analyzing multi-response problems in a dynamic system are made to verified the effectiveness of the proposed method.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT863031032
http://hdl.handle.net/11536/63333
Appears in Collections:Thesis