標題: 應用多屬性決策法於動態系統多重品質特性最佳化之研究
Optimization of Multi-Response Problems in a Dynamic System by Using the Multiple Attribute Decision Making Method
作者: 詹雅嵐
Ya-Lan Chan
唐麗英
Lee-Ing Tong
工業工程與管理學系
關鍵字: 田口方法;穩健設計;動態系統;多重品質特性;品質損失;多屬性決策法;理想解類似度順序偏好法;Taguchi method;robust design;dynamic system;multi-response problem;Multiple Attribute Decision Making;Technique for Order Preference by Similarity to Ideal Solution
公開日期: 1998
摘要: 近年來田口方法(Taguchi Method)在改善品質工程方面頗具價值,甚受國內外產業界與學術界人士的肯定。然而由於工業界的生產流程日益複雜,因此一個產品品質的好壞常須由多個品質特性才能決定,而田口方法僅能適用於單一品質特性最佳化的製程,因此必須發展一套能使產品達到多項品質特性同時最佳化的方法。此外,同一系列的產品往往具有多種規格,為了避免浪費成本重複的開發同一系列但不同規格的產品,動態系統最佳化的方法也已成為目前研究發展的趨勢。 目前雖然關於多品質及動態品質特性最佳化的相關研究相當罕見,而少數幾篇這方面的研究所建議之方法又都牽涉到相當複雜的統計模式,以致失去了實用性。因此本研究發展了一套簡單有效的動態系統多重品質特性最佳化之演算流程。本演算流程是根據田口損失函數的精神,計算出動態系統的品質損失,再以多屬性決策法(Multiple Attribute Decision Making,MADM)中的理想解類似度順序偏好法(Technique for Order Preference by Similarity to Ideal Solution,TOPSIS)來整合多重品質特性以達到製程最佳化之目的。本研究最後以一個工研院製備S-CHBE光學活性中間體之化學實驗為例,說明如何應用本研究所發展出的方法來實作動態系統多重品質最佳化的問題及驗證本研究方法之有效性。
Taguchi Method has gained wide acceptance in industries worldwide to enhance product quality. The increasing complexity of product design requires that one simultaneously consider more than one quality characteristic to improve product quality. However, the Taguchi Method can only consider how to optimize a quality characteristic at one time. In addition, similarly manufactured items normally have many specifications. To avoid repetition in developing similar products that differ only in size, how to optimize a dynamic system becomes a crucial issue. Relatively few techniques have been developed for simultaneous multi-response optimization in a dynamic system. However, most of these techniques have complicated statistical models, making it difficult for engineers to use in practice. Therefore, this study presents a simple but effective algorithm based on Taguchi's loss function to simultaneously optimize the multi-response problemin a dynamic system. First, the quality loss of a dynamic system is computed. The technique for order preference by similarity to ideal solution method in multiple attribute decision making is then used to obtain a total quality measurement index. The index is used as an effective means of determining the optimal factor-level combination. Finally, an actual case involving the analysis of a biological reduction of ethyl acetoacetate process experiment is presented, demonstrating the effectiveness of the proposed procedure.
URI: http://140.113.39.130/cdrfb3/record/nctu/#NT870031020
http://hdl.handle.net/11536/63801
顯示於類別:畢業論文