品質管理活動源自於面對顧客需求所做出之對應，只有充分掌握顧客
需求，並且主動適時將之轉換成公司之經營目標者，才能永續生存。從產
品生命週期中品質鏈的觀點來看，研發階段的設計品質對產品的穩健性具
有關鍵性的影響，而品質設計是依據顧客之要求品質，加以推理﹑翻譯﹑
轉換而成為代用之工程特性的活動行為﹔因此，欲達成品質之競爭優勢，
必須同時考慮符合設計規格（內部顧客觀點）與滿足需求（外部顧客觀點）
。而藉由品質機能展開（Quality Function Deployment , QFD） 有助於
改進需求認知，增加研發各階段人員之互動與溝通，田口方法（Taguchi
Methods,TM）品質工程（系統設計﹑參數設計）則可達成需求規格之最適
化。
本研究乃從產品生命週期中品質鏈的觀點，提出以加強型 QFD 與TM
中品質工程整合之穩健品質設計策略（RQDS）模式，首先在品質屋的結構
下，利用模糊分析層級程序法（FAHP）分析需求品質顧客聲音的權重
（Wvoci ），其次以模糊函數觀念推導一評估關鍵品質特性（Ivoej）的
模糊品質屋（FHOQ）演算法，以推論決定關鍵工程品質特性之整體計算公
式，然後利用品質表分析設計參數及針對關鍵品質特性進行田口實驗。最
後以一新產品開發個案，進行產品與製程設計之實證研究，亦針多重品質
特性之參數實驗與整合模式作業之作業之成本效益加以分析評估。並說明
本研究之結論建議與未來研究方向。
研究發現在決定關鍵品質特性時，除了需考慮關係矩陣Rij外，亦有必
要同時考慮各VOE間之相關矩陣（Cjk）。傳統的方式係將使用者心中模糊
的意念以一明確值做為輸出，若以Wasserman 方法與本研究FHOQ方法所得
的技術重要度排序值比較，FHOQ任兩技術特性間具有較小之差異性，表示
本研究結果較具穩健性（Robustness）。而在結果的表達上，傳統方法是
採用計算所得的技術特性重要度分數作為排序的標準，本研究則是採用以
排序群間的最大值及最小值為基準，以所得的Ut值作為排序值，可充分反
應各排序元素的差異性。而整合QFD與TM之穩健品質設計模式對新產品開
發中關鍵因素與目標的研定﹑研發團隊間之互動溝通會產生實質效果，亦
可達到品質﹑成本﹑縮短開發時程之效益。

Quality management is a responding activity when a company
faces its customers'' requirements. From a product life cycle
point of view, the design quality in a product research stage
has a key inference on the product final quality robustness.
However, a design quality is determined according to customer''s
product quality requirements. These requirements should be
inferred, interpreted and translated into some engineer
requirements to achieve quality superiority. Usually, Quality
Function Development (QFD) method is used to improve the requirement
understanding, and the researcher communication in each research
stage.
However, customer (or technical) requirements are translated
to some pre-defined values (the Wasserman method) that can not
reflect the characteristic of linguistic variables. To solve such
problem, this research proposes a RQDS model that combines both
enhanced QFD method and the Taguchi Method to quantify the
customer''s linguistic expression by assigning each linguistic
requirement a fuzzy number. Firstly, we use a fuzzy analytic
hierarchical process (FAHP) to decide the weight of customer
requirements (WVOC) in HOQ. Secondly, some fuzzy contribution
functions are applied to evaluate the weight of technical
requirements (IVOE) in a fuzzy house of quality (FHOQ).
Thirdly, we use a quality table to analyze the relative design
parameters, and to experiment with Taguchi method of the weight
of technical requirements. Finally, we select a new product
manufacturing case to examine the efficiency of the proposed
algorithm.
In traditional QFD, the contributions of product technical
improvement to customer requirements are represented by fixed
values. However, these contributions are usually expressed by
linguistic variables, which are nature of ambiguity and
multiplicity of meaning. To solve this problem, this research
uses a set of fuzzy functions to represent these contributions.
The benefit of this is that it suits the practical application.
In addition, we find that determining the weight of technical
requirements not only the relationship matrix Cij between
customer requirement and design requirement should be considered,
but also the co-relationship matrix (Rjk) among VOEs must take
into account. Comparing the weight of technical requirement (IVOE)
obtained using the Wasserman method with the proposed FHOQ
method, we find that the proposed method has smaller diversity
of IVOE than that of Wasserman method.