| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系碩博士班 === 96 === Quality Function Deployment (QFD) is a well known method for enterprises to analyze the relationship between customer requirements (CRs) and design requirements (DRs), and there has been much research into it. In the processes of constructing the house of quality (HOQ) of QFD, there are several values and quality matrices that must be completed by experts, including the importance of customer requirements, relationship matrix and correlation matrix. In most of the research, the values and matrices are assumed to be known by authors or they only show the result of them. In practice, a group of experts or a team will fill in the values and matrices with crisp or fuzzy numbers, such as utility or linguistic variables; therefore, the process of filling in the values and matrices should be considered as a group decision-making (GDM) problem, and there may be some similarity between different experts＇opinions. In the first section of this study, we will introduce a new clustering algorithm, which we modify the well known clustering algorithm, in which we modify the well known clustering algorithm, Fuzzy C-Means (FCM), in order to cluster experts’crisp or fuzzy opinions and make those in the same cluster have higher similarity. After this, we will aggregate the clusters into a consensus for further use. Second, we will use the new method to cluster and aggregate the experts’ opinions on the CR importance, relationship matrix and correlation matrix in order to construct the HOQ of QFD. Finally, we will use the relationship normalizing method and fuzzy number ranking method to find out the ranking of the importance of DRs.
|
|---|
