Fuzzy multicriteria methodologies and decision support system for quality function deployment

Today, many companies are facing rapid changes brought about by technological innovations and changing customer demands. These companies realize that the effort to develop new products that customers want and continue to purchase is crucial for their survival. Quality function deployment (QFD) provides a specific approach for ensuring quality through each stage of the product development and production process. Notwithstanding the rapid growth of QFD literature, development of systematic procedures for setting the target engineering design levels has scarcely been addressed. This research proposes an integrated approach to formulating and solving the QFD process by developing and illustrating various fuzzy, multiobjective models to aid a designer in choosing target levels for engineering characteristics in several fuzzy environments. Multiattribute value theory combined with fuzzy regression and optimization theory allow the designer to mathematically consider tradeoffs among the various customer attributes as well as the inherent fuzziness in the system. In addition, the modeling approach presented makes it possible to observe separately as well as conjointly the effects of possibility and flexibility on an overall design. Fuzzy regression, which is used to assess the system relationships in our QFD modeling, is compared with conventional statistical regression with respect to their descriptive and predictive capabilities. Moreover, we formally address the interrelationship among the H value, membership function, and spreads of fuzzy parameters in fuzzy regression, and investigate its sensitivity using a family of summed exponential membership functions. We also build a QFD decision support system prototype that may be used in a practical setting. The software system, a self-contained interactive, user-friendly spreadsheet model, incorporates the novel aspects of the proposed fuzzy modeling approach. The results from preliminary laboratory experiments using practitioners and students imply that our modeling approach, combined with the software system, not only accelerates the design process, but also enhances design quality. The fuzzy multiobjective models developed and illustrated may be applied to a wide variety of design problems where multiple design criteria and system functional relationships are interacting and/or conflicting in an uncertain, qualitative, and fuzzy way