Optimizing Technical Target Setting in QFD: A Belief Rule-based Methodology

Quality Function Deployment (QFD) is a customer-driven methodology for translating customer requirements (CRs) into product/engineering design requirements (DRs), and subsequently into parts characteristics, process plans and production requirements to achieve customer satisfaction (CS). The use of QFD for product planning has gained global supports and brought significant benefits to manufacturers. A crucial issue for the successful QFD implementation is setting correct technical targets for DRs, which are directly related to the development of quality products with high CS.Existing approaches for setting technical targets for DRs, however, are suffering from various problems. They either set targets for individual DR separately without optimizing them, or require a linear or fuzzy linear assumption about the relationship between CS and DRs just for simplicity, or utilize a specific nonlinear function such as exponential or quadratic function for the relationship. The linear or fuzzy linear assumption has been found not true in many real-life cases. The specification of a nonlinear function to represent the relationship between CS and DRs is difficult, even if not entirely impossible, because there is no objective evidence to support the selection of any non-linear functions in QFD practices.In order to maximize CS without predetermining any subjective or unrealistic assumptions, there is a clear and strong need to develop a novel methodology to learn the relationship between CS and DRs from the benchmarking and technical competitive analysis data in the House of Quality (HoQ). The learned relationship will be more objective and more accurate than any functions, linear or non-linear, specified subjectively, and can be used to set targets for DRs optimally and more realistically. In the QFD literature and practices, there has been no effort to set targets for DRs in this way. This project is therefore proposed to adopt the belief rule-based (BRB) approach to fill the gap.The project is proposed to investigate: (i) how belief rules and belief rule bases can be used to map the casual relationship between CS and DRs; (ii) how a belief rule-base (BRB) model can be built, trained and validated using benchmarking and technical competitive analysis data in the HoQ; (iii) how the trained and validated BRB model can be utilized to support target setting in QFD. The project results will make significant new contributions to the Hong Kong research base and industrial applications in this area.