A CASE-BASED REASONING SYSTEM FOR BEARING DESIGN A Thesis

A CASE-BASED REASONING SYSTEM FOR BEARING DESIGN Xiaoli Qin Advisor: Dr. William C. Regli Using Case-Based Reasoning (CBR) techniques in engineering design provides a promising methodology for solving design problems in this complex domain. CBR is based on the idea that past problem-solving experiences can be reused and learned from in solving new problems. We have developed a CBR design system to assist human designers in solving bearing design problems in the mechanical engineering domain. It retrieves previously designed cases from a case-base, using certain adaptation techniques to repair them and their solutions to satisfy the current problem requirements. Our approach combines parametric adaptations and constraint satisfaction adaptations. The technique of parametric adaptation considers not only parameter substitution, but also the interrelationships between the problem definition and its solution. The technique of constraint satisfaction adaptation provides a method to globally check the design requirements to assess case adaptability. Currently, our system can be used and tested in the Rolling Bearing design domain.