Development of Key Technologies in a Case-Based Knowledge System for Fixture Design

Case-Based Reasoning is an important method for the reuse of engineering knowledge. A study has been carried out by the authors on the key technologies that are needed in a Case-Based knowledge system to support engineering design. This paper reports a uniform dynamic representation mechanism based on granularity to represent the engineering knowledge, in the form of cases and rules, for the design of fixtures on machine tools. Based on the Nearest-Neighbour algorithm, an algorithm of case retrieval is brought forth to support the dynamic adaptation of index words and to optimize case retrieval by some design rules. It has been identified that case adaptation is one of the most difficult issues of Case-Based Reasoning for fixture design. Design rules have been used to improve the automation and quality of case adaptation. According to the characteristics of fixtures in the CAD data, several mining patterns can be provided, including case cluster analysis, design rules mining, CAD data criterion mining and user application pattern mining. Then an application platform for mechanical CAD data mining based on cases is built, and a feasible way has come up with to settle the problem of design knowledge acquisition. Based on the developed key technologies, a Case-Based knowledge system for fixture design has been developed and has been tested and applied in an aviation corporation.