Use of Data Mining to Support the Development of a Knowledge Intensive CAD System

摘要

In order to compete in the global manufacturing market, agility is the only possible solution to responsernto the fragmented market segments and frequent changed customer requirements. However, manufacturing agilityrncan only be attained through the deployment of knowledge. To embed knowledge into a CAD system to form arnknowledge intensive CAD (KIC) system is one of the possible ways to enhance the design compatibility of arnmanufacturing company. The main obstacle to develop a KIC system is the capitalization of a huge amount ofrnlegacy data to form a knowledge database. In the past, such a capitalization process could only be done solelyrnmanually or semi-automatic. In this paper, a six steps model for automatic design knowledge capitalizationrnthrough the use of data mining is proposed whilst details of how to select and benchmark an appropriate datarnmining algorithm for a specific design task will be discussed. A case study concerning the design of a plasticrntoaster casing was used as an illustration for the proposed methodology. It was found that the average absoluternerror of the predictions form the best algorithm was within 17%.