Categorization of CAD models based on thin part identification

Nowadays, industrial CAD models databases are becoming larger and larger, thus inaugurating new challenges for their management during the product design process. Unfortunately, neither the PDM nor PLM systems are good at indexing and sorting CAD models according to their content, and it is sometime more efficient to redesign a part than to try to find it in a database. Categorizing parts is a first step to solve this problem. This may allow both the re-use of the parts and the retrieval of the methods and processes adopted on them, thus avoiding wasting time restarting from scratch. In this paper, we focus on the characterization and classification of industrial parts with respect to the meshing issue, and notably the meshing of thin parts difficulty handled automatically and which often requires adaptation steps. The concepts of thin object and thin part are introduced together with the mechanisms and criteria used to identify such shape characteristics on CAD models. The final categorization results of a set of tests exploiting a normalized distance distribution associated to specific ratios computed from the bounding box surrounding the object to be classified. The proposed approach has been implemented and validated.