Classification problems in manufacturing of sheet metal parts

Classification involves feature analysis and the design of part hierarchies. The aim of classification is the support of CAD/CAM, for example retrieval for similar parts or the coarse planning of manufacturing. A short insight into conventional classification systems and feature recognition is given. Because of several shortcomings of these systems requirements for a modem classification system are provided. Several classification methods are presented. They work with a special algorithm, Fourier descriptors, neural networks and combinations. A semantic network is used for analysing features and serves as a knowledge base and integrating tool for all methods. A good example for application is the retrieval of similar parts, which avoids redundant work. Based on the described classification methods such a system can be simply realized. (C) 1997 Elsevier Science B.V.