Lehrstuhl für Fertigungstechnologie, Universität Erlangen-Nürnberg

Knowledge Based Analysis and Classification of Sheet Metal Components



Datum: 28.04.1995


Author


Reporter

  • Prof. Dr.-Ing. Dr. h.c. M . Geiger
  • Prof. Dr.-Ing. H. Bley, Saarbrücken

The aim of the thesis has been to build up a concept for a modern classification system, to show the application and potentials of such developments. Conventional methods such as form ordering systems have been considered. With modern, knowledge based methods a prototype for a classification system was realized, particularly for sheet metal parts. The core of the concept is a universal part description method with form features and semantic features. The automatic feature analysis makes a powerful part recognition possible. Both the coarse shape, often a rectangle, and local form features, e.g. slots, have been considered. By using special algorithms, Neural Networks and Fuzzy Logic the deviation of a defined shape can be recognized and quantified. As an integrating tool for knowledge representation and processing a special type of Semantic Network which has advantages as for flexibility and universality, has been developed. Applications are possible with the developed basic knowledge. A universal system for the retrieval of similar parts has been created. The support for computer aided process planning and calculation was also shown.