COLD-FORGING PROCESS DESIGN BASED ON THE INDUCTION OF ANALYTICAL KNOWLEDGE

In this paper, a cold forging process design method based on the induc tion of analytical knowledge is proposed. The analysis engine, which i s a finite-element-based program, is used to analyze various multi-sta ge cold forging processes based on pre-defined process condition param eters and tooling geometry. According to the simulation results, a kno wledge-acquisition procedure is instituted. i.e. a neural network mode l is constructed, in which the multi-layer network and the back-propag ation algorithm are utilized to learn the training examples from the s imulation results. In the last part of this paper, an industrial case study for the multi-stage cold forging process design of a low-carbon steel speaker tip is discussed. The optimal process condition paramete rs, such as the preform punch geometry and the preform punch stroke ar e decided, based on the requirement of homogeneous plastic deformation of the cold-forged product. The proposed method is useful for the sho p floor to decide the cold forging process parameters for producing a sound product within the required minimum quantity of the die set. (C) 1997 Elsevier Science S.A.