Optimum Tool Geometry and Process Parameters Prescribed by a Neural Network Model in the Case of Cylindrical Parts Deep-drawing

Geometrical inaccuracy of sheet metal parts due to springback are the reason for considerable efforts in the tool and process development. Numerous studies have been carried out in order to find the optimum process parameters and tool geometry so that the resulted parts to be within tolerances. In the present paper, the finite element method coupled to the neural network method are used to get the best relation between process parameters and tool geometry in order to minimize the shape deviations of the formed parts, related to the target geometry.