Prediction of metal sheet forming based on a geometrical model approach

The panel production of small batch sizes for the hull of large ships requires a stable and flexible forming process, which is momentarily manually controlled by a system operator. The manual forming press control includes the metal sheet handling above the forming tool for defining the contact point and engagement depth of the sword and subjective monitoring of the forming degree by using the light gap check method. For objectifying the process monitoring and reducing the dependency on the experience of the system operator an automated solution is needed. Within the automated process control the metal sheet deformation behavior has to be predicted in real-time during the forming process. To achieve this, the deformation prognosis for the ship panel’s production is handled inside the described work. Based on a state of art analysis a geometrical approach to describe the metal sheet deformation behavior is developed for the multi-step forming process by three-point-bending. The related geometrical parameters are predicted using a new type of prediction method by means of an artificial neural network. This prediction method requires the network definition and extensive experimental investigations for training the artificial neural network.