Numerical Simulation and Neural Network Prediction the Cold Bending Spring back for Ship Hull Plate

The accurate prediction of the spring back has great significance to the cold bending of plates. Based on the analysis of the square non pressure head CNC bending machine forming principle, the finite element model of the cold forming of the hull plate surface was established using the ANSYS/LS-DYNA finite element software. And the spring back computing research was done on the thickness of 8 mm to 16 mm. The influence rule of the thickness to spring back was analyzed. And the numerical simulation and experimental results of spring back comparison verified the reliability of finite element simulation. Then the prediction model of the plate thickness and the spring back was established using neural network which is based on nonlinear dynamic system and the test sample spring back was predicted. The results of simulation show that the BP neural network can predict the spring back transformation trend very well by comparison with the results of numerical simulation and provides a reliable basis for spring back control. A new idea was proposed for the ship hull plate CNC forming by the application of neural network.