The Precise Prediction of Springback Based on GRNN

The deformation of sheet metal is so complicated that the prediction of springback will cost much time with FEM and the results may not match the facts. So it tried to build a function relationship between the springback and the craft parameters with artificial neural network (ANN) in this paper. And for improving the property of prediction it took research on the ANN. It introduced the GA to solve the problem of base function centers distribution. Finally it applied the neural network presented for sheet metal curling. The results showed that the GRNN (genetic algorithm and radical base function neural network) could predict the springback accurately.