Springback prediction in T-section beam bending process using neural networks and finite element method

In this paper, three point bending method is used for the T-section beam bending process. The prediction model of springback is developed using artificial neural network approach. The corresponding loading stroke that can theoretically eliminate the residual deflection of a beam after springback is determined. Application examples indicate that the proposed approach could achieve an allowable straightness error. Numerical simulations using finite element method are also performed to investigate the effect of material properties on springback. A neural network for identification of material parameters is developed by the simulation data. Besides, the residual stress distributions across the beam section are analyzed. The finite element model is validated with experimental results of springback.