Incorporating feedforward neural network within finite element analysis for L-bending springback prediction

• A generalised neural network has been implemented in the finite element simulation.
• The trained data is applicable to different finite element simulation.
• The method improves the springback prediction as compared to the experimental data.

The use of the latest nonlinear recovery in finite element (FE) analysis for obtaining an accurate springback prediction has become more complicated and requires complex computational programming in order to develop a constitutive model. Thus, the purpose of this paper is to apply an alternative method that is capable of facilitating the modelling of nonlinear recovery with acceptable accuracy. By using the artificial neural network (ANN), the experimental results of monotonic loading, unloading, and reloading can be processed through a back propagation network that is able to detect a pattern and do a direct mapping of elastically-driven change after the plastic forming. FE analysis procedures were carried out for the springback prediction of sheet metal based on an L-bending experiment. The findings of the FE analysis show an improvement in the accuracy of the predictions when compared to the measured data.