Using the artificial neural network to investigate the effect of parameters in square cup deep drawing of aluminum-steel laminated sheets

In this study, the effective parameters involved in the deep drawing of double-layer metal sheets in a die of square cross-section were investigated through artificial neural network (ANN) modeling. For this purpose, first, the deep drawing of double-layer (Al1200 / ST14) sheets was carried out experimentally. Also, the finite element simulation of the process was performed, and the results validated through experimental tests. A set of 46 different experimental data were employed in this paper. The ANN was trained by using a mean square error of 10-4. The input parameters, i.e., punch radius, die radius, blank holder force, clearance, and the permutation layers were set to the network. The surface response method (RSM); was employed to evaluate the results of the ANN model, and the input parameters of the deep drawing process on the thinning of Al1200 and ST14 composite layers were analyzed. The obtained results indicate that the punch edge radius has the most significant influence on the thinning of the Al1200 layer. Increasing the gap between the punch and die to 1/4 of the sheet thickness, increased the cup wall layers thickness of the Al1200 and ST14 respectively by 3.38% and 0.5%. The performance of the ANN model demonstrates that it can estimate the amount of thinning in the composite layers with satisfactory accuracy.