Modelling of rheological behaviour in multistep hot deformation of aluminium alloys by ANNs

The use of artificial neural networks in modelling the rheological behaviour of AA 6082 aluminium alloy under multistep hot deformation conditions has been studied. Feed-forward back-propagation neural networks were trained and tested using experimental data obtained by multistage hot torsion tests carried out under different procedures and conditions. The comparison between experimental and predicted envelope curves has proven that the artificial neural network-based models can be effectively used to predict flow stress under multistep deformation conditions and to capture the effects of dynamic and static process parameters.