Large strain flow curve characterization considering strain rate and thermal effect for 5182-O aluminum alloy

This research aims to characterize the effect of strain rate and temperature on flow behaviours under large plastic deformation for 5182-O aluminum alloy. Experiments are conducted with both dogbone and notched specimens at different strain rates and temperatures. All tests are analyzed by inverse engineering to identify the strain hardening behaviour at large plastic strains up to fracture. The experimental results show the highly coupling effect of strain rate and temperature. The two hardening laws of Swift-Voce model and p-model are calibrated by the finite element model update procedure, and the combination with the smallest error is selected as the input set of the artificial neural network (ANN) model. Then the dynamic hardening behaviours are modelled by the ANN to consider the highly coupled effect. The calibrated ANN model is further applied to ABAQUS/Explicit for numerical simulation under different loading conditions. Taking the finite element calculation time and prediction accuracy into consideration, the ANN model with single hidden layer optimized by particle swarm optimization algorithm is finally selected. The calibration results of the selected ANN model have the best consistency with an acceptable level of numerical computation time.