Microstructure prediction using finite element simulation and artificial neural network for extrusion of AA6XXX aluminum alloy

Grain size evolution in AA6XXX extruded profiles is a critical factor influencing their mechanical, thermal and surface properties. Traditional techniques for microstructure control are often resource-intensive and time-consuming. This study presents an innovative approach that combines Finite Element Method (FEM) simulation with experimental microstructure measurements to train an Artificial Neural Network (ANN) for grain size prediction. A hollow AA6060 profile was extruded and experimentally characterized to determine its grain size distribution. Experimental data were compared with the results of FEM simulations conducted using QForm Extrusion UK software. Based on this comparison, an ANN was trained using the FEM outputs as input data to predict the final microstructure of the extruded profile. The proposed methodology demonstrated good accuracy in predicting the microstructure through the use of FEM simulations and machine learning techniques. This approach provides a faster, more sustainable and more cost-effective alternative to conventional methods, representing a significant advancement in optimizing the extrusion process for AA6XXX alloys.