Prediction of mechanical properties of as-cast and heat-treated automotive Al alloys using artificial neural networks

The desired mechanical properties of cast automotive aluminium alloys depend on the alloy chemistry, casting parameters, mould design, melt treatment, and heat treatment conditions. Despite extensive work in the literature, the large number of these controlling parameters have made it difficult to predict the mechanical properties and to model them using conventional techniques. In the present study, Artificial Neural Networks (ANN) was used to predict the mechanical properties. A database of mechanical properties (UTS, YS and EI%) as a function of chemical composition, heat treatment (solutionizing, quenching and ageing) and casting variables (mould type and melt treatment) was collected from published literature. Several standard multilayer AMN models were then trained using data randomly selected from the database. The outputs of the ANN models were subsequently compared with the remaining data The results indicate that ANN is a suitable modelling technique for predicting mechanical properties and optimising the heat treatment process.