Mechanical and microstructural properties prediction by artificial neural networks in FSW processes of dual phase titanium alloys

Friction Stir Welding (FSW), as a solid state welding process, seems to be one of the most promising techniques for joining titanium alloys avoiding a large number of difficulties arising from the use of traditional fusion welding processes. In order to pursue cost savings and a time efficient design, the development of numerical simulations of the process can represent a valid choice for engineers. In the paper an artificial neural network was properly trained and linked to an existing 3D FEM model for the FSW of Ti–6Al–4V titanium alloy, with the aim to predict both the microhardness values and the microstructure of the welded butt joints at the varying of the main process parameters. A good agreement was found between experimental values and calculated results.