6533b871fe1ef96bd12d11a8

RESEARCH PRODUCT

Using a neural network for predicting the average grain size in friction stir welding processes

Gianluca BuffaLivan FratiniD. Palmeri

subject

Materials scienceArtificial neural networkFSW metallurgy neural networksMechanical EngineeringMetallurgyMicrostructureGrain sizeFinite element methodComputer Science ApplicationsLap jointModeling and SimulationButt jointFriction stir weldingGeneral Materials ScienceFriction weldingComposite materialSettore ING-IND/16 - Tecnologie E Sistemi Di LavorazioneCivil and Structural Engineering

description

In the paper the microstructural phenomena in terms of average grain size occurring in friction stir welding (FSW) processes are focused. A neural network was linked to a finite element model (FEM) of the process to predict the average grain size values. The utilized net was trained starting from experimental data and numerical results of butt joints and then tested on further butt, lap and T-joints. The obtained results show the capability of the AI technique in conjunction with the FE tool to predict the final microstructure in the FSW joints.

yearjournalcountryeditionlanguage
2009-09-01
10.1016/j.compstruc.2009.04.008http://hdl.handle.net/10447/42174