Improving joint quality of hybrid friction stir welded Al/Mg dissimilar alloys by RBFNN-GWO system

Welding of Al/Mg dissimilar alloys faces many challenges. In this study, ultrasonic-stationary shoulder assisted friction stir welding (U-SSFSW) was employed to join the dissimilar alloys of AZ31B Mg and 6061-T6 Al. Radial basis function neural network (RBFNN) was used to model the relationships between the inputs of welding speed, rotating speed and ultrasonic power and the output of ultimate tensile strength (UTS) of U-SSFSW joint. After that, grey wolf optimization (GWO) algorithm was used to explore the maximum UTS and the corresponding optimal process parameters. The maximum UTS reached 158 MPa under the RBFNN-GWO system optimized process parameters. The microstructure and fracture behavior were analyzed to clarify the superiorities of the optimal process parameters and the enhancement mechanism of U-SSFSW technique under the optimized parameters.