Experimental investigation and optimization of quality characteristics during friction stir welding of Al- and Zn-based magnesium alloy using artificial intelligence

Friction stir welding is successfully used to weld different wrought magnesium alloys. This work investigated the mechanical and microstructural behavior of the friction stir-welded AZ31B magnesium alloy. The experiments were conducted as per experimental runs designed by response surface methodology. An artificial neural network model was developed to produce a relationship between process variables (tool rotational speed, welding speed, and tool shoulder diameter) and characteristics of the friction stir-welded joints (tensile strength, percentage elongation, impact strength, microhardness, and grain size). The acceptable range of statistical parameters validated the adequacy of the model. The multi-objective optimization technique, genetic algorithm was used to obtain a set of Pareto optimal solutions. The best-compromised optimum solution for maximum tensile strength (164.2 MPa), percentage elongation (8%) impact strength (3.5 J), microhardness (85 Hv), and minimum grain size (13.1 μm) was validated by confirmation test with <3 percent absolute error percentage. The fractographical analysis has been performed and dimples and torn edges observed in fracture zones.