Machine learning driven prediction of mechanical and electromagnetic properties in dissimilar aluminum friction stir welded joints

This study examines the mechanical and electromagnetic behaviour of dissimilar aluminium alloys, AA5083-O and AA7075-T651, joined using friction stir welding (FSW) under varying tool rotational speeds (800, 1000, 1100, 1200, and 1400 rpm) at a constant traverse speed. Mechanical characterization through tensile and microhardness testing revealed that the joint produced at 1100 rpm exhibited the highest tensile strength of 345 MPa. The presence of intrinsic welding defects was found to significantly influence both mechanical integrity and electromagnetic performance. Electromagnetic properties, including permittivity, permeability, and electrical conductivity, were evaluated across the frequency range of 8.2–12.4 GHz, showing consistent trends with variations in welding parameters. The weld joint fabricated at 1100 rpm exhibited superior electromagnetic performance, with shielding effectiveness values below − 12 dB across the X-band (8.2–12.4 GHz), corresponding to approximately 90% attenuation of incident electromagnetic waves. To model and predict these responses, multiple machine learning(ML) algorithms were employed. Among them, the Gaussian Process Regression (GPR) model demonstrated superior predictive capability, achieving an R² value of 99.85%.