Prediction of anisotropic property of activated metal inert gas welding by employing different supervised machine learning models

Activated Metal Inert Gas (A-MIG) welding of EN10028 (SA516Gr70) steel was investigated using varying current, voltage, and gas flow rate to assess their influence on tensile strength (TS). A total of 100 welded samples were prepared and tested per ASTM standards. Material characterization was performed on samples with the highest and lowest TS to evaluate the correlation between microstructure and strength. Machine learning models Linear Regression, Random Forest Regression, and Support Vector Regression (SVR) were applied to predict TS based on welding parameters. • The SVR model achieved the best predictive performance, with an R² of 0.8750 and a model accuracy of 96.73 %. • The results confirm the potential of SVR for accurately forecasting TS in A-MIG welded EN10028, facilitating process optimization in pressure applications.