Optimization of Short Arc Mode robotic MIG welding through statistical analysis, microstructural characterization, and machine learning

In this study, robotic MIG welding process parameters for IS 2062 E250 steel are optimized using SAM process. An experiment using the Taguchi approach was carried out where the influence of voltage, current, and travel speed on tensile strength was studied. Tensile test and non-destructive tests (visual test, magnetic particle test, and liquid penetrant test) were used to check the quality of welding. The results showed that specimen 8 was observed to have a maximum tensile strength of 498 MPa and Welding voltage was found to have the greatest effect at 64.2%, then welding current, and lastly, travel speed, as per the ANOVA analysis. Optimal welding parameters using S/N ratio analysis were found to be 14.2 V, 130 A, and 90 mm/min. Results from SEM fractographic analysis showed localized particle pull-outs, micro voids, and shallow dimples, which indicated ductile fracture, with excellent metallurgical bonds. XRD analysis showed strong α-Fe (ferrite) peaks at (110) and (200) planes, whereas iron as the major element was confirmed by EDS, with distinct Fe-L, Fe-Kα, and Fe-Kβ peaks. Finally, the Gaussian Process Regression (GPR) model showed the best predictive performance among the evaluated models, with the lowest prediction error, indicating its potential for tensile-strength prediction within the investigated robotic MIG welding conditions.