Two-staged technique for determining ultimate tensile strength in MIG welding of mild steel

Abstract Optimization of welding parameters is highly significant in welding process and intelligent prediction of process parameters leverages data availability towards reducing cost of experimental procedures. In this study, a two-staged technique which integrates Taguchi method and adaptive neurofuzzy inference system (ANFIS) models was proposed to optimize and predict weld tensile strength of AISI1008 Mild steel plates of 3 mm thickness mild steel plates similar butt welds produced through metal inert gas (MIG) welding process. Three process parameters, namely; welding voltage, welding current, and gas flow rate are used as input parameters of the model whereas the tensile strength of the welded mild steel plate is considered as the output parameter. The maximum ultimate tensile strength of the welded joint was found at 99 MPa. The analysis of variance results also shows that welding voltage contributes 57.3%, more than welding current which contributes 20% and gas flow rate contributes 10% in affecting the strength of the weld. The ANFIS model also shows a root mean square error (RMSE) of 0.16, a mean absolute deviation (MAD) of 0.1125 and a variance accounted for (VAF) of 99.99. This further emphasis the effectiveness of ANFIS modeling technique in welding operations. On the overall, Taguchi method is an effective optimization method and an integration of ANFIS technique can reduce the cost and throughput associated with running further experiments.