Performance of predictive models to determine weld bead shape parameters for shielded gas metal arc welded T-joints

Assessing weld bead shape is essential for the quality control of a welded joint. This work presents an attempt to predict the weld bead shape parameters of shielded Gas Metal Arc Welded (GMAW) fillet joints through the application of statistical design techniques and artificial neural network. Extensive tests were performed on low carbon mild steel plates ranging in thickness from 3 mm to 10 mm. Welding voltage, welding current, and moving heat source speed were considered as the welding parameters. To develop empirical equation for defining bead geometry parameters of GMAW, multiple linear regression model (MLR) was formulated. Also, an artificial neural network (ANN) model was developed and individual feature importance was studied using SHapley Additive exPlanations (SHAP). The results show that the ANN-based approach performs better in predictability and error assessment. The predictive model has been used to study the temperature profile of a joint numerically, and an excellent match has been noticed when compared with experimental measurement. This study shows the usefulness of the predictive tools to aid numerical analysis of welding.