Bead Geometry Prediction in Pulsed GMAW Welding: A Comparative Study on the Performance of Artificial Neural Network and Regression Models

Weld bead geometry is a critical factor for determining the quality of welded joints, this welding process input parameters play key role. In study, relationships between variables and size weld produced by pulsed GMAW were investigated neural network trained with Bayesian-Regulation Back Propagation algorithm second degree regression models. A series experiments carried out applying Box-Behnken design experiment. The results showed that both models can predict well geometry. However, model had slightly better performance than second-order model. Both be used further analyses using them may surmount or reduce need experimental procedures especially in thermal analysis validations finite element modelling.