Prediction of bead reinforcement height and width of Gas Tungsten Arc Welded bead-on plate joints using Artificial Neural Network

A number of welding parameters are responsible for the quality of welds. The modeling of weld bead shape is important for predicting the quality of welds. In this paper, an attempt has been made to develop a back-propagation neural network (BPNN) model for the prediction of reinforcement height and width of bead in GTA bead-on plate welding process. The experimental results were used as testing sample for BPNN model. Welding current and welding speed were considered as the input parameters and bead reinforcement height and bead width were response parameters in the development of the BPNN model. The percentage errors (%) for all the samples were calculated to validate BPNN model. The result was found that the BPNN model developed in the present research work can predict the responses selected with good agreement.