Weld Quality Prediction of PAW by Using PSO Trained RBFNN

Abstract

Selection of process parameters in welding environment is more complex while determining the weld bead quality. For the selection of best parameters, artificial intelligence tools like neural networks blended with stochastic optimization technique like particle swarm optimization (PSO) proved to be very effective. In this paper, an attempt has been made to predict the weld bead quality using neural network trained PSO. Bead on plate experiments was conducted using plasma arc welding on superalloy Inconel material. Multiple regression mathematical equations developed by response surface methodology (RSM) were used for the analysis. The developed methodology will be very useful for automation. � 2020, Springer Nature Singapore Pte Ltd.