Prediction of process parameters for gas metal arc welding by multiple regression analysis

Abstract

The purpose of this study is to find a prediction regression equation of the welding process parameters in order to obtain the desired geometry of the back-bead in butt welding, a form of gas metal arc welding in which a groove gap exists. The regression model equation is obtained from welding process parameters through the correlation of the parameters of the back-bead, to which an inverse transformation is performed. From the parameters of the back-bead, the correlation between the welding process parameters was found in a regression model equation. The accuracy of the regression model was proven through the mean error rate for prediction under conditions of data for the analysis level and the verification level respectively. As a result, the maximum prediction error rate in the forward process was under 9.5 per cent, and the prediction error rate in the inverse process, which is the prediction for the welding process parameters, was under 6.5 per cent, proving that a regression equation of the model for the prediction system of the back-bead can be obtained through both processes.