Linear regression equations for modeling the submerged-arc welding process

Abstract

Traditionally, curvilinear regression equations have been used for modeling the geometrical features of the submerged-arc welding process. However, recent investigation has shown that, due to low tolerance in some cases, it is not possible to include all variables in the regression analysis. This investigation was therefore carried out to study the feasibility of using linear regression equations instead of curvilinear techniques to model the weld features. The linear regression equations are found to give correlation coefficients similar to those obtainable from curvilinear regression equations. The average mean, standard deviation, minimum value and maximum value of the deviations between the measured features and those computed by use of the linear regression equations are only slightly (1.38–5.40%) higher. Linear regression equations are equally suitable, therefore, for modeling the sub-merged-arc welding process.