Combined Artificial Neural Network and Least Squares Method for Exploring Relationships between Welding Conditions and Weld Characteristics

Abstract

In this study, a method was developed for the automatic generation of input-output relationship models that clearly indicate the role of input factors in determining the outputs. This proposed method was used to model the relationships between the welding conditions and penetration dimensions of TIG welds. The models developed using the proposed method clearly and explicitly indicate the effects of the welding current, arc length, and welding speed on the penetration dimensions. In addition, the errors of the predicted weld penetration dimensions obtained using the developed model were less than 3%.