A comparative study for weldability prediction of AHSS stackups

Abstract

Resistance welding is the most commonly implemented method to join steel sheets in the automobile industry. To increase the efficiency and to maintain or reduce the weight of vehicles, advanced high-strength steel (AHSS) has been developed as a build material for vehicular structures. This paper aims to exploit the ability of prediction for nugget sizes in resistance spot welding by using support vector regression (SVR) and artificial neural network (ANN) model. In this study, the nugget size will be predicted according to parameters in resistance spot welding, such as welding current, welding time and welding force, etc. using machine learning methods. In addition to considering important process parameters for resistance spot welding, some design parameters, such as the thickness of the materials and the coating, are also considered. As the experimental results, SVR shows better performance in the prediction of nugget size when compared to ANN.