Hybrid heat source model designing and parameter prediction on tandem submerged arc welding

Abstract

In submerged arc welding simulation, the heat source parameters are always decided by experience, and it usually leads to a high simulation error. This work is aimed to develop a methodology for the estimation of the heat source in submerged arc welding. A hybrid heat source model was applied on submerged arc welding simulation. The new heat source model was combined by a surface heat source model and the double ellipsoid heat source model. The surface heat source model was designed based on the Gaussian heat source model. The width and penetration of the weld pool were simulated and compared with the measurement results, and the width at 2 mm depth from the top surface was also considered to describe the shape of the weld pool more accurately. In order to reduce the simulation complexity, the sensitivity of heat source parameters was discussed. The heat source parameter corresponding to different experimental processes was obtained by modified pattern search method. The artificial neural network algorithm and the support vector machine algorithm were applied to predict the relationship between all possible process and the heat source parameters. The validation experiment showed that the prediction model was accurate.