X. Zhang, A. Daneryd, B. Holmgren, ABB Corporate Research, Vasteras, Sweden; D. Skarin, ABB AB, Robotics, Vasteras, Sweden
This paper describes a research and development activity carried out in ABB Company for automobile industry application. Current focus is GMAW robot welding with zinc coated thin steel plates in 0.8, 1 and 2 mm thickness. In practice, it is more difficult to weld thin zinc coated steel plates than uncoated steel plates due to its special property of galvanized steel. In this study, an arc welding system model is built up to analyze arc currents and arc voltages. The weld profiles are predicted based on neural network analysis. A large number of arc welding experiments were done in order to obtain enough data to build up the neural network models. A new version of robot welding design tool has been developed based on arc physical and neural network analysis. The Arc and weld profiles from the experiments and the design tool are compared. Results show that the design tool is reliable for predicting arc and weld profile under specified conditions. It is a useful tool for welding engineers to make their design before welding.
Summary: In this study, The weld profiles for GMAW robot welding with zinc coated thin steel plates are predicted based on neural network analysis. The Arc and weld profiles from experiments and the developed design tool are compared.
