Laser Welding Process And Weld Shape Prediction For Titanium Alloy

In the field of aviation and aerospace industries, a greatly increasing attention is being paid to the application of laser welding titanium alloys. It becomes the first thing to control the quality of joints effectively. The weld shape directly influences the weld quality, so it is important to predict the weld shape or establish the mapping between the welding process parameters and the weld shape. However, laser welding, which involves a series of influential factors, is a complicated process, The mathematic models based on simplification and the empirical formulas deduced by regress method are both incompetent, but the artificial neural network is very suitable to deal with nonlinear systems like laser welding because of its inherent characteristics. Taking BT20 titanium alloy as the object, an artificial neural network method is introduced in this paper. Based on a large number of tests, network models of prediction of weld shape for laser welding this alloy were designed by means of MATLAB software. The models used laser power, weld speed and focus position as inputs. The outputs were the weld shape defined by eight geometric parameters. The results described the weld section figure fully and simply. The optimal network configuration and parameters were obtained by designing choicely and training repeatly. Verification tests shown that a reasonable precision of predicting was obtained for penetration welds and partial penetration welds except the weld reinforcement. Visual shapes made the results to be accepted easily. Besides, wire-feeding laser welding and laser/high frequency induction hybrid welding were also investigated in this paper. Two groups of matching relation of the weld speed and the wire feed rate were found in a condition of invariable laser power and focus position. The degree of undercutting was reduced because of the fed wire. The high-frequency induction heat source made the improvement of the weld penetration. An estimation method of the weld shape was presented at last.