Abstract
Laser welding of titanium alloys is attracting increasing interest as an alternative to traditional joining techniques for industrial applications, with particular reference to the aerospace sector, where welded assemblies allow for the reduction of the buy-to-fly ratio, compared to other traditional mechanical joining techniques. In this research work, an investigation on laser welding of Ti–6Al–4V alloy plates is carried out through an experimental testing campaign, under different process conditions, in order to perform a characterization of the produced weld bead geometry, with the final aim of developing a cognitive methodology able to support decision-making about the selection of the suitable laser welding process parameters. The methodology is based on the employment of artificial neural networks able to identify correlations between the laser welding process parameters, with particular reference to the laser power, welding speed and defocusing distance, and the weld bead geometric features, on the basis of the collected experimental data.
Highlights
Titanium alloys are widely employed in the aerospace sector, due to high strength in combination with low density and good tensile properties; resistance to corrosion allows for applications in chemically aggressive conditions
An investigation of laser welding of Ti–6Al–4V alloy plates is carried out through an experimental testing campaign under different process conditions, in order to perform a characterization of the produced weld bead geometry, with the final aim of developing a cognitive methodology able to support decision making regarding the selection of the suitable laser welding process parameters, in order to obtain a weld bead with defined geometry
To evaluate the performance of the trained Artificial neural networks (ANN), a testing set was built to employ the average values of the measured geometric features (CW, root width (RW), heat-affected zone zone on the upper surface (HAZup), heat-affected zone on the lower surface (HAZlow), and fused zone (FZ)) of the weld bead, which were computed over the three repetitions of each experimental testing condition
Summary
Titanium alloys are widely employed in the aerospace sector, due to high strength in combination with low density and good tensile properties; resistance to corrosion allows for applications in chemically aggressive conditions. A reduced mean grain size in the fusion zone is achieved; the overall mechanical quality is improved, considering that the grain growth is deemed to be one of the reasons for the reduction of tensile strength upon welding For these reasons, a number of scientific studies are available in the literature dealing with laser beam welding of titanium alloys [6,7], which are widely used in aerospace thanks to their high strength in combination with low density and good tensile properties; medical and surgical devices are even produced using these alloys, thanks to high biocompatibility [8]. In many manufacturing process applications, ANN have been employed in a forward manner, i.e., to calculate the response in terms of output quality parameters, based on the values of specified input process parameters This is the case, for instance, with the work carried out in [9], where an artificial neural network was designed in order to calculate penetration-to-fuse-widths and penetration-to-haz-widths for different laser powers, welding speeds, and focal positions. The measured geometric features of the weld bead obtained from the experimental testing campaign were employed as input to train the ANN, while the estimated process parameters represented the ANN output
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
