:: Artigo completo: https://rdcu.be/dj68R
Abstract
GMA welding process stability and robustness, as well as weld quality, depend strongly on metal transfer behavior. Evaluating in detail the phenomenon of droplet transfer, considering the detachment of metal from the tip of the electrode until it reaches the weld pool and considering the forces acting in the process, is a field of great interest for those looking to control these parameters to achieve the desired characteristics in welding. Currently, there are better high-speed filming systems, which allow the acquisition of better-quality images and open the possibility for the application of computer vision algorithms with the aim of better understanding the dynamics of the metallic transfer. The objective of this research was to develop an algorithm that allows the detection of metal droplets in the footage of the metal transfer process. For this, exploratory work performed an algorithm that managed to detect the drops with high precision. Several methods using findContours, Canny, Laplacian, Sobel, and a weightless neural model were adapted and compared in the detection and measurement of the projected areas, as well as with regard to the processing times. The results generated by the algorithm allowed the detection and monitoring of the metal transferred in the form of drops, as well as the calculation of the projected area and the displacement, allowing for a better understanding of the metal transfer phenomena and, consequently, future conceptions of welding manufacturing control methods.
Keywords: Droplet transfer; MIG/MAG; Vision sensor system; Manufacturing; Welding
Referência:
SOLANO, J.L.O., MORENO-URIBE, A.M., JAIMES, B.R.A. et al. Detection and characterization of metal transfer in GMAW using computational vision algorithms. Int J Adv Manuf Technol (2023). https://doi.org/10.1007/s00170-023-12180-9