Unsupervised droplet identification during the pulsed laser enhanced GMAW process

Abstract

Pulsed laser-enhanced gas metal arc welding (GMAW) is an innovative arc welding process developed recently at the University of Kentucky. It uses the recoil pressure force generated by a pulsed laser to provide an additional force to reduce the needed electromagnetic detaching force (which is produced by the current) to assure the detachment of the droplet and the arc stability. The reduction in the current needed to detach droplets and maintain the arc stability improves the controllability of the most widely used arc welding process—GMAW—and its application range. To accurately control the detachment of the droplets, the force generated by the pulsed laser needs to be computed. Since this force is proportional to the distance from the center of the droplet to the welding wire, the problem can thus be changed to compute the distance between the droplet and the wire. To compute this distance, image processing method is the most effective way. Hence, different well-known image processing algorithms are implemented to address this problem and their performances are evaluated in this paper. Considering the robustness, processing speed, and automation, none of the evaluated image processing methods produce an acceptable result. To solve this specific problem, a novel image processing method is proposed. It is unsupervised and fast. Experimental results indicate that this proposed method can also achieve adequate computation accuracy.