Image Processing of the Weld Pool in Robotic Plasma Welding System

Abstract

The formation of stable penetration beads in joining of the thick materials is important in order to achieve high quality welded metal joints. Since plasma welding employs the high welding current density, it is suitable in joining of the thick materials. The relationship between the voltage and the keyhole is unclear. It is important to keep the standoff in a high quality of the plasma welding. But it is difficult to estimate the standoff by using the voltage and the welding current. The authors tried to observe the weld pool on top side of the base metal, directly. For this purpose, the authors tried to apply a CCD camera with an external trigger. Moreover, the keyhole depends on the standoff of the plasma torch. When the base metal was inclined due to the setting in the robotic welding, a good quality of welding cannot be obtained. In the teaching of the robot, it took much time to adjust the standoff, since the electrode of the plasma torch was hidden into the plasma tip. In order to save the teaching time, the authors tried to control standoff of torch by processing the weld pool images taken with the CCD camera during the welding.

Adaptive Control by Convolutional Neural Network in Plasma Arc Welding System

“…The authors took the weld pool image with the keyhole using a CMOS camera from the top side. 19) If there is no gap in the workpieces with the groove, the quality of PAW will depend on the tracking of the welding line. 21) Due to heat distortion, the gap occurs in the fabrication of the thick plates.…”