Vision sensor is one of the most powerful sensors int the non-contact sensory feedback of monitoring and controlling the welding process. In the butt welding process of thin plates, there exist some changes of the weldment geometry arising from the thermal distortion and pre-setting change of the plates to be welded. These changes cause a variation of the gap size, which is the main reason of automatic seam tracking and adaptive control of welding parameters.
The main subject of this study is the construction of an automatic welding system which has the function of tracking the weld seam and selecting the welding parameters in GMA welding of butt joint weldments like the container side panel. In this study, a visual sensing system for the thin plate arc welding process was constructed by the combination of the diode laser, zoom lens and close-up lens. By adopting a zoom lens in this visual sensing system, the resolution of the system could be improved up to 70㎛ per one pixel. To detect the gap information such as gap center and gap size, the secondary differential method was used. Misalignment of the weld seam line was calculated by using a first order curve fitting method. To improve the weld quality, the optimum value of the welding current according to the variation of gap size should be adaptively selected during welding. In this study, the optimum welding current for the various gap sizes were also investigated through a number of experiments of thin plate butt welding.