| The position of tungsten, weld groove and wire tip are very sensitive to thin stainless steel weld. As the requirement of DAIKIN ltd., a project was established to research relationship of weld quality with the above sensitive characteristics, which are extracted by image processing. These characteristics can be used to refine weld technology and image processing algorithm can be used to do weld real time control.In order to reach the error requirement, a new weld visual sensing system was developed by Zhao Liang-qiang, which including CCD sensor, lens, CCD camera and movable welding platform.According to above system, an image processing algorithm was developed to extract A, B, C, D, E and F characteristics from weld pool images which were obtained under 0.9mm thin stainless steel weld at 2 pulse. After 20 images were processed, the changing trends were analyzed.And another image processing algorithm was put forward to adjust the position of tungsten and weld groove in real time. According to experiment, the calculation precision could reach to 0.2mm.What is more, 2 methods to extracted weld pool contour were applied to 2mm thin stainless steel weld at 7Hz, which including canny operator and sudden change detection at weld pool contour pixels.Through the development of image processing algorithms under different thickness and pulse for thin stainless steel TIG weld, reliable data and calculation method were provided for visual characteristics analysis and thin stainless steel automotive weld. |
PDF Full Text Request