Research On Automatic Weld Seam Tracking System Based On Machine Vision

At present,the welding structure have a wide range of applications in all walks of life.However,the quality of welding processing is often difficult to guarantee due to the large labor intensity,poor working environment,low efficiency,uncontrollable weld position caused by welding thermal deformation,and operator's human factors.In recent years,with the development of artificial intelligence technology,real-time,adaptive welding intelligent production has become possible.The weld tracking technology as one of the most effective solution for welding path control is the hot spot of domestic and foreign welding industry.However,to achieve the automatic tracking of welds,we should first solve the problem of automatic tracking of welds and the feature extraction of welds,which is the main research content of this paper.The research of this project was supported by the project of "Key Technology Research on Intelligent Welding of Arc Welding Robot Based on Vision" in Suzhou Industry Prospecting Application Research(Industry).In the machine vision-based seam tracking system,the visual sensing system is the core of the whole system.In this paper,the hardware of the visual sensing system,that is,the light source,the visual sensor and the optical system are analyzed.According to the characteristics of the arc light in the welding working condition,the spectral characteristics of the structural light and the imaging characteristics of the industrial camera,the light source and the workpiece in the optical path system relative position,to ensure that the visual sensor can clearly capture the weld location and minimize the arc of the image acquisition interference under the premise of the design of Ethernet-based digital CMOS industrial camera and structural light Method for image acquisition and analysis of visual sensor hardware system program.The relationship between the actual coordinate of the weld,the coordinates of the weld image and the coordinates of the robot is obtained by integrating the visual system calibration and the robot-camera system calibration technology.The image features are analyzed,and the median filter is used to smooth the noise.And then the Canny operator is used to extract the edges of the image,and the skeleton method is used to obtain the single pixel skeleton line.The least squares method is used to fit the points on the skeletal line and get through the straight line intersection method.Five characteristic points that characterize the actual weld characteristics.Construction of Robot Weld Automatic Tracking System Test Platform Based on Machine Vision.The three main hardware of the visual sensor,the industrial computer and the welding robot are connected through the ethernet,and the TCP / IP protocol communication is used.On the basis of the integrated development of the camera and the robot communication software,the weld image processing,Finally,respectively,in the short circuit transition,jet drop and jet transition process conditions,according to the deviation "from small to large",teaching point "from sparse to intensive" to the system to test the test system,The The experimental results show that the system can overcome the interference of different intensity arc light during the conventional welding process and can correct the deviation between 0.5mm and 3mm.