Research On Weld Seam Identification And Penetration Of High-efficiency Deep Penetration Keyhole TIG Welding Based On Visual Sensing

Based on traditional Tungsten Inert Gas Welding(TIG),high-efficiency deep penetration keyhole TIG welding use the large current to generate the high energy density,high stiffness and strong penetrating power welding arc,forming the “lock effect” in the molten pool and therefore implementing a process method of medium-thick plate welding.With the rapid development of visual technology,the scope of vision for welding is gradually expanding.Through controlling the parameter information of weld and molten pool,obtained by the visual sensor in the welding process,it provides a basis for real-time prediction and control of welding process.This paper sets up a high-efficiency and deep penetration keyhole TIG welding system,in which it uses the visual sensor detection system,with high width range Charge Coupled Device(CCD)camera as the core,to conduct edge detection of weld and molten pool images respectively.The clamping device,connecting the welding gun and CCD camera,can realize photographing images of weld and molten pool at multiple angles.Based on the CCD camera calibration at different tungsten needle heights,it can determine the transformation relation between the image coordinate system and the physical coordinate system.In the research process of the medium-thick plate weld visual inspection system,it puts forward a reasonable identification and detection algorithm according to the characteristics of welding seam,whose position can be identified by the HOG-SVM algorithm automatically.And then it uses the curvature search method to obtain the edge information of the welding seam.Finally,the weld seam deviation is calculated.From different welding directions and angles,it uses the CCD camera to shoot the molten pool image and select the best shooting perspective to process the gathered image of molten pool and the front keyhole entrance of welding parts.The paper makes analysis on the characteristics of molten pool and uses the grayscale transformation,fitting and other methods to obtain the edge information of the molten pool and the keyhole entrance.Finally,the detection accuracy of the molten pool edge algorithm is analyzed through experiments.Based on high-efficiency deep enetration keyhole TIG welding process testing,this paper designs three processes,from welding arc to welding stability,incomplete penetration to full penetration and stable penetration of welding parts,to analyze the variation rules between welding parameter and the geometric feature parameters of molten pool and the keyhole entrance,as well as the relationship between welding parameters and penetration and penetration degree.Finally,under different welding process parameters,the paper sets up a Back Propagation(BP)neural network prediction model according to the geometric characteristic parameters variation rules between the molten pool and the keyhole entrance.And the model takes the geometric characteristic parameter of molten pool and keyhole entrance as the input,the penetration of seam in the back of the weld as the output.The analysis procedure lays the foundation for realizing the intelligent control of the welding process of the high-efficiency and deep penetration keyhole TIG in the future.