The Study On Identification And Positioning Of Welding Seam For Pipe

With the development of the petroleum and petrochemical industry, pipelineconstruction is being in the peak-hour development. Pipeline automatic weldingtechnology has made a significant progress, and the application of welding technologyhas good prospects in industrial automation. The important factors affecting the qualityof pipe welding line are welding position and welding parameters. After getting the crackwelding line information, it can control the welding parts to lead automatic welding. So itcan greatly reduce the labor intensity and improve the welding quality.In the background of the application, this paper mainly study automobile exhaustpipe welding seam and positioning method of designed and implemented，includinggathering cracking pipeline image information, taking it for filter pretreatment. Cuttingthe crack useful information and restructured and drawing the key weld relativecoordinate information. This paper take a FPGA as core, use Verilog to program ahigh-speed interface controller and then control the CMOS image sensor to acquire pipewelding seam image information. A complete weld line information figure was formedafter cutting the each picture from the CMOS, which accords to geometric relationship ofcamera’s field angle and the distance between the steel tube and CMOS sensor.Aftergetting the weld line information. Using iterative algorithm transferred to binarygathering information, at the same time signed and picked up the largest connecteddomain in the image.After getting effective throat area, the Canny operator used toextract edge weld contour information,and extracted its centerline for weld trackinginformation, and then Hough operator taked to optimized weld coordinate information.Inthis paper, adopted high precision ultrasonic ranging module measure steel pipe weldaround two pipeline to the relative distance s1and s2of cameras. On this basis,combining with the relationship of the camera’s field of view angle can realized theaccurate localization of welding line information.In this paper, we take advantages of the FPGA high-speed processing performance incollecting image data and storing in external SRAM to improve the speed of data acquisition, and no occupation in the processor resources. In addition, we have realized theCutting of the effective coverage of weld line，and then construct them to a plan of thesteel to processing. Through pixel density parameters, the actual size was related to thesize of the image, enhanced the precision of weld recognition，in this paper.