Researching On Control Algorithms Of Welding Quality Of Straight Seam Of Pipe Based On Line Structured Light

Straight seam welded pipe are widely used in pipeline transportation, national defense construction, etc. Straight seam welded pipe welded by Pulse Tungsten Inert Gas Welding have high quality welding seam. However, the automation level of producing straight seam welded pipe by using Pulse Tungsten Inert Gas in our country at present is not high, mainly reflected in the weld quality depend on manual detection and being unable to implement feedback control of welding process parameters through welding quality. In this regard, this paper has carried on the research on the control algorithm of Welding Quality of straight seam of pipe based on line structured light.Weld appearance molding is the key factor that determines the strength of welded joint and the related performance. In this paper, weld width and excess weld metal of weld reverse side molding are used to evaluate the welding quality of longitudinal welded pipe. Firstly, an on-line measuring system is designed by using line structured light imaging technology to measure weld width and excess weld metal. Through the system we acquire the structured light image and process it preparatively, then extract the center of structured light strip and calculate weld width and excess weld metal. In this process, we consider synthetically the extraction speed and accuracy of stripe center and then propose a method for extracting the subpixel center of strip based on the skeleton. Then combined with the advantages of both Kalman filter model and dynamic GM(1,1) prediction model, we propose a Kalman filter prediction model based on dynamic GM(1,1) prediction model that can predict the current weld width and excess weld metal. Through the actual data test the validity of the model is proved. Finally, according to the forecast data of the current time and the detection data of the few moments before, the feedback control scheme of welding process parameters is proposed. And then we analyze the time-consuming of entire process from welding quality on-line detection to the feedback control, and use CUDA to accelerate the serious time-consuming module to reach the processing time of the system requirements. The validity of the control scheme is proved by analyzing real data acquired by the actual welding with our scheme.