Research On Characteristic Of Divisional Weld Pool Visual Image In Spray Transfer Welding

The zoning image of welding pool was studiede in this subject , based on the clear visual image of MIG spray transfer pool had been obtained. Detail characteristic information, such as invert-image-angle, the dross dynamics and the" rough and uneven "parameter of the welding pool, were discussed in three corresponding zones—arc,dross and the tail of the welding pool. The results provide new theory and useful data for the Intelligent Control of the welding quality.For the zone of the arc, seven characteristic parameters and the corresponding algorithms were defined and designed, including invert-image-angle of the arc, the lean parameter of the arc, etc. Studies showed that the invert-image-angle would rush to 110°and the area of the arc would rush to take up 50% of the whole welding pool area if the the flow rate of protective gas was less than 5L/min. The arc would lean if the blowtorch deviated from the weld seam.For the zone of the dross, after the dross dynamics being analyzed, the normal state of the dross was found and the "Tuowei" phenomenon was defined. A lot of experiments showed that the less the flow rate of the protective gas was, the more probability the dross would be broke up and the "tuowei" phenomenon would be more obvious.When the "Tuowei" phenomenon appeared, the pore was more possible to appear. The "tuowei" phenomenon could also be found when the rust cleaning were not performed before welding, at this time the proportion of the dross area in the whole welding pool area was increased. Studies also showed that the dross would lean to the same direction if the blowtorch deviated from the weld seam.For the tail of the welding pool, the" rough and uneven "parameter and the lean angle of the tail of the welding pool were, for the first time, defined in this paper. Detailed relationship between" rough and uneven "parameter and the welding quality was discussed in the polynomial fitting method. The range of it was acquired under different flow rate of protective gas. The" rough and uneven "parameter would rush to 10 if the flow rate of the gas was under 5L/min. The algorithm of the lean angle was programmed after discussing the tail of the welding pool when the blowtorch leant to one side. It provides new ideas for Seam Tracking."One against one" clustering methods and radial basis function were selected in SVM-Modeling. According to the results of the X-ray detection , the evaluation software of the welding quality grade was compiled which could forecast the pore and slag of a welding joint and the accuracy of the detecting could reach up to 97.757%.