Online Detection Method Of Weld Penetration Based On Vision For Fiber Laser Welding

The welding of thick plate is widely used in shipbuilding,aerospace and other manufacturing industries,and laser welding can effectively improve the efficiency and the quality of weld seam.Fiber laser welding showed unique advantages in the welding of thick plate compared to traditional arc welding,including high depth-to-width ratio,narrow heat affected zone,and small welding deformation.However,due to the increase of plate thickness,the energy transfer and rheological behavior in the molten pool are more complicated compared with thin plate,the stability and consistency of welding process are more important and difficult to keep for thick plate.In particular,penetration fluctuation is a typical problem that impedes the extensive application of high power laser in heavy industry to some extent.A partially penetrated weld seam will greatly reduce the weld quality.Therefore,monitoring the penetration state of the welding process can detect and promptly modify the problems in time to ensure the quality of welding,and has important significance in the application of thick plate automation welding.Among many kinds of welding process monitoring sensors,the visual sensor is widely concerned and studied because of its informative,intuitive and none contact acquisition status.In this paper,a dual camera synchronous trigger acquisition system consisted the Basler industrial CCD camera was established to detect the temperature field during welding process.And the image of welding process in different penetration status were collected by the high speed image acquisition system consisted a high speed CMOS camera produced by Photonfocus.The temperature field of molten pool is calculated by using the method of colorimetric temperature measurement and the dual camera synchronous trigger acquisition system.It was found that in the process of laser welding,the flow pattern of molten pool would change in welding status,which will affect the shape and temperature field of molten pool.When the welding process was partially penetrated,the rear of weld pool end would show a shrinkage narrow shape,and the average surface temperature would decrease compared to the fully penetrated welding process,in which the average temperature of the weld pool is much higher,and the contour is smoother.According to the temperature field and the morphologic information of molten pool profile,a model of molten pool flow pattern during different penetration status was proposed based on the influence factor of molten metal flow pattern.And the effect of penetration status on the molten pool morphology was analyzed.The welding process image was collected by high speed image acquisition system.According to the characteristic of the molten pool in different penetration status,the characteristic parameters of welding process were designed and extracted.The four image features of welding process are:the radiation intensity of metallic vapor?RIMV?,the area of molten pool end?AMPE?,the rear angle of molten pool end?RAMPE?,and the aspect ratio of molten pool end?ARMPE?.To describe the process characteristics of each feature,4 process parameters were extracted from every image feature.The 4 process parameters are:mean value?M?,variation coefficient?CV?,relative range?R?and the frequency ratio?F?.The principal component analysis was used to reduce the dimension of the 16 process parameters.4 principal components of partial penetration made by thickness increasing or laser power decreasing were extracted respectively.And two comprehensive indexes Y₁ and Y₂ were obtained by further calculation of principal components.Experimental results showed that the on-line detection method was effective to recognize the penetration status of high power laser welding through the image feature and characteristic parameters extracted from the image of welding process.