Study On On-line Detection Method And System Based On The Characteristic Evaluation Of Molten Pool In Fiber Laser Welding

In industrial production,welding is an important connection method for large and complex equipment parts,and the strength of the welding structure is of great significance to the safety of industrial production.There are complex physical information in the laser welding process,in which the dynamic characteristics of the molten pool directly affect the welding forming process,which also determines the stability of the welding process and the welding quality.If the molten pool characteristics can be detected by imaging to judge the occurrence of welding defects,it is of great significance for real-time evaluation of welding quality and feedback of status information.Therefore,this topic is based on image technology to evaluate the characteristics of the molten pool,study the relationship between welding process parameters and welding defects,molten pool characteristics,real-time online monitoring of the molten pool state,and then predict the quality of the weld,provide a theory for optimizing the welding process and improving the welding quality in accordance with.The specific research content is as follows:(1)The influence of molten pool characteristics on the dynamic behavior of molten pool is studied.Based on the built-in molten pool characteristic evaluation visual detection system,real-time online monitoring of the molten pool dynamic behavior,using bilateral filtering,Otsu threshold segmentation and morphological processing and other preprocessing algorithms and Canny edge detection algorithm to extract the length,width,area and trailing angle of the molten pool such as characteristic parameters.Study the variation law of molten pool characteristic quantity at different welding process parameters.The results show that with the increase of laser power and the decrease of welding speed,the width,length,width,and area of the molten pool increase approximately linearly,the trailing angle of the molten pool decreases,the molten pool movement is more intense,and the flow form is from laminar to vortex change,the solidification time becomes longer.The larger the aspect ratio and area of the molten pool and the smaller the trailing angle,the more vigorous the molten pool movement and the more eddy current movement.The aspect ratio of the molten pool is 1.5-2.4,the area is 8.74-20.12 mm~2,and the tailing angle of the molten pool is between38°-50°to achieve moderate penetration welding.The molten pool fluctuates sharply and flows to the tail of the molten pool in the form of waves to achieve the most excellent welding.(2)The influence mechanism of the molten pool dynamic behavior on the formation of porosity defects was studied.The keyhole at the front of the molten pool is unbalanced or the energy distribution is uneven,which causes the keyhole to periodically vibrate,generating a large number of bubbles.Depth fluctuation period of depth of the molten pool is used to characterize the velocity of the molten pool.The test results show that the laser power increases and the welding speed decreases,the penetration period of the molten pool is shorter,which lead to the keyhole vibration period shorten and the velocity of molten pool lengthen.The results show that when other welding parameters remain unchanged,with the increase of laser power,the fluctuation period of penetration depth of molten pool decreases,the flow velocity of the molten pool becomes faster,the bubbles easily overflow from the molten pool,and the probability of turning into pores is reduced,and the porosity rate of the weld show an trend of firstly increasing and then decreasing.With the increase of welding speed,the fluctuation period of the molten pool penetration depth and the flow velocity of molten pool show opposite trends.The solidification speed of the molten pool is faster,the probability of the bubbles changing into porosity increases,and the porosity of the welded show a decreasing trend.(3)A prediction method of welding porosity defects based on convolutional neural network of dynamic behavior of molten pool is proposed.In view of the characteristics of molten pool dynamic images,an improved AlexNet deep convolutional neural network including 5 convolutional layers,3 pooling layers,and 3 fully connected layers is constructed.The initial model is trained based on the ImageNet database,and the target training set and back propagation training to obtain a classification model based on molten pool images to predict porosity defects.The test results show that the prediction accuracy of porosity defects based on static molten pool images reaches92.7%,and the overall prediction accuracy rate is as high as 90%,while the prediction accuracy of online detection of weld porosity defects through molten pool images reaches 90.6%,and the overall prediction accuracy rate is 86.6%.In this paper,online detection method of weld porosity defects based on convolutional neural network is proposed,which realizes online prediction of weld porosity defects based on the dynamic characteristics of the molten pool,with high accuracy and good detection effect,which provides a theoretical basis for laser weld online detection of weld defects.