Research On Weld Tracking And Welding Quality Evaluation Method In Laser Welding

As a key procedure of industrial production,welding is mostly affected by subjective of the operators,lead to the poor welding quality and low efficiency.Therefore,other solutions are urgently needed to replace the manual work of welding.Welding automation can significantly improve welding efficiency and welding quality.The weld contour tracking and welding quality assessment are two important parts during automatic welding process.Therefore,the research on weld contour tracking and welding quality assessment is of great significance to increase the degree of welding automation.In laser welding,the weld gap is small or even zero.At present,the commonly used laser vision detection method can not recognize this kind of weld arrangement.In order to solve this problem,a passive vision recognition device based on linear array camera is built in this thesis.Combined with the distance increasing ring,high magnification imaging of the weld area is realized,which provides a reliable data basis for realizing weld contour tracking.According to the characteristics of the complex texture structure of the captured image,the weld recognition algorithm is built.The accuracy of the recognition algorithm is verified through welding and metallographic experiments.On this basis,the recognition effect of the recognition algorithm on butt welds of materials with different gaps,welding speeds and reflectivity is studied.The laser 2D profiler was used to scan the welded seam,and the profile data of the workpiece surface was obtained,and the effect of different filtering methods on the noise suppression in the data was studied as well.As the 3D reconstruction of the weld profile was affected by non-perpendicular alignment between the line laser and the weld or uneven placement of the workpiece itself,the profile data calibration method is studied in this thesis also.The identification algorithm for extracting the residual height and width of the weld is studied,and the repeated acquisition accuracy of the sensor and the identification effect of the algorithm on welds with different shapes are verified.The results showing that:The position of the weld can be identified by the eigenvalues of the four dimensions of the image gray value,the gray value change in the welding direction,the gray gradient and the width of the weld area,and the identification tolerance is less than0.05 mm.The welding seam tracking algorithm proposed in this thesis has a good recognition effect on welds with a gap less than 0.3mm,a better recognition effect on diffuse reflective materials,and a small recognition tolerance when the scanning speed is low.Using the average filter with the size of 5 to denoise the contour line can suppress the noise while retaining the details of the weld contour line.The equation of the plane where the base metal is located can be obtained by twice fitting the data.The weighted least square method can be used to accurately fit the weld center line,and the corrected three-dimensional shape of the weld can be obtained by rotating and translating the point cloud data.The three-dimensional reconstruction results are in good agreement with the actual shape of the weld.The 13 th order polynomial can be used to accurately fit the weld contour curve.Through the zero point of the first derivative and the extreme point of the second derivative of the fitting curve,the location of the characteristic points of the reinforcement and weld toe can be found,and then the reinforcement and width of the weld can be identified as well.The repeated experimental results show that for one weld,the maximum tolerance of weld width is 0.061 mm and the maximum tolerance of weld residual height is 0.011 mm.The recognition algorithm has good recognition effect on the weld with positive and negative residual height,and even the large-size weld formed by laser welding with filler wire.