Research On Lap Weld Identification Based On Linear Structured Light Vision Sensor

Welding manufacturing technology is one of the most important technologies in industrial production,and its intelligent development is an important development direction of current manufacturing technology.At present,most domestic enterprises in the welding manufacturing,the use of manual welding or traditional teaching welding robots for welding production operations.However,the traditional teachable welding robot needs to be programmed in advance,and the robot can only weld according to the preset trajectory,which lacks flexibility and stability.In particular,due to tooling and machining errors,as well as thermal deformation of the workpiece during the welding process,the position of the weld seam previously obtained by manual teaching may deviate from the actual position.Vision sensing improves the key technology of welding robot stability,flexibility and intelligence,especially the line structure light vision sensor has the advantages of high recognition accuracy,fast acquisition speed and strong resistance to noise interference,which is the best choice of vision sensor for welding robot at present.In this academic dissertation,the key technology of lap weld recognition is studied for the application scenario of lap weld,and the main research work includes:(1)Calibration of the linear structured light vision sensor system.The calibration of the camera parameters is achieved by using the Zhang Zhengyou calibration method and a single-strain matrix-based calibration algorithm for the calibration of the line structured light plane,which achieves the calibration of the equations of the line structured light plane by solving the single-strain matrix between the light plane and the image plane(2)Research on light strip center extraction.A contour tracking-based anti-spatter center extraction method is proposed,which can extract the center of the light strip more quickly than other center extraction algorithms.For the problem of spatter in the welding process,this academic dissertation effectively removes spatter interference by comparing the grayscale values of pixels in two adjacent frames,while retaining the complete light bar information and reducing the impact of spatter on the center extraction accuracy.(3)Research on weld seam feature point extraction method.A contour corner point extraction method based on chord-to-point distance accumulation(CPDA)is proposed,while the polygon approximation technique is used to merge the corner points and remove the influence of pseudo corner points,which improves the accuracy and stability of weld seam feature point extraction.(4)Weld seam identification guidance experiments.The extracted weld seam feature point coordinates are converted by hand-eye calibration,the converted coordinates are passed to the welding robot by communication,and the welding robot guidance experiments are conducted,and the results show that the welding head can be accurately guided for well welding along the lap seam.