The detection of three-dimensional morphological parameters of weld seams of pressure pipelines is the guarantee for the safe operation of pressure pipelines.Aiming at the problems of low efficiency and poor accuracy of manual detection methods,we propose a three-dimensional morphological parameter detection method of longitudinal welds on the outer surface by combining vision and structured light,adopting the triangulation method of structured laser and CMOS camera,and proposing the structure of feature-point extraction network based on depth convolution to extract the feature points of weld seams and complete the detection of reinforcement,width and uncut parameters of weld seams.The detection of the parameters of the weld seam,misalignment,the method of fitting the standard circle of the inspected cylindrical parts,completing the detection of the amount of weld seam misalignment and the peaking,the design of the wire cutting processing simulation of the welded parts to evaluate the error of the whole measurement system,the measurement error of the 5 parameters of the weld seam is within 0.1 mm.
维普
万方数据