Weld seam and weld seam defect detection method based on deep learning

Abstract

The invention discloses a weld seam and weld seam defect detection method based on deep learning. A YOLOV3 network is used to realize weld seam and / or weld seam defect detection. A network training step comprises the following steps of using a positioning frame to select and mark the weld seams in a workpiece image and taking as a training data set; using the positioning frame to select weld defects and mark defect types in a weld seam image, and taking as a training data set I; acquiring coordinates xp and yp of the positioning frame, and width and height sizes wp and hp; initializing a network; randomly acquiring an input tensor aj for training calculation, and outputting a detection result; calculating an error function loss of a prediction result by using the detection result; and adjusting a weight W and an offset value b in combination with a gradient descent method, repeating in the way, and acquiring a trained network. The multiple weld seams and the multiple defect types can be synchronously detected, weld seam identification positioning and defect detection can be realized in one measurement, and measurement efficiency and accuracy are effectively increased.

Description

Technical field [0001] The invention relates to the field of defect detection, in particular to a welding seam and a welding seam defect detection method based on deep learning. Background technique [0002] With the development of automation technology, industrial welding robots have been widely used in the field of manufacturing and have become the main automation equipment. New industrial welding robots use remote laser welding technology to overcome the limitations of traditional welding (such as arc welding robot posture The electric welding gun is limited by the size of the workpiece), which has the advantages of fast workpiece welding and small thermal deformation. [0003] In contrast, an efficient welding seam quality inspection method is needed to match the processing cycle requirements. The conventional structured light sensor has high measurement accuracy and can measure three-dimensional parameters. Its scanning working method is suitable for multiple welding seams on ...

AI Technical Summary

Problems solved by technology

[0004] During the welding seam detection process, due to problems such as image deformation, image shooting quality, and angle, the existing two-dimensional image detection methods such as template matching cannot effectively detect the specific defect types and locations in the weld seam; in order to solve the above problems, The present invention proposes an intelligent weld defect detection method based on the principle of deep learning, which detects multiple welds and various defect types synchronously, realizes weld identification and positioning and defect detection in one measurement, and effectively improves measurement efficiency

Images