Prediction method for quantitative deviation of additive manufacturing cladding layer

Abstract

The invention relates to a prediction method for the quantitative deviation of an additive manufacturing cladding layer, and belongs to the technical field of precision welding. The method comprises the steps of after bottoming work of the first four layers of an additive material is completed, using a molten pool visual sensor for collecting a molten pool image of the deviation of a fifth cladding layer of the additive material, and carrying out ROI area selection on the molten pool image; calculating the deviation corresponding to each collected molten pool image according to the Pythagorean theorem, making a data set, and determining an arc starting point and an arc extinguishing point of the fifth layer of the additive material according to the data set; carrying out classification experiments under deviation on different cladding layers, and classifying by adopting a deep residual network; performing a prediction experiment on the linearly changed cladding layer deviation, and converting a classification task of different deviation of a welding seam into a regression task; and verifying the generalization ability of the network. According to the method, the cladding layer deviation in the additive process is quantitatively predicted by utilizing the visual information of the molten pool, and the actual position of a welding gun is adjusted in the additive manufacturing process, so that a good molten pool form is obtained, and the welding quality is improved.

Description

technical field [0001] The invention relates to a method for quantitatively predicting cladding layer offset in additive manufacturing, and belongs to the technical field of precision welding. Background technique [0002] Additive manufacturing technology is a new manufacturing technology based on the principle of layer-by-layer discrete and layer-by-layer accumulation to realize material forming through "bottom-up". This technology can quickly and accurately manufacture objects with complex shapes, and has the advantages of high material utilization rate, good forming effect, low technical cost and high production efficiency. It solves the process bottleneck of rapid manufacturing of complex structural parts. It has been widely used in various industrial fields such as automobiles, aerospace, and medical treatment. This technology greatly improves the production capacity of welding, and can effectively guarantee the welding quality of the welded parts. [0003] In actua...

AI Technical Summary

Problems solved by technology

For passive vision technology, since it directly monitors the weld centerline and welding torch at the arc position, it will not produce the problem of advanced detection error similar to active vision because it misses the field of view. Some scholars have done some work on passive weld seam tracking. However, there are few reports on seam tracking during the additive process

[0005] With the development of computer technology and the rise of big data, deep learning has been widely used in various industrial fields, including welding, but there are few reports on the use of deep learning to monitor passive weld seam offset in additive manufacturing.

Images