Molten Pool Vision And Welding Seam Three-dimensional Information Acquisition Technology

Closed-loop intelligent monitoring of additive manufacturing is an important research direction in the field of additive.The weld pool vision monitors the welding process by the shape and fluidity before forming.The three-dimensional measurement of weld seam monitors the deviation of each layer of the additive and design model by the shape information such as the degree of curvature and the amount of collapse after forming.Clear weld pool image and high-precision weld three-dimensional information is the key to the closed-loop control system of additive manufacturing.This paper has conducted in-depth research on how to obtain these two parts of information.The main research contents are as follows:(1)Optimal imaging band selection method for weld pool imaging based on spectrum analysis.For the disunity of existing weld pool imaging band selection and lacking theoretical guidance,optimal imaging band selection method for weld pool imaging based on spectrum analysis is proposed.Based on the spectrum of self-emitted radiation of weld pool and the quantum efficiency of the camera,this paper deduces the relative response intensity of the camera to the self-emitted radiation of the weld pool,combined with arc spectrum properties,the proposed method calculates the optimal imaging band of the molten pool.through analyzing the relationship between welding current and light intensity,this paper designs a trigger method for weld pool image acquisition.Based on the method,we select 850 nm high-pass filter and 660 nm band-pass filter to capture weld pool images of stainless steel GMAW plate welding.The captured images have clear texture,high signal-to-noise ratio,and abundant details.(2)Binocular stereo vision based on stripe order recognition.For the difficulty of calibration of the projector in the fringe projection profilometry and the difficulty of image matching in the binocular stereo camera,a binocular stereoscopic three-dimensional reconstruction system based on stripe order recognition is built.The system uses the phase information and the binocular epipolar constraint to achieve sub-pixel matching,which makes the 3D reconstruction accuracy reach 0.11 mm,and realizes the 3D reconstruction of the step standard and the steel plate weld.For problems of effective point in 3D reconstruction,the method of segmentation of background and target region in image is proposed.the background and target segmentation are realized by K-means clustering algorithm,so that the 3D reconstruction process only needs to process the target area,which significantly improves the speed of 3D reconstruction.