CN-122023338-A - Defect detection method, device and medium of tritium-related component

CN122023338ACN 122023338 ACN122023338 ACN 122023338ACN-122023338-A

Abstract

The invention relates to the field of tritium-related component detection, and discloses a method, a device and a medium for detecting defects of a tritium-related component, wherein the method comprises the steps of filling substitute gas into the tritium-related component; the method comprises the steps of detecting a certain point of a tritium-related component, obtaining a plurality of frames of welding line images of the tritium-related component, respectively carrying out difference feature analysis on welding line images of adjacent frames to obtain a virtual vector matrix, and determining defects of areas corresponding to the welding line images according to the welding line images of each frame and the virtual vector matrix. The invention can realize accurate, rapid and non-contact defect detection on tritium-related parts.

Inventors

GUO BIN
RONG JUNHAO
HU JIANSHENG
ZHOU JINXUAN
WU GANG
TANG YUQING

Assignees

中国科学院合肥物质科学研究院

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. The defect detection method of the tritium-related component is applied to the tritium-related component and is characterized by comprising the following steps of: Filling replacement gas into the tritium-related component; acquiring a plurality of frame welding line images of the tritium-related component at a certain point of detecting the tritium-related component; respectively carrying out differential feature analysis on weld images of adjacent frames to obtain a virtual vector matrix; and determining the defects of the areas corresponding to the weld images according to the weld images of each frame and the virtual vector matrix.
2. The method for detecting defects of tritium-related components according to claim 1, wherein determining defects of a region corresponding to each weld image according to each frame of the weld image and the virtual vector matrix comprises: the modes of each virtual vector matrix are collected to obtain a dimensionless matrix of the mode; determining a weld mean image according to each frame of weld image; Performing product operation on the dimensionless matrix of the model and the weld joint mean image to obtain a background schlieren synthetic image; And determining whether a through defect exists in the region corresponding to the welding line image according to the background schlieren synthesized image.
3. The method for detecting a defect of a tritium component according to claim 2, wherein determining whether a through defect exists in a region corresponding to the weld image according to the background schlieren composite image comprises: If the module length of the background schlieren synthetic image is smaller than a preset first threshold value, determining that a region corresponding to the welding line image has no through defect; If the mode length of the background schlieren synthetic image is in a preset first interval, determining that a through defect exists in a region corresponding to the welding line image, wherein the left end point of the first interval is larger than the first threshold value.
4. The method for detecting the defect of the tritium component according to claim 2, further comprising: Acquiring an original welding line image; And training a graph convolution nerve model by taking the original weld image, the background schlieren synthesized image without the through defect and the background schlieren synthesized image with the through defect as a training set and a verification set, wherein the graph convolution nerve model outputs defect types of tritium-related components, and the defect types comprise defect-free, surface defect and through defect.
5. The method for detecting the defects of tritium components according to claim 1, further comprising removing outlier elements of each virtual vector matrix by median filtering after the virtual vector matrix is obtained, and replacing the outlier elements with the median of the arrangement of the elements in each virtual vector matrix.
6. The tritium component defect detection method according to claim 1, further comprising the steps of obtaining a plurality of frames of welding line images of the tritium component, carrying out digital image correction on each frame of welding line image, and carrying out translation on each corrected frame of welding line image to achieve registration of each frame of welding line image.
7. The method for detecting defects of tritium-related components according to claim 1, wherein the performing differential feature analysis on the weld images of adjacent frames respectively to obtain a virtual vector matrix comprises: and for each frame of welding seam image, performing cross-correlation calculation or optical flow algorithm on the frame of welding seam image and the welding seam images of adjacent frames to obtain the virtual vector matrix.
8. A device for detecting defects of a tritium-related component, applied to the tritium-related component, for performing the method according to any one of claims 1 to 7, comprising: Replacing the gas cylinder, the image acquisition equipment and the controller; the image acquisition equipment is arranged at a certain point of the tritium-related component and is used for acquiring a plurality of frame welding line images of the tritium-related component; the substitute gas cylinder is used for filling substitute gas into the tritium-related component; The controller is respectively connected with the valve of the substitute gas cylinder and the image acquisition equipment and is used for receiving a plurality of frames of welding line images sent by the image acquisition equipment, respectively carrying out difference characteristic analysis on the welding line images of adjacent frames to obtain a virtual vector matrix, and determining the defects of the areas corresponding to the welding line images according to the welding line images of each frame and the virtual vector matrix.
9. The device for detecting defects of tritium components of claim 8, the device further comprising: The light source is connected with the controller and is used for receiving the switch instruction of the controller and providing light for the weld image; And the amplifying lens component is arranged between the image acquisition equipment and the tritium-related component and is used for amplifying the weld image.
10. A storage medium having stored thereon computer-executable instructions, the computer executable instructions are for performing the method of any one of claims 1 to 7.

Description

Defect detection method, device and medium of tritium-related component Technical Field The invention relates to the field of tritium-related component detection, in particular to a method, a device and a medium for detecting defects of tritium-related components. Background In fusion plants, the stack, tritium plant and the vital components are interacted with each other by a gaseous tritium circuit. In view of the permeability of tritium in metallic materials, large-scale and complex tritium operation procedures place extremely high demands on the tightness of tritium-related components. Therefore, a strict weld quality check must be performed during the manufacturing process of tritium-related components in order to minimize the risk of leakage of tritium-containing aerosols. The existing detection methods for tritium-related parts have further perfected space on the precision and time consumption of defect detection of welding seams, and the problems of insufficient precision and high time consumption are generally caused. Disclosure of Invention The present invention aims to solve at least one of the technical problems existing in the prior art. Therefore, the invention provides a defect detection method of a tritium-related component, which can accurately, rapidly and contactlessly detect the defect of the tritium-related component. The invention also provides a device and a medium with the tritium component defect detection method. According to an embodiment of the first aspect of the invention, a defect detection method of a tritium-related component is applied to the tritium-related component, and comprises the following steps: Filling replacement gas into the tritium-related component; acquiring a plurality of frame welding line images of the tritium-related component at a certain point of detecting the tritium-related component; respectively carrying out differential feature analysis on weld images of adjacent frames to obtain a virtual vector matrix; and determining the defects of the areas corresponding to the weld images according to the weld images of each frame and the virtual vector matrix. The tritium-related component defect detection method has the advantages that the welding line background image, namely the welding line image, is utilized to obtain the virtual vector matrix representing the difference of the adjacent frames through the adjacent frames, and then the defects of the areas corresponding to the welding line images are determined through the welding line images and the virtual vector matrix of each frame. According to some embodiments of the invention, the determining the defect of the region corresponding to the weld image according to each frame of the weld image and the virtual vector matrix includes: the modes of each virtual vector matrix are collected to obtain a dimensionless matrix of the mode; determining a weld mean image according to each frame of weld image; Performing product operation on the dimensionless matrix of the model and the weld joint mean image to obtain a background schlieren synthetic image; And determining whether a through defect exists in the region corresponding to the welding line image according to the background schlieren synthesized image. According to some embodiments of the invention, the determining whether a penetration defect exists in a region corresponding to the weld image according to the background schlieren composite image includes: If the module length of the background schlieren synthetic image is smaller than a preset first threshold value, determining that a region corresponding to the welding line image has no through defect; If the mode length of the background schlieren synthetic image is in a preset first interval, determining that a through defect exists in a region corresponding to the welding line image, wherein the left end point of the first interval is larger than the first threshold value. According to some embodiments of the invention, the method further comprises: Acquiring an original welding line image; And training a graph convolution nerve model by taking the original weld image, the background schlieren synthesized image without the through defect and the background schlieren synthesized image with the through defect as a training set and a verification set, wherein the graph convolution nerve model outputs defect types of tritium-related components, and the defect types comprise defect-free, surface defect and through defect. According to some embodiments of the invention, the method further comprises removing outlier elements of each virtual vector matrix by median filtering after obtaining the virtual vector matrices, and replacing the outlier elements with the median of the arrangement of the elements in each virtual vector matrix. According to some embodiments of the invention, the method further comprises the steps of carrying out digital image correction on each frame of welding line image after acquir