Machine vision-based high-efficiency evaluation method for sensitivity of hot crack of welding seam

Abstract

The invention belongs to the technical field of weld joint hot crack evaluation, and particularly relates to an efficient evaluation method for weld joint hot crack sensitivity based on machine vision. A welding hot crack is manufactured by carrying out a three-point bending test on the back of a welding seam in the later solidification period, the welding seam part is transversely covered and scanned by means of a laser rectangular surface light source, a high-resolution CCD is used for collecting a welding seam picture under laser irradiation in real time, and edge information of a crack and shadow combination is recognized through an image processing algorithm. And calculating the crack rate to evaluate the heat crack sensitivity of the weld. The technical problems that an existing hot crack evaluation method is tedious in procedure and low in detection efficiency are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of weld hot crack evaluation, and in particular relates to an efficient weld hot crack sensitivity evaluation method based on machine vision. Background Art

[0002] Cracks are one of the most harmful defects in welding. Their existence will seriously affect the mechanical properties and service life of welded structures. Hot cracks are a common type of cracks. They are often generated during or immediately after welding and can appear in welded structures of various materials such as high-strength steel, stainless steel, and high-temperature alloys. It is generally believed that hot cracks are caused by the segregation of low-melting-point eutectics during the solidification of the molten metal in the weld, which leads to weak grain boundary bonding and is eventually formed under the driving force of welding tensile stress. At present, conventional hot crack evaluation methods mainly include hot crack sensitivity index method (indirect evaluation method) and FISCO and adjustable restraint test (direct evaluation method). Among them, the indirect evaluation method only considers the effect of chemical composition on crack tendency, but ignores the influence of factors such as organization and stress, and its evaluation accuracy is poor. The test device of the FISCO method is complex, the test efficiency is low, and the probability of crack generation is small. The adjustable restraint test can simulate hot cracks, but the detection of weld cracks needs to be completed by professional non-destructive testing means (commonly used penetration testing) after applying strain, which prolongs the evaluation cycle. CN 111604624 A proposes a test device and evaluation method for obtaining weld thermal cracks, which can easily obtain weld thermal cracks, but the detection method of weld cracks still adopts traditional non-destructive testing, and the efficiency problem is still prominent. Based on the above problems, it is urgent to propose an efficient evaluation method for weld thermal crack sensitivity. Summary of the invention

[0003] The purpose of the present invention is to provide an efficient evaluation method for weld thermal crack sensitivity based on machine vision, so as to solve the technical problems of the existing thermal crack evaluation method, such as complicated procedures and low detection efficiency.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is:

[0005] An efficient evaluation method for weld thermal crack sensitivity based on machine vision is proposed. A three-point bending test is carried out on the back of the weld in the late solidification stage to produce welding thermal cracks. The weld area is scanned horizontally with the help of a laser rectangular surface light source. A high-resolution CCD is used to collect real-time weld photos under laser irradiation. The edge information of the crack and shadow combination is identified through an image processing algorithm, and the crack rate is calculated to evaluate the weld thermal crack sensitivity.

[0006] Prefabricate a single-sided groove on the plate to be tested in advance; clean the oil on the groove surface and nearby areas before starting the welding; use a template to perform a three-point bending experiment on the back of the weld to create thermal cracks during the solidification of high-temperature liquid metal, and use a laser rectangular surface light source to cover and scan the weld area perpendicular to the weld direction; use a high-resolution CCD to collect real-time photos of the weld under laser irradiation; use threshold segmentation and edge detection to identify the edge information of welding thermal cracks and shadow combinations; count the sum of the identified crack lengths, calculate the crack rate, and evaluate the sensitivity of thermal cracks.

[0007] The single-sided groove type is single-sided V-shaped or Y-shaped.

[0008] When welding, leave a gap of 1-2mm.

[0009] The laser scanning speed is 120-240 mm / min.

[0010] The image sampling frequency is 5-10fps.

[0011] The calculation formula of crack rate is: Where Cs is the surface crack rate, Ls is the sum of the detected surface crack lengths, and L is the weld length.

[0012] The TIG welding process is nickel-based welding wire.

[0013] Set a reasonable threshold to binarize the image and select the photo with the largest area of ​​grayscale 0 / grayscale 255.

[0014] The canny operator is used to detect the edges in the image, and the contours with an aspect ratio greater than 6 are identified as thermal cracks, and the crack rate is statistically calculated.

[0015] The beneficial effects achieved by the present invention are:

[0016] The present invention eliminates post-weld nondestructive testing through machine vision and image processing technology, greatly simplifies the operating procedures of traditional welding thermal crack evaluation methods, reduces the influence of ambient light on the accuracy of crack detection at the test site, and greatly reduces the crack missed detection rate.

[0017] By using machine vision and image processing technology, the post-weld non-destructive testing process has been eliminated, greatly simplifying the operating procedures of the traditional welding thermal crack evaluation method; using high-brightness, high-purity, low-attenuation uniform rectangular laser as the active light source will greatly reduce the impact of ambient light at the test site on the accuracy of crack detection; using the shadow under laser irradiation as a whole to detect its edge is equivalent to expanding the width direction of the crack, but the length dimension does not change much, which greatly reduces the crack missed detection rate and has little effect on the calculation of the crack rate by statistical crack length. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A schematic diagram of an efficient evaluation method for weld thermal crack sensitivity based on machine vision;

[0019] Figure 2 Schematic diagram of rectangular laser scanning method. DETAILED DESCRIPTION

[0020] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] The present invention discloses an efficient evaluation method for weld thermal crack sensitivity based on machine vision, which manufactures weld thermal cracks by carrying out a three-point bending test on the back of the weld in the late solidification stage, and laterally covers and scans the weld part with the help of a laser rectangular surface light source, and uses a high-resolution CCD to collect weld photos under laser irradiation in real time, and identifies the edge information of the crack and shadow combination through an image processing algorithm, and calculates the crack rate to evaluate the weld thermal crack sensitivity. The present invention is specifically implemented as follows:

[0022] 1. Prefabricate a single-sided groove on the plate to be tested in advance, and the groove type is a single-sided V-shaped or Y-shaped groove;

[0023] 2. After cleaning the oil stains on the groove surface and nearby areas, start welding and leave a gap of 1-2mm during welding;

[0024] 3. During the solidification process of high-temperature liquid metal, a three-point bending experiment is performed on the back of the weld using a template to create thermal cracks. At the same time, a laser rectangular surface light source is used to cover and scan the weld area perpendicular to the weld direction, and the laser scanning speed is 120-240mm / min;

[0025] 4. Use high-resolution CCD to collect real-time photos of welds under laser irradiation, with an image sampling frequency of 5-10fps;

[0026] 5. Use image processing algorithms to detect and identify edge information of welding thermal cracks and shadow combinations, where the image processing algorithms mainly include threshold segmentation and edge detection;

[0027] 6. Count the sum of the crack lengths after identification, calculate the crack rate, and evaluate the sensitivity of thermal cracks.

[0028] The present invention provides an efficient evaluation method for weld thermal crack sensitivity based on machine vision, such as Figure 1 As shown, it involves components such as a laser source, a CCD, a pressing block, and a master mold. Specifically, it includes the following steps:

[0029] Step 1: Prefabricate a single-sided groove on the plate to be tested;

[0030] Step 2: Clean the oil stains on the groove surface and nearby areas before starting welding;

[0031] Step 3: During the solidification process of the high-temperature liquid metal, a three-point bending experiment is performed on the back of the weld using a template to create a thermal crack, and a laser rectangular surface light source is used to cover and scan the weld area perpendicular to the weld direction;

[0032] Step 4: Use high-resolution CCD to collect real-time photos of the weld under laser irradiation;

[0033] Step 5: Use image processing algorithms to detect and identify edge information of welding thermal cracks and shadow combinations;

[0034] Step 6: Count the sum of the identified crack lengths, calculate the crack rate, and evaluate the sensitivity of thermal cracks.

[0035] Furthermore, the groove type is a single-sided V-shape or a Y-shape.

[0036] Furthermore, a gap of 1-2 mm is reserved during the welding in step 2.

[0037] Furthermore, in step three, the laser scanning speed is 120-240 mm / min, and the image sampling frequency is 5-10 fps.

[0038] Furthermore, the image processing algorithm in step five is mainly threshold segmentation and edge detection.

[0039] Furthermore, the calculation formula of the crack rate is

[0040]

[0041] Where Cs is the surface crack rate, Ls is the sum of the detected surface crack lengths, and L is the weld length.

[0042] Example: Test the welding hot crack sensitivity of Inconel600. Prepare 6mm thick plate, process Y-shaped groove, blunt edge 1mm, gap 1mm, and use nickel-based wire TIG welding process. The laser source is as follows Figure 1 The image is scanned from left to right in a moving manner. Each frame of the image can cover the entire length of the weld. The weld image acquisition frequency is 6f / s. A reasonable threshold is set to binarize the image, and the photo with the largest area of ​​grayscale 0 / grayscale 255 is selected. The canny operator is used to detect the edges in the image. Based on experience, the contours with an aspect ratio greater than 6 are identified and regarded as thermal cracks, and the crack rate is calculated statistically.

Claims

1. An efficient evaluation method for weld thermal crack sensitivity based on machine vision, characterized by: A three-point bending test is carried out on the back of the weld in the late stage of solidification to create welding thermal cracks. The weld area is scanned horizontally with the help of a laser rectangular surface light source. A high-resolution CCD is used to collect real-time photos of the weld under laser irradiation. The edge information of the crack and shadow combination is identified through an image processing algorithm, and the crack rate is calculated to evaluate the thermal cracking sensitivity of the weld.

2. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 1 is characterized in that: Prefabricate a single-sided groove on the plate to be tested in advance; clean the oil on the groove surface and nearby areas before starting the welding; use a template to perform a three-point bending experiment on the back of the weld to create thermal cracks during the solidification of high-temperature liquid metal, and use a laser rectangular surface light source to cover and scan the weld area perpendicular to the weld direction; use a high-resolution CCD to collect real-time photos of the weld under laser irradiation; use threshold segmentation and edge detection to identify the edge information of welding thermal cracks and shadow combinations; count the sum of the identified crack lengths, calculate the crack rate, and evaluate the sensitivity of thermal cracks.

3. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2 is characterized in that: The single-sided groove type is single-sided V-shaped or Y-shaped.

4. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2 is characterized in that: When welding, leave a gap of 1-2mm.

5. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2 is characterized in that: The laser scanning speed is 120-240 mm / min.

6. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2 is characterized in that: The image sampling frequency is 5-10fps.

7. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2, characterized in that: The calculation formula of crack rate is: Where Cs is the surface crack rate, Ls is the sum of the detected surface crack lengths, and L is the weld length.

8. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2 is characterized in that: The TIG welding process is nickel-based welding wire.

9. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2, characterized in that: Set a reasonable threshold to binarize the image and select the photo with the largest area of ​​grayscale 0 / grayscale 255.

10. The method for efficiently evaluating weld thermal crack sensitivity based on machine vision according to claim 2, characterized in that: The canny operator is used to detect the edges in the image, and the contours with an aspect ratio greater than 6 are identified as thermal cracks, and the crack rate is statistically calculated.

Images