Apparatus and method for detecting of faults in surfaces of metal tube

Abstract

An apparatus and method for detecting defects on the surface of a boiler tube are disclosed. An apparatus for detecting defects on the surface of a boiler tube according to one aspect of the present invention comprises: an image receiving unit for receiving a boiler tube surface image; a cavity detection unit for detecting a cavity by applying an object detection algorithm to the boiler tube surface image; and a cavity verification unit for determining the presence or absence of a cavity using the boiler tube surface image and the cavity detection image in which a cavity is detected by the cavity detection unit. The cavity verification unit may include: an image preprocessing unit for converting the boiler tube surface image and the cavity detection image into a grayscale image; an image feature extraction unit for extracting a directional component by performing a convolution operation between the grayscale image and a directional filter; a probability calculation unit for calculating a probability value based on the directional component by cropping a certain area of ​​the image from which the directional component is extracted around the cavity location; and a determination unit for determining the presence or absence of a cavity by comparing the calculated probability value with a preset threshold.

Description

Technology Field

[0001] The present invention relates to a boiler tube surface defect detection device and method, and more specifically, to a boiler tube surface defect detection device and method capable of automatically detecting cavities in a boiler facility. Background Technology

[0002] Metal boiler equipment gradually develops defects due to the heat applied over prolonged periods during operation. To ensure stable operation, it is necessary to periodically monitor the extent of defects and perform repairs or replacements at the appropriate time before a breakdown occurs. Therefore, accurately assessing the condition of the equipment to predict the timing of failure and respond accordingly plays a crucial role in efficient equipment management.

[0003] Accordingly, experts assess the degree of defects in boiler equipment through non-destructive methods such as surface replication testing and hardness measurement, and determine whether to repair or replace the equipment based on these findings. Surface replication testing is a non-destructive method in which a replica of the metal is taken and its microstructure is observed using an optical microscope (OM) and a scanning electron microscope (SEM). Through this observation, the presence and number of creep cavities on the metal surface are identified, and the results rely on subjective judgment based on the expert's experiential knowledge, such as comparing them with experimentally obtained reference images. Consequently, there are issues in that the interpretation of results regarding the presence of defects in boiler equipment takes a long time, and there is a high possibility of human error.

[0004] In addition, experts evaluate the degree of creep damage according to the VGB-TW508 (Edition 1992) standard. This standard divides creep damage for DIN 17175 steel into six major levels: Grade 2 for isolated pores, Grade 3 for directional pores, and subdivides the stages of development into microcracks and macrocracks into Grades 4 and 5, respectively. Although the degree of damage is divided into levels 0 to 5, there is a problem in that the classification results vary depending on the expert because the classification criteria are not quantitative and the differences between levels are ambiguous.

[0005] Accordingly, there is a need for the development of technology that can clearly and consistently assess the degree of defect in boiler equipment, as this influences the determination of the replacement timing.

[0006] The background technology of the present invention is disclosed in Korean Registered Patent Publication No. 10-2128066 (published June 30, 2020). The problem to be solved

[0007] The present invention has been devised to improve upon the aforementioned problems, and an objective according to one aspect of the present invention is to provide a boiler tube surface defect detection device and method capable of automatically detecting cavities in a boiler facility.

[0008] The problems that the present invention aims to solve are not limited to the problem(s) mentioned above, and other unmentioned problems will be clearly understood by those skilled in the art from the description below. means of solving the problem

[0009] A boiler tube surface defect detection device according to one aspect of the present invention comprises an image receiving unit for receiving a boiler tube surface image, a cavity detection unit for detecting a cavity by applying an object detection algorithm to the boiler tube surface image, and a cavity verification unit for determining the presence or absence of a cavity using the boiler tube surface image and the cavity detection image in which a cavity is detected by the cavity detection unit, wherein the cavity verification unit may include an image preprocessing unit for converting the boiler tube surface image and the cavity detection image into a grayscale image, an image feature extraction unit for extracting a directional component by performing a convolution operation between the grayscale image and a directional filter, a probability calculation unit for calculating a probability value based on the directional component by cropping a certain area of ​​the image from which the directional component is extracted centered on the cavity location, and a determination unit for determining the presence or absence of a cavity by comparing the calculated probability value with a preset threshold.

[0010] In the present invention, the image feature extraction unit may include a feature extraction module that extracts four directional components through a convolution operation between four directional filters and the gray image, and a binary image conversion module that converts the gray image from which the four directional components have been extracted into a binary image.

[0011] In the present invention, the four directional filters may be directional filters of 0 degrees, 45 degrees, 90 degrees, and 135 degrees.

[0012] In the present invention, the binary image conversion module can convert the gray image into a binary image by using an adaptive threshold in which the threshold value changes according to the region of the gray image.

[0013] In the present invention, the probability calculation unit divides each directional component binary image into a fixed area of ​​N x N centered on the cavity location, calculates the number of white pixels in each divided area, and can calculate the calculated number of white pixels as a probability value.

[0014] In the present invention, the determination unit can compare the probability value for each region with the threshold, and if the probability value is greater than or equal to the threshold, it can finally determine that it is a cavity.

[0015] A method for detecting defects on the surface of a boiler tube according to another aspect of the present invention comprises: a step in which a cavity detection unit detects a cavity by applying an object detection algorithm to an image of the surface of a boiler tube; a step in which a cavity verification unit converts the image of the surface of the boiler tube and the image of the cavity detection into a grayscale image; a step in which the cavity verification unit extracts a directional component by performing a convolution operation between the grayscale image and a directional filter; a step in which the cavity verification unit crops a certain area of ​​the image from which the directional component has been extracted around the location of the cavity and calculates a probability value based on the directional component; and a step in which the cavity verification unit determines the presence or absence of a cavity by comparing the calculated probability value with a preset threshold.

[0016] In the present invention, the step of extracting the directional components may include the step of the cavity verification unit extracting four directional components through a convolution operation between four directional filters and the gray image, and the step of the cavity verification unit converting the gray image from which the four directional components have been extracted into a binary image.

[0017] In the present invention, the four directional filters may be directional filters of 0 degrees, 45 degrees, 90 degrees, and 135 degrees.

[0018] In the step of converting to the binary image, the cavity verification unit can convert the gray image into a binary image using an adaptive threshold in which the threshold value changes according to the region of the gray image.

[0019] In the step of calculating the probability value, the cavity verification unit divides each directional component binary image into a fixed area of ​​N x N centered on the cavity location, calculates the number of white pixels in each divided area, and can calculate the calculated number of white pixels as a probability value.

[0020] In the step of determining the presence or absence of the cavity, the cavity verification unit compares the probability value for each region with the threshold, and if the probability value is greater than or equal to the threshold, it can finally determine that it is a cavity. Effects of the invention

[0021] A boiler tube surface defect detection device and method according to one embodiment of the present invention can rapidly process a large amount of image data by automatically detecting cavities in boiler equipment and can obtain consistent and highly reliable results in determining the degree of metal defect occurrence by reducing false detections that may occur due to judgments based on the experience of existing experts.

[0022] The boiler tube surface defect detection device and method according to one embodiment of the present invention has the effect of improving the cavity detection performance of a boiler facility by reflecting the condition that 'the cavity must be located above the grain line or triple point,' which is difficult to determine solely by object detection technology.

[0023] Meanwhile, the effects of the present invention are not limited to those mentioned above, and various effects may be included within the scope obvious to a person skilled in the art from the contents described below. Brief explanation of the drawing

[0024] FIG. 1 is a block diagram schematically showing a boiler tube surface defect detection device according to one embodiment of the present invention. FIG. 2 is a block diagram specifically showing the cavity verification section illustrated in FIG. 1. Figure 3 is a block diagram illustrating the image feature extraction unit illustrated in Figure 2. FIG. 4 is an illustrative diagram for explaining a method for extracting four directional component features according to one embodiment of the present invention. FIG. 5 is an example diagram showing a directional component binary image divided into a fixed area of ​​61 x 61 centered on the cavity location according to one embodiment of the present invention. FIG. 6 is an exemplary diagram illustrating a method for finally determining whether a cavity exists using a probability value according to an embodiment of the present invention. FIG. 7 is a flowchart illustrating a method for detecting boiler tube surface defects according to an embodiment of the present invention. Specific details for implementing the invention

[0025] Hereinafter, a boiler tube surface defect detection device and method according to an embodiment of the present invention will be described with reference to the attached drawings. In this process, the thickness of lines or the size of components shown in the drawings may be exaggerated for clarity and convenience of explanation. Furthermore, the terms described below are defined considering their functions in the present invention, and these may vary depending on the intention or convention of the user or operator. Therefore, the definitions of these terms should be based on the content throughout this specification.

[0026] The implementations described herein may be implemented, for example, as methods or processes, devices, software programs, data streams, or signals. Even if discussed only in the context of a single form of implementation (e.g., discussed only as a method), the implementation of the discussed features may also be implemented in other forms (e.g., devices or programs). Devices may be implemented in appropriate hardware, software, and firmware, etc. Methods may be implemented in devices such as processors, which generally refer to processing devices including, for example, computers, microprocessors, integrated circuits, or programmable logic devices. Processors also include communication devices such as computers, cell phones, portable / personal digital assistants ("PDAs"), and other devices that facilitate the communication of information between end-users.

[0028] FIG. 1 is a block diagram schematically illustrating a boiler tube surface defect detection device according to an embodiment of the present invention; FIG. 2 is a block diagram specifically illustrating a cavity verification unit illustrated in FIG. 1; FIG. 3 is a block diagram for explaining an image feature extraction unit illustrated in FIG. 2; FIG. 4 is an illustrative diagram for explaining a method for extracting four directional component features according to an embodiment of the present invention; FIG. 5 is an illustrative diagram illustrating a directional component binary image divided into a fixed area of ​​61 x 61 centered on the cavity location according to an embodiment of the present invention; and FIG. 6 is an illustrative diagram for explaining a method for finally determining whether a cavity exists using a probability value according to an embodiment of the present invention.

[0029] Referring to FIG. 1, a boiler tube surface defect detection device (100) according to one embodiment of the present invention may include an image receiving unit (110), a cavity detection unit (120), and a cavity verification unit (130).

[0030] The image receiving unit (110) receives image data acquired by an image sensor such as a camera, scanner, and microscope. At this time, the image data may be an image of the surface of a boiler tube.

[0031] The cavity detection unit (120) receives image data from the image receiving unit (110) and can detect a cavity in the image data using an object detection algorithm. Here, a cavity may refer to a defect, damage, etc.

[0032] That is, the cavity detection unit (120) can detect a cavity in the boiler tube surface image using an object detection algorithm. At this time, the object detection algorithm may be, for example, an algorithm such as Yolov3.

[0033] By using an object detection algorithm, it is possible to detect all objects in cracks formed by similar shapes (circular, elliptical, polygonal, etc.) or multiple cavities by training on objects classified as creep cavities by experts.

[0034] However, the reliability of object detection algorithms is low, as they often mistake objects detected using algorithms for carbons or dust that resemble creep cavities. To improve accuracy in such cases, the system must be improved by additionally considering expert analysis results indicating that actual creep cavities occur along the metal's grain lines or at the center of triple points.

[0035] Accordingly, the cavity verification unit (130) can determine whether there is a cavity by using the boiler tube surface image and the cavity detection image in which the cavity is detected by the cavity detection unit (120).

[0036] This cavity verification unit (130) may include an image preprocessing unit (132), an image feature extraction unit (134), a probability calculation unit (136), and a judgment unit (138).

[0037] The image preprocessing unit (132) can convert the boiler tube surface image and the cavity detection image into a grayscale image. The boiler tube surface image and the cavity detection image may be RGB images, and the image preprocessing unit (132) can convert the RGB images into a grayscale image.

[0038] The image feature extraction unit (134) can extract a directional component image by performing a convolution operation on the grayscale image converted by the image preprocessing unit (132) and a directional filter.

[0039] To determine whether a cavity exists on the grain line, a filter capable of effectively extracting directional features of the image must be designed and selected.

[0040] Meanwhile, the surface image of the boiler tube exhibits a characteristic where grain line components become blurred depending on the high temperature and operating time. Through the analysis of such characteristics, the present invention can extract four directional component features after designing four directional filters. For example, four directional component features can be extracted using directional filters of 0 degrees, 45 degrees, 90 degrees, and 135 degrees.

[0041] The image feature extraction unit (134) can extract directional component features for the boiler tube surface image using four directional filters.

[0042] This image feature extraction unit (134) may include a feature extraction module (134a) and a binary image conversion module (134b).

[0043] The feature extraction module (134a) can extract four directional components through convolution operations between four directional filters and a grayscale image.

[0044] A method for extracting four directional component features by a feature extraction module (134a) will be explained with reference to FIG. 4. The feature extraction module (134a) can perform convolution operations on a grayscale image such as (a) of FIG. 4 and directional filters such as (b) at 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively. Then, the feature extraction module (134a) can extract features such as (c) at 0 degrees, 45 degrees, 90 degrees, and 135 degrees, respectively.

[0045] As described above, in order to calculate the probability values ​​of the four directional components extracted by the convolution operation of the grayscale image and the directional filter, it is necessary to convert the grayscale image into a binary image.

[0046] Accordingly, the binary image conversion module (134b) can convert a grayscale image from which four directional components have been extracted into a binary image. At this time, the binary image conversion module (134b) can convert the grayscale image into a binary image by using an adaptive threshold in which the threshold value changes according to the region of the grayscale image.

[0047] Generally, a fixed threshold is used to convert a grayscale image into a binary image, but since boiler equipment images generate noise such as blur, it is not easy to convert the grayscale image into a binary image using a fixed threshold. For this reason, the binary image conversion module (134b) can use an adaptive threshold in which the threshold changes depending on the region, rather than a fixed threshold.

[0048] The probability calculation unit (136) can calculate a probability value for determining the boiler tube defect state (whether a cavity exists) using the binary image converted by the binary image conversion module (134b). At this time, the probability calculation unit (136) can crop a certain area of ​​the binary image centered on each cavity location and calculate a directional component-based probability value for each.

[0049] Specifically, the probability calculation unit (136) divides the directional component binary image into N x N fixed regions centered on each cavity location, calculates the number of white pixels in each divided region, and calculates the calculated number of white pixels as a probability value for the grain line. At this time, the probability calculation unit (136) can divide the directional component binary image-based cavity into fixed regions such as 51 x 51, 55 x 55, 61 x 61, 65 x 65, etc.

[0050] Preferably, the probability calculation unit (136) can divide the 0-degree directional component binary image, the 45-degree directional component binary image, the 90-degree directional component binary image, and the 135-degree directional component binary image, respectively, into a fixed area of ​​61 x 61 centered on the cavity, as shown in FIG. 5.

[0051] If the crop size is set small, the area of ​​the grain line is small, which may affect defect detection performance. Also, if the crop size is set large, additional noise may be generated in addition to the directional component of the grain line. In other words, it may be sensitive to image noise. Accordingly, the present invention divides the cavity into a fixed area of ​​61 x 61.

[0052] The probability calculation unit (136) can calculate the number of white pixels, which are directional components, in a 61 x 61 area as shown in FIG. 5. The reason for calculating white pixels is to determine how much each directional component is included in the four directional binary images. That is, in a binary image of a certain area centered on each cavity, a large number of directional components means that the cavity exists on the grain line in that area. Conversely, a small number of directional components means that the probability of the cavity existing on the grain line is low. Accordingly, in the present invention, the cavity is defined as existing on the grain line, and the Not Cavity is defined as not existing on the grain line.

[0053] The determination unit (138) can determine whether there is a cavity by comparing the probability value calculated by the probability calculation unit (136) with a preset threshold value (e.g., 30%). At this time, the determination unit (138) can compare the probability value for each region with the threshold value, and if the probability value is greater than or equal to the threshold value, it can make a final determination that it is a cavity.

[0054] For example, let us explain the case where a probability value as in (b) is calculated from a directional component binary image as in (a) of FIG. 6. In this case, it can be seen that the probability values ​​of the 0-degree directional component binary image, the 45-degree directional component binary image, and the 135-degree directional component binary image are greater than or equal to the threshold value of 30%, and the 90-degree directional component binary image is less than the threshold value of 30%. Therefore, the judgment unit (138) can make a final determination that the 0-degree directional component binary image, the 45-degree directional component binary image, and the 135-degree directional component binary image are cavities, and the 90-degree directional component binary image is not a cavity.

[0056] FIG. 7 is a flowchart illustrating a method for detecting boiler tube surface defects according to an embodiment of the present invention.

[0057] Referring to FIG. 7, the cavity detection unit (120) detects a cavity in a boiler tube surface image using an object detection algorithm (S710). At this time, the object detection algorithm may be, for example, an algorithm such as Yolov3. The cavity detection unit (120) can transmit the boiler tube surface cavity detection image in which the cavity is detected and the original image to the cavity verification unit (130).

[0058] When step S710 is performed, the cavity verification unit (130) converts the boiler tube surface cavity detection image and the original image into a grayscale image (S720), and extracts a directional component image by performing a convolution operation between the converted grayscale image and a directional filter (S730). At this time, the cavity verification unit (130) can extract directional component features for the boiler tube surface image using four-directional directional filters.

[0059] When step S730 is performed, the cavity verification unit (130) converts the gray image from which four-directional components have been extracted into a binary image (S740). At this time, the cavity verification unit (130) can convert the gray image into a binary image using an adaptive threshold in which the threshold value changes for each region of the gray image.

[0060] When step S740 is performed, the cavity verification unit (130) crops a certain area of ​​the binary image centered on each cavity location and calculates a directional component-based probability value for each (S750). That is, the cavity verification unit (130) divides the directional component binary image centered on each cavity location into a certain area of ​​N x N, calculates the number of white pixels in each divided area, and can calculate the calculated number of white pixels as a probability value for the grain line.

[0061] When step S750 is performed, the cavity verification unit (130) determines whether the calculated probability value is greater than or equal to a preset threshold value (S760).

[0062] As a result of the judgment in step S760, if the probability value is greater than or equal to the threshold value, the cavity verification unit (130) makes a final determination that it is a cavity (S770), and if the probability value is not greater than or equal to the threshold value, it makes a determination that it is not a cavity (S780).

[0064] As described above, the boiler tube surface defect detection device and method according to one embodiment of the present invention can rapidly process a large amount of image data by automatically detecting cavities in boiler equipment and can obtain consistent and highly reliable results in determining the degree of metal defect occurrence by reducing false detections that may occur due to judgments based on the experience of existing experts.

[0065] The boiler tube surface defect detection device and method according to one embodiment of the present invention has the effect of improving the cavity detection performance of a boiler facility by reflecting the condition that 'the cavity must be located above the grain line or triple point,' which is difficult to determine solely by object detection technology.

[0066] Although the present invention has been described with reference to the embodiments illustrated in the drawings, this is merely illustrative, and those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible therefrom.

[0067] Therefore, the true technical scope of protection of the present invention should be determined by the following patent claims. Explanation of the symbols

[0068] 110 : Video receiver 120 : Cavity detection unit 130 : Cavity Verification Unit 132 : Image Preprocessing Unit 134 : Image Feature Extraction Unit 136 : Probability Calculation Unit 138 : Judgment Division

Claims

Claim 1 The system includes: an image receiving unit for receiving a boiler tube surface image; a cavity detection unit for detecting cavities by applying an object detection algorithm to the boiler tube surface image; and a cavity verification unit for determining the presence or absence of a cavity using the boiler tube surface image and the cavity detection image in which a cavity is detected by the cavity detection unit, wherein the cavity verification unit includes: an image preprocessing unit for converting the boiler tube surface image and the cavity detection image into a grayscale image; an image feature extraction unit for extracting directional components by performing a convolution operation between the grayscale image and a directional filter; a probability calculation unit for calculating a probability value based on the directional components by cropping a certain area of ​​the image from which the directional components are extracted centered on the cavity location; and a determination unit for determining the presence or absence of a cavity by comparing the calculated probability value with a preset threshold, wherein the image feature extraction unit includes a feature extraction module for extracting four directional components through a convolution operation between four directional filters and the grayscale image. A boiler tube surface defect detection device characterized by including a binary image conversion module that converts a grayscale image from which the four directional components are extracted into a binary image. Claim 2 delete Claim 3 A boiler tube surface defect detection device according to claim 1, characterized in that the four directional filters are directional filters of 0 degrees, 45 degrees, 90 degrees, and 135 degrees. Claim 4 A boiler tube surface defect detection device according to claim 1, wherein the binary image conversion module converts the gray image into a binary image using an adaptive threshold in which the threshold value changes according to region of the gray image. Claim 5 A boiler tube surface defect detection device according to claim 1, wherein the probability calculation unit divides each directional component binary image converted by the binary image conversion module into a fixed area of ​​N x N centered on the cavity location, calculates the number of white pixels in each divided area, and calculates the calculated number of white pixels as a probability value. Claim 6 A boiler tube surface defect detection device according to claim 5, wherein the judgment unit compares the probability value for each region with the threshold, and if the probability value is greater than or equal to the threshold, makes a final determination that it is a cavity. Claim 7 A method for detecting defects on a boiler tube surface, characterized by comprising: a step in which a cavity detection unit detects a cavity by applying an object detection algorithm to an image of the boiler tube surface; a step in which a cavity verification unit converts the boiler tube surface image and the cavity detection image into a grayscale image; a step in which the cavity verification unit extracts four directional components through a convolution operation between four directional filters and the grayscale image; a step in which the cavity verification unit converts the grayscale image from which the four directional components have been extracted into a binary image; a step in which the cavity verification unit crops a certain area of ​​the binary image centered on the cavity location to calculate a probability value based on the directional components; and a step in which the cavity verification unit determines the presence or absence of a cavity by comparing the calculated probability value with a preset threshold. Claim 8 delete Claim 9 A method for detecting surface defects in a boiler tube according to claim 7, characterized in that the four directional filters are directional filters of 0 degrees, 45 degrees, 90 degrees, and 135 degrees. Claim 10 A boiler tube surface defect detection method according to claim 7, wherein in the step of converting to a binary image, the cavity verification unit converts the gray image into a binary image using an adaptive threshold in which the threshold value changes according to the region of the gray image. Claim 11 A method for detecting defects on a boiler tube surface according to claim 7, wherein, in the step of calculating the probability value, the cavity verification unit divides each directional component binary image centered on the cavity location into a fixed area of ​​N x N, calculates the number of white pixels in each divided area, and calculates the calculated number of white pixels as a probability value. Claim 12 A method for detecting defects on the surface of a boiler tube according to claim 11, wherein, in the step of determining the presence or absence of the cavity, the cavity verification unit compares the probability value for each region with the threshold value, and if the probability value is greater than or equal to the threshold value, it is finally determined to be a cavity.

Images