CN114994080A - Method and device for detecting surface defects on outer ring of cylindrical fuel rod bundle - Google Patents

Abstract

The invention discloses a method for detecting surface defects of the outer ring of a cylindrical fuel bundle, comprising: collecting the surface image of the end plate at the top of the fuel bundle, coding and positioning the fuel rod; acquiring the inclination of the fuel bundle according to the surface image of the end plate parameters and adjust the fuel bundle to make it in a vertical state; collect the surface image of the outer ring of the fuel bundle to determine the code of the fuel rod and the position of the surface defect in the image; obtain the surface defect area of the fuel rod one by one according to the coding sequence of the fuel rod. The contour point cloud image is analyzed to determine the defect level; a device is also disclosed, which includes: a visual coding module, an image processing module, an adjustment module, an outer ring image acquisition module, and a defect 3D acquisition module. The invention locates the defect positions on the fuel rods with different codes by coding the fuel rods on the fuel rod bundle, and further realizes the rapid and high-precision detection of the surface defects of the outer ring of the fuel rod bundle through 3D non-contact imaging analysis. .

Description

Method and device for detecting surface defects on outer ring of cylindrical fuel rod bundle

technical field

The invention relates to the technical field of fuel rod detection, in particular to a method and a device for detecting surface defects of the outer ring of a cylindrical fuel rod bundle.

Background technique

Nuclear fuel rods are the first safety barrier of a nuclear reactor and play a vital role in preventing nuclear leakage. Excessive defects on the surface of nuclear fuel rod cladding and welding seam may cause damage to nuclear fuel rods, which directly affects the safe operation of nuclear power plant reactors. At present, the surface inspection method of fuel rods is usually manual visual inspection. If defects are found, the depth of the damage is measured with a microscope. The defect level can only be given by referring to the visual comparison of the target sample, and the traditional three-coordinate measurement method has poor flexibility and measurement accuracy. Affected by artificial factors, the deviation is large, and it is difficult to place the cylindrical fuel bundles smoothly on the marble platform, and the measurement benchmark and specific size calculation program are also difficult to determine, resulting in low efficiency and easy to miss detection. In addition to manual visual inspection, ultrasonic and eddy current inspection techniques are also used to detect nuclear fuel rod defects, but they are usually the detection of defects inside the cladding. X-rays can also be used to detect defects in nuclear fuel rods, which are usually to detect pores, penetration, etc. inside the weld.

"Spent fuel assembly measurement device and measurement method" disclosed in the Chinese patent document, its publication number is CN107063101A, and the publication date is 2017-08-18, including: two cameras installed on the support platform at a predetermined height position to measure the length data of the spent fuel assembly; the second measurement module connected to the support bench by moving up and down along the height direction of the support bench, including a plurality of linear displacement sensors arranged on the outer periphery of the spent fuel assembly to be measured and a linear displacement sensor The drive mechanism connected with the displacement sensor, the spent fuel assembly to be tested is fixed on the support base, a plurality of linear displacement sensors are used to measure the width data of the spent fuel assembly to be tested at different vertical positions; and the signal acquisition and processing system, respectively connected with The linear displacement sensor and the camera are connected to calculate the length, width and deformation of the spent fuel assembly to be tested according to the length data and a plurality of width data. The invention can efficiently and comprehensively detect the length of the fuel assembly, the length of the fuel rod, the width of the grid, the torsion and the bending degree of the fuel assembly. However, this technology can only be used to detect the dimensional parameters of fuel rods, and cannot be used to detect surface defects of fuel rods.

SUMMARY OF THE INVENTION

In order to overcome the lack of automatic detection of the surface defects of fuel rods in the prior art, and the low efficiency and large deviation of manual visual inspection of the surface defects of fuel rods, the invention provides a method for detecting the surface defects of the outer ring of cylindrical fuel rod bundles. The method and device of the invention locates the defect positions on the fuel rods of different codes by coding the fuel rods on the fuel rod bundle, and further realizes the detection of the surface defects of the outer ring of the fuel rod bundle through 3D non-contact imaging analysis.

In order to achieve the above object, the present invention adopts the following technical solutions:

A method for detecting surface defects on an outer ring of a cylindrical fuel bundle, comprising:

S1. Collect the surface image of the end plate at the top of the fuel rod bundle, and encode and position the fuel rod; the fuel rod bundle includes several fuel rods;

S2, obtaining the tilt parameter of the fuel bundle according to the surface image of the end plate;

S3. Adjust the fuel bundles according to the inclination parameters, so that the fuel bundles are in a vertical state;

S4. Collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod in the image and the position of the surface defect;

S5. Acquire the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level.

In the present invention, the visual coding module is arranged above the fuel bundle to collect the surface image of the top end plate of the fuel bundle vertically downward; the outer ring image acquisition module is arranged on the side of the fuel bundle to collect the outer ring surface image of the fuel bundle; After analyzing the surface image of the outer ring to locate the defect, the defect 3D acquisition module is also set on the side of the fuel bundle to focus on collecting the contour point cloud image of the defect area to obtain more accurate defect parameters. The outer ring image acquisition module and the defect 3D acquisition module Only the space on the side of the fuel bundle does not collide. Each fuel rod on the fuel bundle can be coded by the surface image of the end plate of the fuel bundle, so that after the defect information is obtained by analyzing the surface image of the outer ring of the fuel bundle, the defect can be associated with the corresponding coded fuel rod. A pairing makes it easy for the defect 3D acquisition module to collect the point cloud image of the defect outline of the fuel rod according to the code of the fuel rod, which can accurately and quickly measure the defect information on each fuel rod and determine the defect level.

Preferably, the S1 includes the following steps:

S11. Real-time acquisition of an image of the end plate surface at the top of the fuel bundle;

S12, locate the center position of the end plate and the center position of the characters on the surface of the end plate according to the image of the end plate surface;

S13. Calculate the angle between the line connecting the center of the end plate to the center of the character and the horizontal direction of the image;

S14. Calculate the included angle between the line connecting the center of each fuel rod to the center of the end plate and the line connecting the center of the character to the center of the end plate;

S15, coding and positioning the fuel rod according to the included angle calculated in S13 and S14.

In the present invention, there will be characters marking the fuel bundles on the surface of the top end plates of the fuel bundles. The line extending from the center of the end plate to the center of the characters can be used as the starting reference line, and the fuel bundles can be aligned clockwise or counterclockwise. The fuel rods on the bundle are encoded; in addition, when the outer ring image acquisition module is used to collect the surface image of the outer ring of the fuel bundle, the outer ring image acquisition module and the fuel bundle need to rotate relative to each other, so the center of each fuel rod needs to be calculated. The included angle between the line connecting to the center of the end plate and the line connecting the center of the character to the center of the end plate, so that the code of the corresponding fuel rod in the image can be determined according to the angle of rotation after image acquisition.

Preferably, the step S2 specifically includes: converting the surface image of the end plate into a grayscale image, obtaining the numerical gradient of the grayscale value at equal intervals in the horizontal axis direction and the vertical axis direction in the grayscale image, and obtaining the numerical gradient of the grayscale value according to the horizontal direction of the grayscale image. The gradient change and the gradient change in the vertical direction are obtained to obtain the inclination angle and inclination direction of the end plate surface.

In the present invention, the fuel bundles are placed vertically when viewed from the surface, and the visual coding module is located directly above the fuel bundles, and the direction of the image acquisition is perpendicular to the horizontal plane; on the surface of the end plate, except for the characters in different colors, the rest of the It is the same. When the fuel bundle is inclined at a small angle, the top end plate surface of the fuel bundle has an inclined angle with the horizontal plane. At this time, the grayscale image of the end plate surface image collected by the visual coding module is because the end plate surface is different. The difference between the vertical distance and the visual coding module will cause the gray value on the grayscale image to gradually change. Therefore, the inclination angle and direction of the top end plate can be calculated by the image processing module according to the grayscale gradient of the grayscale image. The tilt angle and direction of the fuel bundles.

其中

x和y为像素点的空间坐标，f(x,y)是像素点的灰度、亮度或强度，R代表图像的区域，p和q分别代表行坐标和列坐标的次幂。Preferably, the position coordinates of the character center in S12 are the average of the coordinates of all pixel points in the rod bundle character area in the end plate surface image; the end plate center position coordinates (x ₀ , y ₀ ) satisfy

in

x and y are the spatial coordinates of the pixel, f(x, y) is the grayscale, brightness or intensity of the pixel, R represents the area of the image, and p and q represent the power of the row and column coordinates, respectively.

In the present invention, the position of the bar bundle character on the end plate and the center position of the end plate are located by the calculation of the image moment. The center position of the character is calculated according to the average value of the coordinates of all the character pixel points, and the center position of the end plate is It is calculated from the central moment constructed with the centroid of the target area of the image as the center. The calculation of the moment is the centroid of the point in the target area relative to the target area, and has nothing to do with the position of the target area, that is, it has translation invariance.

Preferably, according to the chord length formula, take the center position of the end plate and the center position of the character as the two end points to calculate the inclination angle of the line connecting these two points to obtain the angle α between the line connecting the center of the end plate and the center of the character and the horizontal direction of the image; Calculate the angle θ _i between the line connecting the center of any fuel rod to the center of the end plate and the horizontal direction of the image, then the angle between the line connecting the center of any fuel rod to the center of the end plate and the line connecting the center of the character to the center of the end plate is β _i =θ _i -α.

In the present invention, the line connecting the center position of the end plate and the center position of the character is used as the length of the hypotenuse of the triangle, and the projection of the line in the horizontal direction of the image and the projection in the vertical direction of the image can be used as two right-angled sides to form a right-angled triangle, then a right-angled triangle can be formed. According to the relationship between the side length and angle of the right triangle, the connecting line from the center of the end plate to the center of the character and the included angle in the horizontal direction of the image can be calculated; the connecting line from the center of the fuel rod to the center of the end plate and the center of the character to the end can be calculated by the same principle. The included angle of the line connecting the center of the plate; the value of these included angles ranges from 0 degrees to 360 degrees. If the calculated included angles are negative, add 360 degrees as the final result.

Preferably, the S4 includes the following steps:

S41, the fuel rod bundle is rotated in the direction of the fuel rod coding sequence;

S42, the outer ring image acquisition module collects the surface image of the corresponding fuel rod within the field of view;

S43, performing Kalman filtering processing on the fuel rod surface image;

S44. Acquire the position and parameter of defects on the surface of the fuel rod according to the filtered surface image of the fuel rod;

S45. Acquire a code corresponding to the fuel rod according to the rotation angle of the fuel rod bundle.

In the present invention, the fuel rod bundle is rotated successively according to the angle between the line connecting the center of the fuel rod to the center of the end plate and the line connecting the center of the character and the center of the end plate calculated before, so that the correspondingly encoded fuel rods are in the position of the outer ring image acquisition module. The obtained image always has a lot of noise or texture at the edge, so it is necessary to use the Kalman filter formula of the linear discrete system for filtering, and the filtered image is obtained to locate the defect. Location information and defect parameters on the image.

Preferably, the following steps are included in the S5:

S51, the defect 3D acquisition module collects the contour point cloud image of the defect area on the surface of the fuel rod;

S52, performing filtering processing on the contour point cloud image by using different edge detection operators in turn; the edge detection operators include Prewitt operator, Sobel operator and Roberts operator;

S53, calculate the identity matrix model according to the filtered contour point cloud image, and calculate the corresponding cost function to optimize; the identity matrix model is a complex or real number set arranged according to a rectangular array;

S54 , after the optimal unitity matrix model is obtained by optimization, calculate the three-dimensional size value of the defect with the minimum function of the corresponding unitity matrix model, and determine the defect level.

In the present invention, the position of the defect and the code of the fuel rod where the defect is located have been located by the surface image of the outer ring, so the corresponding encoded fuel rod can be directly rotated into the field of view of the defect 3D acquisition module to collect the contour point cloud image of the defect area, The image of the defect area can be collected more purposefully and accurately; the cost function in the present invention refers to the error between the number of all contour point cloud data and the projection that satisfy the identity matrix model, and the maximum value is obtained when the cost function obtains the minimum value. The optimal identity matrix model can be used to calculate the three-dimensional size of the defect according to the minimum function of the model.

A device for detecting surface defects on the outer ring of a cylindrical fuel bundle, comprising:

The visual coding module is used to collect the surface image of the end plate at the top of the fuel rod bundle, and to code and position the fuel rod;

The image processing module is used to obtain the tilt parameter of the fuel bundle according to the surface image of the end plate;

The adjustment module is used to adjust the fuel bundle according to the inclination parameter, so that the fuel bundle is in a vertical state;

The outer ring image acquisition module is used to collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod and the position of the surface defect in the image;

The defect 3D acquisition module is used to obtain the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level.

In the present invention, the visual coding module is arranged just above the vertical fuel bundle, and the image processing module is connected with the visual coding module to process the end plate surface image collected by the visual coding module; the adjustment module is set on the positive side of the vertical fuel bundle. The fuel rod bundle is fixed below and the vertical state of the fuel rod bundle is adjusted; the outer ring image acquisition module and the defect 3D acquisition module are arranged on the side of the fuel rod bundle, and both can move up and down in the vertical direction, and the outer ring image acquisition module The module and the defect 3D acquisition module are not coincident in spatial position.

The invention has the following beneficial effects: by coding the fuel rods on the fuel rod bundle, the defect positions on the fuel rods with different codes are located, and further through 3D non-contact imaging analysis, the surface defects of the outer ring of the fuel rod bundle can be detected. Fast, consistent and high-precision measurement; replaces manual visual inspection of fuel rod surface defects, avoiding the problems of easy missed inspection, low accuracy and low efficiency in the manual visual inspection process; through the grayscale image of the surface image of the end plate The gray-scale gradient calculation is used to obtain the tilt angle and direction of the fuel bundles, so as to adjust the vertical state of the fuel bundles to make the collected images more accurate and avoid the influence of the inclined fuel rods on defect detection.

Description of drawings

Fig. 1 is the flow chart of the outer ring surface defect detection method of the present invention;

FIG. 2 is a schematic diagram of encoding fuel rods in an embodiment of the present invention;

In the figure: 1. The connecting line from the center of the end plate to the center of the character; 2. The connecting line from the center of the fuel rod to the center of the end plate; 3. The fuel rod; 4. The character of the end plate.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

As shown in Figure 1, a method for detecting surface defects on the outer ring of a cylindrical fuel bundle includes:

S1, collect the surface image of the end plate at the top of the fuel rod bundle, and encode and position the fuel rod; the fuel rod bundle includes several fuel rods; S1 includes the following steps:

S11. The visual encoding module collects the surface image of the top end plate in real time directly above the fuel bundle;

S12, locate the center position of the end plate and the center position of the characters on the surface of the end plate according to the image of the end plate surface;

S13. Calculate the angle between the connection line from the center of the end plate to the center of the character and the horizontal direction of the image;

S14. Calculate the included angle between the line connecting the center of each fuel rod to the center of the end plate and the line connecting the center of the character to the center of the end plate;

S15, coding and positioning the fuel rod according to the included angle calculated in S13 and S14.

S2. Acquire the tilt parameters of the fuel bundles according to the surface image of the end plate; in S2, convert the surface image of the end plate into a grayscale image, and obtain the value of the grayscale value at equal intervals in the horizontal axis direction and the vertical axis direction in the grayscale image. Gradient, according to the gradient change in the horizontal direction and the gradient change in the vertical direction of the grayscale image, the inclination angle and inclination direction of the end plate surface are obtained.

S3. Adjust the fuel bundles according to the inclination parameters, so that the fuel bundles are in a vertical state.

S4, collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod in the image and the position of the surface defect; S4 includes the following steps:

S41, the fuel rod bundle is rotated in the direction of the fuel rod coding sequence;

S42, the outer ring image acquisition module collects the surface image of the corresponding fuel rod within the field of view;

S43, performing Kalman filtering processing on the fuel rod surface image;

S44. Acquire the position and parameter of defects on the surface of the fuel rod according to the filtered surface image of the fuel rod;

S45. Acquire a code corresponding to the fuel rod according to the rotation angle of the fuel rod bundle.

S5. Acquire the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level; S5 includes the following steps:

S51, the defect 3D acquisition module collects the contour point cloud image of the defect area on the surface of the fuel rod;

S52, performing filtering processing on the contour point cloud image by using different edge detection operators in turn; the edge detection operators include Prewitt operator, Sobel operator and Roberts operator;

S53, calculate the identity matrix model according to the filtered contour point cloud image, and calculate the corresponding cost function to optimize; the identity matrix model is a complex or real number set arranged according to a rectangular array;

S54 , after the optimal unitity matrix model is obtained by optimization, calculate the three-dimensional size value of the defect with the minimum function of the corresponding unitity matrix model, and determine the defect level.

A device for detecting surface defects on the outer ring of a cylindrical fuel bundle, comprising:

The visual coding module is used to collect the surface image of the end plate at the top of the fuel rod bundle, and to code and position the fuel rod;

The image processing module is used to obtain the tilt parameter of the fuel bundle according to the surface image of the end plate;

The adjustment module is used to adjust the fuel bundle according to the inclination parameter, so that the fuel bundle is in a vertical state;

The outer ring image acquisition module is used to collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod and the position of the surface defect in the image;

The defect 3D acquisition module is used to obtain the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level.

In the present invention, the visual coding module is arranged above the fuel bundle to collect the surface image of the top end plate of the fuel bundle vertically downward; the outer ring image acquisition module is arranged on the side of the fuel bundle to collect the outer ring surface image of the fuel bundle; After analyzing the surface image of the outer ring to locate the defect, the defect 3D acquisition module is also set on the side of the fuel bundle to focus on collecting the contour point cloud image of the defect area to obtain more accurate defect parameters. The outer ring image acquisition module and the defect 3D acquisition module Only the space on the side of the fuel bundle does not collide. Each fuel rod on the fuel bundle can be coded by the surface image of the end plate of the fuel bundle, so that after the defect information is obtained by analyzing the surface image of the outer ring of the fuel bundle, the defect can be associated with the corresponding coded fuel rod. A pairing makes it easy for the defect 3D acquisition module to collect the point cloud image of the defect outline of the fuel rod according to the code of the fuel rod, which can accurately and quickly measure the defect information on each fuel rod and determine the defect level.

In the present invention, there will be characters marking the fuel bundles on the surface of the top end plates of the fuel bundles. The line extending from the center of the end plate to the center of the characters can be used as the starting reference line, and the fuel bundles can be aligned clockwise or counterclockwise. The fuel rods on the bundle are encoded; in addition, when the outer ring image acquisition module is used to collect the surface image of the outer ring of the fuel bundle, the outer ring image acquisition module and the fuel bundle need to rotate relative to each other, so the center of each fuel rod needs to be calculated. The included angle between the line connecting to the center of the end plate and the line connecting the center of the character to the center of the end plate, so that the code of the corresponding fuel rod in the image can be determined according to the angle of rotation after image acquisition.

In the present invention, the fuel bundles are placed vertically when viewed from the surface, and the visual coding module is located directly above the fuel bundles, and the direction of image acquisition is perpendicular to the horizontal plane; on the surface of the end plate, except for the characters in different colors, the rest of the It is the same. When the fuel bundle is inclined at a small angle, the top end plate surface of the fuel bundle has an inclined angle with the horizontal plane. At this time, the grayscale image of the end plate surface image collected by the visual coding module is because the end plate surface is different. The difference between the vertical distance from the visual coding module and the visual coding module will cause the gray value on the grayscale image to gradually change. Therefore, the tilt angle and direction of the top end plate can be calculated according to the grayscale gradient of the grayscale image to obtain the fuel bundle. Tilt angle and direction.

In the present invention, the position of the bar bundle character on the end plate and the center position of the end plate are located by the calculation of the image moment. The center position of the character is calculated according to the average value of the coordinates of all the character pixel points, and the center position of the end plate is It is calculated from the central moment constructed with the centroid of the target area of the image as the center. The calculation of the moment is the centroid of the point in the target area relative to the target area, and has nothing to do with the position of the target area, that is, it has translation invariance.

In the present invention, the line connecting the center position of the end plate and the center position of the character is used as the length of the hypotenuse of the triangle, and the projection of the line in the horizontal direction of the image and the projection in the vertical direction of the image can be used as two right-angled sides to form a right-angled triangle, then a right-angled triangle can be formed. According to the relationship between the side length and angle of the right triangle, the connecting line from the center of the end plate to the center of the character and the included angle in the horizontal direction of the image can be calculated; the connecting line from the center of the fuel rod to the center of the end plate and the center of the character to the end can be calculated by the same principle. The included angle of the line connecting the center of the plate; the value of these included angles ranges from 0 degrees to 360 degrees. If the calculated included angles are negative, add 360 degrees as the final result.

In the present invention, the fuel rod bundle is rotated successively according to the angle between the line connecting the center of the fuel rod to the center of the end plate and the line connecting the center of the character and the center of the end plate calculated before, so that the correspondingly encoded fuel rods are in the position of the outer ring image acquisition module. The obtained image always has a lot of noise or texture at the edge, so it is necessary to use the Kalman filter formula of the linear discrete system for filtering, and the filtered image is obtained to locate the defect. Location information and defect parameters on the image.

In the present invention, the position of the defect and the code of the fuel rod where the defect is located have been located by the surface image of the outer ring, so the corresponding encoded fuel rod can be directly rotated into the field of view of the defect 3D acquisition module to collect the contour point cloud image of the defect area, The image of the defect area can be collected more purposefully and accurately; the cost function in the present invention refers to the error between the number of all contour point cloud data and the projection that satisfy the identity matrix model, and the maximum value is obtained when the cost function obtains the minimum value. The optimal identity matrix model can be used to calculate the three-dimensional size of the defect according to the minimum function of the model.

In the present invention, the visual coding module is arranged just above the vertical fuel bundle, and the image processing module is connected with the visual coding module to process the end plate surface image collected by the visual coding module; the adjustment module is set on the positive side of the vertical fuel bundle. The fuel rod bundle is fixed below and the vertical state of the fuel rod bundle is adjusted; the outer ring image acquisition module and the defect 3D acquisition module are arranged on the side of the fuel rod bundle, and both can move up and down in the vertical direction, and the outer ring image acquisition module The module and the defect 3D acquisition module are not coincident in spatial position.

In the embodiment of the present invention, the cylindrical fuel bundle is first vertically placed on a rotatable adjustment module, which can adjust the inclination angle, drive the fuel bundle to rotate, and fix the fuel bundle. The fuel bundle is composed of several fuel rods around a certain center, and the top and bottom ends of the fuel bundle are respectively provided with top end plates and bottom end plates to fix the fuel rods, as shown in Figure 2 is the fuel bundle. The top view of the top end plate, a number of fuel rods 3 are surrounded to form fuel rod bundles, and there are end plate characters 4 on the end plate for marking the fuel rod bundles.

其中x和y为像素点的空间坐标，f(x,y)表示在xy平面内任意空间坐标(x,y)上的像素点的灰度、亮度或强度，R代表图像所在的区域Region，p和q分别代表行坐标和列坐标的次幂。对于字符中心位置坐标：是计算图像矩区域内所有字符所包含的像素点坐标的平均值，其中心点行坐标＝字符区域内所有像素点行坐标相加的和除以面积，中心点列坐标＝字符区域内所有像素点列坐标相加除以面积。对于端板中心位置坐标：坐标(x₀,y₀)满足

通过(p+q)阶矩m_pq中对p和q去不同的值得到的。After the fuel bundles are placed, the visual coding module is set directly above the fuel bundles. The image acquisition direction of the visual coding module is vertically downward from the horizontal plane, and the surface images of the end plates at the top of the fuel bundles are collected in real time. Then, the center position of the characters on the end plate and the center position of the end plate are located by the calculation of the image moment. Define the moment of order (p+q) as

Where x and y are the spatial coordinates of the pixel point, f(x,y) represents the grayscale, brightness or intensity of the pixel point on any spatial coordinate (x,y) in the xy plane, R represents the region where the image is located, p and q represent the powers of the row and column coordinates, respectively. For character center position coordinates: it is to calculate the average value of pixel coordinates contained in all characters in the image moment area, and its center point row coordinates = the sum of all pixel row coordinates in the character area divided by the area, and the center point column coordinates = The column coordinates of all pixel points in the character area are added and divided by the area. For the coordinates of the center position of the end plate: the coordinates (x ₀ , y ₀ ) satisfy

It is obtained by applying different values to p and q in the (p+q) moment m _pq .

对于经过A和B两点的直线其斜率为k，可以得到关系式

如图2所示记端板中心到字符中心的连线和图像水平方向的夹角为α时，可以得到关系式|AB|＝|x_z-x₀|/sinα＝|y_z-y₀|/cosα，从而可以根据已知的A点和B点的坐标求出夹角α的值。对于燃料棒中心到端板中心连线2与图像水平方向的夹角θ_i，其计算方式与夹角α的计算方式相同，对于任意燃料棒中心位置C(x_i,y_i)，可以得到类似的关系式|AC|＝|x_i-x₀|/sinθ_i＝|y_i-y₀|/cosθ_i，从而求出夹角θ_i。根据求解出的这两个夹角，可以得到任意一根燃料棒中心到端板中心连线2和字符中心到端板中心连线1的夹角为β_i＝θ_i-α。在本实施例中由于燃料棒是等间隔环绕形成的燃料棒束，所以相邻两根燃料棒中心到端板中心连线的夹角是等间隔的，其值Δβ＝β_i-β_i-1＝θ_i-θ_i-1，在图中从端板中心到字符中心连线的延伸线通过某一根燃料棒，以该燃料棒作为初始的编码燃料棒b1，按照顺时针方向每间隔Δβ角度对下一根燃料棒进行编码，得b1、b2、…、bi、…bn，可以知道在图中每个燃料棒相对端板字符的位置都是固定的，因此可以通过拍摄不同时刻顶端端板的图像，获取角度α的变化来对不同时刻的每一个编码燃料棒进行定位。After obtaining the character center position coordinates A (x _z , y _z ) and the end plate center position coordinates B (x ₀ , y ₀ ), calculate these two points according to the chord length formula with the end plate center position and the character center position as the two end points The inclination angle of the connecting line can obtain the angle α between the connecting line 1 from the center of the end plate to the center of the character and the horizontal direction of the image; the distance between the two points A and B is

For a straight line passing through points A and B whose slope is k, the relation can be obtained

As shown in Figure 2, when the angle between the line connecting the center of the end plate to the center of the character and the horizontal direction of the image is α, the relational expression |AB|=|x _z -x ₀ |/sinα=|y _z -y ₀ |/cosα, so that the value of the included angle α can be obtained according to the known coordinates of point A and point B. For the included angle θ _i between the line 2 from the center of the fuel rod to the center of the end plate and the horizontal direction of the image, the calculation method is the same as the calculation method of the included angle α. For any center position C(x _i , y _i ) of the fuel rod, we can obtain Similar relational expression |AC|=|x _i -x ₀ |/sinθ _i =|y _i -y ₀ |/cosθ _i , thereby obtaining the included angle θ _i . According to the two angles obtained, the angle between the line 2 from the center of the fuel rod to the center of the end plate and the line 1 from the center of the character to the center of the end plate can be obtained as β _i =θ _i -α. In this embodiment, since the fuel rods are formed by equidistantly surrounding fuel rod bundles, the included angle between the center of the adjacent two fuel rods and the center of the end plate is equally spaced, and its value is Δβ=β _i -β _{i- 1} = θ _i -θ _i-1 , in the figure, the extension line from the center of the end plate to the center of the character passes through a certain fuel rod, and this fuel rod is used as the initial coded fuel rod b1, and every interval in the clockwise direction is The Δβ angle encodes the next fuel rod to obtain b1, b2, ..., bi, ...bn. It can be known that the position of each fuel rod relative to the end plate characters in the figure is fixed, so it can be captured by shooting the top of the fuel rod at different times. The image of the end plate, the change of angle α is obtained to locate each coded fuel rod at different times.

After the coding and positioning of the fuel rods is completed, since the fuel rod bundles may have problems such as inclination when they are initially placed, they are not completely vertical, so they need to be leveled and straightened. The inclination angle and inclination direction of the fuel bundle can be obtained by detecting the inclination angle and inclination direction of the top end plate. For the inclination angle and inclination direction of the top end plate, the grayscale image of the surface image of the top end plate can be obtained. degree gradient to get. In the image processing module, the collected surface image of the end plate is converted into a grayscale image. In this embodiment, the color of the surface of the top end plate except the character part of the end plate is the same, so in the case where the top end plate is horizontal The grayscale image of the lower end plate surface image is uniform, and the grayscale gradient is zero. When the rod bundle is inclined, the vertical distance from each point on the top end plate to the visual encoding module is different. The smaller the vertical distance is, the larger the gray value on the grayscale image is, and the larger the vertical distance is in the The smaller the gray value on the grayscale image is, the numerical gradient of the grayscale value is calculated at equal intervals in the horizontal axis direction and the vertical axis direction in the grayscale image, and the grayscale gradient in the horizontal direction and the grayscale gradient in the vertical direction are obtained. , the direction of gray value reduction is the positive direction pointing to the inclined direction. According to the combination of the horizontal gray gradient vector and the vertical gray gradient vector, the gray gradient vector of the end plate is obtained. The direction of the vector is the inclination direction of the end plate, and the gray gradient vector The decreasing rate of change in degrees corresponds to the angle at which the end plate is tilted. After the fuel bundles are adjusted according to the calculated inclination direction and inclination angle of the end plates, the fuel bundles are kept in a vertical state, that is, the surface of the end plates is in a horizontal state.

其中

为状态量X_k的一步预报估计，

为状态量X_k的最优线性估计，K_k为增益矩阵可通过现有公式推导得到，Z_k为第k次的测量值是一个q维观测向量，D_k为一个q×m阶矩阵，U_k是一个m维控制向量，H_k是一个q×n阶矩阵。对于经过滤波处理后的外环表面图像，可以得到缺陷区域图像内所有像素点的坐标，该像素点的坐标可以采用之前端板表面图像相同的图像矩计算方式获得，在得到缺陷区域像素点的坐标后既可以定位得到缺陷位置坐标，同时可以根据垂直距离计算公式

计算出缺陷的垂直距离长度。同时由于之前可以找到图像中每个燃料棒对应的编码，因此可以将每根编码燃料棒与燃料棒表面的缺陷位置信息一一对应，便于后续缺陷3D采集模块对缺陷区域的定位采集。After the adjustment is completed, the fuel rod bundle is rotated by Δβ angle each time, so that all the fuel rods pass through the field of view of the outer ring image acquisition module in turn; The complete outer ring surface image corresponding to the encoded fuel rod within the range; after completing one acquisition, the fuel rod bundle continues to rotate by Δβ angle, and the outer ring image acquisition module collects the next outer ring surface image. Therefore, the corresponding codes can be found for all the fuel rods in the surface image of the outer ring according to the rotation angle. After the surface image of the outer ring is collected, the image needs to be filtered by the Kalman filter formula of the linear discrete system. The minimum mean square error is used as the best estimation criterion, and the state space model of signal and noise is used. The estimated value and the observed value at the current moment are used to update the estimation of the state variable, and the estimated value at the current moment is obtained. According to the established system equation and observation equation, the signal to be processed is estimated to meet the minimum mean square error. for

in

is the one-step forecast estimate of the state quantity X _k ,

is the optimal linear estimation of the state quantity X _k , K _k is the gain matrix that can be derived from the existing formula, Z _k is the k-th measurement value is a q-dimensional observation vector, D _k is a q × m order matrix, U _k is an m-dimensional control vector, and H _k is a q×n-order matrix. For the filtered outer ring surface image, the coordinates of all pixels in the image of the defect area can be obtained, and the coordinates of the pixel can be obtained by the same image moment calculation method as the front end plate surface image. After the coordinates, the coordinates of the defect position can be located, and the formula can be calculated according to the vertical distance.

Calculate the vertical distance length of the defect. At the same time, since the code corresponding to each fuel rod in the image can be found before, each coded fuel rod can be in a one-to-one correspondence with the defect position information on the surface of the fuel rod, which is convenient for the subsequent defect 3D acquisition module to locate and collect the defect area.

Sobel:

Roberts:

Prewitt算子对左右点的单点噪声不太敏感，将自身考虑乘0，左右两侧分别是-1和1；Sobel算子将矩阵拆分为两个[1,2]^T和[-1,0]，前面的是高斯核，后面的是边缘提取，即在边缘提取之前先做平滑处理；Roberts算子中Mx检测的是135度方向的线，而My检测的是45度方向的线，检测模块考虑的方向是和[0,0]垂直的方向。在进行滤波处理后再通过OpenCV中滤除误匹配对采用RANSAC算法寻找最佳单位性矩阵。单位性矩阵是按照长方阵列排列的复数或实数集合，由于单位性矩阵有8个未知参数，至少需要8个线性方程求解，对应到数据点位置信息上，一组点对可以列出两个方程，则至少需要包含4组匹配点对。因此从轮廓点云图像的所有数据点中随机抽出4个样本并保证这4个样本之间不共线，计算出对应的单位性矩阵，然后利用这个单位性矩阵模型测试轮廓点云图像中的所有数据，并计算满足这个单位性矩阵模型的所有数据点的个数与投影误差，即该单位性矩阵的代价函数。用于确定函数在定义域内的值域边界，为最大值或者最小值。在进行优化后当代价函数取到最小值时，对应的单位性矩阵为最佳单位性矩阵，此时得到单位性矩阵模型的最小函数

公式中的h为数值参数取值，用来对单位性矩阵进行归一化操作。从满足最佳单位性矩阵模型的数据点中可以计算出对应缺陷的三维尺寸数据，根据计算出的三维尺寸数据可以对缺陷等级进行划分。如按照一根燃料棒上存在的缺陷总面积、缺陷的凹陷深度或缺陷的凸起高度等划分缺陷等级。The fuel rod bundle is rotated into the field of view of the defect 3D acquisition module according to the code sequence of the fuel rods. In the field of view, the contour point cloud image of the surface defect area is collected. In the present embodiment, the contour point cloud image is filtered using the Prewitt operator, the Sobel operator and the Roberts operator in turn, wherein Prewitt:

Sobel:

Roberts:

The Prewitt operator is not very sensitive to the single-point noise of the left and right points, and multiplies itself by 0, and the left and right sides are -1 and 1 respectively; the Sobel operator splits the matrix into two [1,2] ^T and [-1 ,0], the front is the Gaussian kernel, and the latter is the edge extraction, that is, smoothing is performed before the edge extraction; Mx in the Roberts operator detects the line in the 135-degree direction, while My detects the line in the 45-degree direction , the direction considered by the detection module is the direction perpendicular to [0,0]. After filtering, the RANSAC algorithm is used to find the best identity matrix by filtering out the false matching in OpenCV. The identity matrix is a set of complex or real numbers arranged in a rectangular array. Since the identity matrix has 8 unknown parameters, at least 8 linear equations are required to solve, corresponding to the position information of the data points, a set of point pairs can list two equation, it needs to contain at least 4 sets of matching point pairs. Therefore, 4 samples are randomly selected from all the data points of the contour point cloud image and ensure that the four samples are not collinear, the corresponding identity matrix is calculated, and then the identity matrix model is used to test the contour point cloud image. All data, and calculate the number and projection error of all data points that satisfy the identity matrix model, that is, the cost function of the identity matrix. It is used to determine the range boundary of the function within the definition domain, which is the maximum or minimum value. After optimization, when the cost function takes the minimum value, the corresponding identity matrix is the optimal identity matrix, and the minimum function of the identity matrix model is obtained at this time.

The h in the formula is the value of the numerical parameter, which is used to normalize the identity matrix. The three-dimensional dimension data of the corresponding defect can be calculated from the data points satisfying the optimal identity matrix model, and the defect grades can be divided according to the calculated three-dimensional dimension data. For example, the defect grades are divided according to the total area of defects existing on a fuel rod, the depth of the depression of the defect or the height of the protrusion of the defect.

The above-mentioned embodiments are further elaboration and description of the present invention for the convenience of understanding, and are not intended to limit the present invention. Any modification, equivalent replacement and improvement made within the spirit and principle of the present invention shall be included in within the scope of protection of the invention.

Claims (9)

1. A method for detecting defects on the outer ring surface of a cylindrical fuel bundle, comprising: S1. Collect the surface image of the end plate at the top of the fuel rod bundle, and encode and position the fuel rod; the fuel rod bundle includes several fuel rods; S2, obtaining the tilt parameter of the fuel bundle according to the surface image of the end plate; S3. Adjust the fuel bundles according to the inclination parameters, so that the fuel bundles are in a vertical state; S4. Collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod in the image and the position of the surface defect; S5. Acquire the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level. 2. The method for detecting surface defects of the outer ring of a cylindrical fuel rod bundle according to claim 1, wherein the S1 comprises the following steps: S11. Real-time acquisition of an image of the end plate surface at the top of the fuel bundle; S12, locate the center position of the end plate and the center position of the characters on the surface of the end plate according to the image of the end plate surface; S13. Calculate the angle between the line connecting the center of the end plate to the center of the character and the horizontal direction of the image; S14. Calculate the included angle between the line connecting the center of each fuel rod to the center of the end plate and the line connecting the center of the character to the center of the end plate; S15, coding and positioning the fuel rod according to the included angle calculated in S13 and S14. 3. The method for detecting surface defects of the outer ring of a cylindrical fuel rod bundle according to claim 1 or 2, wherein the step S2 specifically comprises: converting the surface image of the end plate into a grayscale image, and using the grayscale The numerical gradient of the gray value is obtained at equal intervals in the horizontal and vertical axis directions of the image, and the inclination angle and inclination direction of the end plate surface are obtained according to the gradient change in the horizontal direction and the vertical direction of the gray image. 4 . The method for detecting surface defects on the outer ring of cylindrical fuel bundles according to claim 2 , wherein the coordinates of the character center position in the S12 are all pixels in the character area of the bundle in the end plate surface image. 5 . The average value of the coordinates; the coordinates of the center position of the end plate (x ₀ , y ₀ ) satisfy

in

x and y are the spatial coordinates of the pixel, f(x, y) is the grayscale, brightness or intensity of the pixel, R represents the area of the image, and p and q represent the power of the row and column coordinates, respectively. 5. A method for detecting surface defects of cylindrical fuel rod bundles according to claim 2 or 4, characterized in that, according to the chord length formula, the two points are calculated by taking the center position of the end plate and the center position of the character as the two end points The angle α between the line connecting the center of the end plate and the center of the character and the horizontal direction of the image is calculated by the inclination angle of the connecting line; the angle θ _i between the connecting line from the center of any fuel rod to the center of the end plate and the horizontal direction of the image is calculated in the same way. , then the included angle between the line connecting the center of any fuel rod to the center of the end plate and the line connecting the center of the character to the center of the end plate is β _i =θ _i -α. 6. The method for detecting surface defects on the outer ring of cylindrical fuel rod bundles according to claim 1 or 2, wherein the S4 comprises the following steps: S41, the fuel rod bundle is rotated in the direction of the fuel rod coding sequence; S42, the outer ring image acquisition module collects the surface image of the corresponding fuel rod within the field of view; S43, performing Kalman filtering processing on the fuel rod surface image; S44. Acquire the position and parameter of defects on the surface of the fuel rod according to the filtered surface image of the fuel rod; S45. Acquire a code corresponding to the fuel rod according to the rotation angle of the fuel rod bundle. 7. The method for detecting surface defects of cylindrical fuel rod bundles according to claim 1 or 2, wherein the S5 comprises the following steps: S51, the defect 3D acquisition module collects the contour point cloud image of the defect area on the surface of the fuel rod; S52, performing filtering processing on the contour point cloud image using different edge detection operators in turn; S53, calculate the identity matrix model according to the filtered contour point cloud image, and calculate the corresponding cost function for optimization; S54 , after the optimal unitity matrix model is obtained by optimization, calculate the three-dimensional size value of the defect with the minimum function of the corresponding unitity matrix model, and determine the defect level. 8 . The method for detecting surface defects on the outer ring of cylindrical fuel rod bundles according to claim 7 , wherein the edge detection operator comprises a Prewitt operator, a Sobel operator and a Roberts operator, and the unit A property matrix model is a collection of complex or real numbers arranged in a rectangular array. 9. A device for detecting surface defects on the outer ring of a cylindrical fuel bundle, suitable for the method according to claim 1 or 2 or 4 or 8, wherein the device comprises: The visual coding module is used to collect the surface image of the end plate at the top of the fuel rod bundle, and to code and position the fuel rod; The image processing module is used to obtain the tilt parameter of the fuel bundle according to the surface image of the end plate; The adjustment module is used to adjust the fuel bundle according to the inclination parameter, so that the fuel bundle is in a vertical state; The outer ring image acquisition module is used to collect the surface image of the outer ring of the fuel rod bundle, and determine the code of the fuel rod and the position of the surface defect in the image; The defect 3D acquisition module is used to obtain the contour point cloud images of the defect areas on the surface of the fuel rods one by one according to the coding sequence of the fuel rods, and analyze and determine the defect level.

Images