Billet element distribution analysis method and device, electronic equipment and storage medium

Abstract

This application provides a method, apparatus, electronic device, and storage medium for analyzing the elemental distribution of steel billets, belonging to the field of metallic material testing technology. The method includes: acquiring a visualized image of the elemental distribution corresponding to a target area of ​​a steel billet sample and the original surface point cloud data of the steel billet sample; dividing the target area into grids on the visualized elemental distribution image to obtain each target grid; extracting the original elemental content data of pixels in each target grid, using each target grid as an estimation block, and employing anisotropic kriging interpolation to determine the average elemental content corresponding to each target grid; determining the projection area corresponding to each target grid in three-dimensional space according to the target mapping relationship, and obtaining surface point cloud data containing elemental content information, using this surface point cloud data as a three-dimensional elemental distribution cloud map. This application can improve the accuracy, efficiency, and consistency of elemental distribution in steel billets.

Description

Technical Field

[0001] This application belongs to the field of metal material testing technology, and more specifically, relates to a method and apparatus for elemental distribution analysis of steel billets, electronic equipment, and storage medium. Background Technology

[0002] The steel industry is a pillar of the national economy, and the uniformity of element distribution in steel billets directly affects the performance of high-end products.

[0003] Traditional "from sample cutting to in-situ analysis" processes can generate elemental distribution cloud maps using in-situ analyzers. However, manually marking areas visually is prone to misalignment, leading to deviations between the boundaries and the actual physical boundaries, exacerbating quantification errors, and creating a conflict between efficiency and consistency in element location. Furthermore, traditional elemental content quantification methods typically use simple arithmetic averaging of pixels within a grid or employ interpolation methods that ignore spatial correlations, resulting in low accuracy in estimating the average elemental content and failing to accurately reflect the severity of segregation bands. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, electronic device, and storage medium for analyzing the elemental distribution of steel billets, which can improve the accuracy, efficiency, and consistency of elemental distribution. To achieve the above objective, the technical solution provided by this application is as follows: Firstly, a method for analyzing the elemental distribution of steel billets is provided, including: Obtain a visualized image of the elemental distribution corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample; The target mapping relationship is determined by comparing the pixel coordinate system of the element distribution visualization image with the three-dimensional spatial coordinate system of the original surface point cloud data. The target region is divided into grids on the element distribution visualization image to obtain each target grid; Extract the raw data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the average element content of each target grid using anisotropic kriging interpolation. Based on the target mapping relationship, the projection area corresponding to each target grid in three-dimensional space is determined, and the average content of each element is assigned to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information. This surface point cloud data is used as a three-dimensional element distribution cloud map, which is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

[0005] Secondly, a billet elemental distribution analysis device is provided, comprising: The data acquisition module is used to acquire a visualized image of the element distribution corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample. The mapping relationship determination module is used to determine the target mapping relationship based on the pixel coordinate system of the element distribution visualization image and the three-dimensional spatial coordinate system of the original surface point cloud data. The target grid determination module is used to divide the target area into grids on the element distribution visualization image to obtain each target grid; The element average content determination module is used to extract the raw data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the element average content corresponding to each target grid by using anisotropic kriging interpolation. The element content analysis module is used to determine the projection area of ​​each target grid in three-dimensional space according to the target mapping relationship, and assign the average content of each element to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information. This surface point cloud data is used as a three-dimensional element distribution cloud map, which is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

[0006] Thirdly, embodiments of this application also provide an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the billet element distribution analysis method provided by any possible implementation of the first aspect.

[0007] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the billet element distribution analysis method provided by any possible implementation of the first aspect.

[0008] The beneficial effects of the technical solution provided in this application are as follows: Compared with related technologies, the billet element distribution analysis method, apparatus, electronic device, and storage medium provided in this application fundamentally solve the problem of "artificial visual marker offset" mentioned in the background technology by establishing a mapping relationship between the pixel coordinate system and the three-dimensional spatial coordinate system and aligning the grid boundary with the physical boundary in the image. This reduces the element content quantification error caused by boundary misalignment. Based on grid division, anisotropic Kriging interpolation is introduced to determine the average element content of the grid, which can more realistically reflect the average element content of the overall situation of the target grid area, improving the accuracy and robustness of element content estimation. This embodiment also maps the average element content to a three-dimensional spatial point cloud, generating a brand-new three-dimensional element distribution cloud map, which can intuitively display the real distribution state of elements in the uneven area of ​​the sample surface, providing more comprehensive spatial information for analyzing segregation bands, inclusion distribution, etc. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below.

[0010] Figure 1 A schematic flowchart illustrating the elemental distribution analysis method for steel billets provided in this application embodiment; Figure 2 A structural block diagram of the billet element distribution analysis device provided in the embodiments of this application; Figure 3 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0011] The embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the embodiments described below with reference to the accompanying drawings are exemplary descriptions for explaining the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions of the embodiments of this application.

[0012] Those skilled in the art will understand that, unless otherwise stated, the singular forms “a,” “an,” and “the” used herein may also include the plural forms. It should be further understood that the terms “comprising” and “including” as used in embodiments of this application mean that the corresponding feature can be implemented as the presented feature, information, data, step, operation, element, and / or component, but do not exclude implementation as other features, information, data, step, operation, element, component, and / or combinations thereof supported by the art. It should be understood that when we say that an element is “connected” or “coupled” to another element, the one element can be directly connected or coupled to the other element, or it can mean that the one element and the other element establish a connection relationship through an intermediate element. Furthermore, “connected” or “coupled” as used herein can include wireless connection or wireless coupling. The term “and / or” as used herein indicates at least one of the items defined by the term; for example, “A and / or B” can be implemented as “A,” or as “B,” or as “A and B.” When describing multiple (two or more) items, if the relationship between the multiple items is not explicitly defined, the multiple items can refer to one, several or all of the multiple items. For example, the description of "parameter A includes A1, A2, A3" can be implemented as parameter A includes A1 or A2 or A3, or it can be implemented as parameter A includes at least two of the three items A1, A2 and A3.

[0013] This application provides a method for analyzing the elemental distribution of steel billets, which can be executed by electronic devices, such as... Figure 1 As shown, the method may include: S101: Obtain a visualized image of the element distribution corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample.

[0014] In this embodiment, the billet sample refers to a block-shaped sample cut from a continuously cast billet or rolled billet for elemental distribution analysis, and its size can be determined according to specific needs. The target region refers to the specific area on the surface of the billet sample that needs to be analyzed for elemental distribution. This region can be the entire sample surface, or it can be one or more regions of interest delineated based on prior knowledge (e.g., subcutaneous cracks or central segregation-prone areas in continuously cast billets).

[0015] In this embodiment, the billet sample to be analyzed is stably placed on the common working platform of the in-situ analyzer and the 3D scanner. To ensure the accuracy of subsequent data fusion, at least three high-contrast circular markers that are not on the same straight line can be pasted on the surface of the billet sample, or obvious geometric features of the sample itself (such as chamfers and holes) can be used as common reference points.

[0016] The in-situ elemental analysis equipment is activated via electronic control. Appropriate scanning parameters are set based on the type of element to be analyzed and the expected segregation scale. For example, if the main analysis focuses on macroscopic segregation in steel billets (segregation band widths are typically in the millimeter range), the scanning step size can be set to 100μm-200μm; if analyzing microscopic intergranular segregation, the step size can be reduced to 20μm-50μm. The in-situ elemental analysis equipment excites the target area of ​​the steel billet sample point-by-point according to a preset path (e.g., grating scanning), detecting the characteristic elemental spectra at each point, and converting the spectral intensity into elemental content values ​​using an internal algorithm. After scanning, one or more two-dimensional elemental distribution visualization images (corresponding to different elements) of the target area are automatically generated and stored.

[0017] In this embodiment, an electronic device controls the activation of a 3D scanning device to scan the same target area of ​​the same steel billet sample. Scanning parameters can be set according to the sample surface material (e.g., metallic reflectivity) and the required accuracy. The 3D scanning device projects structured light or laser onto the sample surface and, using a binocular camera or phase measurement technology, calculates the 3D coordinates of millions of spatial points on the sample surface, thereby generating raw surface point cloud data covering the entire target area. This raw surface point cloud data is typically exported in a standard point cloud format (e.g., PLY, PCD, ASC).

[0018] S102: Determine the target mapping relationship based on the pixel coordinate system of the element distribution visualization image and the three-dimensional spatial coordinate system of the original surface point cloud data.

[0019] In this embodiment, image processing techniques (e.g., grayscale centroid method, Hough circle detection, or corner detection) are used to locate each feature point in the element distribution visualization image and obtain its precise coordinates in the pixel coordinate system, with an accuracy reaching the sub-pixel level.

[0020] Identify the same set of feature points in the original surface point cloud data. Specifically, for the aforementioned marked points (or common reference points), the marked point regions are segmented using a point cloud reflection intensity threshold (e.g., intensity values ​​greater than 200 are considered marked points), and their centroid coordinates are calculated as three-dimensional coordinates; for unmarked points, the corresponding three-dimensional coordinates can be obtained through point cloud curvature analysis or feature fitting.

[0021] The precise coordinates in the pixel coordinate system are mapped one-to-one with the 3D coordinates according to the marker point numbers, forming a set of matching point pairs. Based on all matching point pairs, a mapping function is determined using a numerical fitting method (e.g., least squares method). During the fitting process, the coefficients of each term of the function are determined with the objective of minimizing the mapping error of all matching points, thereby obtaining the complete target mapping relationship. The mapping function in this embodiment is typically in polynomial form (e.g., a second-order polynomial), which can describe the nonlinear transformation relationship between pixel coordinates and 3D coordinates.

[0022] As can be seen from the above, this embodiment effectively eliminates positioning deviations caused by billet sample placement, imaging distortion, and scanning errors by arranging common reference features and automatically extracting their coordinates, and using numerical fitting methods to accurately solve the mapping relationship. Compared with manual coarse alignment, this embodiment significantly improves the registration accuracy between two-dimensional images and three-dimensional point clouds, providing a highly reliable coordinate transformation benchmark for subsequent mesh projection and three-dimensional element reconstruction, and ensuring the overall accuracy and consistency of billet element distribution analysis.

[0023] S103: Divide the target region into grids on the element distribution visualization image to obtain each target grid.

[0024] In this embodiment, the boundaries of each target grid are aligned with the physical boundaries of the billet sample in the element distribution visualization image.

[0025] In one embodiment of this application, the target region includes multiple sub-target regions. The target region is divided into grids on an element distribution visualization image to obtain each target grid, including: For each sub-target region, a corresponding initial mesh size is preset to form the initial basic mesh of the target region; the initial basic mesh consists of multiple initial basic mesh elements. Based on the prior geometric location of the segregation band in the billet sample, the segregation sensitive band is delineated in the initial basic grid, and the initial basic grid cells falling into the segregation sensitive band are refined to obtain the first grid set. Based on the first grid set, calculate the local gradient of the element distribution visualization image, and iteratively subdivide the grid cells in the first grid set whose local gradient values ​​exceed a preset gradient threshold to obtain the second grid set; The physical boundary of the billet sample in the element distribution visualization image is extracted based on the edge detection algorithm. All grid nodes in the second grid set that cross the physical boundary are adjusted to the physical boundary to obtain the adjusted grid node coordinates. Based on the adjusted grid node coordinates, a set of two-dimensional grid cells covering the target area is generated, which serves as each target grid.

[0026] In this embodiment, the initial mesh size refers to the side length of each mesh cell when dividing the initial basic mesh (in pixels). Segregation zone refers to a band-shaped region in the billet where element distribution is uneven due to the solidification process; common types include central segregation zone and subcutaneous segregation zone. Its location and width are closely related to continuous casting process parameters. Prior geometric location refers to the spatial range of possible segregation zones, pre-determined based on metallurgical process theory or historical testing data. Segregation sensitive zone refers to a band-shaped region defined in the initial basic mesh based on the prior geometric location; initial mesh cells falling into this region are considered to require priority subdivision to improve the resolution of segregation zones. Gradient threshold is used to determine whether mesh cells need further refinement, characterizing the degree of elemental variation within this region. Physical boundary refers to the actual contour line of the billet sample in the elemental distribution visualization image.

[0027] In this embodiment, the target area is divided into several sub-target areas according to the analysis purpose of the current scenario. For example, the cross-section of the continuously cast billet is divided into a central area, a middle area, and an edge area. For each sub-area, a corresponding initial mesh size is preset. Considering the possibility of segregation in the central area, middle area, and edge area, corresponding sizes are set separately. A mesh set composed of several rectangular elements is generated on the target area according to the corresponding size of each area, which is denoted as the initial basic mesh.

[0028] This embodiment can determine the a priori geometric location of the segregation zone based on metallurgical process experience, and identify the segregation sensitive zone in each initial base grid. For each initial base grid, all initial base grid cells are traversed, and cells falling within the sensitive zone are subdivided, that is, each cell is divided into four sub-cells in each of the two directions (the side length is halved). After refinement, a new grid set is obtained, called the first grid set.

[0029] This embodiment uses a differential operator to calculate the local gradient of the element distribution visualization image. For example, for each pixel, the difference in element content between it and its neighboring pixels is calculated to obtain the gradient magnitude. Each grid cell in the first grid set is traversed, and the average gradient value (or maximum gradient value) of all pixels within that grid cell is calculated. If the average value exceeds a preset gradient threshold (e.g., 0.8% per pixel), it is determined that the element changes within that cell are drastic and further subdivision is required. The cell is iteratively subdivided, with each iteration halving the cell's side length and recalculating the gradient value within the newly generated sub-cells, until the gradients of all sub-cells are below the preset gradient threshold or reach the minimum grid size limit (e.g., 5 pixels). The iteration is then stopped, and the grid set obtained after iterative subdivision is used as the second grid set.

[0030] This embodiment employs an edge detection algorithm (such as the Canny operator) to extract the physical boundary of the billet sample from the element distribution visualization image, obtaining a closed contour composed of a series of boundary pixels. All grid nodes in the second grid set are traversed, determining whether the grid cell containing each node crosses the physical boundary (i.e., part of the cell is inside the sample, and part is in the background). For cells that cross the boundary, their nodes located outside the boundary are adjusted to the nearest boundary point, ensuring that the adjusted node coordinates accurately fall on the physical boundary. After node adjustment, all grid cells are constrained within the sample, and the edges of the boundary cells coincide with the physical boundary.

[0031] In this embodiment, the specific method for determining whether the grid cell containing each node crosses a physical boundary can be as follows: First, traverse all mesh nodes. For each node, determine its location on the physical boundary using the classic "ray casting method" or "angle sum method," thus identifying whether it is an internal or external point. Specifically, launch a horizontal ray from the node to the right (or in any direction) and count the number of intersections between the ray and the physical boundary profile. If the number of intersections is odd, the node is located inside the sample; if the number of intersections is even, the node is located outside the sample (background area).

[0032] Secondly, for each mesh element (enclosed by four nodes), check the properties of its four nodes: if all four nodes are internal points, the element is completely inside the sample and does not cross the boundary, so no adjustment is needed; if all four nodes are external points, the element is completely in the background area and can be discarded or marked as an invalid element; if there are both internal and external points among the four nodes, the element crosses the physical boundary and node adjustment is required.

[0033] In this embodiment, a set of two-dimensional mesh cells covering the entire target area is regenerated based on the adjusted coordinates of all mesh nodes. Each mesh cell in this set is used as a target mesh, whose boundary is precisely aligned with the physical boundary. Furthermore, adaptive subdivision is achieved in different regions based on prior knowledge and local gradients.

[0034] As can be seen from the above, through the multi-criteria adaptive mesh generation method, this embodiment initially refines the segregation-sensitive region based on the prior geometric position to ensure that the key parts have sufficient spatial resolution; secondly, iterative subdivision of the region with drastic element changes is carried out by using the local gradient threshold, which can adaptively capture the segregation boundary and micro-uniformity features; thirdly, the mesh nodes are adjusted to the physical boundary by edge detection, which eliminates the calculation error caused by the mesh crossing the sample boundary.

[0035] The mesh generation method provided in this embodiment balances computational efficiency and analytical accuracy. It avoids redundant calculations that uniformly refine the entire region while ensuring high resolution in regions with complex element distribution. This provides high-quality statistical units for subsequent anisotropic kriging interpolation, thereby improving the accuracy and reliability of billet element distribution analysis.

[0036] In one embodiment of this application, after generating a set of two-dimensional mesh cells covering the target area as each target mesh, the method further includes: For each grid cell in the target grid, calculate the corresponding quality index based on the shape rules of that grid cell; If there are mesh cells with quality indicators lower than the preset quality threshold, they are marked as quality defect mesh cells; The defective mesh cells and their neighboring meshes are locally re-divided until the quality index of all mesh cells in the target mesh is not lower than the preset quality threshold. The target mesh is updated based on all the locally optimized mesh cells to obtain the updated target mesh, which is used to determine the average element content.

[0037] In this embodiment, the preset quality threshold refers to a pre-set critical value for judging whether a mesh cell is acceptable in terms of quality. If the quality index of a mesh cell is lower than this threshold, it is determined that the mesh cell has severe shape distortion and needs to be optimized. The threshold in this embodiment can be determined according to the analysis accuracy requirements and the mesh type. For example, for a quadrilateral mesh, if all interior angles are not less than 30° and not greater than 150°, the corresponding normalized quality threshold can be 0.5.

[0038] This embodiment iterates through all grid cells in the current target grid and calculates the corresponding quality index for each cell. Taking the smallest interior angle as an example, for a quadrilateral cell, the angle values ​​of the four interior angles are calculated sequentially (which can be obtained by the vector dot product of the node coordinates), and the minimum value is taken as the quality index of the cell.

[0039] In this embodiment, the minimum interior angle threshold is set to 30° (corresponding to a normalized quality value of 0.33), indicating that the minimum interior angle of all mesh elements is not less than 30°. By traversing all mesh elements, mesh elements with a minimum interior angle less than 30° are marked as quality defect mesh elements. If other quality indicators are also used (e.g., aspect ratio not exceeding 3), then mesh elements that simultaneously meet the minimum interior angle threshold and other quality indicators are considered quality qualified mesh elements; otherwise, they are marked as quality defect mesh elements.

[0040] For each defective mesh cell, it is merged with all adjacent mesh cells to form an irregularly shaped local region to be optimized. In this embodiment, "adjacent" refers to cells that share at least one mesh edge. The original mesh connections within this local region to be optimized are cleared, and the mesh is re-divided based on the polygonal boundary of this local region to generate new mesh cells with regular shapes and meeting quality standards, ensuring that the nodes of the new mesh remain continuous with those of the external adjacent meshes.

[0041] In this embodiment, after local re-division, the quality index of the newly generated mesh cells is recalculated. If there are still quality defective mesh cells with a quality threshold, the local area to be optimized is re-divided, or the quality defective mesh cell is merged with a wider neighborhood and then divided again. Through multiple iterations, all quality defective mesh cells are gradually eliminated. In this embodiment, optimization stops when the quality index of all mesh cells is not lower than the preset quality threshold.

[0042] This embodiment combines all the optimized mesh elements (including those not optimized and newly generated elements) to form an updated target mesh. This target mesh is the final mesh set used for subsequent calculations of elemental average content.

[0043] As can be seen from the above, this embodiment eliminates distorted mesh cells (such as extremely small interior angles and extremely large aspect ratios) by evaluating the quality and optimizing the initially divided mesh, thus avoiding numerical instability or error amplification in subsequent anisotropic kriging interpolation calculations due to poor cell shape. Secondly, it ensures that all mesh cells have good geometric regularity, so that each mesh has similar statistical characteristics when used as an estimation block, thereby improving the consistency and reliability of the element average content estimation. This embodiment also maintains computational efficiency while improving mesh quality through local optimization, avoiding redundant operations of re-subdividing the entire target region.

[0044] S104: Extract the raw data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the average element content corresponding to each target grid by using anisotropic kriging interpolation.

[0045] In this embodiment, for each target mesh, the pixel coordinates of its four nodes are known. Based on these four nodes, the quadrilateral region corresponding to the target mesh is determined. Each pixel within the bounding rectangle of the target mesh is traversed. A ray tracing method can be used to determine whether the pixel's center point is inside the quadrilateral. A horizontal ray is emitted to the right from the pixel's center point, and the number of intersections with the quadrilateral's sides is counted. If the number of intersections is odd, the pixel is determined to be inside the quadrilateral region; if the number of intersections is even, the pixel is determined to be outside the quadrilateral region.

[0046] In this embodiment, the element distribution visualization image can be stored in the electronic device as a two-dimensional array. The rows of this array correspond to the image height, and the columns correspond to the width. Each element value in the array corresponds to the element content value of a pixel. In this embodiment, based on the determined pixel coordinates, the corresponding element position in the array is directly accessed, and the stored element content value is read as the original element content data of the pixel.

[0047] In this embodiment, if the boundary of the target mesh passes through a pixel, the pixel center attribution method can be used to determine whether the center point of the pixel is located inside the quadrilateral, without considering the case where the pixel partially falls inside.

[0048] In one embodiment of this application, after extracting the original data of element content of pixels in each target grid, the method further includes: For each pixel, the local spatial neighborhood of the pixel is determined based on a preset neighborhood radius, with the pixel as the center. Obtain the absolute value of the difference between the original data of the element content of the pixel and the mean value of the element content of all pixels in the local spatial neighborhood, as well as the standard deviation of the element content in the local spatial neighborhood; If the absolute value of the difference is greater than the product of the standard deviation and the preset outlier threshold coefficient, the pixel is determined to be a discrete outlier and is removed. If the absolute value of the difference is not greater than the product of the standard deviation and the preset abnormal threshold coefficient, then the pixel is determined to be a normal pixel. The effective pixel set is constructed from all normal points, and the effective pixel set is used to determine the average element content of each target grid.

[0049] In this embodiment, the preset neighborhood radius refers to a parameter used to define the size of the local spatial neighborhood, and the unit is pixels. The preset anomaly threshold coefficient refers to a constant used to control the sensitivity of anomaly detection, and can be set according to the noise level and analysis accuracy requirements of the element content data.

[0050] In this embodiment, for each pixel in the element distribution visualization image, a circular neighborhood (or a square neighborhood) is determined based on the coordinates of that pixel (e.g., the current pixel is point P) in the image coordinate system, according to a preset neighborhood radius. This circular neighborhood includes all pixels whose Euclidean distance to the center is not greater than the preset neighborhood radius. If the circular neighborhood extends beyond the image boundary, only valid pixels within the image are considered, but the number of pixels in the neighborhood must be at least the minimum statistical sample size (e.g., no less than 10). Otherwise, the preset neighborhood radius can be appropriately increased or the pixel can be skipped (in practice, anomaly detection in image boundary regions can be handled separately or directly retained). In this embodiment, the range of the circular neighborhood is not limited to the target grid to which the current pixel belongs; it can cross grid boundaries to ensure the local spatial continuity of the statistics.

[0051] In this embodiment, the original data of element content of all pixels in the circular neighborhood are obtained, and their arithmetic mean and standard deviation are calculated.

[0052] For example, if there is a circular neighborhood N Each pixel has an element content value of [value]. If i = 1, 2, ..., N, then the arithmetic mean of the pixels within the circular neighborhood can be calculated as follows: ,in, This represents the arithmetic mean of all pixels within a circular neighborhood; the standard deviation of each pixel within this circular neighborhood can be calculated as follows: ,in, It represents the standard deviation of each pixel within the circular neighborhood.

[0053] This embodiment calculates the element content value of the current pixel (point P). The absolute value of the difference from the mean The absolute value of this difference is compared with the preset anomaly threshold coefficient. k Compare with the product of standard deviation: like > If point P is a discrete outlier, it is removed from the pixel list of its target grid (i.e., it will not be used again in subsequent statistics of that grid); if ≤ If the condition is met, then point P is considered a normal point and should be retained.

[0054] In this embodiment, after performing the above steps on all pixels, the set of pixels determined to be normal points within each target grid is taken as the effective pixel set. This set will be used as input data for subsequent anisotropic kriging interpolation to calculate the average element content of each grid.

[0055] As can be seen from the above, this embodiment effectively eliminates isolated outliers caused by measurement noise, surface contamination, or accidental interference by removing discrete outliers through local spatial neighborhood statistics, preventing them from misleading subsequent interpolation calculations and avoiding misjudging local noise as segregation features. This embodiment uses a dynamic threshold based on local mean and standard deviation, which can adapt to the fluctuations in element content in different regions and avoid misjudgment or omission caused by using a global fixed threshold. The effective pixel set formed by this embodiment retains the true distribution characteristics of the original data, providing a purer and more reliable data source for anisotropic Kriging interpolation, thereby improving the accuracy, robustness, and repeatability of billet element distribution analysis.

[0056] In one embodiment of this application, using each target grid as an estimation block, anisotropic kriging interpolation is employed to determine the average element content corresponding to each target grid, including: Based on the target mapping relationship, the pixel coordinates of all effective pixels in each target grid are converted into physical coordinates on the surface of the billet sample to obtain the physical position coordinates of each effective pixel and the corresponding element content data. Determine the anisotropic principal axis direction angle based on the solidification direction of the steel billet sample; Based on the anisotropic principal axis direction angle, the element content data of all effective pixels and their physical location coordinates, the anisotropic parameters of the variogram are obtained by fitting. For each target grid, based on the element content data of the effective pixels within the target grid, the average element content of the grid is calculated to obtain the average element content of the physical region corresponding to the target grid. The calculation of the average operation content of the grid includes: The target grid is used as the block to be estimated. Based on the anisotropy parameter of the variogram, the variogram value between every two valid pixels in the target grid and the average variogram value between each valid pixel and all valid pixels in the block to be estimated are calculated. Construct a block Kriging equation system based on the variogram values ​​and the average variogram values; Solve the block kriging equations to obtain the weights corresponding to each effective pixel in the target grid. The average element content of the target grid is determined based on the element content data of each effective pixel in the target grid and the weight corresponding to each effective pixel.

[0057] In this embodiment, for all valid pixels within each target grid, their pixel coordinates are converted into physical coordinates of the billet sample surface using the established target mapping relationship. Simultaneously, the elemental content data corresponding to each point is retained. Each converted valid pixel is used for subsequent variogram analysis and interpolation estimation.

[0058] This embodiment determines the anisotropic principal axis direction angle based on the solidification process and sampling direction of the steel billet. For billet samples taken from the cross-section of a continuously cast slab, the columnar crystal growth direction is usually perpendicular to the width of the slab, i.e., along the thickness direction. If a physical coordinate system is established with the width direction as the X-axis and the thickness direction as the Y-axis, then the principal axis direction angle... =90° (along the thickness direction). For billet samples taken from a longitudinal section (parallel to the drawing direction), the solidification direction may be parallel to the drawing direction. =0°. If the principal axis direction angle cannot be directly determined, it can be automatically identified by fitting anisotropic variograms: calculate the experimental variograms for multiple directions (e.g., 0°, 45°, 90°, 135°), select the direction with the smallest range as the principal axis direction (because the element changes the fastest along the solidification direction and the correlation distance is the shortest), or determine the principal axis direction angle by fitting an ellipse.

[0059] This embodiment fits a variogram model based on the physical coordinates and element content data of all valid pixels (global data or locally representative data can be used). Specifically, it includes: For a given principal axis direction angle Calculate the semivariance between all pixel pairs that are approximately spaced at a preset interval distance; select a suitable theoretical variogram model (e.g., a spherical model) and fit the model parameters using the least squares method. These model parameters may include the sill value (total variance), nugget constant (microscopic measurement error), and range parameters under anisotropic conditions.

[0060] For anisotropy, the range of the principal axis directions can be fitted separately. and the range of motion perpendicular to the principal axis ,pass and The ratio of these ratios determines the intensity of anisotropy. Anisotropy parameters are usually automatically fitted based on the data distribution characteristics, and the fitting results must meet the positive definiteness requirement of the model.

[0061] In this embodiment, for the target mesh (the block to be estimated) to be processed, the variogram value between each pair of effective pixels i and j can be determined based on the actual distance between each pair of effective pixels i and j and the angle between them and the principal axis direction, using the obtained variogram model. .

[0062] For each valid pixel i, determine the relationship between that point and the entire block to be estimated. V The average variogram value among (i.e., the physical region covered by the target mesh) Specifically, the block to be estimated is discretized into several sub-regions, and the average value of the variogram from pixel i to all points within the block to be estimated is determined. In this embodiment, the block to be estimated can be treated as a rectangular region, and the average variogram value can be calculated using numerical integration or an approximate formula (e.g., using the center point of the block to be estimated to represent the entire block).

[0063] Based on the determined variogram values ​​and average variogram values, a block kriging equation system is established, which can be expressed as: ,in, This represents the weight of each pixel to be determined. Represents the Lagrange multipliers. n This represents the total number of pixels in the block to be estimated.

[0064] Based on the weights and element content data of each pixel obtained from the solution, and using the formula for estimating the average element content, the estimated value of the average element content of the physical region corresponding to the target mesh is determined, and this estimated value is used as the average element content of the target mesh. In this embodiment, the formula for estimating the average element content can be: ,in, This represents the estimated average element content of the physical region corresponding to the target grid. This represents the original data of the element content of the j-th valid pixel.

[0065] Repeatedly traverse all target grids and perform the operation of calculating the average content of each grid in turn to obtain the average content of elements corresponding to each target grid.

[0066] As can be seen from the above, this embodiment estimates the average content of the grid by using anisotropic kriging interpolation, which fully considers the spatial anisotropic characteristics of element segregation during the solidification process of the billet (i.e., the change is drastic along the solidification direction and gradual perpendicular to the solidification direction), making the interpolation results more consistent with the actual physical metallurgical laws and avoiding the estimation bias caused by the isotropic assumption. This embodiment directly estimates the average value of the physical region where the target grid is located based on the block kriging method, rather than estimating the point values ​​first and then averaging them, which reduces the estimation variance and improves the accuracy of the average content calculation. This embodiment also utilizes the effective pixel set and variogram model to adaptively allocate weights, making the estimation result statistically optimal (unbiased and with the smallest variance), thereby effectively suppressing noise interference and improving the accuracy and reliability of element distribution analysis.

[0067] S105: Based on the target mapping relationship, determine the projection area corresponding to each target grid in three-dimensional space, and assign the average content of each element to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information, and use the surface point cloud data as a three-dimensional element distribution cloud map.

[0068] In this embodiment, the three-dimensional element distribution cloud map is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

[0069] In this embodiment, for each target mesh, the pixel coordinates of its four nodes are known. Based on the aforementioned target mapping relationship, the pixel coordinates of these four nodes are converted into three-dimensional spatial coordinates, obtaining the four corner points of the mesh in three-dimensional space. The spatial quadrilateral (which may be a curved surface) enclosed by the four corner points is used as the projection area of ​​the target mesh onto the surface of the three-dimensional point cloud. In this embodiment, the point cloud is a discrete set of points; therefore, this projection area is actually a spatial range.

[0070] The process iterates through all data points in the original surface point cloud data. For each data point, a parameterized method based on grid corners is used to determine whether it lies within the projection region corresponding to the current target grid, thus identifying all point cloud data points falling within the projection region. Specifically, the data points are projected onto the projection region, and their affiliation is determined by calculating the centroid coordinates of the data points within the projection region or by assessing the inclusion relationship between the points and the grid boundaries.

[0071] For all data points belonging to the current target grid, the average element content of the target grid is assigned as an attribute value to each data point. After assignment, each data point not only contains the original 3D coordinate information but also gains the element content attribute.

[0072] In this embodiment, the steps from determining the projection area to assigning the average element content to all data points within the projection area are repeated for all target grids until all grids have been assigned, thereby obtaining surface point cloud data covering the entire target area with element content information.

[0073] The obtained point cloud data containing elemental attributes is visualized and rendered. Different colors are assigned to each point based on its elemental content value (e.g., areas with elemental content higher than a preset value are displayed in red, and areas with content lower than a preset value are displayed in blue), thus generating a 3D elemental distribution cloud map. This cloud map can intuitively display the elemental content and spatial distribution characteristics at various 3D coordinate positions on the surface of the steel billet sample. The preset region content can be determined based on the scenario and experience.

[0074] As can be seen from the above, this embodiment accurately maps the average element content obtained from two-dimensional mesh statistics onto the surface of a three-dimensional point cloud, realizing the fusion of elemental chemical information and sample geometry. Compared with traditional methods, this mapping is based on precise coordinate transformation relationships, avoiding errors from manual alignment; at the same time, assigning values ​​using the target mesh as the unit retains the adaptive refinement advantage of mesh division while ensuring efficient generation of three-dimensional data.

[0075] In one embodiment of this application, after obtaining surface point cloud data with elemental content information, the method further includes: In the three-dimensional element distribution cloud map, multiple sampling lines are obtained along the rolling direction, width direction and thickness direction of the billet sample, respectively; For each sampling line, the element content values ​​of each sampling point passed through by the sampling line are extracted to form a sequence of element content values ​​along the line, and the element content gradient between adjacent sampling points is calculated; the sampling point is the spatial location point on the sampling line selected for reading the element content value. Based on a preset segregation gradient threshold, the spatial region corresponding to the continuous region in the element content value sequence where the gradient value exceeds the preset segregation gradient threshold is taken as a segregation band of the sampling line; wherein, one sampling line corresponds to zero or one or more segregation bands; Obtain the quantitative indicators for each segregation band, including the width, peak value, and segregation degree of the segregation band.

[0076] In this embodiment, the sampling line refers to a virtual straight line or curve set along a certain direction in the three-dimensional elemental distribution cloud map, used to extract the elemental content values ​​of each data point on the line or curve. The segregation gradient threshold is defined as the critical gradient value used to determine whether a sudden change in elemental content marks the boundary of a segregation zone. This threshold can be set according to the steel grade, process requirements, and measurement noise level.

[0077] In this embodiment, multiple sampling lines are set along the rolling direction, width direction, and thickness direction of the billet sample in the three-dimensional elemental distribution cloud map. The number of sampling lines can be determined based on the sample size and the uniformity of segregation distribution. The start and end points of each sampling line should cover the entire target area, and the sampling lines should be parallel to each other to comprehensively evaluate the segregation distribution. In this embodiment, the spatial path of the sampling lines can be generated by selecting two endpoints on the point cloud data and interpolating at equal intervals to create a series of sampling points. The spacing between each data point can be set to the average spacing of the point cloud (e.g., 0.1 mm) to maintain resolution.

[0078] For each sampling line, traverse each sampling point along it, find the nearest point cloud data point in the 3D element distribution cloud map (or obtain it through interpolation), and read the element content value of that point. Since the point cloud is discrete, nearest neighbor interpolation or radial basis interpolation can be used to obtain the content value at the sampling point. Arrange the content values ​​of all sampling points in order along the line to obtain the element content value sequence of the sampling line.

[0079] For each pair of adjacent data points in the element content value sequence, the difference in element content between the latter and former data points in the pair is compared. This difference is divided by the spatial distance between the two data points to determine the element content gradient value between the pair of adjacent data points. The element content value sequence is then determined based on each element content gradient value. In this embodiment, a higher element content gradient value indicates a more drastic change in element content at that location; a lower element content gradient value indicates a more gradual change in element content at that location.

[0080] This embodiment compares a preset segregation gradient threshold with the obtained element content value sequence. The entire element content value sequence is traversed, and all points where the element content gradient value exceeds the preset segregation gradient threshold are designated as segregation band boundary locations. If multiple consecutive points in the actual data have gradients exceeding the threshold, these consecutive points are merged, and their center position is taken as the accurate segregation band boundary. Each pair of adjacent start and end boundaries defines a segregation band. If no gradient on a sampling line exceeds the threshold, it is determined that there is no obvious segregation band in that direction.

[0081] For each identified segregation band, its width, peak value, and segregation degree are determined. In this embodiment, the width refers to the spatial distance between the start and end boundaries of the segregation band; the peak value refers to the maximum (if it is positive segregation) or minimum (if it is negative segregation) value of the element content among all sampling points within the segregation band. In this embodiment, the segregation degree is calculated as follows: the average element content of the adjacent regions on both sides of the segregation band (e.g., within a certain distance on both sides outside the band) is used as the matrix level, and the degree of deviation of the peak value from the matrix level is used as the segregation degree. This segregation degree (in percentage form) is used to characterize the severity of segregation.

[0082] As can be seen from the above, this embodiment achieves quantitative extraction and analysis of the elemental distribution characteristics of steel billets by setting sampling lines along multiple directions and automatically identifying segregation bands based on gradient thresholds. Compared with traditional visual observation and manual grading, this embodiment can accurately locate the spatial position and boundary of segregation bands and provide multi-dimensional quantitative indicators such as width, peak value, and segregation degree, making the assessment of segregation severity more scientific and precise. This guides the optimization of continuous casting process parameters and improves the uniformity and stability of the internal quality of the steel billet.

[0083] In one embodiment of this application, after obtaining the quantitative indicators of each segregation band, the method further includes: If the width of the segregation band exceeds the preset width threshold, or the segregation degree exceeds the preset segregation degree threshold, then the segregation degree of the current billet sample is determined to be beyond the allowable range. Based on the spatial location of the segregation band that exceeds the allowable range, determine the corresponding continuous casting process stage and generate a parameter adjustment strategy for the continuous casting process stage. The relationship between output parameter adjustment strategies and 3D element distribution cloud maps.

[0084] In this embodiment, the preset width threshold refers to the upper limit of the pre-set allowable width of the segregation band, expressed in millimeters (mm). This threshold can be determined according to the steel grade, product standards, and subsequent processing requirements. The preset segregation degree threshold refers to the upper limit of the pre-set allowable segregation degree, usually expressed as a percentage (%). This threshold can be determined according to the steel grade and its application.

[0085] In this embodiment, the width and segregation degree of each segregation band are compared one by one with preset width thresholds and segregation degree thresholds. If the width of any segregation band exceeds the preset width threshold, or its segregation degree exceeds the preset segregation threshold, the current billet sample is determined to have a segregation degree exceeding the allowable range, and is considered a non-conforming product or a product requiring warning. If the width and segregation degree of all segregation bands are within the corresponding preset threshold range, the product is determined to be a qualified product.

[0086] In this embodiment, for each non-conforming product or product requiring early warning, the continuous casting process stage that the segregation zone may correspond to is determined based on its spatial location in the three-dimensional element distribution cloud map (i.e., the coordinate region where the segregation zone is located), combined with the solidification characteristics of the continuous casting process and prior knowledge.

[0087] Specifically, if the segregation band is located in the central region of the cross-section of the billet sample, it may be related to the light reduction process at the end of solidification or the tendency of central segregation; if the segregation band is at a predetermined depth from the surface of the billet sample (e.g., 5 mm to 15 mm from the surface), it may be related to the liquid level fluctuation in the crystallizer, the entrainment of protective slag, or the early cooling intensity of the secondary cooling zone; if the segregation band exhibits a periodic distribution, it may be related to the electromagnetic stirring or vibration parameters of the crystallizer.

[0088] This embodiment can initially identify the process steps that need to be adjusted by mapping spatial location to process stages.

[0089] This embodiment generates a specific parameter adjustment strategy for each continuous casting process stage corresponding to each excessive segregation zone, taking into account the severity of segregation (e.g., width and degree of segregation) and the type of segregation (positive or negative segregation). This adjustment strategy includes at least the adjustment target, adjustment direction, and adjustment magnitude. The adjustment magnitude can be determined based on experience or similar cases in a process database.

[0090] Specifically, if the width of the central segregation zone exceeds the threshold, it is recommended to "increase the total reduction of the light pressing at the end of solidification by 1.5 mm" or "advance the starting position of the light pressing 2 m towards the direction of billet pulling"; if the segregation of the subcutaneous segregation zone is too high, it is recommended to "reduce the water volume of the first stage of the secondary cooling zone by 10% to reduce the regenerative effect" or "optimize the electromagnetic stirring frequency of the crystallizer from 3 Hz to 2.5 Hz"; if the segregation zone is related to the crystallizer, it is recommended to "check the fluctuation range of the liquid level in the crystallizer and control the fluctuation within ±3 mm" or "adjust the viscosity of the protective slag from 0.15 Pa·s to 0.12 Pa·s".

[0091] The relationship between the parameter adjustment strategy and the corresponding 3D element distribution cloud map is generated to form a comprehensive output. The output can include a cloud map, annotations of excessive segregation zones, and a report document with corresponding process adjustment suggestions. In the 3D cloud map visualization interface, the segregation zone areas are highlighted and accompanied by floating text annotations explaining the corresponding process stage and adjustment suggestions. The strategy is stored as metadata along with the cloud map file for easy retrieval and analysis in the process management system later.

[0092] As can be seen from the above, this embodiment achieves closed-loop management from "segregation detection" to "process optimization" by comparing the segregation quantification index with the corresponding preset threshold and tracing the segregation spatial location back to the specific continuous casting process stage. It establishes an objective segregation qualification judgment standard and avoids the subjectivity of human experience. Secondly, through the correlation mapping between spatial location and process stage, segregation problems can be quickly located to specific process links, improving the efficiency of problem diagnosis. Thirdly, the generated parameter adjustment strategy is targeted and operable, providing clear optimization directions for on-site process personnel and helping to continuously improve the internal quality and production stability of continuous casting billets.

[0093] Based on the same principle as the billet elemental distribution analysis method provided in the embodiments of this application, the embodiments of this application also provide a billet elemental distribution analysis device, such as... Figure 2 As shown, the billet element distribution analysis device 20 may specifically include: a data acquisition module 21, a mapping relationship determination module 22, a target grid determination module 23, an element average content determination module 24, and an element content analysis module 25.

[0094] Among them, the data acquisition module 21 is used to acquire the element distribution visualization image corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample. The mapping relationship determination module 22 is used to determine the target mapping relationship based on the pixel coordinate system of the element distribution visualization image and the three-dimensional spatial coordinate system of the original surface point cloud data. The target mesh determination module 23 is used to divide the target area into meshes on the element distribution visualization image, obtain each target mesh, and align the boundary of each target mesh with the physical boundary of the billet sample in the element distribution visualization image. The element average content determination module 24 is used to extract the original data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the element average content corresponding to each target grid by using anisotropic kriging interpolation. The element content analysis module 25 is used to determine the projection area of ​​each target grid in three-dimensional space according to the target mapping relationship, and assign the average content of each element to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information. The surface point cloud data is used as a three-dimensional element distribution cloud map, which is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

[0095] In one embodiment of this application, the target region includes multiple sub-target regions. When dividing the target region into grids on an element distribution visualization image to obtain each target grid, the target grid determination module 23 is specifically used for: For each sub-target region, a corresponding initial mesh size is preset to form the initial basic mesh of the target region; the initial basic mesh consists of multiple initial basic mesh elements. Based on the prior geometric location of the segregation band in the billet sample, the segregation sensitive band is delineated in the initial basic grid, and the initial basic grid cells falling into the segregation sensitive band are refined to obtain the first grid set. Based on the first grid set, calculate the local gradient of the element distribution visualization image, and iteratively subdivide the grid cells in the first grid set whose local gradient values ​​exceed a preset gradient threshold to obtain the second grid set; The physical boundary of the billet sample in the element distribution visualization image is extracted based on the edge detection algorithm. All grid nodes in the second grid set that cross the physical boundary are adjusted to the physical boundary to obtain the adjusted grid node coordinates. Based on the adjusted grid node coordinates, a set of two-dimensional grid cells covering the target area is generated, which serves as each target grid.

[0096] In one embodiment of this application, after extracting the raw data of element content of pixels in each target grid, the element average content determination module 24 is specifically used for: For each pixel, the local spatial neighborhood of the pixel is determined based on a preset neighborhood radius, with the pixel as the center. Obtain the absolute value of the difference between the original data of the element content of the pixel and the mean value of the element content of all pixels in the local spatial neighborhood, as well as the standard deviation of the element content in the local spatial neighborhood; If the absolute value of the difference is greater than the product of the standard deviation and the preset outlier threshold coefficient, the pixel is determined to be a discrete outlier and is removed. If the absolute value of the difference is not greater than the product of the standard deviation and the preset abnormal threshold coefficient, then the pixel is determined to be a normal pixel. The effective pixel set is constructed from all normal points, and the effective pixel set is used to determine the average element content of each target grid.

[0097] In one embodiment of this application, when determining the average element content corresponding to each target grid using anisotropic kriging interpolation with each target grid as an estimation block, the element average content determination module 24 is specifically used for: Based on the target mapping relationship, the pixel coordinates of all effective pixels in each target grid are converted into physical coordinates on the surface of the billet sample to obtain the physical position coordinates of each effective pixel and the corresponding element content data. Determine the anisotropic principal axis direction angle based on the solidification direction of the steel billet sample; Based on the anisotropic principal axis direction angle, the element content data of all effective pixels and their physical location coordinates, the anisotropic parameters of the variogram are obtained by fitting. For each target grid, based on the element content data of the effective pixels within the target grid, the average element content of the grid is calculated to obtain the average element content of the physical region corresponding to the target grid. Specifically, in the operation of calculating the average content of the grid, the element average content determination module 24 is also used for: The target grid is used as the block to be estimated. Based on the anisotropy parameter of the variogram, the variogram value between every two valid pixels in the target grid and the average variogram value between each valid pixel and all valid pixels in the block to be estimated are calculated. Construct a block Kriging equation system based on the variogram values ​​and the average variogram values; Solve the block kriging equations to obtain the weights corresponding to each effective pixel in the target grid. The average element content of the target grid is determined based on the element content data of each effective pixel in the target grid and the weight corresponding to each effective pixel.

[0098] In one embodiment of this application, after obtaining surface point cloud data with elemental content information, the elemental content analysis module 25 is specifically used for: In the three-dimensional element distribution cloud map, multiple sampling lines are obtained along the rolling direction, width direction and thickness direction of the billet sample, respectively; For each sampling line, the element content values ​​of each sampling point passed through by the sampling line are extracted to form a sequence of element content values ​​along the line, and the element content gradient between adjacent sampling points is calculated; the sampling point is the spatial location point on the sampling line selected for reading the element content value. Based on a preset segregation gradient threshold, the spatial region corresponding to the continuous region in the element content value sequence where the gradient value exceeds the preset segregation gradient threshold is taken as a segregation band of the sampling line; wherein, one sampling line corresponds to zero or one or more segregation bands; Obtain the quantitative indicators for each segregation band, including the width, peak value, and segregation degree of the segregation band.

[0099] In one embodiment of this application, after generating a set of two-dimensional mesh cells covering the target area as each target mesh, the target mesh determination module 23 is specifically used for: For each grid cell in the target grid, calculate the corresponding quality index based on the shape rules of that grid cell; If there are mesh cells with quality indicators lower than the preset quality threshold, they are marked as quality defect mesh cells; The defective mesh cells and their neighboring meshes are locally re-divided until the quality index of all mesh cells in the target mesh is not lower than the preset quality threshold. The target mesh is updated based on all the locally optimized mesh cells to obtain the updated target mesh, which is used to determine the average element content.

[0100] In one embodiment of this application, after obtaining the quantitative indicators of each segregation band, the elemental content analysis module 25 is specifically used for: If the width of the segregation band exceeds the preset width threshold, or the segregation degree exceeds the preset segregation degree threshold, then the segregation degree of the current billet sample is determined to be beyond the allowable range. Based on the spatial location of the segregation band that exceeds the allowable range, determine the corresponding continuous casting process stage and generate a parameter adjustment strategy for the continuous casting process stage. The relationship between output parameter adjustment strategies and 3D element distribution cloud maps.

[0101] The apparatus in this application embodiment can execute the method provided in this application embodiment, and the implementation principle is similar. The actions performed by each module in the apparatus of each embodiment of this application correspond to the steps in the method of each embodiment of this application. For detailed functional descriptions of each module of the apparatus, please refer to the descriptions in the corresponding methods shown above, which will not be repeated here.

[0102] Figure 3 A schematic diagram of the structure of an electronic device to which this application embodiment applies is shown, such as... Figure 3 As shown, the electronic device can be used to implement the methods provided in any embodiment of this application.

[0103] like Figure 3 As shown, the electronic device 300 may primarily include at least one processor 301. Figure 3 The diagram shows components such as a memory 302, a communication module 303, and an input / output interface 304. Optionally, these components can be connected and communicate with each other via a bus 305. It should be noted that... Figure 3 The structure of the electronic device 300 shown is merely illustrative and does not constitute a limitation on the electronic devices to which the methods provided in the embodiments of this application are applicable.

[0104] The memory 302 can be used to store operating systems and applications, etc. The applications can include computer programs that implement the methods shown in the embodiments of this application when invoked by the processor 301, and can also include programs for implementing other functions or services. The memory 302 can be ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, RAM (Random Access Memory) or other types of dynamic storage devices that can store information and computer programs, or it can be EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited thereto.

[0105] Processor 301 is connected to memory 302 via bus 305 and implements corresponding functions by calling the application programs stored in memory 302. Processor 301 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 301 can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0106] Electronic device 300 can connect to a network via communication module 303 (which may include, but is not limited to, components such as a network interface) to communicate with other devices (such as user terminals or servers) through the network and achieve data interaction, such as sending data to or receiving data from other devices. Communication module 303 may include wired network interfaces and / or wireless network interfaces, meaning the communication module may include at least one of wired or wireless communication modules.

[0107] The electronic device 300 can connect to necessary input / output devices, such as a keyboard and display device, via the input / output interface 304. The electronic device 300 itself may have a display device, and other display devices can also be connected externally via the interface 304. Optionally, a storage device, such as a hard drive, can also be connected via the interface 304 to store data from the electronic device 300, read data from the storage device, or store data from the storage device in the memory 302. It is understood that the input / output interface 304 can be a wired interface or a wireless interface. Depending on the actual application scenario, the device connected to the input / output interface 304 can be a component of the electronic device 300 or an external device connected to the electronic device 300 when needed.

[0108] The bus 305 used to connect the components may include a path for transmitting information between the components. The bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Depending on its function, the bus 305 may be divided into an address bus, a data bus, a control bus, etc.

[0109] Optionally, for the solution provided in the embodiments of this application, the memory 302 can be used to store a computer program that executes the solution of this application, and the processor 301 runs the computer program. When the processor 301 runs the computer program, it implements the operation of the method or apparatus provided in the embodiments of this application.

[0110] Based on the same principle as the method provided in the embodiments of this application, the embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the corresponding content of the aforementioned method embodiments.

[0111] It should be noted that the terms "first," "second," "third," "fourth," "1," "2," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in a sequence other than that shown in the figures or text.

[0112] In the embodiments of this application, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0113] It should be understood that although arrows indicate various operation steps in the flowcharts of this application's embodiments, the order in which these steps are implemented is not limited to the order indicated by the arrows. Unless explicitly stated herein, in some implementation scenarios of this application's embodiments, the implementation steps in each flowchart can be executed in other orders as required. Furthermore, some or all steps in each flowchart, based on the actual implementation scenario, may include multiple sub-steps or multiple stages. Some or all of these sub-steps or stages can be executed at the same time, and each sub-step or stage can also be executed at different times. In scenarios where execution times differ, the execution order of these sub-steps or stages can be flexibly configured according to requirements, and this application's embodiments do not limit this.

[0114] The above are only optional implementation methods for some implementation scenarios of this application. It should be noted that for those skilled in the art, other similar implementation methods based on the technical concept of this application without departing from the technical concept of this application also fall within the protection scope of the embodiments of this application.

Claims

1. A method for analyzing the elemental distribution of steel billets, characterized in that, include: Obtain a visualized image of the element distribution corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample; The target mapping relationship is determined based on the pixel coordinate system of the element distribution visualization image and the three-dimensional spatial coordinate system of the original surface point cloud data. The target region is divided into grids on the element distribution visualization image to obtain each target grid. Extract the raw data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the average element content of each target grid using anisotropic kriging interpolation. Based on the target mapping relationship, the projection area corresponding to each target grid in three-dimensional space is determined, and the average content of each element is assigned to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information. This surface point cloud data is used as a three-dimensional element distribution cloud map, which is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

2. The method as described in claim 1, characterized in that, The target region includes multiple sub-target regions. The step of dividing the target region into grids on the element distribution visualization image to obtain each target grid includes: For each sub-target region, a corresponding initial grid size is preset to form the initial basic grid of the target region; the initial basic grid includes multiple initial basic grid cells. Based on the prior geometric position of the segregation band in the billet sample, a segregation sensitive band is defined in the initial basic grid, and the initial basic grid cells falling into the segregation sensitive band are refined to obtain the first grid set. Based on the first grid set, calculate the local gradient of the element distribution visualization image, and iteratively subdivide the grid cells in the first grid set whose local gradient values ​​exceed a preset gradient threshold to obtain a second grid set; The physical boundary of the billet sample in the element distribution visualization image is extracted according to the edge detection algorithm, and all grid nodes in the second grid set that cross the physical boundary are adjusted to the physical boundary to obtain the adjusted grid node coordinates; Based on the adjusted grid node coordinates, a set of two-dimensional grid cells covering the target area is generated, which serves as the target grids.

3. The method as described in claim 1, characterized in that, After extracting the raw data of element content of pixels in each target grid, the process also includes: For each pixel, the local spatial neighborhood of the pixel is determined based on a preset neighborhood radius, with the pixel as the center. Obtain the absolute value of the difference between the original data of the element content of the pixel and the mean value of the element content of all pixels in the local spatial neighborhood, as well as the standard deviation of the element content in the local spatial neighborhood; If the absolute value of the difference is greater than the product of the standard deviation and the preset anomaly threshold coefficient, then the pixel is determined to be a discrete anomaly and the pixel is removed. If the absolute value of the difference is not greater than the product of the standard deviation and the preset abnormal threshold coefficient, then the pixel is determined to be a normal pixel. The effective pixel set is constructed from all normal points, and the effective pixel set is used to determine the average element content of each target grid.

4. The method as described in claim 3, characterized in that, The step of determining the average element content corresponding to each target grid using anisotropic kriging interpolation, with each target grid as an estimation block, includes: Based on the target mapping relationship, the pixel coordinates of all effective pixels in each target grid are converted into physical coordinates of the steel billet sample surface to obtain the physical position coordinates of each effective pixel and the corresponding element content data. Determine the anisotropic principal axis direction angle based on the solidification direction of the steel billet sample; Based on the anisotropic principal axis direction angle, the element content data of all effective pixels and their physical position coordinates, the anisotropic parameters of the variogram are obtained by fitting. For each target grid, based on the element content data of the effective pixels within the target grid, the average element content of the grid is calculated to obtain the average element content of the physical region corresponding to the target grid. The operation of calculating the average content of the grid includes: The target grid is used as the block to be estimated. Based on the anisotropy parameter of the variogram, the variogram value between every two valid pixels in the target grid and the average variogram value between each valid pixel and all valid pixels in the block to be estimated are calculated. Construct a block Kriging equation system based on the variogram value and the average variogram value; Solve the block Kriging equations to obtain the weights corresponding to each effective pixel in the target grid. The average element content of the target grid is determined based on the element content data of each effective pixel in the target grid and the weight corresponding to each effective pixel.

5. The method as described in claim 1, characterized in that, After obtaining the surface point cloud data with elemental content information, the process further includes: In the three-dimensional element distribution cloud map, multiple sampling lines are obtained along the rolling direction, width direction and thickness direction of the billet sample, respectively; For each sampling line, the element content values ​​of each sampling point passed through by the sampling line are extracted to form a sequence of element content values ​​along the line, and the element content gradient between adjacent sampling points is calculated; the sampling point is a spatial location point on the sampling line selected for reading the element content value. Based on a preset segregation gradient threshold, the spatial region corresponding to the continuous region in the element content value sequence where the gradient value exceeds the preset segregation gradient threshold is taken as a segregation band of the sampling line; wherein, one sampling line corresponds to zero or one or more segregation bands; The quantitative indicators of each segregation band are obtained, including the width, peak value, and segregation degree of the segregation band.

6. The method as described in claim 2, characterized in that, After generating the set of two-dimensional mesh cells covering the target region as the target meshes, the method further includes: For each grid cell in the target grid, calculate the corresponding quality index based on the shape rules of that grid cell; If there are mesh cells with quality indicators lower than the preset quality threshold, they are marked as quality defect mesh cells; The quality defect grid cell and its neighboring grid are locally re-divided until the quality index of all grid cells in the target grid is not lower than the preset quality threshold. The target mesh is updated based on all locally optimized mesh cells to obtain an updated target mesh, which is used to determine the average element content.

7. The method as described in claim 5, characterized in that, After obtaining the quantitative indicators of each segregation band, the method further includes: If the width of the segregation band exceeds the preset width threshold, or the segregation degree exceeds the preset segregation degree threshold, then the segregation degree of the current billet sample is determined to be beyond the allowable range. Based on the spatial location of the segregation band that exceeds the allowable range, determine the corresponding continuous casting process stage and generate a parameter adjustment strategy for the continuous casting process stage. Output the correlation between the parameter adjustment strategy and the three-dimensional element distribution cloud map.

8. A billet elemental distribution analysis device, characterized in that, include: The data acquisition module is used to acquire a visualized image of the element distribution corresponding to the target area of ​​the billet sample, as well as the original surface point cloud data of the billet sample. The mapping relationship determination module is used to determine the target mapping relationship based on the pixel coordinate system of the element distribution visualization image and the three-dimensional spatial coordinate system of the original surface point cloud data. The target grid determination module is used to divide the target region into grids on the element distribution visualization image to obtain each target grid. The element average content determination module is used to extract the raw data of element content of pixels in each target grid, and use each target grid as an estimation block to determine the element average content corresponding to each target grid by using anisotropic kriging interpolation. The element content analysis module is used to determine the projection area of ​​each target grid in three-dimensional space according to the target mapping relationship, and assign the average content of each element to all three-dimensional coordinate points in the corresponding projection area to obtain surface point cloud data with element content information. The surface point cloud data is used as a three-dimensional element distribution cloud map, which is used to reflect the element content at each three-dimensional coordinate position on the surface of the billet sample.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the billet element distribution analysis method according to any one of claims 1 to 7 when running the computer program.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the billet element distribution analysis method according to any one of claims 1 to 7.

Images