Aluminum alloy casting air hole defect grading detection method and system

Abstract

The invention discloses an aluminum alloy casting air hole defect grading detection method and system, and relates to the technical field of casting detection. The method comprises the following steps: performing multi-angle imaging scanning on the aluminum alloy casting to obtain a cross-sectional image sequence; performing image enhancement on the cross-sectional image sequence, extracting pore defect initial feature information for segmentation identification, obtaining pore defect distribution information for morphological analysis, and obtaining pore morphological defect parameters; performing data bidirectional verification according to the pore defect distribution information in combination with pore morphological defect parameters, constructing a pore multi-stage defect array for grading, and determining a defect grading result; and the aluminum alloy castings are subjected to quality grading according to the defect grading result, and automatic sorting is conducted according to the quality grades. The technical problems that in the prior art, the aluminum alloy casting air hole defect detection precision is low, effective grading evaluation is difficult to achieve, quality control is inaccurate, and the sorting efficiency is insufficient are solved, and the technical effect of improving the detection precision and the sorting efficiency is achieved.

Description

Technical Field

[0001] This invention relates to the field of casting inspection technology, specifically to a method and system for classifying and detecting porosity defects in aluminum alloy castings. Background Technology

[0002] Aluminum alloy castings are crucial basic materials in the automotive, rail transportation, and other fields, and their internal quality directly affects the product's service life and safety performance. During the casting process, porosity defects frequently occur inside the castings due to gas entrapment, metal shrinkage, or insufficient process control. Porosity not only reduces the mechanical properties of the castings but can also trigger fatigue cracks, seriously threatening the product's service reliability. Existing inspection methods mostly employ single-angle X-ray fluoroscopy or ultrasonic testing, which often suffer from problems such as vague defect characteristics, high missed detection rates, and inability to achieve effective classification and evaluation. This results in insufficient precision in casting quality control and low subsequent sorting efficiency. Summary of the Invention

[0003] This application provides a method and system for classifying and detecting porosity defects in aluminum alloy castings, which solves the technical problems of low detection accuracy and difficulty in achieving effective classification and evaluation of porosity defects in aluminum alloy castings in the prior art, resulting in inaccurate quality control and insufficient sorting efficiency.

[0004] The first aspect of this application provides a method for graded detection of porosity defects in aluminum alloy castings, the method comprising: Multi-angle imaging scanning is performed on aluminum alloy castings to acquire tomographic image sequences. Image enhancement is applied to these tomographic image sequences to extract initial feature information of porosity defects for segmentation and identification. Porosity defect distribution information is obtained, and morphological analysis is performed to acquire porosity morphological defect parameters. Data bidirectional verification is conducted based on the porosity defect distribution information and the porosity morphological defect parameters to construct a multi-level porosity defect array for grade classification, determining the defect grading results. The aluminum alloy castings are then quality-rated based on the defect grading results, and automatic sorting is performed according to the quality level.

[0005] A second aspect of this application provides a grading detection system for porosity defects in aluminum alloy castings, the system comprising: Image acquisition module: performs multi-angle imaging scanning on aluminum alloy castings to acquire tomographic image sequences; Analysis module: enhances the tomographic image sequences, extracts initial feature information of porosity defects for segmentation and recognition, obtains porosity defect distribution information for morphological analysis, and obtains porosity morphology defect parameters; Grading module: performs bidirectional data verification based on the porosity defect distribution information and the porosity morphology defect parameters, constructs a multi-level porosity defect array for grading, and determines the defect grading results; Casting sorting module: performs quality rating on aluminum alloy castings based on the defect grading results, and automatically sorts them according to the quality level.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, multi-angle imaging scanning is performed on the aluminum alloy casting to acquire a tomographic image sequence. Next, image enhancement is applied to the tomographic image sequence to extract initial feature information of porosity defects for segmentation and identification. Porosity defect distribution information is obtained, and morphological analysis is performed to acquire porosity morphology defect parameters. Then, based on the porosity defect distribution information and porosity morphology defect parameters, bidirectional data verification is performed to construct a multi-level porosity defect array for grading and determining the defect classification results. Finally, the aluminum alloy casting is quality-rated based on the defect classification results, and automatic sorting is performed according to the quality level. This solves the technical problems of low detection accuracy and difficulty in achieving effective grading and evaluation of porosity defects in aluminum alloy castings in existing technologies, leading to inaccurate quality control and insufficient sorting efficiency. It achieves the technical effect of improving detection accuracy and sorting efficiency. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic diagram of the process for classifying and detecting porosity defects in aluminum alloy castings provided in an embodiment of this application. Figure 2 A schematic diagram of the structure of the aluminum alloy casting porosity defect classification detection system provided in the embodiments of this application.

[0009] Figure labeling: Image acquisition module 11, analysis module 12, grading module 13, casting sorting module 14. Detailed Implementation

[0010] This application provides a method and system for classifying and detecting porosity defects in aluminum alloy castings, which solves the technical problems in the prior art of low detection accuracy and difficulty in achieving effective classification and evaluation of porosity defects in aluminum alloy castings, resulting in inaccurate quality control and insufficient sorting efficiency.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0013] Example 1, as Figure 1 As shown, this application provides a method for graded detection of porosity defects in aluminum alloy castings, wherein the method includes: Multi-angle imaging scans were performed on aluminum alloy castings to obtain tomographic image sequences.

[0014] In this embodiment, a two-dimensional planar coordinate system is established for the aluminum alloy casting, and key point analysis is performed on the outer contour of the casting to determine multiple effective imaging field points. Based on the multiple effective imaging field points, center point clustering is performed to obtain the effective field center position. Using X-ray or industrial CT imaging equipment, the casting is scanned in a step-rotation manner with the effective field center as the reference. During the rotation, multi-angle projection image sequences are acquired at preset angle intervals. The multi-angle projection image sequences are converted into cross-sectional tomographic images of the aluminum alloy casting through a back-projection reconstruction algorithm, thereby obtaining a tomographic image sequence covering the overall internal structure of the casting.

[0015] Furthermore, multi-angle imaging scanning of aluminum alloy castings is performed to obtain tomographic image sequences. The methods include: A two-dimensional plane coordinate system for the aluminum alloy casting is constructed. Key point analysis is performed by traversing the two-dimensional plane coordinate system to determine multiple effective imaging field points. Center point clustering is performed based on the multiple effective imaging field points to determine the effective field center. The aluminum alloy casting is scanned according to the effective field center and the multiple effective imaging field points to construct an initial two-dimensional projection image. The initial two-dimensional projection image is scanned by step rotation according to the effective field center to obtain a multi-angle image acquisition sequence. Cross-sectional image reconstruction is performed on the aluminum alloy casting according to the multi-angle image acquisition sequence to construct the tomographic image sequence.

[0016] Specifically, the aluminum alloy casting to be inspected is fixed on a rotating support platform of an industrial CT scanning device. The scanning device includes an X-ray source and a flat panel detector, which are arranged opposite to each other. A two-dimensional planar coordinate system for the aluminum alloy casting is constructed using the two-dimensional imaging plane of the detector. Edge detection (e.g., using the Canny operator) is performed on the acquired contour projection image to extract multiple key points of the casting's outer contour. Based on the two-dimensional planar coordinate system, the key points are traversed and analyzed to generate a candidate point set covering the imaging area. Multiple effective imaging field points are then selected based on the spatial distribution characteristics of the candidate points. Subsequently, a clustering algorithm (such as K-means clustering) is used to cluster the multiple effective imaging field points, determining the cluster center as the effective field center. The scanning reference range is determined by combining the multiple effective imaging field points. Under the constraint of the effective field center, the X-ray source and detector are controlled to perform local scanning of the aluminum alloy casting to acquire an initial two-dimensional projection image. Next, using the center of the effective field of view as the rotation reference point, the rotating support stage is driven to rotate 360° in 0.5°~1° increments, acquiring a projected image at each angular position to obtain a multi-angle image acquisition sequence. Finally, the multi-angle image acquisition sequence is input into a filtered back-projection algorithm or algebraic reconstruction technology to reconstruct the cross-section of the aluminum alloy casting layer by layer, obtaining a tomographic image sequence covering the overall internal structure.

[0017] Image enhancement is performed on the tomographic image sequence, initial feature information of pore defects is extracted for segmentation and identification, distribution information of pore defects is obtained for morphological analysis, and pore morphology defect parameters are obtained.

[0018] Furthermore, the method involves image enhancement of the tomographic image sequence, extraction of initial feature information of pore defects for segmentation and identification, and obtaining pore defect distribution information. The method includes: The tomographic image sequence is traversed and nonlocal mean filtering is performed to extract multiple image edge structures. Contrast enhancement is then applied to the tomographic image sequence according to these multiple image edge structures to generate a contrast-enhanced image set. Based on the contrast-enhanced image set, the tomographic image sequence is combined with historical porosity defect images for initial feature screening to determine initial porosity defect feature information. Boundary recognition is performed based on the initial porosity defect feature information to determine the porosity region boundaries. Segmentation and recognition are then performed according to the porosity region boundaries to determine multiple porosity regions. Porosity traversal and recording are then performed to obtain porosity defect distribution information.

[0019] Specifically, nonlocal mean filtering is applied frame-by-frame to the tomographic image sequence to reduce image noise while preserving edge texture features, thereby extracting multiple image edge structures. Based on these extracted edge structures, the tomographic image sequence undergoes contrast enhancement processing, such as using histogram equalization or adaptive histogram equalization, to obtain a contrast-enhanced image set. This contrast-enhanced image set is then compared with historical stomatal defect image samples, and a convolutional neural network classification model or support vector machine discriminant model is used to initially screen candidate stomatal region blocks, determining initial stomatal defect feature information. Based on this initial stomatal defect feature information, boundary recognition is performed on the tomographic image sequence to determine stomatal region boundaries. The tomographic image sequence is then segmented and identified according to these boundaries, identifying multiple stomatal regions, which are then traversed and recorded one by one to form stomatal defect distribution information.

[0020] Furthermore, based on the initial feature information of the pore defects, boundary identification is performed to determine the boundary of the pore region. The method includes: Local neighborhood grayscale analysis is performed based on the initial feature information of the pore defects to determine the segmentation threshold; the tomographic image sequence is segmented according to the segmentation threshold to generate a binarized image for pore identification, and multiple suspected pore regions are identified; connectivity is determined for the multiple suspected pore regions, and connected suspected pore regions are used as seed points; the grayscale gradient of surrounding pixels is extracted with the seed points as the center for region growing to determine the boundary of the pore region.

[0021] Specifically, based on the initial feature information of pore defects, the pixel grayscale value distribution is extracted in the local neighborhood of the tomographic image sequence. Grayscale histogram statistical analysis is performed on the neighborhood, and a segmentation threshold is determined based on the principle of variance maximization or mean segmentation. The tomographic image sequence is segmented according to the segmentation threshold, separating regions with pixel values ​​higher than the threshold from regions with pixel values ​​lower than the threshold, generating a binary image. Morphological processing (such as erosion and dilation operations) is performed on the binary image to eliminate isolated noise and enhance the defect contour, thereby obtaining multiple suspected pore regions. Based on this, connected component analysis is performed on multiple suspected pore regions. Spatially adjacent and pixel-connected regions are determined, and suspected regions determined to be connected are merged into the same connected component, with the center pixel of the connected component selected as the seed point. Centered on the seed point, a region growing algorithm is executed based on the grayscale gradient information of neighboring pixels. During the growing process, neighboring pixels are expanded layer by layer, and it is judged whether the grayscale gradient threshold condition is met, until no further expansion is possible, thereby obtaining the boundary of the pore region.

[0022] Furthermore, morphological analysis is performed to obtain pore defect distribution information and pore morphology parameters. Methods include: The area of ​​each of the multiple pore regions is calculated to determine the pixel area data of the multiple regions; pixel conversion is performed based on the boundaries of the pore regions to obtain the total number of boundary pixels; the pixel area data of the multiple regions and the total number of boundary pixels are converted according to the spatial calibration coefficient to obtain the actual physical area and the actual physical perimeter; the morphological analysis of the pores is performed based on the actual physical area and the actual physical perimeter to calculate the equivalent circle diameter of the pores and obtain the pore morphological defect parameters.

[0023] Specifically, the process involves traversing multiple pore regions, counting the number of pixels within each region, and multiplying this number by the actual physical area of ​​each pixel to obtain the pixel area data for that pore region. Based on the pore region boundaries, boundary pixels are extracted and counted to obtain the total number of boundary pixels. Then, using the spatial calibration coefficients of the scanning system, the pixel area data and the total number of boundary pixels are converted into actual physical area and actual physical perimeter, respectively, thus obtaining physical quantities reflecting the true geometric scale of the pores. Furthermore, based on the actual physical area and actual physical perimeter, morphological calculation formulas are used to analyze the pore morphology, and the equivalent circle diameter of the pore can be obtained using the formula... The calculations are performed, where A represents the actual physical area of ​​the pores. Simultaneously, the roundness and compactness of the pores are calculated based on the actual physical perimeter. Roundness can be determined using the formula... The calculation is performed, where P is the actual physical circumference of the pore, used to measure how closely the pore shape approximates a standard circle; compactness can be determined using the formula... Calculations are performed to characterize the complexity and irregularity of the pore boundaries. Finally, multiple indicators such as the equivalent circle diameter, actual physical area, actual physical perimeter, and roundness of the pore are combined to form pore morphology defect parameters, which are used to characterize the geometric properties of the pores.

[0024] Based on the porosity defect distribution information and the porosity morphology defect parameters, the data is verified bidirectionally to construct a multi-level porosity defect array for classification and to determine the defect classification results.

[0025] Based on the pore distribution information, the spatial coordinates of each pore region are labeled, and the spatial distribution density of pores within the 3D mesh is calculated. Simultaneously, based on pore morphology defect parameters, the equivalent circle diameter, roundness, and compactness of the pores are statistically analyzed to form a parameter set reflecting the size and morphology of the pores. Subsequently, a first verification rule and a second verification rule are established. The first verification rule uses a threshold judgment based on the deviation of the equivalent circle diameter of the pores, while the second verification rule uses a reasonableness judgment based on the spatial distribution density of the pores. For each pore region, its morphological parameters and distribution information are imported into the two types of verification rules for cross-verification. When both the first and second verification results satisfy the rules, the pore is determined to be a valid defect region; if either verification fails, it is marked as invalid data and discarded. After obtaining the valid defect regions, a multi-level pore defect array is constructed: pores are divided into different size levels according to the numerical range of the equivalent circle diameter, and the aluminum alloy casting is divided into multiple density levels according to the spatial distribution density. The size levels and density levels are correlated and mapped to form a multi-dimensional defect array. Finally, a comprehensive evaluation of the aluminum alloy castings is performed based on a multi-level defect array, and the porosity defect classification results are output.

[0026] Furthermore, a multi-level porosity defect array is constructed by performing bidirectional data verification based on the porosity defect distribution information and the porosity morphology defect parameters. The method includes: Based on the porosity defect distribution information, porosity locations are identified, and their spatial coordinates are determined. A first verification rule based on the porosity morphology defect parameters and a second verification rule based on the porosity defect distribution information are established. The porosity defect distribution information is then subjected to bidirectional cross-verification with the porosity morphology defect parameters: S1: Deviation analysis is performed based on the equivalent circle diameter of the porosity to determine the deviation parameters. Verification is then performed according to the first verification rule to determine the first verification result. S2: Distribution calculation is performed based on the spatial coordinates of the porosity to determine the spatial distribution density. Verification is then performed according to the second verification rule to determine the second verification result. A bidirectional verification judgment is then performed based on the first and second verification results, and the multi-level porosity defect array is constructed according to the judgment results.

[0027] Based on the distribution information of pore defects, the location of each pore region is identified, its coordinate value in the three-dimensional reconstruction space is extracted, and the spatial location coordinates of the pores are determined accordingly.

[0028] Two types of verification rules are established: a first verification rule based on pore morphology defect parameters, used to constrain pore geometric characteristics; and a second verification rule based on pore defect distribution information, used to constrain pore spatial distribution characteristics. Specifically, this includes: using the equivalent circle diameter of the pore as the main indicator, calculating its deviation from a set threshold range; when the deviation meets the tolerance range, the first verification rule is passed, and the first verification result is obtained. Based on the spatial coordinates of the pores, a mesh generation method is used to calculate the number of pores per unit volume, obtaining the pore spatial distribution density; then, the pore spatial distribution density is compared with a set density threshold range; when the distribution density is within a reasonable range, the second verification rule is passed, and the second verification result is obtained.

[0029] The results of the first and second verifications are combined for bidirectional verification: when both are valid, the pore is confirmed as a real defect and included in the multi-level pore defect array; if either verification fails, it is marked as invalid data and removed.

[0030] Furthermore, based on the first verification result and the second verification result, a bidirectional verification judgment is performed, and the multi-level defect array of pores is constructed according to the judgment result. The method includes: When both the first verification result and the second verification result pass bidirectional verification, the porosity defect distribution information and the porosity morphology defect parameters are determined to be valid data. Then: S1: Set a numerical range according to the equivalent circle diameter of the porosity, and divide multiple porosity size levels according to the numerical range; S2: Perform regional analysis on the aluminum alloy casting according to the spatial coordinates of the porosity, and determine the key areas to divide into multiple porosity density levels; S3: Associate and map the multiple porosity size levels with the multiple porosity density levels to construct the multi-level porosity defect array.

[0031] When both the first and second verification results are deemed valid, the porosity defect distribution information and porosity morphology defect parameters are confirmed as reliable data, and a multi-level porosity defect array is constructed accordingly. Specifically, using the equivalent circle diameter d of the porosity as an indicator, the porosity is divided into grades according to a preset diameter range. For example, porosity d < 50 μm is classified as size grade L1, porosity 50 μm ≤ d < 100 μm is classified as size grade L2, and so on, thereby obtaining multiple porosity size grades.

[0032] Based on the spatial coordinates of pores, the three-dimensional model of the aluminum alloy casting is divided into regional meshes, and the number of pores in each mesh unit is counted to calculate the spatial distribution density. By combining the sensitivity of key functional areas (such as stress areas, joint surfaces, etc.) and setting different density thresholds, the pore density is divided into low density level, medium density level and high density level, resulting in multiple pore density levels.

[0033] Multiple pore size levels are correlated and mapped with pore density levels, for example, in the form of a two-dimensional matrix, where the horizontal axis represents the pore size level and the vertical axis represents the pore density level, with different cross-units corresponding to different defect levels. This constructs a multi-level pore defect array, enabling multi-dimensional quantitative characterization of pore defects.

[0034] Furthermore, a multi-level defect array of pores is constructed for classification, and the defect classification results are determined. The methods include: Based on the volume discretization analysis of aluminum alloy castings, a three-dimensional spatial mesh unit is constructed. The multi-level porosity defect array is mapped to the three-dimensional spatial mesh unit to count the number of porosity defects, obtaining the total number of porosity defects in multiple mesh units. The multi-level porosity defect array is mapped to the three-dimensional spatial mesh unit for porosity size analysis. Based on the equivalent circle diameter of porosity in multiple mesh units, the equivalent circle diameter of porosity in the first order is extracted as the maximum equivalent diameter. Based on the total number of porosity defects in multiple mesh units and the maximum equivalent diameter, defect influence analysis is performed to generate a defect influence factor. The multiple mesh units in the three-dimensional spatial mesh unit are weighted according to the defect influence factor to obtain the local defect severity index of multiple mesh units. The aluminum alloy casting is then evaluated to obtain an overall defect severity score. Based on the overall defect severity score, the porosity of the aluminum alloy casting is classified to determine the defect classification result.

[0035] A three-dimensional reconstruction model of an aluminum alloy casting is used to discretize its overall volume. A uniform mesh generation method is employed to decompose the casting volume into multiple three-dimensional spatial mesh units, each with a fixed volume size, such as 100μm × 100μm × 100μm. Then, a multi-level porosity defect array is mapped onto these three-dimensional spatial mesh units. The porosity defects within each mesh unit are statistically analyzed to obtain the total number of porosity defects. Simultaneously, the equivalent circle diameter of all pores within each mesh unit is extracted using the mapping results and sorted in descending order. The diameter of the first-order pore is selected as the maximum equivalent diameter of that unit. Based on this, a defect impact analysis is performed by combining the total number of porosity defects and the maximum equivalent diameter, constructing a defect impact factor F, which can be expressed by the formula F = αN + βd. max Where N is the total number of porosity defects in the unit, and d max The maximum equivalent diameter is given by α and β, which are weighting factors set empirically based on the application scenario of the casting.

[0036] The local defect severity index is obtained by weighting each mesh element according to the defect impact factor. The weighting method can be based on the sensitivity of functional areas, for example, higher weights for critical stress areas and lower weights for non-critical areas. The weighted sum of the local defect severity indices of all mesh elements is then used to obtain the overall defect severity score of the aluminum alloy casting.

[0037] Based on the overall defect severity score and the preset grading threshold range, porosity defects in aluminum alloy castings are classified into different levels. For example: a score of 0.2 or less is classified as Level 1 (minor defect), 0.2 to 0.5 as Level 2 (moderate defect), 0.5 to 0.8 as Level 3 (serious defect), and a score of 0.8 or more as Level 4 (scrap).

[0038] Furthermore, the local defect severity index of multiple mesh elements is obtained to evaluate aluminum alloy castings and obtain an overall defect severity score. The methods include: The multiple grid cells are traversed and labeled with regional risks according to their functions to obtain multiple regional risk levels; the multiple grid cells are weighted according to the multiple regional risk levels to determine the weight coefficients of the multiple grid cells; based on the weight coefficients of the multiple grid cells and the local defect severity index of the multiple grid cells, local defect analysis is performed on the multiple grid cells to obtain the weighted contribution value of the multiple grid cells; the aluminum alloy casting is evaluated according to the weighted contribution value to obtain the overall defect severity score.

[0039] Preferably, multiple grid cells are traversed, and functional areas are divided according to the structural characteristics of the aluminum alloy casting. For example, load-bearing areas, connection areas, and non-critical areas are labeled as different functional areas. Based on preset functional area importance weighting coefficients (e.g., 1.0 for load-bearing areas, 0.7 for connection areas, and 0.3 for non-critical areas), a regional risk level is assigned to each grid cell. Subsequently, the grid cells are weighted according to the regional risk level to obtain the weighting coefficient for each grid cell. The weighting coefficient can be assigned linearly or hierarchically; for example, high-risk areas have larger weights, and low-risk areas have smaller weights. On this basis, the local defect severity index of each grid cell is multiplied by its weighting coefficient to obtain the weighted contribution value of that grid cell. Finally, the weighted contribution values ​​of all grid cells are accumulated and normalized to obtain the overall defect severity score of the aluminum alloy casting.

[0040] Weighted contribution value: ,in, Let i be the weighted contribution value of the i-th grid cell. These are the weighting coefficients. This is the severity index of local defects.

[0041] Overall defect severity rating for aluminum alloy castings: , where S is the overall defect severity score, with a value ranging from [0, 1].

[0042] The aluminum alloy castings are graded according to the defect classification results, and then automatically sorted according to the quality level.

[0043] The overall defect severity score is compared with a preset quality rating standard. For example, castings with a score in the range of 0 to 0.2 are rated as Grade A (high quality), those in the range of 0.2 to 0.5 are rated as Grade B (usable), those in the range of 0.5 to 0.8 are rated as Grade C (substandard), and those in the range of 0.8 to 1 are rated as Grade D (scrap). This yields the quality rating level for each casting.

[0044] Based on this, aluminum alloy castings are automatically sorted according to their quality rating. Specifically, an automatic conveyor system is installed on the production line, using barcodes or RFID tags to associate the inspection results with the corresponding castings. When a casting enters the sorting station, the control system triggers an actuator (such as a pneumatic pusher, robotic arm, or sorting baffle) based on the rating result, guiding castings of different ratings to their corresponding collection channels or material frames. For example, Grade A castings enter the qualified product channel, Grade B castings enter the repairable area, Grade C castings enter the defective product area, and Grade D castings enter the scrap recycling area, thus achieving automatic sorting according to quality rating.

[0045] In summary, the embodiments of this application have at least the following technical effects: First, multi-angle imaging scanning is performed on the aluminum alloy casting to acquire a tomographic image sequence. Next, image enhancement is applied to the tomographic image sequence to extract initial feature information of porosity defects for segmentation and identification. Porosity defect distribution information is obtained, and morphological analysis is performed to acquire porosity morphology defect parameters. Then, based on the porosity defect distribution information and porosity morphology defect parameters, bidirectional data verification is performed to construct a multi-level porosity defect array for grading and determining the defect classification results. Finally, the aluminum alloy casting is quality-rated based on the defect classification results, and automatic sorting is performed according to the quality level. This solves the technical problems of low detection accuracy and difficulty in achieving effective grading and evaluation of porosity defects in aluminum alloy castings in existing technologies, leading to inaccurate quality control and insufficient sorting efficiency. It achieves the technical effect of improving detection accuracy and sorting efficiency.

[0046] Example 2, based on the same inventive concept as the aluminum alloy casting porosity defect classification detection method in the previous examples, such as... Figure 2 As shown, this application provides a grading detection system for porosity defects in aluminum alloy castings, wherein the system includes: Image acquisition module 11: Performs multi-angle imaging scanning on the aluminum alloy casting to acquire a tomographic image sequence; Analysis module 12: Enhances the tomographic image sequence, extracts initial feature information of porosity defects for segmentation and recognition, obtains porosity defect distribution information for morphological analysis, and obtains porosity morphology defect parameters; Grading module 13: Performs bidirectional data verification based on the porosity defect distribution information and the porosity morphology defect parameters, constructs a multi-level porosity defect array for grading, and determines the defect grading result; Casting sorting module 14: Grades the aluminum alloy casting according to the defect grading result and automatically sorts it according to the quality level.

[0047] Furthermore, the image acquisition module 11 is used to perform the following methods: A two-dimensional plane coordinate system for the aluminum alloy casting is constructed. Key point analysis is performed by traversing the two-dimensional plane coordinate system to determine multiple effective imaging field points. Center point clustering is performed based on the multiple effective imaging field points to determine the effective field center. The aluminum alloy casting is scanned according to the effective field center and the multiple effective imaging field points to construct an initial two-dimensional projection image. The initial two-dimensional projection image is scanned by step rotation according to the effective field center to obtain a multi-angle image acquisition sequence. Cross-sectional image reconstruction is performed on the aluminum alloy casting according to the multi-angle image acquisition sequence to construct the tomographic image sequence.

[0048] Furthermore, the analysis module 12 is used to perform the following methods: The tomographic image sequence is traversed and nonlocal mean filtering is performed to extract multiple image edge structures. Contrast enhancement is then applied to the tomographic image sequence according to these multiple image edge structures to generate a contrast-enhanced image set. Based on the contrast-enhanced image set, the tomographic image sequence is combined with historical porosity defect images for initial feature screening to determine initial porosity defect feature information. Boundary recognition is performed based on the initial porosity defect feature information to determine the porosity region boundaries. Segmentation and recognition are then performed according to the porosity region boundaries to determine multiple porosity regions. Porosity traversal and recording are then performed to obtain porosity defect distribution information.

[0049] Furthermore, the analysis module 12 is used to perform the following methods: Local neighborhood grayscale analysis is performed based on the initial feature information of the pore defects to determine the segmentation threshold; the tomographic image sequence is segmented according to the segmentation threshold to generate a binarized image for pore identification, and multiple suspected pore regions are identified; connectivity is determined for the multiple suspected pore regions, and connected suspected pore regions are used as seed points; the grayscale gradient of surrounding pixels is extracted with the seed points as the center for region growing to determine the boundary of the pore region.

[0050] Furthermore, the analysis module 12 is used to perform the following methods: The area of ​​each of the multiple pore regions is calculated to determine the pixel area data of the multiple regions; pixel conversion is performed based on the boundaries of the pore regions to obtain the total number of boundary pixels; the pixel area data of the multiple regions and the total number of boundary pixels are converted according to the spatial calibration coefficient to obtain the actual physical area and the actual physical perimeter; the morphological analysis of the pores is performed based on the actual physical area and the actual physical perimeter to calculate the equivalent circle diameter of the pores and obtain the pore morphological defect parameters.

[0051] Furthermore, the level classification module 13 is used to perform the following method: Based on the porosity defect distribution information, porosity locations are identified, and their spatial coordinates are determined. A first verification rule based on the porosity morphology defect parameters and a second verification rule based on the porosity defect distribution information are established. The porosity defect distribution information is then subjected to bidirectional cross-verification with the porosity morphology defect parameters: S1: Deviation analysis is performed based on the equivalent circle diameter of the porosity to determine the deviation parameters. Verification is then performed according to the first verification rule to determine the first verification result. S2: Distribution calculation is performed based on the spatial coordinates of the porosity to determine the spatial distribution density. Verification is then performed according to the second verification rule to determine the second verification result. A bidirectional verification judgment is then performed based on the first and second verification results, and the multi-level porosity defect array is constructed according to the judgment results.

[0052] Furthermore, the level classification module 13 is used to perform the following method: When both the first verification result and the second verification result pass bidirectional verification, the porosity defect distribution information and the porosity morphology defect parameters are determined to be valid data. Then: S1: Set a numerical range according to the equivalent circle diameter of the porosity, and divide multiple porosity size levels according to the numerical range; S2: Perform regional analysis on the aluminum alloy casting according to the spatial coordinates of the porosity, and determine the key areas to divide into multiple porosity density levels; S3: Associate and map the multiple porosity size levels with the multiple porosity density levels to construct the multi-level porosity defect array.

[0053] Furthermore, the level classification module 13 is used to perform the following method: Based on the volume discretization analysis of aluminum alloy castings, a three-dimensional spatial mesh unit is constructed. The multi-level porosity defect array is mapped to the three-dimensional spatial mesh unit to count the number of porosity defects, obtaining the total number of porosity defects in multiple mesh units. The multi-level porosity defect array is mapped to the three-dimensional spatial mesh unit for porosity size analysis. Based on the equivalent circle diameter of porosity in multiple mesh units, the equivalent circle diameter of porosity in the first order is extracted as the maximum equivalent diameter. Based on the total number of porosity defects in multiple mesh units and the maximum equivalent diameter, defect influence analysis is performed to generate a defect influence factor. The multiple mesh units in the three-dimensional spatial mesh unit are weighted according to the defect influence factor to obtain the local defect severity index of multiple mesh units. The aluminum alloy casting is then evaluated to obtain an overall defect severity score. Based on the overall defect severity score, the porosity of the aluminum alloy casting is classified to determine the defect classification result.

[0054] Furthermore, the level classification module 13 is used to perform the following method: The multiple grid cells are traversed and labeled with regional risks according to their functions to obtain multiple regional risk levels; the multiple grid cells are weighted according to the multiple regional risk levels to determine the weight coefficients of the multiple grid cells; based on the weight coefficients of the multiple grid cells and the local defect severity index of the multiple grid cells, local defect analysis is performed on the multiple grid cells to obtain the weighted contribution value of the multiple grid cells; the aluminum alloy casting is evaluated according to the weighted contribution value to obtain the overall defect severity score.

[0055] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0056] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0057] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for classifying and detecting porosity defects in aluminum alloy castings, characterized in that, The method includes: Multi-angle imaging scanning of aluminum alloy castings was performed to obtain tomographic image sequences; Image enhancement is performed on the tomographic image sequence, initial feature information of pore defects is extracted for segmentation and recognition, distribution information of pore defects is obtained for morphological analysis, and morphological defect parameters of pores are obtained. Based on the porosity defect distribution information and the porosity morphology defect parameters, bidirectional data verification is performed to construct a multi-level porosity defect array for classification and to determine the defect classification results. The aluminum alloy castings are graded according to the defect classification results, and then automatically sorted according to the quality level.

2. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 1, characterized in that, Multi-angle imaging scanning of aluminum alloy castings to obtain tomographic image sequences, the method includes: A two-dimensional plane coordinate system for aluminum alloy castings is constructed, and key point analysis is performed by traversing the two-dimensional plane coordinate system to determine multiple effective imaging field points; Based on the multiple effective imaging field-of-view points, cluster the center points to determine the effective field-of-view center; The aluminum alloy casting is scanned according to the effective field of view center and the multiple effective imaging field of view points to construct an initial two-dimensional projection image; The initial two-dimensional projection image is scanned by step rotation along the effective field of view center to obtain a multi-angle image acquisition sequence; The cross-sectional images of the aluminum alloy casting are reconstructed according to the multi-angle image acquisition sequence to construct the tomographic image sequence.

3. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 1, characterized in that, The method includes performing image enhancement on the tomographic image sequence, extracting initial feature information of pore defects for segmentation and identification, and obtaining pore defect distribution information. The tomographic image sequence is traversed and nonlocal mean filtering is performed to extract multiple image edge structures; The tomographic image sequence is contrast-enhanced according to the multiple image edge structures to generate a contrast-enhanced image set. Based on the contrast-enhanced image set, the tomographic image sequence is combined with historical porosity defect images to perform initial feature screening and determine the initial feature information of porosity defects. Boundary identification is performed based on the initial feature information of the pore defects to determine the boundary of the pore region. The pore regions are segmented and identified according to their boundaries. Multiple pore regions are identified and pore traversal records are made to obtain pore defect distribution information.

4. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 3, characterized in that, Boundary identification based on the initial feature information of the pore defects to determine the boundary of the pore region includes the following method: Local neighborhood grayscale analysis is performed based on the initial feature information of the porosity defects to determine the segmentation threshold; The tomographic image sequence is segmented according to the segmentation threshold to generate a binarized image for stomatal identification, thereby identifying multiple suspected stomatal regions. Connectivity determination is performed on the multiple suspected stomatal regions, and the connected suspected stomatal regions are used as seed points; Using the seed point as the center, the grayscale gradient of the surrounding pixels is extracted for region growth to determine the boundary of the pore region.

5. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 3, characterized in that, Methods include: obtaining morphological analysis of porosity defect distribution information to obtain porosity morphological defect parameters; and other methods. The area of ​​the multiple pore regions is calculated by traversing them to determine the pixel area data of the multiple regions. Pixel conversion is performed based on the boundary of the pore region to obtain the total number of boundary pixels. The pixel area data of the multiple regions and the total number of the multiple boundary pixels are converted according to the spatial calibration coefficient to obtain the actual physical area and the actual physical perimeter. Based on the actual physical area and the actual physical perimeter, the morphological analysis of the pores is performed to calculate the equivalent circle diameter of the pores, thereby obtaining the pore morphological defect parameters.

6. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 5, characterized in that, A multi-level pore defect array is constructed by performing bidirectional data verification based on the pore defect distribution information and the pore morphology defect parameters, the method including: Based on the porosity defect distribution information, the location of the pores is identified, and the spatial coordinates of the pores are determined. Establish a first verification rule based on the pore morphology defect parameters and a second verification rule based on the pore defect distribution information, and perform bidirectional cross-validation on the pore defect distribution information in conjunction with the pore morphology defect parameters: S1: Based on the equivalent circle diameter of the pores, perform deviation analysis, determine the deviation parameters, verify them according to the first verification rule, and determine the first verification result; S2: Based on the spatial coordinates of the pores, perform distribution calculations to determine the spatial distribution density, verify according to the second verification rule, and determine the second verification result; Based on the first verification result and the second verification result, a bidirectional verification judgment is performed, and the multi-level defect array of pores is constructed according to the judgment result.

7. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 6, characterized in that, Based on the first verification result and the second verification result, a bidirectional verification judgment is performed, and the multi-level pore defect array is constructed according to the judgment result. The method includes: When both the first verification result and the second verification result pass bidirectional verification, the porosity defect distribution information and the porosity morphology defect parameters are determined to be valid data. S1: Set a numerical range based on the equivalent circle diameter of the pores, and divide multiple pore size levels according to the numerical range; S2: Based on the spatial coordinates of the pores, perform regional analysis on the aluminum alloy casting to determine the key areas and divide them into multiple pore density levels; S3: Associate and map the multiple pore size levels with the multiple pore density levels to construct the multi-level pore defect array.

8. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 1, characterized in that, A multi-level array of pore defects is constructed for classification, and the defect classification results are determined. The methods include: Based on the volume discretization analysis of aluminum alloy castings, a three-dimensional spatial mesh element is constructed. The porosity multi-level defect array is mapped to the three-dimensional spatial grid cell to count the number of porosity defects, and the total number of porosity defects in multiple grid cells is obtained. The multi-level defect array of pores is mapped to the three-dimensional spatial grid cell for pore size analysis. The equivalent circle diameter of the pores in multiple grid cells is sorted in descending order, and the equivalent circle diameter of the first-order pore is extracted as the maximum equivalent diameter. Based on the total number of pore defects in multiple grid cells and the maximum equivalent diameter, a defect impact analysis is performed to generate a defect impact factor. The local defect severity index of the multiple grid cells in the three-dimensional spatial grid cell is obtained by weighting the defect impact factor according to the defect influence factor, and the aluminum alloy casting is evaluated to obtain the overall defect severity score. The porosity of the aluminum alloy casting is classified according to the overall defect severity score, and the defect classification result is determined.

9. The method for graded detection of porosity defects in aluminum alloy castings as described in claim 8, characterized in that, To evaluate aluminum alloy castings, the local defect severity index of multiple mesh elements is obtained to obtain an overall defect severity score. The method includes: By traversing the multiple grid cells and labeling the regional risks according to their functions, multiple regional risk levels are obtained; The weights of the multiple grid cells are assigned according to the risk levels of the multiple regions, and the weight coefficients of the multiple grid cells are determined. Based on the weighting coefficients of multiple grid cells and the local defect severity index of multiple grid cells, local defect analysis is performed on multiple grid cells to obtain the weighted contribution value of multiple grid cells; The aluminum alloy casting is evaluated based on the weighted contribution value to obtain the overall defect severity score.

10. A grading and detection system for porosity defects in aluminum alloy castings, characterized in that, The system for implementing the graded detection method for porosity defects in aluminum alloy castings according to any one of claims 1-9, the system comprising: Image acquisition module: Performs multi-angle imaging scans on aluminum alloy castings to acquire tomographic image sequences; Analysis module: performs image enhancement on the tomographic image sequence, extracts initial feature information of pore defects for segmentation and recognition, obtains pore defect distribution information for morphological analysis, and obtains pore morphological defect parameters; Grading module: Based on the porosity defect distribution information and the porosity morphology defect parameters, the data is verified bidirectionally to construct a multi-level porosity defect array for grading and to determine the defect grading result; Casting sorting module: Based on the defect classification results, aluminum alloy castings are graded for quality and automatically sorted according to the quality level.

Images