Degradation performance analysis method for characterizing material defect details through ossification tree

Abstract

The invention relates to the field of material damage mechanics, in particular to a deterioration performance analysis method for characterizing material defect details through a ossification tree. The method comprises the following steps: acquiring three-dimensional shape data of internal defects of a material through microscopic CT scanning, extracting defect features in combination with an image processing algorithm, converting the defect features into an ossification tree structure, and constructing a defect heterogeneous tensor model; a single-defect sample is accurately reproduced by adopting a 3D printing technology, a rigidity degradation index equation is established through a mechanical test and symbolic regression analysis, and the influence of defects on the material performance is quantified. According to the method, on the basis of K-means clustering and an auto-encoder network, auto-encoding fingerprints of defects are generated through a feature weight matrix and an adjacent matrix, and clustering analysis of six types of defects such as annular defects and sickle-shaped defects is achieved; the 3D printing technology is adopted, the original shape of the defect is accurately re-engraved, and preparation of a single-defect material is achieved. And through symbolic regression, a one-dimensional bubble rigidity deterioration index equation is provided, and the influence of material defects on material deterioration is quantified.

Description

Technical Field

[0001] The present invention relates to the field of material damage mechanics, and in particular to a degradation performance analysis method for characterizing material defect details using an ossification tree. Background Art

[0002] Traditional material defect analysis primarily relies on the following methods: Microscopic imaging techniques, such as SEM (scanning electron microscopy), TEM (transmission electron microscopy), and X-ray tomography (CT), are used to observe the morphology and distribution of defects. Nondestructive testing techniques, such as ultrasonic testing, eddy current testing, and infrared thermal imaging, are used to identify internal defects. Computational simulations, such as finite element analysis (FEA) and phase field methods, simulate defect evolution. Statistical methods, based on the Weibull distribution and fractal theory, quantify the randomness and complexity of defects.

[0003] Traditional techniques have drawbacks. Modeling relies on Euclidean geometry simplification, resulting in a loss of geometric detail. Noise in CT images can be misinterpreted as defect branches during morphological processing, leading to redundant skeleton nodes or incorrect connections. Fractal dimension calculation relies on box counting, which can overlook the local curvature of defects, resulting in a weak correlation between fractal dimension and stress concentration areas. Skeletonization algorithms are sensitive to image noise and edge discontinuities, potentially distorting the topology of the ossification tree. Furthermore, the highly irregular nature of defect morphology conflicts with traditional parametric modeling methods, resulting in a lack of a defect characterization system that integrates multi-scale features. Furthermore, experimental verification cannot precisely control a single defect variable. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a degradation performance analysis method for characterizing the details of material defects by using an ossification tree. The present invention obtains three-dimensional morphological data of internal defects of the material through micro-CT scanning, extracts defect features and converts them into an ossification tree structure in combination with an image processing algorithm, and constructs a defect heterogeneous tensor model. 3D printing technology is used to accurately reproduce single defect specimens. Through mechanical testing and symbolic regression analysis, a stiffness degradation index equation is established to quantify the impact of defects on material properties.

[0005] The technical solutions adopted are:

[0006] A degradation performance analysis method for characterizing material defect details using an ossification tree comprises the following steps:

[0007] (1) Use an X-ray three-dimensional microscope scanning platform to perform CT scanning on the interior of the material;

[0008] (2) Using the threshold segmentation algorithm, defects are screened and seed point regions are segmented to extract material defects;

[0009] (3) Perform grayscale processing, corrosion and dilation morphological processing on the image material to eliminate noise in the image and fill small holes;

[0010] (4) Construct a defect heterogeneous tensor model and use the skeletonization algorithm to further describe the defects in the composite material and skeletonize the defects into a tree;

[0011] (5) The skeleton of the defect is regarded as the trunk or branch of the tree, and the connection points of the skeleton are regarded as the nodes of the tree to form a tree network; based on K-means clustering and autoencoder network, the autoencoder fingerprint of the defect is generated through the feature weight matrix and adjacency matrix to realize defect clustering analysis;

[0012] (6) For a single defect in 3D printing, a magnification relationship is constructed based on the printing resolution and the minimum curvature radius of the defect;

[0013] (7) Conduct tensile tests on single defects and finally perform degradation performance analysis.

[0014] Preferably, in step (1), the resolution is 20 to 32 μm, and the material is fixed on the scanning stage.

[0015] Preferably, a threshold segmentation algorithm is used to divide the pixels in the image into defective solid skeletons. Assuming the original image is I and the threshold is T, the binary image B is calculated by the following formula:

[0016]

[0017] Where (x,y) represents the pixel position in the image.

[0018] Preferably, seed point region segmentation selects one or more seed points as the initial region, determines the criteria for whether adjacent pixels belong to the same region, and adds adjacent pixels that meet the similarity criteria to the current region until no new pixels can be added;

[0019] Let the initial seed point set be S, and the initial state of region R be R0 = S. The similarity criterion can be defined as:

[0020]

[0021] Where: I(x,y) is the grayscale value of pixel (x,y); mean(R k ) is the region R k The average gray value of; T is the similarity threshold;

[0022] The region expansion process can be expressed as:

[0023] R k+1 =R k ∪{(x,y)|C(x,y,R k)=1};

[0024] Until R k+1 =R k , that is, no new pixels can be added, so a single defect is extracted.

[0025] Preferably, the defect performance is converted into a node representation and an edge representation of an ossification tree; the image skeleton is obtained by iteratively deleting pixels on the boundary of the target object until no more pixels can be deleted, and the skeleton structure is obtained by a skeleton extraction algorithm. The defect skeleton tree image is effectively distinguished by introducing node cosine similarity to effectively distinguish the role of the node in the network. The K-means clustering algorithm is used to implement cluster analysis of defects through degree centrality. The expression is as follows:

[0026]

[0027] Where V is the representation of the node clustering formula; argmin is the search for all clusters to find the cluster that minimizes a certain function; n is the number of nodes; c is the number of nodes. i is the degree centrality of the node; μ c is the center of the cluster, which is the average value of all nodes; k i is the number of existing edges connected to node i.

[0028] Preferably, construct a defect heterogeneous tensor:

[0029] S=W[D,AR,F,C,r min ];

[0030] Where D is the defect diameter, which refers to the maximum size of the defect, usually the distance from one edge of the defect to the other edge; AR is the defect aspect ratio, which refers to the ratio of the longest vertical chord of the defect to the diameter; F is the defect fractal dimension, which can quantitatively describe extremely irregular shapes and can be used to study defects with multi-level and multi-scale fractal characteristics; C is the degree centrality, which is used to describe the importance of the node; r min is the minimum radius of curvature.

[0031] Preferably, each defect heterogeneous tensor is as follows:

[0032]

[0033] Heterogeneous tensors representing ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped and long strip defects respectively.

[0034] Preferably, when 3D printing a single defect, the defect is embedded in the sample model to form a bubble defect, which is then loaded into the slicing software. The slicing software is used to divide the 3D model into thin layers and set the printing parameters to generate a two-dimensional image of each layer so that the subsequent light source can cure the resin layer by layer; secondly, the resin is introduced into the resin tank of the printer to ensure that there are no impurities; finally, printing is started, and the printer projects the two-dimensional image of each slice onto the liquid resin through the projection system, which is exposed to the light source to trigger a photocuring reaction and solidify the layer; the printing platform gradually moves upward, and the slice image of the next layer is projected and cured, and so on, stacked layer by layer until the entire model is printed.

[0035] Preferably, the minimum print size is determined in combination with the printer resolution, and the formula for the magnification M is:

[0036]

[0037] Where r min Minimum curvature radius; A is the printing area size; P is the printer resolution.

[0038] Preferably, the influence of different defects on the mechanical properties of polymer materials is analyzed, and the degradation degree curves of different defects are determined by the chi-square distribution probability density function χ 2 (ε,b), exponential term e -dε The regression equation RDIE is established by the three terms of constant term k, and its expression is:

[0039] R die =a·χ 2 (ε,b)+c·e -dε +k

[0040] Where a is the coefficient of the chi-square distribution probability density function, b is the degree of freedom parameter of the chi-square distribution, c is the coefficient of the exponential function, d is the exponent of the exponential function, and k is the constant term;

[0041] χ 2 (ε,b) is the chi-square distribution probability density function, and its mathematical expression is as follows:

[0042]

[0043] Γ(x) is the gamma function, and its mathematical expression is as follows:

[0044]

[0045] The stiffness degradation exponential equations for ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped, and long strip defects are R die :

[0046]

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] The present invention extracts micron-scale defect features through CT scanning and skeletonization algorithms, and combines heterogeneous tensor models to achieve cross-scale mapping from microscopic morphology to macroscopic mechanical properties, successfully quantifying the influence of large-diameter defects on the overall stiffness of the specimen.

[0049] Based on K-means clustering and an autoencoder network, the method generates an autoencoder fingerprint of defects through the feature weight matrix and adjacency matrix, enabling cluster analysis of six types of defects, including ring-shaped and sickle-shaped defects. Through symbolic regression, a one-dimensional bubble stiffness degradation index equation is proposed to quantify the impact of material defects on material degradation.

[0050] A skeletonized tree structure is generated through a skeletonization algorithm. A tensor model is constructed by combining five types of heterogeneous parameters, including diameter, aspect ratio, fractal dimension, degree centrality, and minimum curvature radius, to achieve a multi-scale fusion representation of geometric shape, topological network, and mechanical properties.

[0051] The first similarity principle of 3D printing technology can accurately replicate the original shape of defects and realize the preparation of single defect materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Flowchart of CT reconstruction, image feature extraction and bubble defect skeletonization of the present invention.

[0053] Figure 2 Schematic diagram of defect extraction and skeletonization of the present invention.

[0054] Figure 3 Schematic diagram of tree heterogeneous tensor parameters of the present invention.

[0055] Figure 4 The stiffness degradation index equation of the present invention is fitted into a curve.

[0056] Figure 5 Overall flow chart of the present invention. DETAILED DESCRIPTION

[0057] The accompanying drawings are only used for illustrative purposes. The present invention will be further described below in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are only preferred examples of the present invention and do not represent the entire technical content of the present invention.

[0058] Example 1

[0059] The present invention proposes a method for characterizing material defect details and degradation performance analysis using an ossification tree, which aims to characterize and analyze the internal defects of composite materials. The method specifically includes the following steps:

[0060] Step S1: Obtain the three-dimensional morphology data of the internal defects of the material through micro-CT scanning, and place it on the X-ray three-dimensional microscope scanning stage with a resolution of 32μm. In order to ensure the accuracy of the scan, the carbon fiber should be kept as fixed as possible during the rotation of the sample stage. Figure 1 As shown (the image is a cross-section of the test material), the gray part is the carbon fiber solid material, the black part is the irregular pore / crack structure, and the material porosity is 2.14%;

[0061] Step S2: To avoid processing a huge amount of CT data, an area with more defect types is selected for cropping research, and the local cube area of ​​the defect in the sample is selected as the research object;

[0062] Step S3: Multi-step processing of the material defect image:

[0063] (1) Use the threshold segmentation algorithm to divide the pixels in the image into foreground (defects) and background (solid skeleton). Let the original image be I and the threshold be T, then the binary image B can be calculated by the following formula:

[0064]

[0065] Where (x,y) represents the pixel position in the image.

[0066] (2) Region growing is a segmentation method based on seed points. It forms regions by gradually expanding pixels that meet the similarity criteria, thus extracting single defects. The details are as follows:

[0067] Select one or more seed points in the image as the initial region, determine the criteria for whether adjacent pixels belong to the same region (such as the gray value difference is less than a certain threshold), and add adjacent pixels that meet the similarity criteria to the current region until no new pixels can be added. Let the initial seed point set be S, and the initial state of the region R be R0 = S. The similarity criterion can be defined as:

[0068]

[0069] Where: I(x,y) is the grayscale value of pixel (x,y). mean(R k ) is the region R k The average gray value of . T is the similarity threshold. The region expansion process can be expressed as:

[0070] R k+1 =R k ∪{(x,y)|C(x,y,R k )=1};

[0071] Until R k+1 =R k, that is, no new pixels can be added, so a single defect is extracted.

[0072] (3) The extracted single defect image is then grayscaled. There are three common methods for image grayscale processing: maximum method, average method, and weighted average method. Here, the three components are weighted averaged with different weights according to importance and other indicators. Since the human eye is most sensitive to green and least sensitive to blue, a more reasonable grayscale image can be obtained by weighted averaging the three RGB components according to the following formula:

[0073] I(x,y)=0.299·R(x,y)+0.587·G(x,y)+0.114·B(x,y).

[0074] (4) Through the combined operation of corrosion followed by expansion, the opening operation is realized:

[0075]

[0076] in, Represents the erosion operation on the image, Represents the dilation operation on the image, where B is the structure element.

[0077] The opening operation can effectively eliminate isolated noise points outside the defect area (such as burrs and stray defects smaller than the structural elements), while keeping the geometric shape of the main area unchanged, meeting the conformal requirements, and further improving the image quality.

[0078] Step S4: Using the skeleton algorithm of image processing, a new concept of defect representation, the "ossification tree", is proposed. The similarity between complex networks, fractals, and defects is used to establish a "tree heterogeneous tensor" containing diameter, aspect ratio, fractal dimension, degree centrality, and minimum curvature radius, etc., and the defect properties are converted into node representation and edge representation of the ossification tree;

[0079] After image preprocessing, we obtain a binary image of the material defect. After removing some points from the binary image, the remaining points can still maintain the original shape. By iteratively deleting pixels on the boundary of the target object until no more pixels can be deleted, we can obtain the skeleton of the image. The specific operation is as follows:

[0080] Assume that the neighboring pixels of the current pixel P1 are P2, P3...P9 arranged in a clockwise direction, completing the following two stages:

[0081] Phase 1:

[0082] a)2≤B(P1)≤6;

[0083] b) A(P1) = 1;

[0084] c) P2*P4*P6=0;

[0085] d) P4*P6*P8=0;

[0086] Where B(P1) is the number of non-zero neighboring pixels of P, and A(P1) is the number of times the value of P changes from 0 to 1 in the 8-neighboring pixels of P.

[0087] The pixels that meet the above conditions are marked as pixels to be deleted and their pixel values ​​are set to 0.

[0088] Phase 2:

[0089] a)2≤B(P1)≤6;

[0090] b) A(P1) = 1;

[0091] c) P2*P4*P8=0;

[0092] d) P2*P6*P8=0;

[0093] The only differences between the two phases are steps (c) and (d). Phase 1 (c) and (d) remove points on the east and south boundary lines, as well as corner points in the northwest corner. Phase 2 (c) and (d) remove points on the west and north boundary lines, as well as corner points in the southeast corner. For a binary image, the algorithm repeatedly iterates through phases 1 and 2 until no new pixels are removed at a certain point, at which point the algorithm terminates.

[0094] Assume B k (x,y) is the image after the kth iteration, M1 and M2 are the sets of labeled pixels in the two stages respectively, then:

[0095]

[0096] Assume that the input defect binary image is The skeleton structure is obtained through the skeleton extraction algorithm:

[0097] S=Skeleton(D)

[0098] Its mathematical expression is:

[0099] S = {p∈D|p is a skeleton pixel that satisfies the skeleton definition conditions}

[0100] In the process of characterizing material defects, complex tree structures often contain a large number of irrelevant or redundant details and may even form loops, which leads to data redundancy and increased computational complexity. Therefore, the MST algorithm is introduced to simplify the geometric representation of defects and break the loop structure in the network. The MST algorithm removes redundant edges to eliminate all loops and reduce redundant information while retaining key topological features. The specific formula is as follows:

[0101]

[0102] In the formula, argmin means finding an optimal edge set E' so that the sum of the selected edge weights is minimized; Represents the edge set of the minimum spanning tree, which is the edge set E of the original graph 0 subset of ; w(e) represents the weight of an edge e.

[0103] Convert the skeleton S into an acyclic tree structure: T = (V, E), where:

[0104] It is a set of nodes, corresponding to the endpoints, branch points and key points of the skeleton;

[0105] is an edge set, representing the skeleton connection relationship;

[0106] Node V i The coordinates are Edge ij The Euclidean length is l ij =‖x i -x j ‖2.

[0107] By introducing node cosine similarity, we can effectively distinguish the role of nodes in the network, namely, whether node A is at the intersection of edges AB and AC or located somewhere on edge BC. The introduction of cosine similarity provides a mathematical basis for accurately determining node positions by analyzing the angular relationship between vectors. Two key thresholds, 10° and 170°, are set to determine whether AB and AC are collinear. If 10° < θ < 170°, point A is a node; if 0° < θ < 10° or 170° < θ < 180°, point A is not a node. The thresholds of 10° and 170° are chosen because they effectively ensure the positional relationship of node A and provide high precision and sensitivity. 10°, as the upper limit for collinearity, accurately identifies nearly parallel vectors, ensuring that node A is considered to be on edge BC when the angle is less than this value. The lower limit of 170° covers the case of reverse collinearity, preventing these special cases from being missed. Setting these two thresholds avoids ambiguous judgments, ensuring clear boundaries while also accounting for the geometric characteristics of real-world scenarios, making the algorithm more practical and flexible.

[0108]

[0109] Where V AB ·V AC Expressed as the dot product of edge vectors; |V AB ||V AC | represents the product of the edge vector norms.

[0110] The defect skeleton tree image is clustered using the K-means clustering algorithm to implement defect clustering analysis through degree centrality. The expression is as follows:

[0111]

[0112] Where V is the representation of the node clustering formula; argmin is the search for all clusters to find the cluster that minimizes a certain function; n is the number of nodes; c is the number of nodes. i is the degree centrality of the node; μ c is the center of the cluster, which is the average value of all nodes; k i is the number of existing edges connected to node i.

[0113] Parameters including but not limited to diameter, aspect ratio, fractal dimension, degree centrality, and minimum curvature radius are used to construct the defect heterogeneous tensor:

[0114] S=W[D,AR,F,C,r min ];

[0115] Where D is the defect diameter, which refers to the maximum size of the defect, usually the distance from one edge of the defect to the other edge; AR is the defect aspect ratio, which refers to the ratio of the longest vertical chord of the defect to the diameter; F is the defect fractal dimension, which can quantitatively describe extremely irregular shapes and can be used to study defects with multi-level and multi-scale fractal characteristics; C is the degree centrality, which is used to describe the importance of the node; r min is the minimum radius of curvature.

[0116] 1) Node V i The local area diameter is defined as the neighborhood N(v i ) The distance between the farthest two points:

[0117]

[0118] Where N(v i ) is based on v i The largest inscribed circle neighborhood centered at .

[0119] 2) AR is the defect aspect ratio, which refers to the ratio of the longest vertical chord of the bubble to its diameter:

[0120]

[0121] Where H is the longest chord of the bubble in the vertical direction (i.e., the length of the minor axis), and D is the diameter of the bubble (i.e., the length of the major axis).

[0122] 3) Use the box counting method to calculate the fractal dimension of the node neighborhood:

[0123]

[0124] where N(δ) is the number of boxes with side length δ covering the node area.

[0125] 4) Since there are many nodes in the ossified tree network, the degree centrality is calculated by selecting the most important nodes:

[0126]

[0127] Among them, e ij is the edge between nodes i and j.

[0128] 5) The minimum curvature radius is the radius of the skeleton curve at v i The curvature k(v i ) is the reciprocal of:

[0129]

[0130] Where y′ and y″ are the first and second order derivatives respectively.

[0131] Step S5: Treat the defect skeleton as the trunk or branches of a tree, and the connection points of the skeleton as the nodes of the tree, forming a tree network. The ossification tree representation method not only highly restores the shape characteristics of the defect, but also the tree adjacency matrix can achieve heterogeneous representation of defect details.

[0132] Parameters are shown as follows Figure 3 As shown, the heterogeneous tensors of each defect are as follows:

[0133]

[0134] Heterogeneous tensors representing ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped and long strip defects respectively.

[0135] Step S6: 3D printing a single defect, the preparation of the test material is completed through the digital light processing (DLP) technology in 3D printing technology. First, the reconstructed defect is embedded in the specimen model to form a bubble defect, and then loaded into the slicing software. The slicing software is used to divide the 3D model into thin layers and set the printing parameters, including layer height, curing time, etc. The slice size is usually between tens and hundreds of microns, which will generate a two-dimensional image of each layer so that the subsequent light source can cure the resin layer by layer. Secondly, the resin is introduced into the resin tank of the printer to ensure that there are no impurities. Finally, the printing begins. The printer projects the two-dimensional image of each slice onto the liquid resin through the projection system, exposes it to the light source, triggers the photocuring reaction, and solidifies the layer. The printing platform gradually moves upward, and the slice image of the next layer is projected and cured. And so on, stacked layer by layer until the entire model is printed.

[0136] In order to study the influence of different types of bubble defects on the degradation performance of the matrix material, we define a magnification when the defect is embedded inside the matrix material. Based on the combination of Canny edge detection and contour tracing algorithm, we find the contour boundary in the defect cross-section contour image and extract it. The minimum curvature radius r of the contour is calculated. min Calculate and mark its position, so that we get the formula to determine the minimum print size combined with the printer resolution. The formula for the magnification M is:

[0137]

[0138] Where r min Minimum curvature radius; A is the printing area size; P is the printer resolution.

[0139] Step S7: The test uses a servo-controlled tensile testing machine with a model of AI-7000-MU1 and a maximum load of 5KN. The relevant tests are carried out in accordance with the test standards. The sample length is 150mm, the clamping end width is 20mm, the middle stretching part width is 10mm, the total height is 6mm, and the mark distance is 90mm. The measured sample is clamped with the clamp of the tensile testing machine to ensure that the clamp is aligned with the mark distance. The test is carried out at room temperature (20°C) at a stretching speed of 1mm / min. When the load reaches a certain level, the load borne by the defective part reaches the limit, resulting in fracture. The fracture data is measured, and the effects of different defects on the mechanical properties of the polymer material are analyzed. Through the integration and processing of the test data, the defective sample result analysis curve during the stretching process is made as shown in the figure. Through the integration and processing of the test data, the degradation degree curve of different defects is determined with the chi-square distribution probability density function χ 2 (ε,b), exponential term e -dε The regression equation RDIE is established by the three terms of constant term k, and its expression is:

[0140] R die =a·χ 2 (ε,b)+c·e -dε +k;

[0141] Where a is the chi-square distribution probability density function coefficient, b is the chi-square distribution degree of freedom parameter, c is the exponential function coefficient, d is the exponent of the exponential function, and k is the constant term.

[0142] χ 2 (ε,b) is the chi-square distribution probability density function, and its mathematical expression is as follows:

[0143]

[0144] Γ(x) is the gamma function, and its mathematical expression is as follows:

[0145]

[0146] The stiffness degradation exponential equations for ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped, and long strip defects are R die :

[0147]

[0148] Introducing Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Coefficient of Determination (R 2 ) and Mean Absolute Percentage Error (MAPE) are used to verify the effect of fitting. The formulas for these indicators are as follows:

[0149]

[0150] Where n is the amount of data, is the actual value of the i-th data, and σ i is the corresponding fitted value, is the average value of the data. The smaller the MSE, RMSE and MAPE, the better the R 2 The closer it is to 1, the better the fitting effect. Combining the curves, we quantified the impact of a single material defect on material degradation.

[0151] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the scope of protection of the present invention.

Claims

1. A degradation performance analysis method for characterizing material defect details using an ossification tree, characterized in that: The steps include: (1) Use an X-ray three-dimensional microscope scanning platform to perform CT scanning on the interior of the material; (2) Using the threshold segmentation algorithm, defects are screened and seed point regions are segmented to extract material defects; (3) Perform grayscale processing, corrosion and dilation morphological processing on the image material to eliminate noise in the image and fill small holes; (4) Construct a defect heterogeneous tensor model and use the skeletonization algorithm to further describe the defects in the composite material and skeletonize the defects into a tree; (5) The skeleton of the defect is regarded as the trunk or branch of the tree, and the connection points of the skeleton are regarded as the nodes of the tree to form a tree network; based on K-means clustering and autoencoder network, the autoencoder fingerprint of the defect is generated through the feature weight matrix and adjacency matrix to realize defect clustering analysis; (6) For a single defect in 3D printing, a magnification relationship is constructed based on the printing resolution and the minimum curvature radius of the defect; (7) Conduct tensile tests on single defects and finally perform degradation performance analysis.

2. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 1, characterized in that: In the step (1), the resolution is 20 to 32 μm, and the material is fixed on the scanning stage.

3. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 1, characterized in that: Use the threshold segmentation algorithm to divide the pixels in the image into defect solid skeletons. Let the original image be I and the threshold be T. Then the binary image B is calculated by the following formula: Where (x,y) represents the pixel position in the image.

4. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 1, characterized in that: Seed point region segmentation, select one or more seed points as the initial region, determine the criteria for whether adjacent pixels belong to the same region, and add adjacent pixels that meet the similarity criteria to the current region until no new pixels can be added; Let the initial seed point set be S, and the initial state of region R be R0 = S. The similarity criterion can be defined as: Where: I(x,y) is the grayscale value of pixel (x,y); mean(R k ) is the region R k The average gray value of; T is the similarity threshold; The region expansion process can be expressed as: R k+1 =R k ∪{(x,y)∣C(x,y,R k )=1}; Until R k+1 =R k , that is, no new pixels can be added, so a single defect is extracted.

5. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 1, characterized in that: Convert defect performance into node representation and edge representation of ossification tree; By iteratively deleting pixels on the boundary of the target object until no more pixels can be deleted, the skeleton of the image is obtained. The skeleton structure is obtained through the skeleton extraction algorithm. The defect skeleton tree image is effectively distinguished by introducing the node cosine similarity to effectively distinguish the role of the node in the network. The K-means clustering algorithm is used to realize the clustering analysis of defects through degree centrality. The expression is as follows: Where V is the representation of the node clustering formula; argmin is the search for all clusters to find the cluster that minimizes a certain function; n is the number of nodes; c is the number of nodes. i is the degree centrality of the node; μ c is the center of the cluster, which is the average value of all nodes; k i is the number of existing edges connected to node i.

6. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 1, characterized in that: Construct defect heterogeneous tensor: S=W[D,AR,F,C,r min ]; Where D is the defect diameter, which refers to the maximum size of the defect, usually the distance from one edge of the defect to the other edge; AR is the defect aspect ratio, which refers to the ratio of the longest vertical chord of the defect to the diameter; F is the defect fractal dimension, which can quantitatively describe extremely irregular shapes and can be used to study defects with multi-level and multi-scale fractal characteristics; C is the degree centrality, which is used to describe the importance of the node; r min is the minimum radius of curvature.

7. The degradation performance analysis method for characterizing material defect details by an ossification tree according to claim 6 is characterized in that The heterogeneous tensors of each defect are as follows: Heterogeneous tensors representing ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped and long strip defects respectively.

8. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 7, characterized in that: When 3D printing a single defect, the defect is embedded in the specimen model to form a bubble defect. The sample is then loaded into the slicing software, which divides the 3D model into thin layers and sets printing parameters to generate a two-dimensional image of each layer so that the subsequent light source can cure the resin layer by layer. Next, introduce the resin into the printer's resin tank to ensure there are no impurities; Finally, printing begins. The printer projects a two-dimensional image of each slice onto the liquid resin through a projection system. When exposed to a light source, it triggers a photocuring reaction and solidifies the layer. The printing platform gradually moves upward, and the slice image of the next layer is projected and solidified, and so on, stacked layer by layer until the entire model is printed.

9. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 8, characterized in that: The minimum print size is determined in combination with the printer resolution. The formula for the magnification M is: Where r min Minimum curvature radius; A is the printing area size; P is the printer resolution.

10. The degradation performance analysis method for characterizing material defect details using an ossified tree according to claim 9, characterized in that: Analyze the influence of different defects on the mechanical properties of polymer materials. By integrating and processing the test data, the analysis curve of defective specimens during the tensile process is made as shown in the figure. By integrating and processing the test data, the chi-square distribution probability density function χ is determined for the degradation degree curves of different defects. 2 (ε,b), exponential term e -dε The regression equation RDIE is established by the three terms of constant term k, and its expression is: R die =a·x 2 (e,b)+c·e -dε +k Where a is the coefficient of the chi-square distribution probability density function, b is the degree of freedom parameter of the chi-square distribution, c is the coefficient of the exponential function, d is the exponent of the exponential function, and k is the constant term; χ 2 (ε,b) is the chi-square distribution probability density function, and its mathematical expression is as follows: Γ(x) is the gamma function, and its mathematical expression is as follows: The stiffness degradation exponential equations for ring-shaped, sickle-shaped, hump-shaped, V-shaped, cone-shaped, and long strip defects are R die :

Images