Method for generating two-dimensional finite element random model of composite material image with controllable parameters

Abstract

The invention relates to a parameter-controllable composite material image two-dimensional finite element random model generation method, which comprises the following process steps of: 1) acquiring an initial state XCT (X-ray Computed Tomography) image and preprocessing to finish manual labeling of each component in a source image; 2) qualitatively selecting characteristic parameters of each component and carrying out quantitative analysis; (3) an initial yarn model is created, and random deformation is given; 4) adding contact attributes based on a numerical simulation method; and 5) endowing the yarn with defect attributes, inserting the yarn into pores, and performing simulation calculation based on a numerical simulation method. According to the parameter-controllable two-dimensional finite element random model generation method for the composite material image, the action mechanism of a single variable on the mechanical property and damage evolution process of the material can be revealed on the microscale by controlling the structural parameters, and then an effective evaluation means can be provided for predicting the macroscopic mechanical property of the material.

Description

Technical field

[0001] The present invention relates to a method for generating a random image model, in particular to a method for generating a parameter-controllable two-dimensional finite element random model of a composite material image, and belongs to the technical field of finite element analysis. [Background Technology]

[0002] Basalt fiber-reinforced composites (BFRP) are composite products made with basalt fiber as reinforcement, typically using a resin transfer molding (RTM) process. With excellent functional structural characteristics such as lightweight and customizable performance, BFRPs are gradually replacing some metal and alloy materials and finding widespread application in transportation, aerospace, and military defense. They are of great significance to the upgrading of my country's manufacturing industry and the transformation of its lightweight industry.

[0003] X-ray Computed Tomography (XCT) scanning is an X-ray-based nondestructive testing technique that provides high-resolution information about a material's internal structure. XCT scanning can reveal the microstructure of basalt fiber composites. XCT scanning allows for nondestructive characterization of the composite's internal microstructure. This provides an important means of acquiring image data for further characterization of the composite's internal microstructure.

[0004] At present, ideal geometric models are mainly generated through parameterized forms. However, yarns generated based on ideal geometric models often show significant characteristics such as simplified deformation and repeated local features. This is because the deformation characteristics of yarns are usually limited by mathematical functions, and the local compression deformation caused by the interaction between yarns or yarns and the matrix during the manufacturing process is seriously ignored. In addition, the mechanical properties and damage evolution mechanism of materials are also closely related to the morphological characteristics, volume fraction and distribution of pores. Most existing research systems focus on the microscale and usually define the pore geometry as idealized features such as rectangles, triangles and circles, thereby ignoring the randomness and authenticity of pore characteristics.

[0005] Therefore, in order to solve the above problems, it is necessary to provide an innovative parameter-controllable two-dimensional finite element random model generation method for composite material images to overcome the above defects in the prior art. [Summary of the invention]

[0006] The purpose of the present invention is to provide a parameter-controllable two-dimensional finite element random model generation method for composite material images, which can be used to quickly generate random models with real properties. By controlling the structural parameters, the mechanism of action of a single variable on the mechanical properties and damage evolution process of the material can be revealed at the microscopic scale, thereby providing an effective evaluation method for predicting the macroscopic mechanical properties of the material.

[0007] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a method for generating a two-dimensional finite element random model of a composite material image with controllable parameters, which includes the following process steps:

[0008] 1) Acquire the initial XCT image and preprocess it, and complete the manual annotation of each component in the source image;

[0009] 2) qualitatively select characteristic parameters of each component and conduct quantitative analysis;

[0010] 3), creating an initial yarn model and giving it random deformation;

[0011] 4) Adding contact properties based on numerical simulation methods;

[0012] 5), assign yarn defect properties and insert pores, and perform simulation calculations based on numerical simulation methods.

[0013] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the step 1) specifically comprises:

[0014] 1-1), obtaining complete XCT source image data, and cropping and filtering the XCT scan source image data; that is, using an XCT device to scan a composite material sample to obtain complete XCT source data, and then importing it into Avizo software for edge cropping, and using mean, median, and non-local mean filters to reduce noise;

[0015] 1-2), based on the cropped and filtered XCT source image data, one image was extracted from every ten images for manual annotation. Specifically, Avizo 2019 software was used to manually annotate every ten images, resulting in 138 images with a pixel length of 1431 × 1119;

[0016] 1-3), use the Boolean operations in the OpenCV library to extract the manually annotated connected areas of the annotated images and establish a label dataset.

[0017] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the step 2) is specifically:

[0018] 2-1), using the average width w of the warp yarn and the standard deviation σ of the width width and length Δx Length Determine specific warp yarn geometric characteristics;

[0019] 2-2), using the weft main axis length d M , the major-minor axis ratio k axis , tilt angle θ and offset ratio kr Determine weft yarn geometric characteristics;

[0020] 2-3), using the width W of the circumscribed rectangle of the aperture v and long L v Determine pore geometry characteristics;

[0021] 2-4), perform frequency distribution quantitative statistics on all warp yarns, weft yarns and pore areas in the label data set, and use the algorithm to perform adaptive function fitting to obtain the final function quantitative statistical results.

[0022] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the specific implementation method of step 2-1) is:

[0023] According to the different positions of the scanned image, the warp yarns are subdivided into warp yarn A, warp yarn B, warp yarn C and warp yarn D. The contour of the warp yarn cross section pixel area is obtained using the cv2.findContours function in the OpenCV library, and its highest point (x top ,y i ) and the lowest point (x bottom ,y i ) is defined as the length of Δx Length ; Calculate the point on the left boundary (x i ,y i,left ) and the right boundary (x i ,y i,right ), the Gaussian distribution is used to count the width of each pixel of the warp yarn in the x direction, and then the average width w and width standard deviation σ of each warp yarn are calculated. width , the geometric shape parameters are calculated as follows:

[0024] Δx Length =|x top -x bottom |

[0025]

[0026] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the specific implementation method of step 2-2) is:

[0027] The cv2.findContours function in the OpenCV library is used to obtain the pixel area contour of the weft yarn cross section, and the main axis length d is calculated based on the following formula M and the major-minor axis ratio k axis count:

[0028]

[0029] Among them, (x n ,yn ) and (x m ,y m ) are the coordinates of two different points on the weft yarn cross-section profile, with (x n ,y n ) and (x m ,y m ) The line segment formed by the two points is the perpendicular bisector, and the line segment where the perpendicular bisector intersects with the weft yarn contour is used to determine the short axis length d of the weft yarn. minor ;

[0030] The inclination angle θ is based on the ellipse fitting function in the OpenCV library. The least squares method is used to fit the weft yarn profile, obtain the major axis, minor axis and inclination angle parameters of the fitted ellipse, and combine the cv2.moments() function in the OpenCV library to calculate the coordinates of the center point of the fitted ellipse. The calculation process of the ellipse equation is as follows:

[0031]

[0032]

[0033] Offset ratio k r The calculation is performed according to the following formula:

[0034]

[0035] Among them, ra is the actual contour of the weft yarn (x n ,y n ) to (x c ,y c ) and rf is the distance between the weft yarn fitting contour (x n ,y n ) to (x,y).

[0036] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the step 3) is specifically:

[0037] 3-1), randomly selecting corresponding geometric parameters based on the quantitative statistical results of the warp and weft yarns in step 2);

[0038] 3-2), changing the shape characteristics of warp and weft yarns by random deformation; the specific implementation method is:

[0039] During the random generation of warp yarns, a fitted rectangular outline is generated based on the geometric profile characteristics of the warp yarns; the length, width, and standard deviation of the fitted rectangular outline are obtained from step 2-1); the warp yarns are processed in the x and y directions by applying a pixel offset elastic transformation matrix to obtain an irregular warp yarn outline, and a bending technique is used to cause a slight deformation of the warp yarn when it is in the preform;

[0040] Similarly, in the random generation process of weft yarn, the major axis and major-minor axis ratio of the weft yarn are obtained from the statistical results of step 2-2) to generate a fitted elliptical contour; a large number of sampling points are uniformly selected on the elliptical contour, and then these sampling points are offset according to the Gaussian distribution offset ratio to generate an irregular contour; then, the midpoint of two consecutive sampling points is taken to replace the old contour, and the median filtering algorithm is combined to remove the burrs on the elliptical contour; finally, a statistical parameter tilt angle is applied to generate weft yarns with different direction angles.

[0041] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the step 4) is specifically:

[0042] 4-1), using image meshing algorithm to convert the image model into a mesh model;

[0043] 4-2), apply boundary conditions and perform numerical simulation calculations; that is, by establishing contact relationships with the lower surface of the movable plate and the upper surface of the fixed plate on the upper and lower surfaces of the finite element model, and establishing rigid body coupling with reference points A and B, the degrees of freedom of reference point A are set to U1=0, U2=Lc and UR3=0, where Lc is the compression distance; the degrees of freedom of reference point B are set to U1=0, U2=0 and UR3=0; the U1 degree of freedom of the warp yarn is set to 0, and the other degrees of freedom are not constrained; the mesh model with the boundary condition settings defined is imported into the Abaqus 2021 finite element analysis software for compression simulation;

[0044] 4-3), use the grid imaging algorithm to convert the grid model into an image model.

[0045] The parameter-controllable two-dimensional finite element random model generation method of the composite material image of the present invention is further as follows: the step 4-1) is specifically: first, the XCT image is downsampled five times, and the node contours of the region of interest are extracted using the cv2.findContours function; then, a contour smoothing algorithm is used to identify contour right angles, and the quadrilateral pixel units in the image are smoothed into two triangles or a combination of a triangle and a quadrilateral; finally, the smoothed node coordinate values ​​are used as input, the Abaqus underlying command function is called to read the coordinate values ​​and proportionally divide the region of interest, and an appropriate grid size is set to divide the matrix area into CPS3 units and the yarn area into CPS4R units.

[0046] The parameter-controllable composite material image two-dimensional finite element random model generation method of the present invention further comprises: the step 5) is specifically:

[0047] 5-1), cropping and corroding the yarn image to add yarn defect features;

[0048] 5-2), randomly select corresponding geometric parameters based on the quantitative statistical results of pores and insert randomly generated pores;

[0049] 5-3), converting the image model into a mesh model and inserting cohesion units;

[0050] 5-4), perform numerical simulation calculations, that is, import the grid model with the boundary condition settings defined into the Abaqus2021 finite element analysis software for quasi-static compression simulation calculations.

[0051] The parameter-controllable two-dimensional finite element random model generation method of the composite material image of the present invention is further as follows: the specific implementation method of the step 5-3) is: the random image model is downsampled five times to maximize the preservation of image information, the image is converted into a grid model, and cohesive units are inserted at the contact interfaces between yarns and yarns and yarns and substrates to simulate the delamination damage formed when the material is loaded.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. The parameter-controlled two-dimensional finite element random model generation method for composite images of the present invention can not only achieve rapid generation based on the control of structural parameters, but also provide a high-fidelity initial defect model for composite damage evolution simulation. In combination with the finite element numerical simulation method, it helps to conduct research on the influence of single structural parameters on material mechanical properties and failure mechanism analysis at a low cost.

[0054] 2. The parameter-controllable two-dimensional finite element random model generation method of the composite material image of the present invention constructs a two-dimensional microscopic finite element model containing initial defects, which is conducive to quickly studying the influence of the statistical characteristics of microscopic structural parameters such as pore and yarn content, position distribution, and geometric morphological characteristics on the damage evolution characteristics of composite materials, and deeply reveals the action mechanism of a single factor.

Brief Description of the Drawings

[0055] Figure 1 The present invention is a flow chart of a method for generating a two-dimensional finite element random model of a composite material image with controllable parameters.

[0056] Figure 2 is a schematic diagram of the manual annotation results of the XCT image in step 1) of the present invention;

[0057] Figure 3 It is a schematic diagram of the qualitative description of the warp and weft yarn features in the XCT image in step 2) of the present invention.

[0058] Figure 4 This is a schematic diagram of the qualitative description of pore characteristics in the XCT image in step 2) of the present invention.

[0059] Figure 5 3) is a schematic diagram of the detailed steps of the random yarn generation algorithm of the present invention.

[0060] Figure 6 4-1) is a schematic diagram of the detailed steps of the image gridding algorithm of the present invention.

[0061] Figure 7 4-3) is a schematic diagram of the detailed steps of the grid imaging algorithm of the present invention.

[0062] Figure 8 Schematic diagram of the detailed steps of the random pore generation algorithm of step 5-2) of the present invention.

[0063] Figure 9 This is a diagram showing the result of random image generation in step 5-2) of the present invention.

[0064] Figure 10 This is a diagram showing the boundary condition definition of the random model in step 5-3) of the present invention.

[0065] Figure 11 It is a display diagram of simulation results of a random model generated by the method of the present invention. [Specific implementation method]

[0066] Please refer to the instruction manual Figure 1 As shown, the present invention is a method for generating a two-dimensional finite element random model of a composite material image with controllable parameters, comprising the following steps:

[0067] 1) Acquire the initial XCT image and preprocess it to complete the manual annotation of pores in the source image; this includes the following steps:

[0068] 1-1), obtaining complete XCT source image data, and cropping and filtering the XCT scan source image data. In this embodiment, a composite material sample is scanned using an XCT device to obtain complete XCT source data. The data is then imported into Avizo software for edge cropping. Mean, median, and non-local mean filters are used to reduce noise and ensure the quality of manual segmentation.

[0069] 1-2), based on the cropped and filtered XCT source image data, one image was extracted every ten images for manual annotation. Specifically, after the XCT source image data was cropped and filtered, Avizo 2019 software was used to manually annotate one image every ten images, resulting in 138 images with a pixel length of 1431×1119. The annotation results are shown in the attached figure. Figure 2 shown.

[0070] 2) Qualitatively select characteristic parameters of each component and perform quantitative analysis; this specifically includes the following steps:

[0071] 2-1), using the average width w of the warp yarn and the standard deviation of the width σ width and length Δx Length Determine specific warp yarn geometry characteristics.

[0072] As attached Figure 3 As shown in the figure, according to the different positions of the scanned image, the warp yarn can be divided into warp yarn type A, warp yarn type B, warp yarn type C and warp yarn type D. Specifically, the cv2.findContours function in the OpenCV library is used to obtain the contour of the warp yarn cross-section pixel area and calculate its highest point (x top ,y i ) and the lowest point (x bottom ,y i ) is defined as the length of Δx Length ; Calculate the point on the left boundary (x i ,y i,left ) and the right boundary (x i ,y i,right ), the Gaussian distribution is used to count the width of each pixel of the warp yarn in the x direction, and then the average width w and width standard deviation σ of each warp yarn are calculated. width , the geometric shape parameters are calculated as follows:

[0073] Δx Length =|x top -x bottom |

[0074]

[0075] 2-2), using the weft main axis length d M , the major-minor axis ratio k axis , tilt angle θ and offset ratio k r Determine the weft yarn geometry.

[0076] Since the weft yarn cross section is flat, the geometric characteristics of the weft yarn are described based on the ellipse equation, that is, the main axis length d M , the major-minor axis ratio k axis , tilt angle θ and offset ratio k r , the cv2.findContours function in the OpenCV library is also used to obtain the pixel area contour of the weft yarn cross section, and the main axis length d is calculated based on the following formula M and the major-minor axis ratio k axis count:

[0077]

[0078] Among them, (xn ,y n ) and (x m ,y m ) are the coordinates of two different points on the weft yarn cross-section profile, with (x n ,y n ) and (x m ,y m ) The line segment formed by the two points is the perpendicular bisector, and the line segment where the perpendicular bisector intersects with the weft yarn contour is used to determine the short axis length d of the weft yarn. minor .

[0079] The inclination angle θ is calculated based on the ellipse fitting function in the OpenCV library. The least squares method is used to fit the weft yarn profile, obtain the major axis, minor axis and inclination angle parameters of the fitted ellipse, and combine it with the cv2.moments() function in the OpenCV library to calculate the coordinates of the center point of the fitted ellipse. The calculation process of the ellipse equation is as follows:

[0080]

[0081] Offset ratio k r The calculation is performed according to the following formula:

[0082]

[0083] Among them, when the angle is constant, r a is the actual outline of the weft yarn (x n ,y n ) to (x c ,y c ) distance, r f is the weft yarn fitting contour (x n ,y n ) to (x,y).

[0084] 2-3), respectively using the width W of the circumscribed rectangle of the aperture v and long L v Determine pore geometry characteristics.

[0085] like Figure 4 As shown, pore characteristics consist of the pore outline and pore area. Furthermore, the pore outline must intersect the circumscribed rectangle at four points, thus dividing the complete pore outline into four nonlinear curve segments. Between adjacent intersection points, there is at least one inflection point, further dividing the nonlinear curve segment into shorter nonlinear curve segments. These short curve segments can behave as either concave or convex functions, and a sufficient number of equally spaced points can approximate the pore outline. Therefore, the main geometric characteristics of the pore can be determined based on the following formula:

[0086]

[0087] Where EI is the aspect ratio of the pore, W v With L v are the width and length of the circumscribed rectangle of the pore, respectively.

[0088] 2-4), perform frequency distribution quantitative statistics on all warp yarns, weft yarns and pore areas in the label data set, and use the algorithm to perform adaptive function fitting to obtain the final function quantitative statistical results.

[0089] 3) Creating an initial yarn model and assigning random deformation, which specifically includes the following steps:

[0090] 3-1), randomly selecting corresponding geometric parameters based on the quantitative statistical results of the warp and weft yarns. In this embodiment, the geometric parameter values ​​of the warp and weft yarns are randomly selected based on the quantitative analysis results of step 2-4) to determine the main geometric characteristics of the warp and weft yarns.

[0091] 3-2) The shape characteristics of the warp and weft yarns are changed by random deformation. The specific implementation method is as follows:

[0092] like Figure 5 As shown in (a), during the random generation of warp yarns, a fitted rectangular outline is generated based on the geometric outline features of warp yarn type A. The length, width and standard deviation of the fitted rectangular outline are obtained from the statistical results of step 2-1). By applying the pixel offset elastic transformation matrix to process the warp yarns in the x and y directions respectively, an irregular warp yarn outline is obtained, and a bending technique is used to make it produce a slight deformation when it is in the preform. Similarly, as Figure 5 As shown in (b), during the random weft yarn generation process, the ratio of the major axis to the major axis is obtained from the statistical results of step 2-2) to generate a fitted elliptical profile. A large number of sampling points are uniformly selected on the elliptical profile, and then these sampling points are offset according to a Gaussian distribution offset ratio to generate an irregular profile. Next, the midpoint of two consecutive sampling points is taken to replace the old profile, and a median filter algorithm is used to remove burrs from the elliptical profile. Finally, a statistical parameter tilt angle is applied to generate weft yarns with different orientation angles.

[0093] 4) Adding contact properties based on numerical simulation methods, which specifically includes the following steps:

[0094] 4-1), use the image meshing algorithm to convert the image model into a mesh model.

[0095] by Figure 6 As an example, the XCT image consists of four phases: warp, weft, matrix, and pores, and each component is composed of several pixels with unique thresholds. First, the XCT image is downsampled five times, and the node contours of the region of interest are extracted using the cv2.findContours function ( Figure 6 (b)), however, the node contours have sharp right angles, which can easily lead to interruptions in the numerical simulation process. Therefore, it is necessary to use a contour smoothing algorithm to identify the contour right angles and smooth the quadrilateral pixel units in the image into two triangles or a combination of a triangle and a quadrilateral ( Figure 6 Finally, the smoothed node coordinates are used as input to call the Abaqus underlying command function to read the coordinates and divide the region of interest in equal proportions. The appropriate grid size is set to divide the matrix region into CPS3 units and the yarn region into CPS4R units (e.g. Figure 6 (d)).

[0096] 4-2), apply boundary conditions and perform numerical simulation calculations.

[0097] The specific implementation method of this step is:

[0098] By establishing contact relationships with the lower surface of the movable plate and the upper surface of the fixed plate on the upper and lower surfaces of the finite element model, and establishing rigid body coupling with reference points A and B, the degrees of freedom of reference point A are set to U1=0, U2=Lc, and UR3=0 (Lc is the compression distance), and the degrees of freedom of reference point B are set to U1=0, U2=0, and UR3=0. In addition, in order to fix the relative position of the warp yarn, the U1 degree of freedom of the warp yarn is set to 0, and the remaining degrees of freedom are not constrained. The mesh model with the boundary condition settings defined is imported into the Abaqus 2021 finite element analysis software for compression simulation, thereby ensuring that the yarn can form contact and deformation under the action of external loads.

[0099] 4-3), use the grid imaging algorithm to convert the grid model into an image model.

[0100] As attached Figure 7 As shown, taking the two-dimensional CPS4R unit as an example, the unit has four nodes (Ip a (x a ,y a ), Ip b (x b ,y b ), Ip c (x c ,y c ) and IP d (x d ,y d )), adjacent nodes together constitute and Forming a closed area S abcd , and the pixel-type circumscribed rectangle S0 can be constructed through four nodes, each pixel N eq Determine the unique center point Cp eq(x e ,y q )( Figure 7 (a) in the above example. Figure 7 As shown in (b) to (f) in the figure, by calculating and Vector cross product, the filter center point is located at Left pixel area S ab , and then respectively and Repeat the above operation to find the pixel area S bc 、S cd and S da Finally, S ab 、S bc 、S cd and S da The pixel areas are superimposed and the common pixel areas are screened out to approximate S abcd (like Figure 7 The specific calculation formula is as follows:

[0101]

[0102] Wherein, e and q are the pixel length and pixel width of the circumscribed rectangle S0 respectively.

[0103] 5) Assigning yarn defect attributes and inserting pores, and performing simulation calculations based on numerical simulation methods; the specific steps include:

[0104] 5-1), the yarn image is cropped and eroded to add yarn defect features. Specifically, the yarn image is appropriately cropped and eroded to produce an incomplete weft yarn cross section and warp yarn defects (warp yarn type B, warp yarn type C, and warp yarn type D) in the yarn image.

[0105] 5-2), randomly select corresponding geometric parameters based on the quantitative statistical results of pores and insert randomly generated pores. Specifically:

[0106] According to the pore length and width distribution model statistically calculated in step 2-3), the length L of the pore circumscribed rectangle is arbitrarily selected. v With width W v , place the bounding rectangle in the Cartesian coordinate system, and randomly select four intersection points (Ip1(x i1 ,y i1 ),Ip2(x i2 ,y i2 ),Ip3(x i3 ,y i3 ) and Ip4(x i4 ,y i4)) to determine its main characteristics (such as Figure 8 The adjacent intersections form a rectangular region, in which at most one inflection point is randomly selected (for more complex features, multiple inflection points can be selected). The rectangular region formed by the adjacent intersections is defined as follows:

[0107]

[0108] However, considering the rectangular area formed by adjacent intersection points ( Figure 8 Randomly selecting inflection points in the light-colored area (b) may produce interference areas ( Figure 8 The yellow area in (b) in the figure should be limited to Figure 8 In the dark area in (b), the inflection point Fp1(x f1 ,y f1 ) and Fp2(x f2 ,y f2 ), this area is defined as follows:

[0109]

[0110] Wherein, a, b, c, and d represent constants used to adjust the range of inflection point generation.

[0111] Take Ip1, Fp1 and Ip2 as an example. Assume that the arc formed by Ip1 and Fp1 is a convex function, and the arc formed by Fp1 and Ip2 is a concave function. Connect the three points to form two line segments Ip1Fp1 and Fp1Ip2. Generate two perpendicular bisectors for each line segment. For the vectors of these two line segments, and The calculation method is as follows:

[0112]

[0113] If a convex arc is generated between Ip1 and Fp1, calculate The angle between the two straight line segments on the right; if a concave arc is generated between Fp1 and Ip2, calculate The angle between the two straight line segments on the left is obtained by min , select the appropriate adjacent straight line segment (red, Figure 8 (c) in the figure, ensure that the generated arc does not exceed the small rectangular area formed between two adjacent points. Then, for the convex arc, make a Left perpendicular line; for concave arc, draw a perpendicular line through the straight line segment. The two perpendicular lines on the right intersect with the corresponding perpendicular bisectors at O1(x o1 ,y o1 ) and O2(xo2 ,y o2 ), and get the shortest radius r1 and r2 (purple line, Figure 8 The calculation method of (c)) in is as follows:

[0114]

[0115] Since the center of each arc is located on the perpendicular line, the arc center can be moved to O1'(x o1 ',y o1 ') and O2'(x o2 ',y o2 '), thereby randomly changing the shape of the arc, such as Figure 8 As shown in (d) in the figure. At this time, the angle θ o1'1 ,θ o1'2 ,θ o2'1 and θ o2'2 In addition, for each arc, a corresponding local polar coordinate system is established to calculate the points E uniformly distributed at equal angular intervals. spv (x esv ,y esv ) coordinate values, such as Figure 8 The specific calculation process is as follows:

[0116]

[0117] Among them, v is the order of equally spaced points. In this example, its maximum value is set to 3, but it is recommended to use a larger value to use more points to approximate the arc, thereby depicting the outline of the pore more smoothly.

[0118] Will Figure 8 (c) to Figure 8 The step (e) in the above is repeated to generate all the points that approximate the hole contour. However, since there will inevitably be local sharp areas between adjacent arcs ( Figure 8 (f) in the figure, therefore, it is necessary to offset these sharp points appropriately to achieve smoothing. The result after processing is as follows Figure 8 Finally, using the coordinate values ​​of all the points that approximate the pore outline as input, use cv2.fillPoly to fill the internal area of ​​the pore, as shown in (g). Figure 8 As shown in (h).

[0119] 5-3), use the method in step 4-1) to convert the image model into a mesh model and insert cohesion units. The specific implementation method is:

[0120] After following steps 3-1) to 5-2) in sequence, random image model generation is achieved, and the diverse generation results are as follows Figure 9 Then the random image model (such as Figure 10 The image is downsampled five times as shown in (a) above, and the pore features are not processed in any way to maximize the preservation of image information. The image is converted into a mesh model using the method in step 4-1) again, and cohesive units are inserted at the contact interfaces between yarns and yarns and between yarns and the matrix to simulate the delamination damage formed when the material is loaded, as shown in Figure 4-1. Figure 10 (b) shown.

[0121] 5-4), based on the method of step 4-2), a numerical simulation calculation is performed, that is, by establishing contact relationships with the lower surface of the movable plate and the upper surface of the fixed plate on the upper and lower surfaces of the finite element model, and establishing rigid body coupling with reference points A and B, the degrees of freedom of reference point A are set to U1=0, U2=Lc and UR3=0 (Lc is the compression distance), and the degrees of freedom of reference point B are set to U1=0, U2=0 and UR3=0, thereby establishing boundary conditions similar to the actual experimental conditions. The mesh model with the boundary condition settings defined is imported into the Abaqus 2021 finite element analysis software for quasi-static compression simulation calculation. The simulation results are as follows: Figure 11 shown.

[0122] The above specific implementation methods are only preferred embodiments of this creation and are not intended to limit this creation. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this creation should be included in the scope of protection of this creation.

Claims

1. A parameter-controllable two-dimensional finite element random model generation method for composite material images, characterized by: The process steps include: 1) Acquire the initial XCT image and preprocess it, and complete the manual annotation of each component in the source image; 2) qualitatively select characteristic parameters of each component and conduct quantitative analysis; 3), creating an initial yarn model and giving it random deformation; 4) Adding contact properties based on numerical simulation methods; 5), assign yarn defect properties and insert pores, and perform simulation calculations based on numerical simulation methods.

2. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 1, characterized in that: The step 1) is specifically as follows: 1-1), obtaining complete XCT source image data, and cropping and filtering the XCT scan source image data; that is, using an XCT device to scan a composite material sample to obtain complete XCT source data, and then importing it into Avizo software for edge cropping, and using mean, median, and non-local mean filters to reduce noise; 1-2), based on the cropped and filtered XCT source image data, one image was extracted from every ten images for manual annotation. Specifically, Avizo 2019 software was used to manually annotate every ten images, resulting in 138 images with a pixel length of 1431 × 1119; 1-3), use the Boolean operations in the OpenCV library to extract the manually annotated connected areas of the annotated images and establish a label dataset.

3. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 1, characterized in that: The step 2) is specifically as follows: 2-1), average width of warp yarn used Width standard deviation σ width and length Δx Length Determine specific warp yarn geometric characteristics; 2-2), using the weft main axis length d M , the major-minor axis ratio k axis , tilt angle θ and offset ratio k r Determine weft yarn geometric characteristics; 2-3), using the width W of the circumscribed rectangle of the aperture v and long L v Determine pore geometry characteristics; 2-4), perform frequency distribution quantitative statistics on all warp yarns, weft yarns and pore areas in the label data set, and use the algorithm to perform adaptive function fitting to obtain the final function quantitative statistical results.

4. The method for generating a two-dimensional finite element random model of a composite material image with controllable parameters according to claim 3, characterized in that: The specific implementation method of step 2-1) is: According to the different positions of the scanned image, the warp yarns are subdivided into warp yarn A, warp yarn B, warp yarn C and warp yarn D. The contour of the warp yarn cross section pixel area is obtained using the cv2.findContours function in the OpenCV library, and its highest point (x top ,y i ) and the lowest point (x bottom ,y i ) is defined as the length of Δx Length ; Calculate the point on the left boundary (x i ,y i,left ) and the right boundary (x i ,y i,right ), the Gaussian distribution is used to count the width of each pixel of the warp yarn in the x direction, and then the average width w and width standard deviation σ of each warp yarn are calculated. width , the geometric shape parameters are calculated as follows: Δx Length =|x top -x bottom | 5. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 3, characterized in that: The specific implementation method of step 2-2) is: The cv2.findContours function in the OpenCV library is used to obtain the pixel area contour of the weft yarn cross section, and the main axis length d is calculated based on the following formula M and the major-minor axis ratio k axis count: Among them, (x n ,y n ) and (x m ,y m ) are the coordinates of two different points on the weft yarn cross-section profile, with (x n ,y n ) and (x m ,y m ) The line segment formed by the two points is the perpendicular bisector, and the line segment where the perpendicular bisector intersects with the weft yarn contour is used to determine the short axis length d of the weft yarn. minor ; The inclination angle θ is based on the ellipse fitting function in the OpenCV library. The least squares method is used to fit the weft yarn profile, obtain the major axis, minor axis and inclination angle parameters of the fitted ellipse, and combine the cv2.moments() function in the OpenCV library to calculate the coordinates of the center point of the fitted ellipse. The calculation process of the ellipse equation is as follows: Offset ratio k r The calculation is performed according to the following formula: Among them, r a is the actual outline of the weft yarn (x n ,y n ) to (x c ,y c ) distance, r f is the weft yarn fitting contour (x n ,y n ) to (x,y).

6. The method for generating a two-dimensional finite element random model of a composite material image with controllable parameters according to claim 3, characterized in that: The step 3) is specifically as follows: 3-1), randomly selecting corresponding geometric parameters based on the quantitative statistical results of the warp and weft yarns in step 2); 3-2), changing the shape characteristics of warp and weft yarns by random deformation; the specific implementation method is: During the random generation of warp yarns, a fitted rectangular outline is generated based on the geometric profile characteristics of the warp yarns; the length, width, and standard deviation of the fitted rectangular outline are obtained from step 2-1); the warp yarns are processed in the x and y directions by applying a pixel offset elastic transformation matrix to obtain an irregular warp yarn outline, and a bending technique is used to cause a slight deformation of the warp yarn when it is in the preform; Similarly, in the random generation process of weft yarn, the major axis and major-minor axis ratio of the weft yarn are obtained from the statistical results of step 2-2) to generate a fitted elliptical contour; a large number of sampling points are uniformly selected on the elliptical contour, and then these sampling points are offset according to the Gaussian distribution offset ratio to generate an irregular contour; then, the midpoint of two consecutive sampling points is taken to replace the old contour, and the median filtering algorithm is combined to remove the burrs on the elliptical contour; finally, a statistical parameter tilt angle is applied to generate weft yarns with different direction angles.

7. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 6, characterized in that: The step 4) is specifically as follows: 4-1), using image meshing algorithm to convert the image model into a mesh model; 4-2), apply boundary conditions and perform numerical simulation calculations; that is, by establishing contact relationships with the lower surface of the movable plate and the upper surface of the fixed plate on the upper and lower surfaces of the finite element model, and establishing rigid body coupling with reference points A and B, the degrees of freedom of reference point A are set to U1=0, U2=Lc and UR3=0, where Lc is the compression distance; the degrees of freedom of reference point B are set to U1=0, U2=0 and UR3=0; the U1 degree of freedom of the warp yarn is set to 0, and the other degrees of freedom are not constrained; the mesh model with the boundary condition settings defined is imported into the Abaqus 2021 finite element analysis software for compression simulation; 4-3), use the grid imaging algorithm to convert the grid model into an image model.

8. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 7, characterized in that: The step 4-1) is specifically as follows: first, the XCT image is downsampled fivefold, and the node contours of the region of interest are extracted using the cv2.findContours function; then, a contour smoothing algorithm is used to identify contour right angles, and the quadrilateral pixel units in the image are smoothed into two triangles or a combination of a triangle and a quadrilateral; finally, the smoothed node coordinate values ​​are used as input, the Abaqus underlying command function is called to read the coordinate values ​​and proportionally divide the region of interest, and an appropriate grid size is set to divide the matrix area into CPS3 units and the yarn area into CPS4R units.

9. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 1, characterized in that: The step 5) is specifically as follows: 5-1), cropping and corroding the yarn image to add yarn defect features; 5-2), randomly select corresponding geometric parameters based on the quantitative statistical results of pores and insert randomly generated pores; 5-3), converting the image model into a mesh model and inserting cohesion units; 5-4), perform numerical simulation calculations: import the mesh model with the boundary condition settings defined into the Abaqus 2021 finite element analysis software for quasi-static compression simulation calculations.

10. The parameter-controllable two-dimensional finite element random model generation method for composite material images according to claim 9, characterized in that: The specific implementation method of step 5-3) is as follows: downsampling the random image model by a factor of five to maximize the preservation of image information, converting the image into a mesh model, and inserting cohesive units at the contact interfaces between yarns and yarns and between yarns and the matrix to simulate the delamination damage formed when the material is loaded.

Images