A Mesoscopic Identification Modeling Method for Ceramic Matrix Composites with Complex Structures

Abstract

The present invention discloses a mesoscopic identification and modeling method for ceramic matrix composites with complex structures, belonging to the field of three-dimensional model reconstruction. The method comprises the following steps: S1, extracting the "front view"; S2, marking the warp and weft yarns with different colors; views in the left view and top view directions; S3, respectively extracting the weft yarns and warp yarns; S4, excerpting; S5, respectively coloring each fiber bundle with different colors to enable the colors of the same fiber bundle on different slices to correspond; S6, according to different colors, respectively building models of all single fiber bundles in space and combining them to obtain the mesoscopic model of the entire test piece. The present invention has the advantages of good effect, high efficiency, accurate identification and reconstruction, etc.

Description

Technical Field

[0001] The present invention belongs to the field of three-dimensional model reconstruction, and relates to a modeling method for ceramic matrix composites, in particular to a mesoscopic identification and modeling method for ceramic matrix composites with complex structures. Background Art

[0002] Ceramic matrix composites are widely used in aerospace and other fields due to their good performance. Their mechanical properties have particularly attracted the attention of relevant practitioners. Therefore, the requirements for research methods and equipment for the mechanical properties of test pieces made of ceramic matrix composites are also constantly increasing. In order to obtain the mechanical properties of test pieces, it is necessary to perform model reconstruction on them and then perform simulation calculations.

[0003] The existing mesoscopic modeling methods have the following drawbacks: On the one hand, since ceramic matrix composites are mostly used in structures such as tail nozzles and turbine rotor blades in the aviation field, these structures have complex shapes and diverse weaving methods. Compared with flat plate components woven with the same material, their internal mesoscopic structures are more complex and the anisotropy of physical parameters is more prominent. If the flat plate modeling method is used, it is very likely that the initial identification of the mesoscopic structure cannot be completed in many parts. On the other hand, the volume of such structures is large and the processing time is long, making it unsuitable to use the traditional manual marking method for modeling. On the other hand, there is currently a mesoscopic modeling method (Jia Yunfa, Zhang Sheng, Song Yingdong, Gao Xiguang, Liu Chenyang, Wang Yuxuan, A Three-Dimensional Reconstruction Method for Variable-Thickness Curved Woven Ceramic Matrix Composites) that considers calculating the edges and angle changes of all fiber bundles in each CT slice, with a huge amount of computation and a long time-consuming. And the function of marking the cross-sections of the same fiber bundle with the same number in different CT slices mentioned in it is only applicable to processing a part of regularly arranged fiber bundles in actual operation. When the spatial movement amplitude of the fiber bundles is large, or when the phenomenon of adding or reducing yarns occurs, this method cannot solve the problem well.

[0004] Therefore, in order to measure the true mechanical properties of test pieces in simulation tests, it is necessary to research a method that can be applied to the mesoscopic identification and modeling of complex ceramic matrix composites and can also effectively reduce the time for three-dimensional reconstruction of materials. Summary of the Invention

[0005] The present invention provides a mesoscopic identification and modeling method for ceramic matrix composites with complex structures to overcome the defects of the prior art.

[0006] To achieve the above object, the present invention provides a mesoscopic identification and modeling method for ceramic matrix composites with complex structures, having the following characteristics: including the following steps:

[0007] S1. Scan the ceramic matrix composite test piece to be tested, and export several slices from the front view direction according to the scan results, denoted as "front view".

[0008] S2. Mark the warp and weft yarns of all slices in the "front view" obtained in S1 with different colors to obtain several new slices, denoted as "front view identification".

[0009] According to the size of the "front view identification", export several left view and top view direction views, denoted as "front-derived left" and "front-derived top".

[0010] S3. Extract the weft and warp yarns from the "front-derived left" and "front-derived top" slices obtained in S2 respectively, to obtain slices containing only several elliptical cross-sections of weft yarns and elliptical cross-sections of warp yarns, denoted as "front-derived left weft yarn" and "front-derived top warp yarn" respectively.

[0011] S4. Select a part of the slices from the "front-derived left weft yarn" and "front-derived top warp yarn" obtained in S3 respectively, denoted as "weft yarn excerpt" and "warp yarn excerpt".

[0012] S5. Color each fiber bundle in the "weft yarn excerpt" and "warp yarn excerpt" with different colors: Take the previous slice as the reference slice, compare each fiber bundle in this slice with each fiber bundle in the reference slice. If it is found that they belong to the same fiber bundle, give the same number according to the number of the fiber bundle in the reference slice. If it belongs to a new fiber bundle after comparison (that is, after comparison, it does not belong to the same fiber bundle as each fiber bundle in the reference slice), give a new number; The slice with the completed numbering is used as the reference slice to number each fiber bundle of the next slice; When the first slice is used as the reference slice, directly number each fiber bundle of it; Use the number as the independent variable for coloring processing, so that the same fiber bundle is marked with the same color in different slices.

[0013] S6. According to different colors, build the models of all single fiber bundles in space respectively, and combine them to obtain the mesoscopic model of the entire test piece.

[0014] Furthermore, the present invention provides a mesoscopic identification and modeling method for ceramic matrix composites with complex structures, which may also have the following characteristics: Among them, the specific method of S1 is: Use a computer tomography device to scan the ceramic matrix composite test piece to be tested to obtain the original three-dimensional data with a spatial size of x*y*z; Open the original three-dimensional data in VG software, and export several slices from the front view direction. The thickness of each slice is equal, denoted as "front view", where the gray part is the fiber bundle and the black part is the pore.

[0015] Furthermore, the present invention provides a method for microscopic identification and modeling of complex structure ceramic-based composite materials, which may also have the following characteristics: wherein, the method for obtaining the front view identification in S2 is: all slices in the "front view" obtained in S1 are marked using the gradient structure tensor method, and due to the different gradients, the warp yarns are marked in red and the weft yarns are marked in blue, and several new slices are obtained, which are recorded as "front view identification".

[0016] Furthermore, the present invention provides a method for microscopic identification and modeling of complex structure ceramic-based composite materials, which may also have the following characteristics: wherein, in S3, before extraction, all slices in the "positive left" and "positive top" obtained in S2 are subjected to histogram equalization processing to make the color contrast between the warp and weft yarns more obvious.

[0017] Furthermore, the present invention provides a method for microscopic identification and modeling of complex structure ceramic-based composite materials, which may also have the following characteristics: wherein, in S3, different colors of weft yarns and warp yarns are extracted respectively by setting different HSV color space thresholds.

[0018] Furthermore, the present invention provides a method for microscopic identification and modeling of complex structure ceramic-based composite materials, which may also have the following characteristics: wherein, in S4, when slices are selected from the "positive left weft yarn" and "positive downward warp yarn" obtained in S3, the spacing between each slice in the original slice group (i.e., "positive left weft yarn" / "positive downward warp yarn") is equal.

[0019] Furthermore, the present invention provides a method for microscopic identification and modeling of complex structure ceramic-based composite materials, which may also have the following characteristics: wherein, in S4, after excerpting, the slices are corrected, including: removing noise points, repairing gaps, and separating yarns that are stuck together.

[0020] Furthermore, the present invention provides a mesoscopic identification and modeling method for complex structure ceramic matrix composites, which may also have the following features: In S5, the specific method for coloring each fiber bundle in "weft yarn extraction" and "warp yarn extraction" with different colors is as follows: Use the connected region labeling method to label each slice, and each fiber bundle cross-section of each slice obtains a region number and centroid coordinates; Take the first slice as the reference slice, and store the region number, centroid abscissa, and centroid ordinate of each fiber bundle cross-section into a "common array", denoted as "table[][]"; When reading the second and subsequent slices, generate a new array to store the region number, centroid abscissa, and centroid ordinate of each fiber bundle cross-section of the current slice, denoted as "tb[][]"; Starting from the elements in the second and third columns of the first row of the array tb[][], compare them with the elements in all corresponding columns of the array table[][] in turn, and find the row element in the array table[][] with the smallest coordinate difference, and its corresponding fiber bundle number is xi; If the horizontal and vertical coordinate differences are both within a set threshold range, it is considered that the corresponding fiber bundle cross-sections on these adjacent two slices are the same fiber bundle, and at this time, change the first column element of the current row of the array tb[][] to xi; If the horizontal and vertical coordinate differences are not within the set threshold range, it is considered that a new fiber bundle appears. If the total number of fiber bundle cross-sections on the previous slice is K, then change the first column element of the current row of the array tb[][] to K + 1; Operate like this until the last row element of the array tb[][] is compared, completely overwrite the common array table[][] with the updated array tb[][], and then compare it with the array of the next slice, and compare it in turn until the last slice; Finally, set a segmented color function, use the first column element of the current row of the updated array tb[][] as the independent variable for coloring processing, mark the same fiber bundle in different slices with the same color, and the colors between each fiber bundle are different from each other, so as to realize the identification of fiber bundles.

[0021] Furthermore, the present invention provides a mesoscopic identification and modeling method for complex structure ceramic matrix composites, which may also have the following features: The specific method of S6 is as follows: Adopt the marching cubes algorithm, and build models of all single fiber bundles in space according to the retrieved different colors, and combine them to obtain the mesoscopic model of the entire test piece.

[0022] Furthermore, the present invention provides a mesoscopic identification and modeling method for complex structure ceramic matrix composites, which may also have the following features: In S6, the retrieved colors are RGB channel pixel values.

[0023] The beneficial effects of the present invention are as follows:

[0024] 1. When separating different types of fiber bundles, the present invention extracts colors in the HSV space, which has a better effect than extracting colors in the RGB space;

[0025] 2. The present invention takes a part of the slices at equal intervals for operation, shortening the calculation amount on the premise of not significantly affecting the recognition result;

[0026] 3. After the connected regions of each slice are labeled, the present invention stores the region numbers in an array for comparison, without directly changing the region numbers on the graph, without modifying the picture information, and considering all the results of the spatial changes of the fiber bundles, making the recognition less error-prone;

[0027] 4. During the recognition process, the present invention does not need to consider the edge shape of the fiber bundle, nor does it need to idealize the cross-section of the fiber bundle;

[0028] 5. The present invention colors the fiber bundles by using a color function. The color difference between each fiber bundle is relatively small and the change is regular, avoiding randomness and greatly shortening the time for retrieving color values (pixel values) during the modeling process. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 is the "front view" of the ceramic matrix composite test piece after CT scanning;

[0030] Figure 2 is the "left view" of the ceramic matrix composite test piece after CT scanning;

[0031] Figure 3 is the "top view" of the ceramic matrix composite test piece after CT scanning;

[0032] Figure 4 is the "front view recognition" after being marked by the gradient structure tensor method;

[0033] Figure 5 is the "front-to-left derivation" of the left view derived from the "front view recognition";

[0034] Figure 6 is the "front-to-top derivation" of the top view derived from the "front view recognition";

[0035] Figure 7 is the "front-to-left weft yarn extraction" after extracting the weft yarn (blue);

[0036] Figure 8 is the "front-to-top warp yarn extraction" after extracting the warp yarn (red);

[0037] Figure 9 is the "weft yarn extraction after optimization and correction";

[0038] Figure 10It is the "warp excerpt" after optimization and correction;

[0039] Figure 11 It is a flowchart for realizing that the colors of the same fiber bundle on different slices can correspond to functions;

[0040] Figure 12 It is a schematic diagram of the common two-dimensional array table[][] for storing benchmark slice information;

[0041] Figure 13 It is a schematic diagram of the two-dimensional array tb[][] for storing current slice information;

[0042] Figure 14 It is a schematic diagram of two types of array information updates, where a is the information update of the same fiber bundle, and b is the information update of a new fiber bundle. Specific implementation manner

[0043] The following will describe the specific implementation manner of the present invention with reference to the accompanying drawings.

[0044] This embodiment provides a mesoscopic recognition and modeling method for complex structure ceramic matrix composites, including the following steps:

[0045] S1. Use a computer tomography device (CT) to scan the ceramic matrix composite test piece to be tested to obtain the original three-dimensional data, whose spatial size is x*y*z. Open the three-dimensional data in the VG software, and export x / N, y / N, and z / N slices from three different view directions respectively. N is the set thickness value of a single slice, denoted as "front view", "left view", and "top view", as shown in Figure 1 , Figure 2 , Figure 3 respectively, where the gray part is the fiber bundle and the black part is the pore.

[0046] S2. Mark all the slices in the obtained "front view" using the gradient structure tensor method. Due to the difference in gradients, the warp is marked red and the weft is marked blue to obtain x / N new slices, denoted as "front view recognition", as shown in Figure 4 . This operation can only be completed on the front view, and the left view and top view cannot achieve the marking effect. According to the size of the "front view recognition", export the views in the other two directions, denoted as "front to left" and "front to top", as shown in Figure 5 and Figure 6 respectively, and the quantities are y / N and z / N.

[0047] S3. Perform histogram equalization on all the obtained "positive left" and "positive top" slices to make the color contrast between the warp and weft more obvious. Then set different HSV color space thresholds from the "positive left" and "positive top" slices to extract the weft (blue) and warp (red) respectively, and obtain a group of images containing only a number of weft elliptical sections and warp elliptical sections, which are recorded as "positive left weft" and "positive top warp", respectively. Figure 7 , Figure 8 shown.

[0048] S4, Figure 7 , Figure 8 The fiber bundles extracted from the slices shown are still relatively rough, and the shape recognition is not high. It is necessary to select a part of the slices from the "positive left weft yarn" and "positive downward warp yarn" sections respectively. The spacing of each slice in the original image group is equal. Assume that the spacing is a and b respectively. Then, correction is performed to remove noise, fill in gaps, and separate the yarns that are stuck together. y / aN and z / bN new slices are obtained, which are recorded as "weft yarn selection" and "warp yarn selection", respectively. Figure 9 , Figure 10 shown.

[0049] S5. Take the picture group of "Weft Selection" as an example. The operation steps are shown in Figure 11 As shown, each slice is marked using the connected component marking method, and each slice obtains y i The first slice is used as the reference slice, and the information of each region (fiber bundle cross section) (region number, horizontal coordinate of the center of gravity, vertical coordinate of the center of gravity) is stored in a "common array" recorded as "table[][]", such as Figure 12 When reading the second and subsequent slices, a new array is generated to store the information (region number, horizontal coordinate of the center of gravity, vertical coordinate of the center of gravity) of each region (fiber bundle cross section) of the current slice, recorded as "tb[][]", as shown in Figure 13 As shown. Starting from the second and third column elements of the first row of array tb[][] (the horizontal coordinate of the center of gravity and the vertical coordinate of the center of gravity), compare them with the elements of all corresponding columns of array table[][] in turn, and find the row of elements in array table[][] with the smallest coordinate difference. The corresponding fiber bundle number is xi. If the horizontal and vertical coordinate differences are both within a set threshold range, it is considered that the corresponding fiber bundle cross-sections on the two adjacent slices are the same fiber bundle. At this time, change the first column element (region number) of the current row of array tb[][] to xi, as shown Figure 14as shown in a; if the difference between the horizontal and vertical coordinates is not within the set threshold range, it is regarded as a new fiber bundle. If the total number of cross-sectional areas of the fiber bundle in the previous slice is K, then the first column element (area number) of the current row of the array tb[][] is changed to K + 1, as Figure 14 shown in b. This operation is performed until the last row element of the array tb[][] is compared, and the updated array tb[][] completely overwrites the common array table[][]. Then, it is compared with the information of the next slice until the comparison of the last slice is completed.

[0050] After that, a segmented color function is set, and the first column element (area number) of the current row of the updated array tb[][] is used as the independent variable for coloring. In this way, the same fiber bundle is marked with the same color in different slices, and the colors of each fiber bundle are different from each other, thus achieving the purpose of fiber bundle recognition. When processing the "warp selection" picture group, the method is the same as the above content.

[0051] S6. The marching cubes algorithm is adopted to build the models of all single fiber bundles in space respectively according to the retrieved different colors (RGB channel pixel values), and the mesoscopic model of the entire test piece is combined.

[0052] In the present invention, unless otherwise specified, the scientific and technical terms used herein have the meanings commonly understood by those skilled in the art. Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A mesoscopic identification and modeling method for ceramic matrix composites with complex structures, characterized by: including the following steps: S1. Scan the test piece of ceramic matrix composite to be tested, and export several slices from the front view direction according to the scanning results, denoted as "front view"; S2. Mark the warp and weft yarns of all slices in the "front view" obtained in S1 with different colors to obtain several new slices, denoted as "front view identification"; According to the size of the "front view identification", export several views in the left view and top view directions, denoted as "front lead left" and "front lead top"; S3. Extract the weft and warp yarns of different colors respectively from the "front lead left" and "front lead top" slices obtained in S2 by setting different HSV color space thresholds, to obtain slices containing only several elliptical cross-sections of weft yarns and elliptical cross-sections of warp yarns, denoted as "front lead left weft yarn" and "front lead top warp yarn" respectively; S4. Select a part of the slices from the "front lead left weft yarn" and "front lead top warp yarn" obtained in S3 respectively, denoted as "weft yarn selection" and "warp yarn selection"; when selecting, the spacing of each slice in the original slice group is equal; S5. Color each fiber bundle in the "weft yarn selection" and "warp yarn selection" with different colors: take the previous slice as the reference slice, compare each fiber bundle in this slice with each fiber bundle in the reference slice, if it is found to belong to the same fiber bundle, give the same number according to the number of the fiber bundle in the reference slice, if it belongs to a new fiber bundle after comparison, give a new number; the slice with the numbering is used as the reference slice to number each fiber bundle of the next slice; when the first slice is used as the reference slice, directly number each fiber bundle of it; perform coloring processing with the number as the independent variable, so that the same fiber bundle is marked with the same color in different slices; S6. Adopt the marching cubes algorithm, and build the models of all single fiber bundles in space respectively according to the retrieved different colors, and combine them to obtain the mesoscopic model of the whole test piece.

2. The mesoscopic identification and modeling method for ceramic matrix composites with complex structures according to claim 1, characterized by: Among them, The specific method of S1 is: Use a computer tomography device to scan the test piece of ceramic matrix composite to be tested to obtain the original three-dimensional data with a spatial size of x*y*z; Open the original three-dimensional data in VG software, and export several slices from the front view direction. The thickness of each slice is equal, denoted as "front view", where the gray part is the fiber bundle and the black part is the pore.

3. The mesoscopic identification and modeling method for ceramic matrix composites with complex structures according to claim 1, characterized by: Among them, The obtaining method of the front view identification in S2 is: Mark all the slices in the "front view" obtained in S1 by using the gradient structure tensor method. Due to the difference in gradients, the warp yarns are marked red and the weft yarns are marked blue to obtain several new slices, denoted as "front view identification".

4. The mesoscopic identification and modeling method for ceramic matrix composites with complex structures according to claim 1, characterized by: Among them, In S3, before extraction, histogram equalization processing is performed on all slices in the "positive lead left" and "positive lead down" obtained in S2, so that the color contrast between warp and weft yarns becomes more obvious.

5. The mesoscopic identification and modeling method for complex structure ceramic matrix composites according to claim 1, characterized in that: Among them, In S4, after extraction, the slices are corrected, including: removing noise, repairing vacancies, and separating the yarns that are stuck together.

6. The mesoscopic identification and modeling method for complex structure ceramic matrix composites according to claim 1, characterized in that: Among them, In S5, the specific method for coloring each fiber bundle in the "weft extraction" and "warp extraction" is as follows: Each slice is marked by the connected region labeling method, and each fiber bundle cross-section of each slice obtains a region number and centroid coordinates; taking the first slice as the reference slice, the region number, centroid abscissa, and centroid ordinate of each fiber bundle cross-section are stored in a "common array", denoted as "table[][]"; when reading the second and subsequent slices, a new array is generated to store the region number, centroid abscissa, and centroid ordinate of each fiber bundle cross-section of the current slice, denoted as "tb[][]"; Starting from the elements in the second and third columns of the first row of the array tb[][], compare them with the elements in all corresponding columns of the array table[][] in turn, and find the row element in the array table[][] with the smallest coordinate difference, and its corresponding fiber bundle number is xi; If the differences in both the abscissa and ordinate are within a set threshold range, it is considered that the corresponding fiber bundle cross-sections on these two adjacent slices are the same fiber bundle. At this time, change the element in the first column of the current row of the array tb[][] to xi; if the differences in the abscissa and ordinate are not within the set threshold range, it is considered that a new fiber bundle appears. If the total number of fiber bundle cross-sections on the previous slice is K, then change the element in the first column of the current row of the array tb[][] to K + 1; Perform such operations until the last row element of the array tb[][] is compared. Completely overwrite the common array table[][] with the updated array tb[][], and then compare it with the array of the next slice, and compare it in turn until the last slice; Finally, set a segmented color function, with the element in the first column of the current row of the updated array tb[][] as the independent variable, perform coloring processing, mark the same fiber bundle in different slices with the same color, and the colors between each fiber bundle are different from each other, so as to realize the identification of fiber bundles.

7. The mesoscopic identification and modeling method for complex structure ceramic matrix composites according to claim 1, characterized in that: Among them, In S6, the retrieved color is the RGB channel pixel value.

Images