CN-122021184-A - Braiding CMCs finite element modeling and analysis method based on real mesoscopic structure

CN122021184ACN 122021184 ACN122021184 ACN 122021184ACN-122021184-A

Abstract

The invention discloses a finite element modeling and analysis method for braided CMCs based on a real microscopic structure, and belongs to the technical field of composite material mechanics simulation. The method comprises the steps of obtaining three-dimensional gray level images of the woven CMCs through X-ray CT scanning, carrying out high-precision automatic segmentation on warp yarns, weft yarns, matrixes, pores, cracks and other components in the images by utilizing a deep learning technology, carrying out boundary smoothing filtering processing on segmentation results, selecting a representative subvolume area according to a volume fraction consistency principle, generating an optimized grid model suitable for finite element calculation based on the real geometric structure of the area, and finally carrying out simulation analysis by endowing material properties and setting boundary conditions to realize accurate prediction on the mechanical properties of the material. The invention solves the problems that the traditional idealized model can not reflect the real microscopic structure, CT image segmentation is difficult, the calculation scale is overlarge, and the like, and obviously improves the precision and the efficiency of finite element simulation of the braiding CMCs.

Inventors

MA WENBING
YU GUOQIANG
QIU XIAOXIAO
ZHOU SHIHAO
Xue beichen
LI JIAMING
Sui Zhengqing
GAO XIGUANG
SONG YINGDONG

Assignees

南京航空航天大学

Dates

Publication Date: 20260512
Application Date: 20260330

Claims (9)

1. The finite element modeling and analyzing method for the braided CMCs based on the real mesostructure is characterized by comprising the following steps of: S1, acquiring a three-dimensional digital gray image of a woven CMCs sample; s2, cutting out a subarea image from the three-dimensional digital gray scale image, and manually marking each material component in the subarea image to form training data; S3, training a deep learning segmentation model based on the training data, wherein the deep learning segmentation model is used for identifying the material components; s4, performing component segmentation on the complete three-dimensional digital gray scale image by using the trained deep learning segmentation model to obtain three-dimensional space distribution data of each component; s5, performing post-processing on the segmented three-dimensional space distribution data, and selecting a subvolume with material component representativeness as a finite element modeling object; S6, generating a corresponding finite element mesh model based on the real microscopic structure data of the subvolume; And S7, importing the finite element mesh model into finite element analysis software, endowing material properties, setting boundary conditions and performing simulation calculation to analyze the mechanical properties of the woven CMCs.
2. The finite element modeling and analysis method for the braided CMCs based on the true mesostructure according to claim 1, wherein the specific step of the step S1 is to scan the braided CMCs by an X-ray computed tomography technology to obtain a three-dimensional digital gray scale image, observe the scanning result and record the component category contained in the three-dimensional gray scale image of the braided CMCs after scanning.
3. The method of modeling and analyzing finite elements of woven CMCs based on true mesostructures of claim 2, wherein the component classes include warp yarns, weft yarns, matrix, voids and cracks.
4. The method for modeling and analyzing finite elements of knitted CMCs based on true mesostructures according to claim 1, wherein in step S3, the deep learning segmentation model is a convolutional neural network.
5. The method for modeling and analyzing finite elements of woven CMCs based on real mesostructures according to claim 1, wherein in step S5, the post-processing includes manually correcting the region of the recognition error in the segmentation result and performing smoothing filtering processing on each component boundary, wherein the smoothing filtering processing adopts recursive gaussian filtering, and the filtering coefficients σ of the recursive gaussian filtering in different directions in the three-dimensional space are independently set according to the component boundary characteristics.
6. The method for modeling and analyzing finite elements of knitted CMCs based on true microstructure according to claim 1, wherein in step S5, the subvolumes are selected to have a volume fraction of each material component that is consistent with a total volume fraction of the corresponding component in the complete three-dimensional image.
7. The method for modeling and analyzing finite elements of knitted CMCs based on real mesostructures according to claim 1, wherein in step S6, when generating the finite element mesh model, differentiated control parameters are set according to geometric complexity of the real mesostructures, the control parameters include mesh type, minimum target length of mesh and maximum mesh length.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method for modeling and analyzing finite elements of knitted CMCs based on a real mesostructure according to any of claims 1 to 7 when executing the program.
9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the method for modeling and analyzing finite elements of knitted CMCs based on a real mesostructure according to any of claims 1 to 7.

Description

Braiding CMCs finite element modeling and analysis method based on real mesoscopic structure Technical Field The method belongs to the field of image processing and finite element modeling of composite materials, and particularly relates to a method for modeling and analyzing finite elements of braided CMCs based on a real mesostructure. Background Ceramic matrix composites (Ceramic Matrix Composites, CMCs for short) are the preferred materials for the hot end components and thermal protection components of aircraft engines due to their excellent mechanical properties. As an anisotropic material, the mechanical properties and damage behavior of CMCs are not only dependent on the material properties of the components, but are also closely related to their microstructure. Thus, accurately predicting the mechanical properties and remaining life of CMCs is very challenging with the mutual coupling of various factors. The finite element Method (FINITE ELEMENT Method, abbreviated as FEM) is one of the important means for researching the mechanical properties of CMCs. In finite element analysis of woven CMCs, the representational volume element method (REPRESENTATIVE VOLUME ELEMENT, RVE for short) is a common technique that performs simulation analysis by constructing periodic element models. However, the RVE method builds an idealized model, and this idealized process deviates from the true mesostructure of the actual CMCs, resulting in a significant increase in the errors that occur in predicting the nonlinear deformation stage and the final failure mode. Meanwhile, local stress concentration effect caused by the manufacturing process cannot be effectively described based on an ideal RVE model, and the stress concentration effect is a main influencing factor of crack initiation and propagation in CMCs. Furthermore, in woven CMCs, the RVE element model has difficulty in accurately reconstructing the three-dimensional curved path of the yarn and the contact area formed by the mutual extrusion thereof, which directly affects the accuracy of the finite element simulation. In contrast, the high-fidelity finite element model based on the real microstructure of the CMCs can remarkably improve the accuracy of simulation analysis. And the high-fidelity finite element model is established based on the real microscopic structure of the braided CMCs, so that the accuracy of finite element simulation analysis of the braided CMCs is greatly improved. While the creation of high fidelity woven CMCs requires accurate acquisition of the microscopic structure within the material. The problems of noise, artifact, gray overlapping and the like of the three-dimensional digital gray images of the braided CMCs obtained based on CT scanning are further increased, and the difficulty of accurately identifying each component is further increased. And the accurately identified woven CMCs can lead to extremely high grid quantity and far exceed conventional calculation capability through a finite element model directly constructed by voxels. Therefore, under the condition of not changing the fidelity of the geometric model, the curved surface high-fidelity reconstruction is realized, and the establishment of the high-fidelity finite element model of the woven CMCs is very important. Therefore, a method of establishing a high-fidelity finite element model for braided CMCs is provided. The method can realize accurate calculation and simulation of the mechanical property and failure behavior of the woven CMCs, and further provides firm support for the use reliability and safety of the woven CMCs. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a finite element modeling and analyzing method for the braided CMCs based on a real microscopic structure. In order to achieve the technical purpose, the invention adopts the following technical scheme: the finite element modeling and analyzing method for the braided CMCs based on the real mesostructure is characterized by comprising the following steps of: S1, acquiring a three-dimensional digital gray image of a woven CMCs sample; s2, cutting out a subarea image from the three-dimensional digital gray scale image, and manually marking each material component in the subarea image to form training data; S3, training a deep learning segmentation model based on the training data, wherein the deep learning segmentation model is used for identifying the material components; s4, performing component segmentation on the complete three-dimensional digital gray scale image by using the trained deep learning segmentation model to obtain three-dimensional space distribution data of each component; s5, performing post-processing on the segmented three-dimensional space distribution data, and selecting a subvolume with material component representativeness as a finite element modeling object; S6, generating a corresponding finite element mesh model based on the real microscopic s