A composite material residual strength prediction method based on intelligent reconstruction

Abstract

The application belongs to the technical field of impact damage tolerance evaluation of fiber reinforced resin matrix composite structure, and particularly relates to a composite material residual strength prediction method based on intelligent reconstruction. A gray-scale image of a damaged area is divided into multiple damage images according to different gray-scale values; a damage contour of the composite material in the damage image is identified, the position information of the damage contour in the thickness direction of the composite material is determined according to the gray-scale value of the damage contour, and the damage contour is coordinated; the damage contour of all damage images is gridded according to the damage contour coordinates; the gridded coordinates of all damage contours are introduced into a pre-established finite element model of the composite material according to the corresponding position information, and the finite element model is three-dimensionally numerically reconstructed; the three-dimensionally numerically reconstructed finite element model is stiffness-decayed according to the damage area distribution characteristics of the three-dimensionally numerically reconstructed finite element model; and the residual strength of the stiffness-decayed finite element model is predicted by using a finite element method.

Description

Technical Field

[0001] This application belongs to the field of impact damage tolerance assessment technology for fiber-reinforced resin matrix composite structures, and specifically relates to a method for predicting the residual strength of composite materials based on intelligent reconstruction. Background Technology

[0002] FRP composite materials possess advantages such as high specific strength, good specific stiffness, fatigue resistance, and corrosion resistance, and are widely used in aerospace, rail transportation, wind power equipment, and other fields. However, unlike metallic materials that absorb impact energy through plastic deformation, composite materials are inherently brittle, and low-velocity impact loads can cause a significant decrease in their mechanical properties, posing serious safety hazards and threats during long-term operation. Therefore, accurately assessing the residual strength of composite material structures containing impact damage is of great significance for the safe service of engineering equipment.

[0003] Existing methods for predicting the residual strength of FRP composites after impact still suffer from the following problems: 1. Macroscopic phenomenological analysis methods directly fit the functional relationship between compressive residual strength and impact energy. This is an empirical method based on a large amount of experimental data, with unclear physical meaning and numerous, difficult-to-determine parameters. 2. Initial damage equivalent analysis methods equate the initial damage caused by impact to a regularly shaped geometric region. This is an artificially simplified approximation method, which is difficult to characterize complex impact damage forms. Furthermore, the accuracy of this method depends on the experience of engineers, and its applicability has not been verified. 3. The impact-post-impact loading full-process analysis method uses finite element technology to simulate the damage evolution of FRP composite structures during impact and post-impact loading. Although this method avoids interference from human factors, it still needs to address issues such as dynamic-static solution coupling, stress-deformation oscillations after impact rebound, the significant impact of impact damage prediction accuracy on residual strength assessment, and low computational efficiency due to numerous analysis steps. More importantly, the full-process analysis method is not suitable for situations where the impact energy is unknown. For composite structures with known internal damage information but unknown impact energy, it is impossible to predict their residual strength. In summary, existing methods for predicting residual strength have limitations in practical engineering applications. How to design a method that can perform refined quantitative processing of impact damage information, achieve full-domain damage characterization and numerical reconstruction, and ultimately predict the residual strength of FRP composite structures has become a technical problem to be solved in this field. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a method for predicting the residual strength of composite materials based on intelligent reconstruction, comprising the following steps:

[0005] Step M1: Convert the ultrasonic C-scan TOF image of the composite material with damaged area into a normalized grayscale image;

[0006] Step M2: Segment the grayscale image into multiple damaged images according to different grayscale values;

[0007] Step M3: Identify the damage contour of the composite material in the damage image, determine the position information of the damage contour in the thickness direction of the composite material based on the gray value of the damage contour, and coordinate the damage contour.

[0008] Step M4: Mesh the damage contours of all damage images according to the damage contour coordinates;

[0009] Step M5: Introduce the meshed coordinates of all damage contours into the pre-established finite element model of the composite material according to their corresponding position information, and perform three-dimensional numerical reconstruction of the finite element model.

[0010] Step M6: Based on the damage area distribution characteristics of the 3D numerically reconstructed finite element model, perform stiffness attenuation on the 3D numerically reconstructed finite element model.

[0011] Step M7: Use the finite element method to predict the residual strength of the finite element model after stiffness decay.

[0012] Preferably, the ultrasound C-scan TOF image includes a color band that reflects the position in the thickness direction.

[0013] Preferably, step M2 specifically includes: dividing the damaged area in the grayscale image into multiple sub-grayscale images according to the grayscale values ​​of each color in the color band, each sub-grayscale image containing multiple color block regions, removing the non-color block regions of each sub-grayscale image, retaining the effective color block regions, and using the effective color block regions as the damage image at the current thickness position of the composite material.

[0014] Preferably, step M3 specifically includes the following steps for mapping the damage contour:

[0015] Step M31: Binarize the damage contour into multiple pixel units, and establish the coordinate information of the damage contour by relating the pixel units to the actual size of the composite material.

[0016] Preferably, step M4 specifically includes:

[0017] Step M41: Traverse the coordinates to obtain the maximum and minimum values ​​of the damage contour in the x and y directions, forming the smallest rectangle containing the damage contour;

[0018] Step M42: Within the minimum rectangle, starting from the position of the minimum x value, and with a step size of one pixel, damage unit information is assigned from the minimum y value to the pixel unit where the maximum y value is located, forming a grid containing damage unit information.

[0019] Preferably, step M6 includes partitioning the finite element model of the three-dimensional numerical reconstruction and then performing stiffness attenuation. The partitioning includes Inclusion Zone I and Inclusion Zone II: Inclusion Zone I is the damage core area containing the entire pit; Inclusion Zone II is the area between Inclusion Zone I and the outer edge of the entire damage area.

[0020] Preferably, the stiffness attenuation specifically includes: for inclusion region I, attenuating the longitudinal elastic modulus E11, the transverse elastic modulus E22, and the in-plane shear modulus G12 to 5% of their initial values; for inclusion region II, attenuating the transverse elastic modulus E22 and the in-plane shear modulus G12 to 20% of their initial values; and for elements exhibiting interface delamination, reducing the stiffness of the corresponding adhesive element to 0.5% of its initial value.

[0021] The advantages of this application include: This technical solution reconstructs experimental measurement data into numerical simulation through a series of methods, providing relatively fast and accurate predictions of residual strength, and obtaining comprehensive information on damage mechanisms and force transmission mechanisms. The application scope of the intelligent damage reconstruction method can be broadened to more general situations; that is, given only the damage image information of the internal C-scan (the damage does not necessarily have to be caused by impact load), the residual strength of the structure can be predicted, enabling the formation of a "virtual experiment" based on fine image segmentation. It achieves intelligent assessment of impact damage to composite material plates, thereby realizing the scientific prediction of the residual strength of composite material plates after impact. Attached Figure Description

[0022] Figure 1 TOF image for lesion detection using ultrasound C-scan;

[0023] Figure 2 Clustered images of impact damage that need to be removed;

[0024] Figure 3 For effective impact damage clustering images;

[0025] Figure 4 This is a schematic diagram of the contour coordinate representation of the grayscale image of tomographic damage.

[0026] Figure 5 This is a schematic diagram of numerical reconstruction of a tomographic image;

[0027] Figure 6 This is a schematic diagram of impact damage zoning in composite materials.

[0028] Figure 7 Comparison of impact damage test results and three-dimensional numerical reconstruction results for composite materials;

[0029] Figure 8 Load-displacement curves of laminates with impact damage obtained from prediction and testing. Detailed Implementation

[0030] To make the technical solution and advantages of this application clearer, the technical solution of this application will be described in a clearer and more complete manner below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some embodiments of this application, and are only used to explain this application, not to limit this application. It should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings. Other related parts can be referred to the general design. In the absence of conflict, the embodiments and technical features in the embodiments of this application can be combined with each other to obtain new embodiments.

[0031] M1: Preprocess the ultrasound C-scan TOF image and convert it into a standardized grayscale image. The ultrasound C-scan TOF image is required to include a gradient color band that can reflect thickness (or relative thickness) information.

[0032] M2: The K-Means++ clustering algorithm for segmenting damaged images also includes the following steps: M201: Based on the proportion r of the color band occupied by the color patch region of the tested specimen, determine the minimum value of the segmentation cluster size (K value) as [N]. p / r], where Np is the number of layers in the test specimen, and the symbol [x] indicates rounding up x; M202: compare the color block area of ​​the test specimen with the sum of the pixels of the color band ratio r, and select the corresponding color block area.

[0033] M3: Identify the valid information image and locate it along the plate thickness tomography, outputting the damage point coordinate matrix: M301: First, superimpose the color block area of ​​the test specimen occupying the proportion r of the color band of the identified valid information gray value image; M302: Locate the color block area of ​​the test specimen occupying the proportion r of the color band along the plate thickness tomography.

[0034] M4: The extracted geometric information of the damage is converted into meshed coordinates of the damage region and introduced into the finite element model to complete the three-dimensional numerical reconstruction of the damage region: M401: Traverse the coordinates to obtain the maximum and minimum values ​​(xmin, xmax, ymin, and ymax) in the x and y directions, and determine the damaged element region; M402: Starting from the element containing the minimum x value (xmin), using a mesh element size as the traversal step size, based on the maximum and minimum y values ​​(ymin and ymax) containing damage information within a single step size interval, the corresponding damaged elements within the traversal step size are sequentially selected, and finally, a set of damaged elements is established. The meshed coordinates are divided into multiple continuous mesh rectangular regions based on the element size. If a discontinuous region is encountered, it is divided into multiple mesh rectangular regions.

[0035] M5: Based on the impact damage distribution characteristics, degrade the mechanical properties of the unidirectional plate: M501: Based on the damage morphology observed after the low-speed impact test, the composite laminate after impact is divided into two softened inclusion zones, where inclusion zone I is the damage core zone containing the entire pit, and inclusion zone II is the area sandwiched between the upper interface layer projection edge and the boundary of inclusion zone I; M502: Combining the actual low-speed impact damage morphology, the impact damage is equivalent to softened inclusions, and the stiffness of the damaged area is directly reduced.

[0036] The stiffness attenuation method is as follows: For inclusion region I, which contains fiber and matrix damage, the longitudinal elastic modulus E11, transverse elastic modulus E22, and in-plane shear modulus G12 are attenuated to 5% of their initial values; for inclusion region II, which contains only matrix damage, the transverse elastic modulus E22 and in-plane shear modulus G12 are attenuated to 20% of their initial values, and for elements with interface delamination, the stiffness of the corresponding adhesive element is reduced to 0.5% of its initial value.

[0037] M6: The prediction of residual strength of composite materials with impact damage using the finite element method includes the following steps: M601: Introducing the damage C-scan image obtained from the impact test and comparing it with the numerically reconstructed CAI model; M602: Performing a finite element simulation of post-impact compression of the specimen with impact damage to obtain the predicted results of the residual strength of the composite material. Figures 1 to 8 As shown, this embodiment provides a method for predicting the residual strength of composite materials based on intelligent damage quantitative assessment. The prediction object is a composite laminate, and the geometric dimensions of the rectangular carbon / epoxy resin laminate are 100×150×4.16mm3. The material system is T700 / M21, and the material parameters are as follows: Figure 1 As shown. The laminate has a ply configuration of [02 / 452 / 902 / -452]S, with a single layer thickness of 0.26 mm. The TOF image of the damage detected by ultrasonic C-scan is shown below. Figure 2 The scanning area is 100mm × 100mm. The testing steps include: performing tomography on the C-scan image using the K-Means++ clustering algorithm; numerical reconstruction method for the impact damage image and simplification strategy for the impact damage; and after embedding the damage information array image into the finite element model, performing intelligent simulation to obtain the residual strength of the composite material plate.

[0038] First, the K-Means++ clustering algorithm is used to perform tomography on the impact damage projection obtained from the C-scan, obtaining images with effective information for tomographic localization along the plate thickness. Layered damage is then located along the plate thickness. Only clusters containing both the left-side color band and the right-side damage portion can reflect the damage contour and location information. Based on this, effective grayscale information images are selected, such as... Figure 2 and Figure 3 As shown.

[0039] By utilizing the correspondence between pixels in the binarized image and the true size of the detection region, the coordinate information of the damaged region's contour can be established, such as... Figure 4 As shown.

[0040] By traversing the coordinates, the maximum and minimum values ​​(xmin, xmax, ymin, and ymax) in the x and y directions are obtained to determine the damaged unit region, such as... Figure 5 The white dashed rectangle area is shown in the diagram. Then, starting from the cell containing the minimum x value (xmin), using a grid cell size as the traversal step size, the damaged cells within the traversal step size are sequentially filtered based on the maximum and minimum y values ​​(ymin and ymax) containing damage information within a single step size interval. Finally, a set of damaged cells is established.

[0041] In this embodiment, the `getByBoundingBox` function from ABAQUS is used for region selection during numerical reconstruction. A 0.5x element size error is applied to all coordinate points; that is, for the coordinates of damage points in the edge region, if they are within half an element size of an adjacent node, damage element information is assigned. Specifically, in numerical reconstruction, for discontinuous damage regions, the corresponding y-axis coordinate information is given based on segmented outlines of the discontinuous damage regions, such as... Figure 5 The three segmented intervals shown are (y1min, y1max), (y2min, y2max), and (y3min, y3max).

[0042] In this embodiment, the typical softening inclusion region caused by impact damage in composite materials is divided into three regions within each layup: inclusion region I containing fiber and matrix damage, inclusion region II containing only matrix damage, and an undamaged region, as shown below. Figure 6 As shown. For interlayer delamination damage, the interlayer interface is divided into two regions: a delaminated region and an undelaminated region, as shown. Figure 6 As shown.

[0043] This embodiment provides a comparison between a damage C-scan image obtained from an impact test using 17J impact energy and a numerically reconstructed CAI model. (See attached image.) Figure 7 As shown, the shape and size of the layered projection are the same as the scanning results, indicating that the numerical reconstruction method matches the scanning results well. Furthermore, it can effectively perform tomography on the layered projection images from the C-scan, achieving three-dimensional layer-by-layer reconstruction of impact damage.

[0044] Finally, the finite element results of the three-dimensional layer-by-layer reconstruction of the compressive load-displacement curve of the laminate, including impact damage and 17J impact energy, were compared with the experimental results, as follows: Figure 8As shown, the finite element method (FEM) predicted a peak strength of 257 MPa, while the experimental result was 242 MPa, with errors of 6.2%. It can be observed that the residual strength calculated by the FEM and the experimentally measured residual strength agree well. The prediction error for the residual strength under 17 J impact energy is around 5%, indicating that this model can accurately predict the residual strength of composite laminates subjected to low-velocity impact loads.

[0045] It should be noted that although the above embodiments have been described herein, this does not limit the scope of patent protection of the present invention. Therefore, any changes and modifications made to the embodiments described herein based on the innovative concept of the present invention, or equivalent structural or procedural transformations made using the content of the present invention's specification and drawings, directly or indirectly applying the above technical solutions to other related technical fields, are all included within the scope of patent protection of the present invention.

Claims

1. A method for predicting the residual strength of composite materials based on intelligent reconstruction, characterized in that, Includes the following steps: Step M1: Convert the ultrasonic C-scan TOF image of the composite material with damaged area into a normalized grayscale image; Step M2: Segment the grayscale image into multiple damaged images according to different grayscale values; Step M3: Identify the damage contour of the composite material in the damage image, determine the position information of the damage contour in the thickness direction of the composite material based on the gray value of the damage contour, and coordinate the damage contour. Step M4: Mesh the damage contours of all damage images according to the damage contour coordinates; Step M5: Introduce the meshed coordinates of all damage contours into the pre-established finite element model of the composite material according to their corresponding position information, and perform three-dimensional numerical reconstruction of the finite element model. Step M6: Based on the damage area distribution characteristics of the 3D numerically reconstructed finite element model, perform stiffness attenuation on the 3D numerically reconstructed finite element model. Step M7: Use the finite element method to predict the residual strength of the finite element model after stiffness decay; Step M4 specifically includes: Step M41: Traverse the coordinates to obtain the maximum and minimum values ​​of the damage contour in the x and y directions, forming the smallest rectangle containing the damage contour; Step M42: Within the minimum rectangle, starting from the position of the minimum x value, and with a step size of one pixel, damage unit information is assigned from the minimum y value to the pixel unit where the maximum y value is located, forming a grid containing damage unit information.

2. The method for predicting the residual strength of composite materials based on intelligent reconstruction as described in claim 1, characterized in that, The ultrasonic C-scan TOF image contains color bands that reflect the position in the thickness direction.

3. The method for predicting the residual strength of composite materials based on intelligent reconstruction as described in claim 2, characterized in that, Step M2 specifically includes: dividing the damaged area in the grayscale image into multiple sub-grayscale images according to the grayscale values ​​of each color in the color band, each sub-grayscale image containing multiple color block regions, removing the non-color block regions of each sub-grayscale image, retaining the effective color block regions, and using the effective color block regions as the damage image at the current thickness position of the composite material.

4. The method for predicting the residual strength of composite materials based on intelligent reconstruction as described in claim 3, characterized in that, Step M3 specifically includes the following steps for mapping the damage contour: Step M31: Binarize the damage contour into multiple pixel units, and establish the coordinate information of the damage contour by relating the pixel units to the actual size of the composite material.

5. The method for predicting the residual strength of composite materials based on intelligent reconstruction as described in claim 1, characterized in that, Step M6 involves partitioning the finite element model of the three-dimensional numerical reconstruction and then performing stiffness attenuation. The partitions include Inclusion Zone I and Inclusion Zone II: Inclusion Zone I is the damage core area containing the entire pit; Inclusion Zone II is the area between Inclusion Zone I and the outer edge of the entire damage area.

6. The method for predicting the residual strength of composite materials based on intelligent reconstruction as described in claim 1, characterized in that, The stiffness reduction specifically includes: for inclusion region I, the longitudinal elastic modulus E11, the transverse elastic modulus E22, and the in-plane shear modulus G12 are reduced to 5% of their initial values; for inclusion region II, the transverse elastic modulus E22 and the in-plane shear modulus G12 are reduced to 20% of their initial values, and for elements with interface delamination, the stiffness of the corresponding adhesive element is reduced to 0.5% of its initial value.

Images