Intelligent surveying and mapping method for residual stress distribution diagram of SiCf / SiC ceramic-based composite material
Through low-temperature pulsed laser cutting and liquid nitrogen cooling technology combined with X-ray diffraction method and convolutional neural network, the problem of low detection accuracy and efficiency in traditional detection methods is solved, and the high-precision and efficient detection of residual stress of SiCf/SiC ceramic matrix composite materials is achieved, which is suitable for stress analysis of complex components.
Patent Information
- Application Number
- CN202510453228.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-11
- Publication Date
- 2025-07-22
AI Technical Summary
Traditional X-ray diffraction method is difficult to accurately detect residual stress differences in SiCf/SiC ceramic matrix composites. Mechanical cutting causes stress relaxation, manual planning of test points is time-consuming and unscientific, which affects detection accuracy and efficiency.
The samples were prepared by low-temperature pulsed laser cutting combined with liquid nitrogen cooling technology, combined with X-ray diffraction method and convolutional neural network algorithm, intelligently planned test points and reconstructed stress fields to generate high-resolution stress distribution cloud maps.
It significantly improves the detection accuracy and efficiency of residual stress distribution of SiCf/SiC ceramic matrix composites, and reduces the number of test points by 20%. It is suitable for stress sampling and analysis of aircraft engines and nuclear reactor components, supporting manufacturing process optimization and life evaluation.
Smart Images
Figure CN120352058A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of detection of ceramic matrix composites, and more specifically, relates to an intelligent mapping method for the residual stress distribution map of SiC f / SiC ceramic matrix composites. Background Art
[0002] SiC f / SiC ceramic matrix composites are composed of continuous SiC fibers and polycrystalline SiC matrices, and have significant mechanical property gradients and thermal expansion mismatch characteristics. The tensile strength along the fiber axis can be increased by 2-3 times compared with the transverse tensile strength, and the difference in the thermal expansion coefficients of the fiber and the matrix (for example, about 4.0×10 -6 / K for the fiber axis and about 5.5×10 -6 / K for the matrix) will result in significant residual stresses at the interface position. In addition, the pores and microcracks between fiber bundles generated during the manufacturing process are prone to become stress concentration sources, further affecting the material reliability. Traditional X-ray diffraction methods have certain limitations in detecting the residual stresses of such materials, and it is difficult to accurately distinguish the residual stress differences in different regions; the local temperature rise (>200°C) generated during mechanical cutting of specimens will cause stress relaxation and result in errors; when drawing the residual stress distribution map, manually planning the test points takes a long time, and the planned distribution of test points is not scientific and reasonable enough. These problems all restrict the realization of high-precision stress mapping. Summary of the Invention
[0003] In view of this, the purpose of the present invention is to provide an intelligent mapping method for the residual stress distribution map of SiC f / SiC ceramic matrix composites, which can significantly improve the detection accuracy and efficiency of the residual stress distribution of SiC f / SiC ceramic matrix composites. The number of test points can be reduced by 20%, which is applicable to stress sampling analysis of complex components such as aero-engine and nuclear reactor components, and can be used to support manufacturing process optimization and life assessment.
[0004] The present invention provides an intelligent mapping method for the residual stress distribution map of SiC f / SiC ceramic matrix composites, including the following steps:
[0005] a) Prepare specimens by using low-temperature pulsed laser cutting combined with liquid nitrogen cooling process;
[0006] b) After surface treatment of the specimens, plan and locate the test points, then use X-ray diffraction method to collect the residual stress data of each test point, and finally reconstruct the stress field by combining the convolutional neural network algorithm to generate a high-resolution stress distribution cloud map.
[0007] Preferably, in step a), the low-temperature pulsed laser cutting uses a pulsed laser cutting system with a wavelength of 1000 nm to 1100 nm and a pulse width of 5 ns to 15 ns; the liquid nitrogen injection pressure for liquid nitrogen cooling is 0.1 MPa to 0.5 MPa, and the flow rate is 4 L / min to 6 L / min; the cutting path is a spiral progressive type, and the thickness of the heat-affected zone is less than 10 μm.
[0008] Preferably, before the surface treatment of the specimen in step b), it further includes:
[0009] After the specimen is ultrasonically cleaned with acetone for 4 min to 6 min, it is cured with a low-shrinkage epoxy resin at 70 °C to 90 °C for 1 h to 3 h.
[0010] Preferably, the process of the surface treatment in step b) is specifically as follows:
[0011] First, sequentially use sandpapers from 500 mesh to 5000 mesh in combination with diamond grinding paste for stage-by-stage wet polishing. The grinding time for each stage ≤ 30 seconds, the normal load is 0.05 MPa to 0.15 MPa, the cooling water flow rate is 0.5 L / min to 1.5 L / min, and the surface temperature rise ≤ 10 °C;
[0012] Then, perform pulsed electrolytic polishing with a 4 wt% to 6 wt% NaOH-ethylene glycol electrolyte for 4 min to 6 min to make the surface roughness Ra ≤ 0.3 μm; verify that there are no microcracks with a length > 20 μm on the surface through a confocal laser microscope.
[0013] Preferably, the process of the test point planning in step b) is specifically as follows:
[0014] Intelligently plan the test point density based on the macroscopic morphology and structural characteristics of the material, and divide the region of interest and the conventional region according to the fiber distribution density and regional curvature on the material surface; the test point density in the region of interest is 10 to 100 points / cm 2 , and the test point density in the conventional region is 1 to 5 points / cm 2 .
[0015] Preferably, the test points in the region of interest are arranged in a hexagonal array with a spacing of 1 mm to 3 mm; the conventional region is arranged in a rectangular array with a spacing of 5 mm to 10 mm.
[0016] Preferably, in step b), the positioning is realized by the linkage of high-precision laser projection positioning and an electromechanical displacement platform to accurately position the test points; the accuracy of the high-precision laser projection positioning is ±10 μm, and the spacing error of the linkage with the electromechanical displacement platform to accurately position the test points ≤ 0.1 mm.
[0017] Preferably, in step b), the X-ray diffraction method uses the (311) crystal plane of the SiC material as the test object, sets the ψ angle to 0°, 25°, 35°, 45°, and the scanning step size to 0.01°.
[0018] Preferably, the convolutional neural network described in step b) includes 2 to 4 convolutional layers and 1 to 3 fully connected layers, uses the mean square error as the training loss function, and outputs the stress tensors σxx, σyy, and σxy.
[0019] Preferably, the specific method for generating the high-resolution stress distribution cloud map in step b) is as follows:
[0020] Generate a stress distribution cloud map with a resolution of 0.1 mm through a three-dimensional interpolation algorithm, and automatically label the dangerous areas with tensile stress > 250 MPa and the dangerous areas with compressive stress < -250 MPa.
[0021] The present invention provides a method for intelligently mapping the residual stress distribution map of SiC f / SiC ceramic matrix composite materials, including the following steps: a) Prepare a specimen using a low-temperature pulsed laser cutting combined with liquid nitrogen cooling process; b) After surface treatment of the specimen, plan and locate the test points, then use the X-ray diffraction method to collect the residual stress data of each test point, and finally reconstruct the stress field by combining the convolutional neural network algorithm to generate a high-resolution stress distribution cloud map. Compared with the prior art, the method for intelligently mapping the residual stress distribution map of SiC f / SiC ceramic matrix composite materials adopts specific process steps and conditions to achieve a good overall interaction: it can significantly improve the detection accuracy and efficiency of the residual stress distribution of SiC f / SiC ceramic matrix composite materials, the number of test points can be reduced by 20%, which is applicable to stress sampling analysis of complex components such as aero-engine and nuclear reactor components, and can be used to support manufacturing process optimization and life assessment. Brief Description of the Drawings
[0022] Figure 1 It is a flow chart of the method for intelligently mapping the residual stress distribution map of SiC f / SiC ceramic matrix composite materials provided by the present invention;
[0023] Figure 2 It is the stress cloud map with a resolution of 0.1 mm generated in Example 1 of the present invention. Detailed Description of the Embodiment
[0024] Next, in combination with the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.
[0025] The present invention provides a method for intelligently mapping the residual stress distribution map of a SiC f / SiC ceramic matrix composite material, comprising the following steps:
[0026] a) Prepare a specimen by using a low-temperature pulsed laser cutting combined with liquid nitrogen cooling process;
[0027] b) After surface treatment of the specimen, perform test point planning and positioning, then collect the residual stress data of each test point by using the X-ray diffraction method, and finally reconstruct the stress field by combining the convolutional neural network algorithm to generate a high-resolution stress distribution cloud map.
[0028] The present invention first prepares a specimen by using a low-temperature pulsed laser cutting combined with liquid nitrogen cooling process to inhibit the cutting heat affected zone.
[0029] In a preferred embodiment of the present invention, specimens with a size of 2.5 cm × 2.5 cm are cut from components or characteristic elements such as the outer ring of an aero-engine turbine.
[0030] In the present invention, the low-temperature pulsed laser cutting preferably uses a pulsed laser cutting system with a wavelength of 1000 nm to 1100 nm and a pulse width of 5 ns to 15 ns, and more preferably uses a pulsed laser cutting system with a wavelength of 1064 nm and a pulse width of 10 ns.
[0031] In the present invention, the liquid nitrogen cooling is continuously sprayed onto the cutting area during the cutting process; the liquid nitrogen spraying pressure of the liquid nitrogen cooling is preferably 0.1 MPa to 0.5 MPa, more preferably 0.3 MPa, and the flow rate is preferably 4 L / min to 6 L / min, more preferably 5 L / min.
[0032] In the present invention, the cutting path for preparing the specimen by using the low-temperature pulsed laser cutting combined with liquid nitrogen cooling is preferably a spiral progressive type, which can generally reduce the heat input, and the thickness of the heat affected zone is preferably less than 10 μm.
[0033] In the present invention, the surface roughness Ra of the cut specimen is preferably < 1.2 μm.
[0034] After that, the present invention performs surface treatment on the above-mentioned cut specimen, then performs test point planning and positioning, collects the residual stress data of each test point by using the X-ray diffraction method, and finally reconstructs the stress field by combining the convolutional neural network algorithm to generate a high-resolution stress distribution cloud map.
[0035] In the present invention, before the surface treatment of the specimen, it preferably further includes:
[0036] After the specimen is ultrasonically cleaned with acetone for 4 min to 6 min, it is cured with a low-shrinkage epoxy resin at 70 °C to 90 °C for 1 h to 3 h;
[0037] More preferably:
[0038] After the specimen is ultrasonically cleaned with acetone for 5 min, it is cured with a low-shrinkage epoxy resin at 80 °C for 2 h.
[0039] In the present invention, the specimen is ultrasonically cleaned with acetone, which can remove surface contaminants, and is embedded and cured with a low-shrinkage epoxy resin, which can avoid introducing additional stress due to curing shrinkage.
[0040] In the present invention, the surface treatment preferably adopts a staged water grinding and polishing process for low-damage surface treatment; the process of the surface treatment is preferably specifically:
[0041] First, sequentially use sandpapers from 500 mesh to 5000 mesh in combination with diamond grinding paste for staged water grinding and polishing. The grinding time for each stage ≤ 30 s, control the normal load of the grinding and polishing machine to be 0.05 MPa to 0.15 MPa, and at the same time maintain the surface temperature through cooling water. The cooling water flow rate is 0.5 L / min to 1.5 L / min, and the surface temperature rise ≤ 10 °C;
[0042] After mechanical polishing, use a 4 wt% to 6 wt% NaOH-ethylene glycol electrolyte as the polishing medium for pulsed electrolytic polishing. The polishing time is 4 min to 6 min to make the surface roughness Ra ≤ 0.3 μm; verify through a confocal laser microscope that there are no microcracks with a length > 20 μm on the surface;
[0043] More preferably:
[0044] First, sequentially use sandpapers from 500 mesh to 5000 mesh in combination with diamond grinding paste for staged water grinding and polishing. The grinding time for each stage is 30 s, control the normal load of the grinding and polishing machine to be 0.1 MPa, and at the same time maintain the surface temperature through cooling water. The cooling water flow rate is 1 L / min, and the surface temperature rise is 7 °C to 9 °C;
[0045] After mechanical polishing, use a 5 wt% NaOH-ethylene glycol electrolyte as the polishing medium for pulsed electrolytic polishing. The polishing time is 5 min to make the surface roughness Ra ≤ 0.3 μm; after the surface treatment of the specimen, it is necessary to verify through a confocal laser microscope that there are no microcracks with a length > 20 μm on the surface.
[0046] In the present invention, it is verified through a confocal laser microscope that there are no microcracks with a length exceeding 20 μm on the surface, ensuring the accuracy of subsequent test data.
[0047] In the present invention, the process of test point planning is preferably specifically as follows:
[0048] Intelligently plan the test point density based on the macroscopic morphology and structural characteristics of the material, and divide the region of interest and the conventional region according to the fiber distribution density and regional curvature of the material surface.
[0049] Based on the macroscopic morphology and structural characteristics of the specimen, the present invention obtains the surface contour image of the sample through an optical microscope, and combines image processing algorithms to extract characteristic information such as fiber distribution density and regional curvature; according to the analysis results, the specimen is divided into a region of interest (such as a possible high stress gradient region, a fiber dense region, etc.) and a conventional region (such as a possible stress stable distribution region, etc.).
[0050] In the present invention, the test point density of the region of interest is preferably 10 - 100 points / cm 2 , and the test point density of the conventional region is preferably 1 - 5 points / cm 2 .
[0051] In the present invention, the test points in the region of interest preferably adopt a hexagonal array layout, and the spacing is preferably 1 mm - 3 mm, more preferably 2 mm; the conventional region preferably adopts a rectangular array layout, and the spacing is preferably 5 mm - 10 mm, more preferably 8 mm.
[0052] The total number of test points in the present invention using the above test point planning method is reduced by more than 20% compared with traditional manual planning, and the total detection time of the whole process is reduced by ≥25%.
[0053] In the present invention, the positioning preferably adopts the linkage of high-precision laser projection positioning and an electromechanical displacement platform to achieve precise positioning of the test points, ensure the matching of the test point density and the stress gradient, and avoid data redundancy or omission; the accuracy of the high-precision laser projection positioning is preferably ±10 μm, and the spacing error for realizing precise positioning of the test points in linkage with the electromechanical displacement platform is preferably ≤0.1 mm.
[0054] The present invention uses the X-ray diffraction method to collect stress data (XRD stress measurement); the X-ray diffraction method preferably takes the (311) crystal plane (2θ = 90.3°) of the SiC material as the test object for testing and data processing, and the ψ angle is preferably set to 0°, 25°, 35°, 45°, and the scanning step is preferably 0.01°.
[0055] The present invention collects the diffraction peak intensity and position information for each test point, uses the conventional X-ray stress measurement method for data processing, obtains the stress tensor data of each test point through multi-angle measurement, and realizes the precise correspondence with the test point position.
[0056] Finally, for the stress field reconstruction and visualization of the present invention, the XRD stress measurement data is input into a convolutional neural network (CNN) for feature extraction; the convolutional neural network (CNN) preferably includes 2 to 4 convolutional layers and 1 to 3 fully connected layers, and more preferably 3 convolutional layers and 2 fully connected layers; the mean square error (MSE) is preferably used as the training loss function, and the stress tensors σxx, σyy, and σxy are output.
[0057] In the present invention, after the CNN outputs the stress tensors σxx, σyy, and σxy, the generation of the high-resolution stress distribution cloud map is preferably specifically as follows:
[0058] A stress distribution cloud map with a resolution of 0.1 mm is generated through a three-dimensional interpolation algorithm, and the dangerous areas with tensile stress > 250 MPa and compressive stress < -250 MPa are automatically marked.
[0059] The present invention can set an alarm threshold to automatically mark the stress dangerous areas in the cloud map, especially the tensile stress dangerous areas; in addition, an analysis algorithm can be used to synchronously generate a stress gradient heat map and a statistical report to intuitively display the stress concentration position and magnitude.
[0060] The present invention provides a multi-modal residual stress mapping method combining cryogenic processing technology, X-ray diffraction technology, and intelligent algorithms, which is used to achieve high-resolution mapping of the residual stress distribution image of SiC f / SiC ceramic matrix composites, and has the following advantages and beneficial effects:
[0061] (1) Cryogenic composite cutting process: The synergistic effect of pulsed laser and liquid nitrogen injection is adopted to suppress the depth of the cutting heat affected zone (the thickness is not greater than 10 μm), and the thermal stress suppression effect is greatly improved;
[0062] (2) Intelligent planning of test points: Based on the macroscopic morphology and structural characteristics of the material, the test point layout is dynamically optimized. For example, the test point density is set to 10 - 100 points / cm in the area with a high stress distribution gradient 2 , and the test point density is set to 1 - 5 points / cm in the area with a low stress distribution gradient 2 . The total number of test points is reduced by more than 20% compared with the traditional manual planning, and the full-process detection time is reduced by ≥ 25%;
[0063] (3) High-precision stress field reconstruction and visualization: Combining the X-ray diffraction stress test data with the convolutional neural network (CNN) algorithm, sub-millimeter-level reconstruction of the stress field is realized. A stress distribution cloud map with a resolution of 0.1 mm is generated through a three-dimensional interpolation algorithm, and the tensile / compressive stress dangerous areas are marked.
[0064] On this basis, this method can be applied to the residual stress detection of the outer ring of aero-engine turbines, nuclear reactor components, porous or micro-curved components, and is especially suitable for the quantitative analysis of the residual stress field after the preparation of high-temperature components made of ceramic matrix composites for aero-engines, which can significantly improve the detection accuracy and efficiency.
[0065] The present invention provides a method for intelligently mapping the residual stress distribution map of SiC f / SiC ceramic matrix composites, comprising the following steps: a) preparing a specimen by using a low-temperature pulsed laser cutting process combined with liquid nitrogen cooling; b) after surface treatment of the specimen, planning and positioning test points, then collecting residual stress data of each test point by using the X-ray diffraction method, and finally reconstructing the stress field by combining a convolutional neural network algorithm to generate a high-resolution stress distribution cloud map. Compared with the prior art, the method for intelligently mapping the residual stress distribution map of SiC f / SiC ceramic matrix composites adopts specific process steps and conditions to achieve good overall interaction: it can significantly improve the detection accuracy and efficiency of the residual stress distribution of SiC f / SiC ceramic matrix composites, the number of test points can be reduced by 20%, which is suitable for stress sampling analysis of complex components such as aero-engine and nuclear reactor components, and can be used to support the optimization of manufacturing processes and life assessment.
[0066] To further illustrate the present invention, the following detailed description is given through the following embodiments.
[0067] Example 1
[0068] Residual stress detection of the outer ring component of an aero-engine turbine, comprising the following steps:
[0069] (1) Specimen preparation:
[0070] Cut a 2.5 cm × 2.5 cm specimen from the outer ring of an aero-engine turbine. Use a pulsed laser cutting system with a wavelength of 1064 nm and a pulse width of 10 ns, in combination with liquid nitrogen injection at 0.3 MPa (flow rate of 5 L / min), and cut along a spiral path. The thickness of the heat-affected zone is 8 μm. After cutting, the surface roughness Ra of the specimen is 1.1 μm. After ultrasonic cleaning with acetone for 5 minutes, it is cured at 80 °C for 2 hours using a low-shrinkage epoxy resin.
[0071] (2) Surface treatment:
[0072] Water grinding and polishing are carried out in stages successively using sandpapers with 500 to 5000 meshes in combination with diamond grinding paste. The grinding time for each grade of sandpaper is 30 seconds, the normal load is 0.1 MPa, the cooling water flow rate is 1 L / min, and the surface temperature rise is about 8 °C. Subsequently, pulse electrolytic polishing is carried out for 5 minutes using a 5wt% NaOH-ethylene glycol electrolyte, and the surface roughness is reduced to Ra = 0.28 μm. A confocal microscope confirms that there are no microcracks with a length > 20 μm.
[0073] (3) Test point planning and data acquisition:
[0074] The surface image of the specimen is obtained through an optical microscope, and the fiber-dense area is identified as the high-stress gradient area. A hexagonal array of test points (spacing 2 mm) is planned, and a rectangular array (spacing 8 mm) is used in the conventional area. The X-ray diffractometer measures with the (311) crystal plane (2θ = 90.3°), the ψ angle is 0°, 25°, 35°, 45°, the scanning step is 0.01°, and a total of 256 test point data are collected. Compared with the traditional manual planning method, the number of test points in this embodiment is reduced by 22.4%.
[0075] (4) Stress field reconstruction and result analysis:
[0076] After normalizing the XRD data, it is input into a CNN network (including 3 convolutional layers and 2 fully connected layers). After training, the stress tensors σxx, σyy, and σxy are output, and a stress nephogram with a resolution of 0.1 mm is generated through three-dimensional interpolation. See Figure 2 as shown. The results show that the maximum tensile stress in the fiber-dense area is 320 MPa, the compressive stress in the matrix area is -150 MPa, and the system automatically marks 3 tensile stress danger zones (> 250 MPa). The total detection time is reduced by 28%. Compared with the traditional mapping method, the resolution of the stress distribution nephogram in this embodiment is increased by 10 times, and the positioning error of the danger zone is ≤ 0.5 mm, meeting the detection requirements of high-temperature components of aeroengines.
[0077] The experimental results show that the present invention achieves the following beneficial effects:
[0078] (1) Improvement in test accuracy: Using the interpolation algorithm significantly improves the spatial resolution of the residual stress distribution map, and the residual stress in the unmeasured area is deduced and obtained, effectively improving the data reliability;
[0079] (2) Optimization of test efficiency: The intelligent planning of test points reduces the total number of test points by more than 20%, effectively reducing the number of ineffective or low-efficiency test points, thereby reducing the total test time of the whole process;
[0080] (3) Strong engineering applicability: It supports the detection of micro-curved components, porous components or defective components, and is convenient for analyzing the stress distribution state near the characteristic structure.
[0081] In summary, the SiC provided by the present inventionf / SiC ceramic matrix composite residual stress distribution map intelligent mapping method suppresses the cutting heat affected zone through low-temperature pulsed laser cutting combined with liquid nitrogen cooling, reduces the influence on the residual stress of the sampling area, and intelligently plans the test point density based on the material morphology characteristics. Combining high-precision laser projection positioning and high-precision electromechanical displacement platform linkage, the X-ray diffraction method is used to test and calculate the residual stress data of each point, and the obtained data and the convolutional neural network (CNN) algorithm are used to realize the sub-millimeter-level reconstruction of the stress field, generating a stress distribution cloud map with a resolution of 0.1 mm; this method can significantly improve the SiC f / SiC ceramic matrix composite residual stress distribution detection accuracy and efficiency, the number of test points can be reduced by 20%, applicable to stress sampling analysis of complex components such as aeroengine and nuclear reactor components, and can be used to support manufacturing process optimization and life assessment; at the same time, this method can also be extended to the residual stress detection of ceramic matrix composites such as C / SiC, providing reliable technical support for the process optimization, life assessment and failure analysis of high-value components such as aeroengine and nuclear reactor components.
[0082] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A kind of SiC f / intelligent mapping method for residual stress distribution map of SiC ceramic matrix composite, characterized in that It includes the following steps: a) Prepare specimens by using cryogenic pulsed laser cutting combined with liquid nitrogen cooling process; b) After surface treatment of the specimens, plan and locate the test points, then collect the residual stress data of each test point by using X-ray diffraction method, and finally reconstruct the stress field by combining with the convolutional neural network algorithm to generate a high-resolution stress distribution contour map.
2. The SiC according to claim 1 f / An intelligent mapping method for the residual stress distribution map of SiC / SiC ceramic matrix composites, characterized in that In step a), the cryogenic pulsed laser cutting uses a pulsed laser cutting system with a wavelength of 1000nm - 1100nm and a pulse width of 5ns - 15ns; the liquid nitrogen injection pressure of the liquid nitrogen cooling is 0.1MPa - 0.5MPa, and the flow rate is 4L / min - 6L / min; the cutting path is spiral progressive, and the thickness of the heat affected zone is less than 10μm.
3. The SiC according to claim 1 f / An intelligent mapping method for the residual stress distribution map of SiC Before the surface treatment of the specimens in step b), it also includes: Ultrasonically clean the specimens with acetone for 4min - 6min, and then cure them with low-shrinkage epoxy resin at 70℃ - 90℃ for 1h - 3h.
4. SiC according to claim 1 f / Intelligent mapping method for residual stress distribution map of SiC / SiC ceramic matrix composite, characterized in that The specific process of the surface treatment in step b) is as follows: First, use sandpapers from 500 mesh to 5000 mesh in combination with diamond grinding paste for staged water grinding and polishing. The grinding time for each stage ≤ 30 seconds, the normal load is 0.05MPa - 0.15MPa, the cooling water flow rate is 0.5L / min - 1.5L / min, and the surface temperature rise ≤ 10℃; Then, carry out pulsed electro-polishing with 4wt% - 6wt% NaOH-ethylene glycol electrolyte for 4min - 6min to make the surface roughness Ra ≤ 0.3μm; verify through a confocal laser microscope that there are no microcracks with a length > 20μm on the surface.
5. The SiC according to claim 1 f / intelligent mapping method for residual stress distribution map of SiC / SiC ceramic matrix composite, characterized in that The specific process of the test point planning in step b) is as follows: Intelligently plan the test point density based on the macroscopic morphology and structural characteristics of the material, and divide the region of interest and the conventional region according to the fiber distribution density and regional curvature on the material surface; the test point density in the region of interest is 10-100 points / cm 2 , and the test point density in the conventional region is 1-5 points / cm 2 .
6. The SiC according to claim 5 f / An intelligent mapping method for the residual stress distribution map of SiC / SiC ceramic matrix composites, characterized in that The test points in the region of interest are arranged in a hexagonal array layout with a spacing of 1mm - 3mm; the conventional region adopts a rectangular array layout with a spacing of 5mm - 10mm.
7. The SiC according to claim 1 f / intelligent mapping method for residual stress distribution map of SiC / SiC ceramic matrix composite material, characterized in that The positioning in step b) is realized by the linkage of high-precision laser projection positioning and an electromechanical displacement platform to accurately position the test points; the accuracy of the high-precision laser projection positioning is ±10μm, and the spacing error of the linkage with the electromechanical displacement platform to accurately position the test points ≤ 0.1mm.
8. The SiC according to claim 1 f / An intelligent mapping method for the residual stress distribution map of SiC / SiC ceramic matrix composites, characterized in that In step b), the X-ray diffraction method uses the (311) crystal plane of SiC material as the test object, sets the ψ angle to 0°, 25°, 35°, 45°, and the scanning step size is 0.01°.
9. The SiC according to claim 1 f An intelligent mapping method for the residual stress distribution map of SiC The convolutional neural network in step b) contains 2 - 4 convolutional layers and 1 - 3 fully connected layers, uses the mean square error as the training loss function, and outputs the stress tensors σxx, σyy, and σxy.
10. The SiC according to claim 1 f An intelligent mapping method for the residual stress distribution map of SiC The specific process of generating the high-resolution stress distribution contour map in step b) is as follows: Generate a stress distribution contour map with a resolution of 0.1mm through a three-dimensional interpolation algorithm, and automatically mark the dangerous areas with tensile stress > 250MPa and the dangerous areas with compressive stress < -250MPa.