Intelligent prediction method and system for structural ceramic bending strength
Through the combination of CT scanning and machine learning models, considering the micropore characteristics of structural ceramics, the randomness and discreteness of ceramic bending strength prediction are solved, and high-precision bending strength prediction is achieved, which improves the reliable application of materials.
Patent Information
- Application Number
- CN202411172579.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-12-31
AI Technical Summary
The prior art is difficult to accurately predict the bending strength of structural ceramics, especially due to the complex internal shape and the random distribution of inherent micropores, which limits the reliable application of materials.
The geometric information data of structural ceramics are obtained through CT scan, and the processing and optimization are carried out to build a feature space that considers the size, morphology, location, orientation and combination of micropores. The extreme gradient enhancement model and adaptive enhancement model are used to predict and build high-precision and machine learning intensity prediction model with excellent generalization.
Accurate prediction of the bending strength of structural ceramics is achieved, the accuracy and reliability of material strength prediction is improved, and the design and preparation process of materials is simplified.
Smart Images
Figure CN119227506A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to an intelligent prediction method and system for the bending strength of structural ceramics, belonging to the technical field of material mechanical property evaluation. Background Art
[0002] Structural ceramics have broad application prospects in high-tech fields such as aerospace, automobile manufacturing, and electronic devices due to their excellent mechanical properties, wear resistance, high temperature resistance, and chemical stability. However, due to their intrinsic brittleness and extremely high defect sensitivity, the strength of structural ceramics is very random and discrete, which seriously restricts their reliable application. Therefore, a deep understanding of the influence of defects on the strength of structural ceramics is crucial for optimizing the material preparation process and evaluating service performance.
[0003] In the past, research on the response relationship between defects and ceramic strength can be divided into two categories: qualitative evaluation and quantitative prediction. Qualitative evaluation has always been difficult to guide the accurate prediction of the bearing limit of specific components, while quantitative prediction is more targeted at simplified problems such as cracks and simple-shaped holes. There is still a large gap from directly guiding engineering practice, because the destruction of structural ceramics in actual applications is mainly caused by inherent micropores with various shapes, random distribution and complex combinations, and it is difficult to simplify it into a simple hole problem that can be described by theoretical models.
[0004] In recent years, machine learning (ML) has become a powerful tool for analyzing large amounts of data and predicting material properties based on the data. Algorithms such as artificial neural networks (ANN), support vector machines (SVR), extreme gradient boosting (XGBoost) and random forests (RF) have been successfully used to predict the mechanical properties of ceramics, thereby guiding material design and preparation. However, the feature space generally only considers elemental composition, additives, raw material properties and sintering process, and has not yet paid attention to the response relationship between microstructure and mechanical properties.
[0005] With the development of nondestructive testing technology, it can basically meet the defect detection of various structural materials. Bao et al. combined SR-μCT and ML technology, based on the defect geometric characteristics, and trained the model through support vector machine (SVM), and found that the geometric parameters of critical defects will significantly affect the fatigue life of Ti-6Al-4V materials. Horňas et al. used the µCT test results to construct a prediction model of defect characteristics and fatigue life of Ti-6Al-4V materials. These works provide effective ideas and feasible paths for quantitative prediction of the strength of ceramics under the action of inherent pores. Summary of the invention
[0006] The purpose of the present invention is to overcome the shortcomings of the prior art and provide an intelligent prediction method and system for the bending strength of structural ceramics. Based on statistical processing and theoretical models, a feature space considering the size, morphology, position, orientation and combination of micropores is constructed. By partitioning and considering the critical size and shape of the pores, a machine learning strength prediction model with high precision and excellent generalization is constructed and optimized, thereby guiding the strength prediction and reliable application of materials.
[0007] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In one aspect, the present invention provides an intelligent prediction method for the bending strength of structural ceramics, comprising: Obtaining geometric information data of the structural ceramics to be tested; Processing and optimizing the geometric information data of the structural ceramic to be tested to obtain the processed and optimized geometric information data of the structural ceramic to be tested; The geometric information data of the structural ceramics to be tested after processing and optimization is used as input, and the bending strength value of the structural ceramics to be tested is obtained based on the output of the pre-trained structural ceramic bending strength intelligent prediction model.
[0008] Furthermore, the geometric information data of the structural ceramic to be tested is obtained by CT scanning, and the CT scanning range is an area extending outward by 20-50% of the span with the geometric center of the structural ceramic to be tested as the center point.
[0009] Furthermore, the geometric information data of the structural ceramic to be tested includes hole size, hole shape, hole position and hole orientation of the structural ceramic to be tested.
[0010] Furthermore, the processing and optimization of the geometric information data of the structural ceramic to be tested to obtain the processed and optimized geometric information data of the structural ceramic to be tested includes: According to the geometric information data of the structural ceramic to be tested, the geometric information data of the holes whose diameter value is greater than the critical defect size of the material is screened and obtained, wherein the expression of the critical defect size of the material is as follows: ; ; in, is the critical defect size, ρ c is the critical defect sharpness of the material, is the fracture toughness, is the geometric correction factor, is the critical stress when the structural ceramic to be tested is destroyed, a is the notch depth, is the characteristic intensity, The load applied to test the three-point bending strength is is the span between the support points, is the width of the structural ceramic to be tested, is the thickness of the structural ceramic to be tested; The geometric information data of the holes with diameter values larger than the critical defect size of the material are divided into regions according to the hole positions of the holes with diameter values larger than the critical defect size of the material, so as to obtain the geometric information data of the structural ceramics to be tested after processing and optimization.
[0011] Furthermore, assuming that the height direction of the structural ceramic to be tested is x axis, the width direction of the structural ceramic to be tested is y Axis, the span direction of the structural ceramic to be tested is z The geometric information data includes the maximum radius of the hole, the average radius of the hole, the standard deviation of the hole sphericity, the hole yz The ratio of the plane projection area to the total surface projection area of the holes, the holes in y and z The average value of the projection length ratio in the direction and the hole center are y The average value and the hole position on the axis coordinate, wherein the standard deviation of the hole sphericity and the hole in y and z The average value of the directional projection length ratio is calculated from the holes with radius values greater than the critical defect sharpness; The calculation expressions of the standard deviation of the hole sphericity and the hole position are as follows: ; in, is the standard deviation of the sphericity of holes with radius values greater than the critical defect sharpness, is the number of holes with radius values greater than the critical defect sharpness, For the The sphericity of the hole with a radius greater than the critical defect sharpness, is the average sphericity of holes with radius values greater than the critical defect sharpness; Indicates the hole location, is the span of the three-point bending strength test, is the thickness of the structural ceramic to be tested, The hole center is The average value on the axis coordinates, The hole center is The mean value on the axis coordinates.
[0012] Furthermore, the geometric information data of holes with diameters greater than the critical defect size of the material are divided into regions according to the hole positions of the holes with diameters greater than the critical defect size of the material, to obtain the geometric information data of the structural ceramic to be tested after processing and optimization, including: According to the hole positions of the holes whose diameters are larger than the critical defect size of the material, the geometric information data of the holes whose diameters are larger than the critical defect size of the material within the area where the tensile stress surface of the structural ceramic to be tested accounts for 10-20% of the thickness of the structural ceramic to be tested during the bending strength test are used as the geometric information data of the structural ceramic to be tested after processing and optimization.
[0013] Furthermore, the pre-training method of the intelligent prediction model for the bending strength of structural ceramics comprises: Building extreme gradient boosting models and adaptive boosting models; Obtain a structural ceramic sample data set with known true values of bending strength, and divide it into a training set and a test set; The training set and test set data are used as inputs in turn, and the bending strength prediction values are obtained through the outputs of the extreme gradient boosting model and the adaptive boosting model respectively. In the process, the model parameters of the extreme gradient boosting model and the adaptive boosting model are tuned by network search; The predicted values of the bending strength of the extreme gradient enhancement model and the adaptive enhancement model are used to calculate the evaluation index with the corresponding true values of the bending strength, and the evaluation results of the extreme gradient enhancement model and the adaptive enhancement model are obtained; The evaluation results of the extreme gradient enhancement model and the adaptive boosting model are compared, and the model with better evaluation results is selected as the intelligent prediction model for the bending strength of structural ceramics, that is, the pre-trained intelligent prediction model for the bending strength of structural ceramics.
[0014] Furthermore, the evaluation indicators include verification score, mean square error and determination coefficient, and their calculation expressions are as follows: ; in, To verify the score, is the sample size, For the The predicted value of bending strength of samples, For the The true value of the bending strength of the samples, is the average bending strength of the samples, is the coefficient of determination, is the mean square error.
[0015] Furthermore, the training set and the test set are randomly divided by a leave-one-out cross-validation method.
[0016] On the other hand, the present invention also provides an intelligent prediction system for the bending strength of structural ceramics, comprising: Memory, for storing computer programs / instructions; A processor is used to execute the computer program / instructions to implement the steps of any of the above-mentioned methods for intelligently predicting the bending strength of structural ceramics.
[0017] Compared with the prior art, the present invention has the following beneficial effects: The present invention uses the geometric information data of the structural ceramic to be tested as input, and can directly predict the bending strength of the structural ceramic material through the intelligent prediction model of the bending strength of the structural ceramic, which is simple and fast, and the prediction accuracy is also greatly improved; Based on statistics and theoretical physics models, the present invention proposes multiple physical quantity characteristics, constructs a feature space that takes into account the pore size, morphology, position, orientation and combination method, and mathematically processes the physical quantity characteristics based on the inherent properties of structural ceramic materials. Then, a machine learning strength prediction model with high precision and excellent generalization is constructed and evaluated and optimized, which achieves convenient and fast prediction while greatly improving the accuracy of bending strength prediction of structural ceramic materials. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 A schematic diagram of the three-dimensional distribution of holes in the non-destructive testing area of the intelligent prediction method for the bending strength of structural ceramics in Example 1 of the present invention; Figure 2 It is a schematic diagram of comparison between the predicted values of bending strength of 13 Si3N4 ceramic rods output by the extreme gradient boosting model (XGBoost) and the adaptive boosting model (AdaBoost) and the corresponding true values in Example 1 of the present invention; Figure 3 Schematic diagram of relative errors between the predicted values of bending strength and the corresponding true values of the two validation sets, namely, Sample 1 and Sample 2, output by the adaptive boosting model (AdaBoost) in Example 1 of the present invention. DETAILED DESCRIPTION
[0019] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.
[0020] Example 1
[0021] The embodiment of the present invention provides a method for intelligently predicting the bending strength of structural ceramics, comprising the following steps: The structural ceramic samples used in this embodiment include 15 Si3N4 ceramic rods from the same batch and the structural ceramics to be tested.
[0022] First, the structural ceramic samples are scanned by CT to obtain geometric information data of the structural ceramic samples. The CT uses a nano-X-ray microscope (nano-CT). The CT scanning range is an area extending 20% to 50% of the span outward from the geometric center of the structural ceramic. In this embodiment, the CT scanning range is an area extending 25% of the span outward from the geometric center of the structural ceramic sample as the center point. Figure 1 As shown, the scanning detection area volume is 5×6×10mm 3 , the span is 40mm.
[0023] Geometric information data includes the size, shape, position, orientation, etc. of the structural ceramic sample, including but not limited to: the radius of the circumscribed sphere of the hole R 、Hole sphericity S w (i.e., the ratio of the surface area of a sphere with the same volume as the hole to the surface area of the hole), the hole location X (i.e. the hole center is x axis coordinate values), hole position Y (i.e. the hole center is y axis coordinate values), hole position Z (i.e. the hole center is z axis coordinate values), hole projection size L y (That is, the hole boundary size in the coordinate system is y Axis projection), hole projection size L z (That is, the hole boundary size in the coordinate system is z Axis projection), hole projection area S xy (i.e. the hole is xy Projected area of the plane), projected area of the hole S xz (i.e. the hole is xz The projected area of the plane) and the projected area of the hole S yz (i.e. the hole is yz projected area of the plane).
[0024] Above The axis is the height direction of the structural ceramic sample (i.e. the bending loading direction), The axis is the width direction of the structural ceramic sample (i.e. the front-to-back direction during bending test). The axis is the span direction of the structural ceramic specimen (i.e., the left-right direction during bending test).
[0025] The geometric information data of the structural ceramic sample is processed and optimized to obtain the geometric information data of the structural ceramic sample after processing and optimization, specifically including: Based on the geometric information data extracted by scanning, holes with diameter values greater than the critical defect size are screened out, where the calculation expression of the critical defect size is as follows: ; ; in, is the critical defect size, ρ c is the critical defect sharpness of the material, is the fracture toughness, is the geometric correction factor, is the critical stress when the structural ceramic to be tested is destroyed, a is the notch depth, is the characteristic intensity, The load applied to test the three-point bending strength is is the span between the support points, is the width of the structural ceramic to be tested, is the thickness of the structural ceramic to be tested.
[0026] The purpose of screening holes with diameters larger than the critical defect size is that the geometric information data of these holes have a greater impact on the bending strength prediction results, while the influence of other micropores can be basically ignored.
[0027] Next, the sphericity standard deviation of holes with radius values greater than the critical defect sharpness is calculated, and the calculation expression is as follows: ; in, is the standard deviation of the sphericity of holes with radius values greater than the critical defect sharpness, is the number of holes with radius values greater than the critical defect sharpness, For the The sphericity of the hole with a radius greater than the critical defect sharpness, It is the average sphericity of holes with radius values greater than the critical defect sharpness.
[0028] According to the hole positions of the holes whose diameters are larger than the critical defect size of the material, the geometric information data of the holes whose diameters are larger than the critical defect size of the material within the area where the tensile stress surface of the structural ceramic to be tested accounts for 10-20% of the thickness of the structural ceramic to be tested during the bending strength test are used as the geometric information data of the structural ceramic to be tested after processing and optimization.
[0029] In this embodiment, the geometric information data of the structural ceramic sample after processing and optimization includes 7 physical quantity features, namely, the maximum radius of the hole, the average radius of the hole, the standard deviation of the sphericity of the hole with a radius value greater than the critical defect sharpness, and the hole The ratio of the plane projection area to the total surface projection area of the holes and the holes with radius values greater than the critical defect sharpness are and The average value of the projection length ratio in the direction and the hole center are The average value and hole position on the axis coordinates, wherein the calculation expression of the hole position is as follows: ; in, Indicates the hole location, is the span of the three-point bending strength test, is the thickness of the structural ceramic to be tested, The hole center is The average value on the axis coordinates, The hole center is The mean value on the axis coordinates.
[0030] Construct the extreme gradient boosting model (XGBoost) and the adaptive boosting model (AdaBoost) and pre-train them. The pre-training methods include: Two Si3N4 ceramic rods were selected from the 15 Si3N4 ceramic rods as the validation set, and then the remaining 13 Si3N4 ceramic rods were randomly divided into a training set and a test set by the leave-one-out cross-validation method. That is, the training set included 12 Si3N4 ceramic rods, and the test set included 1 Si3N4 ceramic rod. The true values of the bending strength of the above 15 Si3N4 ceramic rods were tested by a three-point bending strength test.
[0031] Then, the geometric information data of the 13 Si3N4 ceramic rods after processing and optimization were used as input, and the bending strength prediction values of the 13 Si3N4 ceramic rods were obtained based on the output of the extreme gradient enhancement model and the adaptive boosting model. In the process, the model parameters of the extreme gradient enhancement model and the adaptive boosting model were tuned by network search. The network search was called GridSearch CV in the Scikit-learn library in python. The parameter values of the two models after tuning are shown in Table 1 and Table 2, respectively.
[0032] Table 1: Optimal values of hyperparameters for the Extreme Gradient Boosting model (XGBoost) Hyperparameters Best Value Number of ensemble weak evaluators 50 Maximum Depth 3 Sampling rate 0.6 Learning Rate 0.1 Table 2: Optimal values of hyperparameters for the Adaptive Boosting model (AdaBoost) Hyperparameters Best Value Number of ensemble weak evaluators 50 Learning Rate 0.6 Maximum Depth 2 The evaluation index calculation of the bending strength prediction values of 13 Si3N4 ceramic rods obtained by the extreme gradient boosting model (XGBoost) and the adaptive boosting model (AdaBoost) and the corresponding bending strength true values (i.e., experimental values) was performed to obtain the evaluation results of the extreme gradient boosting model and the adaptive boosting model. Specifically, the evaluation indicators include verification score, mean square error, and determination coefficient, and their calculation expressions are as follows: ; in, To verify the score, is the sample size, For the The predicted value of bending strength of samples, For the The true value of the bending strength of the samples, is the average bending strength of the samples, is the coefficient of determination, is the mean square error.
[0033] In this embodiment, the evaluation results of the extreme gradient boosting model (XGBoost) and the adaptive boosting model (AdaBoost) are shown in Table 3.
[0034] Table 3: Evaluation results of the extreme gradient boosting model (XGBoost) and the adaptive boosting model (AdaBoost) Model <![CDATA[R 2 ]]> RMSE <![CDATA[Q LOOCV ]]> XGBoost 0.92 18.99 360.82 AdaBoost 0.96 12.20 148.83 From the data in Table 3, we can see that the model accuracy and generalization of the adaptive boosting model (AdaBoost) are better. Figure 2 , Figure 2 is a line diagram of the actual value of bending strength (i.e., experimental value) and the predicted value of bending strength of the two models. Figure 2 It can be seen that the predicted value of the adaptive boosting model (AdaBoost) is closer to the true value, and there are no data points with large deviations, which proves that the adaptive boosting model (AdaBoost) is better.
[0035] The two validation set data were input into the adaptive boosting model (AdaBoost), and the outputs were 701.8MPa and 753.5MPa, respectively. The three-point bending strength test was performed on the two validation sets, and the actual values of the bending strength were 718.1MPa and 744.3MPa, respectively. Therefore, it can be calculated that the average absolute percentage error of the adaptive boosting model (AdaBoost) is less than 2%, as shown in Figure 3 As shown, it is proved that the model can accurately predict the flexural strength of structural ceramics.
[0036] Therefore, the adaptive boosting model (AdaBoost) is selected as the intelligent prediction model for the bending strength of structural ceramics, that is, the pre-trained intelligent prediction model for the bending strength of structural ceramics.
[0037] Finally, the geometric information data of the structural ceramics to be tested after processing and optimization is used as input, and the bending strength value of the structural ceramic sample is obtained based on the output of the pre-trained structural ceramic bending strength intelligent prediction model.
[0038] It should be noted that the intelligent prediction model for the bending strength of structural ceramics selected after evaluation can be used to predict the bending strength of any structural ceramic to be tested without the need for re-evaluation.
[0039] Embodiment 2: This embodiment provides an intelligent prediction system for the bending strength of structural ceramics, comprising: Memory, for storing computer programs / instructions; A processor is used to execute the computer program / instructions to implement the steps of the intelligent prediction method for the bending strength of structural ceramics described in Example 1.
[0040] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. An intelligent prediction method for the bending strength of structural ceramics, characterized in that: include: Obtaining geometric information data of the structural ceramics to be tested; Processing and optimizing the geometric information data of the structural ceramic to be tested to obtain the processed and optimized geometric information data of the structural ceramic to be tested; The geometric information data of the structural ceramics to be tested after processing and optimization is used as input, and the bending strength value of the structural ceramics to be tested is obtained based on the output of the pre-trained structural ceramic bending strength intelligent prediction model.
2. The intelligent prediction method for bending strength of structural ceramics according to claim 1 is characterized in that: The geometric information data of the structural ceramic to be tested is obtained by CT scanning, and the CT scanning range is an area extending outward by 20-50% of the span from the geometric center of the structural ceramic to be tested as the center point.
3. The intelligent prediction method for bending strength of structural ceramics according to claim 1, characterized in that: The geometric information data of the structure ceramic to be tested includes the hole size, hole shape, hole position and hole orientation of the structure ceramic to be tested.
4. The intelligent prediction method for bending strength of structural ceramics according to claim 1, characterized in that: The method of processing and optimizing the geometric information data of the structural ceramic to be tested to obtain the processed and optimized geometric information data of the structural ceramic to be tested includes: According to the geometric information data of the structural ceramic to be tested, the geometric information data of the holes whose diameter value is greater than the critical defect size of the material is screened and obtained, wherein the expression of the critical defect size of the material is as follows: ； ； in, is the critical defect size, ρ c is the critical defect sharpness of the material, is the fracture toughness, is the geometric correction factor, is the critical stress when the structural ceramic to be tested is destroyed, a is the notch depth, is the characteristic intensity, The load applied to test the three-point bending strength is is the span between the support points, is the width of the structural ceramic to be tested, is the thickness of the structural ceramic to be tested; The geometric information data of the holes with diameter values larger than the critical defect size of the material are divided into regions according to the hole positions of the holes with diameter values larger than the critical defect size of the material, so as to obtain the geometric information data of the structural ceramics to be tested after processing and optimization.
5. The intelligent prediction method for bending strength of structural ceramics according to claim 4, characterized in that: Assume that the height direction of the structural ceramic to be tested is x axis, the width direction of the structural ceramic to be tested is y Axis, the span direction of the structural ceramic to be tested is z The geometric information data includes the maximum radius of the hole, the average radius of the hole, the standard deviation of the hole sphericity, the hole yz The ratio of the plane projection area to the total surface projection area of the holes, the holes in y and z The average value of the projection length ratio in the direction and the hole center are y The average value and the hole position on the axis coordinate, wherein the standard deviation of the hole sphericity and the hole in y and z The average value of the directional projection length ratio is calculated from the holes with radius values greater than the critical defect sharpness; The calculation expressions of the standard deviation of the hole sphericity and the hole position are as follows: ; in, is the standard deviation of sphericity of holes with radius values greater than the critical defect sharpness, is the number of holes with radius values greater than the critical defect sharpness. For the The sphericity of the hole with a radius greater than the critical defect sharpness, is the average sphericity of holes with radius values greater than the critical defect sharpness; Indicates the hole location, is the span of the three-point bending strength test, is the thickness of the structural ceramic to be tested, The hole center is The average value on the axis coordinates, The hole center is The mean value on the axis coordinates.
6. The intelligent prediction method for bending strength of structural ceramics according to claim 4, characterized in that: The method of dividing the geometric information data of holes with diameters larger than the critical defect size of the material into regions according to the hole positions of the holes with diameters larger than the critical defect size of the material, and obtaining the geometric information data of the structural ceramic to be tested after processing and optimization, includes: According to the hole positions of the holes whose diameters are larger than the critical defect size of the material, the geometric information data of the holes whose diameters are larger than the critical defect size of the material within the area where the tensile stress surface of the structural ceramic to be tested accounts for 10-20% of the thickness of the structural ceramic to be tested during the bending strength test are used as the geometric information data of the structural ceramic to be tested after processing and optimization.
7. The intelligent prediction method for bending strength of structural ceramics according to claim 1, characterized in that: The pre-training method of the intelligent prediction model for the bending strength of structural ceramics comprises: Building extreme gradient boosting models and adaptive boosting models; Obtain a structural ceramic sample data set with known true values of bending strength, and divide it into a training set and a test set; The training set and test set data are used as inputs in turn, and the bending strength prediction values are obtained through the outputs of the extreme gradient boosting model and the adaptive boosting model respectively. In the process, the model parameters of the extreme gradient boosting model and the adaptive boosting model are tuned by network search; The predicted values of the bending strength of the extreme gradient enhancement model and the adaptive enhancement model are used to calculate the evaluation index with the corresponding true values of the bending strength, and the evaluation results of the extreme gradient enhancement model and the adaptive enhancement model are obtained; The evaluation results of the extreme gradient enhancement model and the adaptive boosting model are compared, and the model with better evaluation results is selected as the intelligent prediction model for the bending strength of structural ceramics, that is, the pre-trained intelligent prediction model for the bending strength of structural ceramics.
8. The intelligent prediction method for bending strength of structural ceramics according to claim 7, characterized in that: The evaluation indicators include verification score, mean square error and determination coefficient, and their calculation expressions are as follows: ; in, To verify the score, is the sample size, For the The predicted value of bending strength of samples, For the The true value of the bending strength of the samples, is the average bending strength of the samples, is the coefficient of determination, is the mean square error.
9. The intelligent prediction method for bending strength of structural ceramics according to claim 1, characterized in that: The training set and the test set are randomly divided by a leave-one-out cross-validation method.
10. An intelligent prediction system for bending strength of structural ceramics, characterized in that: include: Memory, for storing computer programs / instructions; A processor is used to execute the computer program / instructions to implement the steps of the intelligent prediction method for the bending strength of structural ceramics according to any one of claims 1 to 9.