C / SiC composite material design method, system and medium

Abstract

The embodiment of the invention provides a C / SiC composite material design method and system and a medium. The method comprises the steps of obtaining literature data, establishing an original data set based on the literature data, preprocessing the original data set, and constructing a sample data set of machine learning; screening the data features based on a screening rule to generate optimized data features; establishing a prediction model; establishing a plurality of parameter combinations based on parameters required by the C / SiC composite material, and analyzing linear ablation performance information of the composite material; performing parameter combination iterative optimization based on a Bayesian optimization algorithm to obtain an optimal parameter combination matched with the linear ablation performance information, and generating composite material processing technology and component conditions; by establishing a machine learning model, the ablation resistance of the C / SiC composite materials with different parameters can be predicted, and the design of the high-ablation-resistance C / SiC composite material can be accelerated and aided.

Description

Technical Field

[0001] This application relates to the field of ceramic matrix composite technology, and more specifically, to a design method, system, and medium for C / SiC composite materials. Background Technology

[0002] C / SiC composite material is a ceramic-based ultra-high temperature thermal structural material with excellent properties. Its outer silicon carbide matrix can form a dense silicon dioxide layer under high-temperature oxidation conditions, exhibiting excellent high-temperature resistance and oxidation resistance; while the inner carbon fiber endows the material with excellent mechanical properties. It has wide applications in aerospace engines, hypersonic vehicle thermal protection systems, high-temperature gas filtration devices, and high-performance optical devices, significantly reducing weight and improving the high-temperature performance of products.

[0003] The ablation properties of C / SiC composites prepared using different weaving structures, preparation processes, and modification measures vary considerably. Therefore, to further improve the performance of C / SiC composites, researchers have combined different preparation processes with the addition of different matrix modifying phases to achieve rapid densification of C / SiC composites while simultaneously enhancing their overall performance, thus promoting the multifunctional applications of C / SiC composites.

[0004] C / SiC composite materials have high preparation and testing costs, while traditional material design methods typically rely on trial and error experiments, usually requiring numerous tests to find the optimal material combination. However, this approach is costly, time-consuming, and involves the uncertainty of trial and error.

[0005] Machine learning aims to extract key features affecting material properties from massive amounts of data. A machine learning model is built based on limited data, experimental parameters are set based on the feedback from the machine learning, recommended candidates are synthesized and tested, new data is added to the training dataset, and the model is iteratively improved, thereby efficiently and accurately designing new materials.

[0006] In conclusion, further research is needed on machine learning methods to improve the ablation resistance of C / SiC composite materials. Summary of the Invention

[0007] The purpose of this application is to provide a design method, system and medium for C / SiC composite materials. By establishing a machine learning model, the ablation resistance of C / SiC composite materials with different parameters can be predicted, which can accelerate the design of highly ablation-resistant C / SiC composite materials.

[0008] This application also provides a C / SiC composite material design method, including: Acquire literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; Extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; A prediction model is established by training the optimized features using multiple machine learning algorithms. Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on the required parameters for C / SiC composite materials, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite materials; The parameter combination is iteratively optimized based on the Bayesian optimization algorithm to obtain the optimal parameter combination that matches the line ablation performance information, and to generate the composite material processing technology and composition conditions.

[0009] Optionally, in the C / SiC composite material design method described in the embodiments of this application, literature data is obtained, an original dataset is established based on the literature data, the original dataset is preprocessed to obtain preprocessed data, and a sample dataset for machine learning is constructed based on the preprocessed data, specifically including: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

[0010] Optionally, in the C / SiC composite material design method described in this application embodiment, data features are extracted from the sample dataset, and the data features are filtered based on screening rules to obtain redundant features. These redundant features are then removed to generate optimized data features, specifically including: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph. Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

[0011] Optionally, in the C / SiC composite material design method described in the embodiments of this application, a prediction model is established by training the optimized features based on multiple machine learning algorithms, specifically including: The optimized data features are randomly divided into training and test sets according to a set ratio. Multiple machine learning models are obtained by iteratively training the model based on the training set. The prediction error of the machine learning model is analyzed based on the test set, and the performance information of the machine learning model is evaluated based on the prediction error. Machine learning models whose performance information is greater than the set performance information are selected as prediction models based on performance information.

[0012] Optionally, in the C / SiC composite material design method described in this application embodiment, the required parameters for the C / SiC composite material are obtained, several parameter combinations are established based on the required parameters for the C / SiC composite material, and the several parameter combinations are input into the prediction model to analyze the linear ablation performance information of the composite material, specifically including: Obtain relevant parameters of C / SiC composite materials, including the braiding structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters of C / SiC composite materials; Based on the relevant parameters, an initial parameter combination is constructed. For the missing feature parameters in the initial parameter combination, a model is built to fill in the missing values ​​based on the sample dataset to obtain the full feature parameter combination. By inputting the full set of feature parameters into the prediction model, the predicted results of the line ablation performance corresponding to the full set of feature parameters are obtained.

[0013] Optionally, in the C / SiC composite material design method described in the embodiments of this application, parameter combination iterative optimization is performed based on the Bayesian optimization algorithm to obtain the optimal parameter combination matching the line ablation performance information, and to generate the composite material processing technology and composition conditions, specifically including: Within a preset parameter value range, multiple sets of full characteristic parameter combinations of C / SiC composite materials are randomly generated; Each set of full feature parameters is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. The Bayesian optimization algorithm is used to iteratively optimize the combination of all feature parameters. When the iteration meets the convergence condition, the optimal parameter combination is output. The processing technology and component conditions of C / SiC composite materials are converted according to the optimal parameter combination.

[0014] Secondly, embodiments of this application provide a C / SiC composite material design system, which includes a memory and a processor. The memory includes a program for a C / SiC composite material design method, and when the program for the C / SiC composite material design method is executed by the processor, it performs the following steps: Acquire literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; Extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; A prediction model is established by training the optimized features using multiple machine learning algorithms. Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on the required parameters for C / SiC composite materials, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite materials; The parameter combination is iteratively optimized based on the Bayesian optimization algorithm to obtain the optimal parameter combination that matches the line ablation performance information, and to generate the composite material processing technology and composition conditions.

[0015] Optionally, in the C / SiC composite material design system described in this application embodiment, literature data is acquired, an original dataset is established based on the literature data, the original dataset is preprocessed to obtain preprocessed data, and a sample dataset for machine learning is constructed based on the preprocessed data, specifically including: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

[0016] Optionally, in the C / SiC composite material design system described in this application embodiment, data features of the sample dataset are extracted, and the data features are filtered based on screening rules to obtain redundant features. These redundant features are then removed to generate optimized data features, specifically including: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph. Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

[0017] Thirdly, embodiments of this application also provide a computer-readable storage medium, which includes a C / SiC composite material design method program. When the C / SiC composite material design method program is executed by a processor, it implements the steps of the C / SiC composite material design method as described in any of the preceding claims.

[0018] As can be seen from the above, the C / SiC composite material design method, system, and medium provided in this application involve acquiring literature data, establishing an original dataset based on the literature data, preprocessing the original dataset to obtain preprocessed data, constructing a machine learning sample dataset based on the preprocessed data, extracting data features from the sample dataset, filtering the data features based on screening rules to obtain redundant features, removing redundant features to generate optimized data features, training the optimized features based on multiple machine learning algorithms to establish a prediction model, acquiring the required parameters for C / SiC composite materials, establishing several parameter combinations based on the required parameters for C / SiC composite materials, inputting several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite material, iteratively optimizing the parameter combinations based on Bayesian optimization algorithm to obtain the optimal parameter combination matching the linear ablation performance information, and generating the composite material processing technology and composition conditions. By establishing a machine learning model, the ablation resistance performance of C / SiC composite materials with different parameters can be predicted, which can accelerate the auxiliary design of highly ablation-resistant C / SiC composite materials. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating the C / SiC composite material design method provided in this application embodiment; Figure 2 The density and box plots of the line ablation rate values ​​were plotted based on the original dataset for the design method of C / SiC composite materials provided in the embodiments of this application. Figure 3 Thermodynamic plot of the Pearson correlation coefficient between any two features in the sample dataset of the C / SiC composite material design method provided in the embodiments of this application; Figure 4 The data results predicted by the machine learning model that has better predictive performance for the C / SiC composite material design method provided in the embodiments of this application; Figure 5 The order of feature importance determined by the machine learning model constructed using the XGBoost algorithm for the C / SiC composite material design method provided in this application embodiment. Detailed Implementation

[0021] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0022] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0023] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a C / SiC composite material design method according to some embodiments of this application. This C / SiC composite material design method is used in a terminal device and includes the following steps: S101, Obtain literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; S102, extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; S103, based on multiple machine learning algorithms, trains the optimized features to establish a prediction model; S104, obtain the required parameters of C / SiC composite material, establish several parameter combinations based on the required parameters of C / SiC composite material, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of composite material; S105 uses a Bayesian optimization algorithm to iteratively optimize parameter combinations, obtaining the optimal parameter combination that matches the line ablation performance information, and generating the composite material processing technology and composition conditions.

[0024] It should be noted that by establishing a machine learning prediction model, the influence of the weaving structure, process parameters, modified components, and ablation conditions on the ablation resistance of C / SiC composite materials can be effectively analyzed, and an intuitive ranking of importance can be provided. By inputting specified parameters, the ablation resistance of the corresponding material can be predicted efficiently, which can serve as a reference and guide for the development of C / SiC composite materials with high ablation resistance. Iterative optimization of experimental parameter combinations based on sample datasets yields optimal design parameters, which are then used to better design experiments. Subsequent experimental data can be used to supplement and expand the sample dataset. This iterative method not only reduces the cost and time of trial-and-error experiments but also enables more accurate prediction of material properties and behavior, providing valuable reference and guidance for material design.

[0025] According to an embodiment of the present invention, literature data is acquired, an original dataset is established based on the literature data, the original dataset is preprocessed to obtain preprocessed data, and a sample dataset for machine learning is constructed based on the preprocessed data, specifically including: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

[0026] It should be noted that the normalization formula is as follows: in The normalized value obtained after MinMaxScaler (minimum-maximum normalization); It is the raw data from the dataset; and These are the maximum and minimum values ​​of the dataset, respectively. The sample dataset includes data on the braided structure of the C / SiC composite material, preparation process parameters, composite material performance parameters, physicochemical properties of the modified phase, and ablation conditions.

[0027] According to an embodiment of the present invention, data features are extracted from a sample dataset, and the data features are filtered based on filtering rules to obtain redundant features. These redundant features are then removed to generate optimized data features. Specifically, this includes: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph, such as... Figure 3 As shown; Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

[0028] The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

[0029] It should be noted that the formula for calculating linear correlation is as follows:

[0030] in: It is the Pearson correlation coefficient; numerator: and The covariance, i.e. ; Denominator: Standard deviation and Standard deviation The product of. , : No. Observations of each sample point; , Sample mean, i.e. , ; Sample size.

[0031] Pearson correlation coefficient The value range is -1 to 1; negative values ​​indicate negative correlation, positive values ​​indicate positive correlation, and an absolute value close to 1 indicates a significant linear correlation. For highly correlated features... Generally, only one of them is kept.

[0032] According to embodiments of the present invention, a prediction model is established by training optimized features based on multiple machine learning algorithms, specifically including: The optimized data features are randomly divided into training and test sets according to a set ratio. Multiple machine learning models are obtained by iteratively training the model based on the training set. The prediction error of the machine learning model is analyzed based on the test set, and the performance information of the machine learning model is evaluated based on the prediction error. Machine learning models whose performance information is greater than the set performance information are selected as prediction models based on performance information.

[0033] It should be noted that the dataset after feature filtering is divided into training and test sets. The algorithm randomly divides the original dataset into training and test datasets according to an 80 / 20 ratio. Hyperparameter search was performed on each algorithm using a Bayesian optimization algorithm. The hyperparameter space was searched using Bayesian optimization based on the hyper_opt algorithm, and the prediction error of each model on the test set was calculated to evaluate the model's performance. The study selects the best-performing model from a variety of established machine learning models. Based on the machine learning models trained on the training set, the best-performing model is selected by comparing their errors on the test set. This study uses mean absolute error (MAE), mean squared error (MSE), and coefficient of determination (CPE). The performance of the model is evaluated using three indicators, and the formula is as follows: in For the sample size, For predicted values, For accurate values, The mean of the accurate values. MAE measures the average absolute difference between the predicted and actual values, while MSE measures the mean squared difference between the predicted and actual values. The smaller these two parameters are, the better the model performance. The value ranges from 0 to 1, and a higher value (closer to 1) indicates a better fit between the model and the input data; Figure 4 The results of several superior machine learning prediction models are shown. The closer the scatter plot distribution is to the 45° diagonal, the better the model's fit. In this example, the XGBoost regression model has the best prediction performance. Feature importance analysis was performed based on the optimal machine learning model, and the contribution of different features to improving the ablation resistance of C / SiC composite materials was obtained based on the analysis results. SHAP analysis was performed based on the optimal machine learning model. A positive SHAP value indicates that the feature enhances the predicted value, exhibiting a positive effect; conversely, a negative SHAP value indicates that the feature reduces the predicted value, exhibiting a negative effect. Based on the plotted SHAP graph, the influence of different features on the ablation resistance of C / SiC composite materials can be intuitively determined.

[0034] According to an embodiment of the present invention, the required parameters for C / SiC composite materials are obtained, several parameter combinations are established based on the required parameters for C / SiC composite materials, and the several parameter combinations are input into a prediction model to analyze the linear ablation performance information of the composite material, specifically including: Obtain relevant parameters of C / SiC composite materials, including the braiding structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters of C / SiC composite materials; Based on the relevant parameters, an initial parameter combination is constructed. For the missing feature parameters in the initial parameter combination, a model is built to fill in the missing values ​​based on the sample dataset to obtain the full feature parameter combination. By inputting the full set of feature parameters into the prediction model, the predicted results of the line ablation performance corresponding to the full set of feature parameters are obtained.

[0035] It should be noted that the relevant parameters of C / SiC composite materials include the weaving structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters. Based on these parameters, an initial parameter combination is constructed. Then, for the missing feature parameters in the initial combination (such as the density and porosity of the composite material), a gap-filling model is established based on the sample dataset to complete the data, forming a full feature parameter combination. Finally, the full feature parameter combination is input into the trained prediction model, and the model outputs the predicted results of the linear ablation performance of the composite material under the corresponding full feature parameter combination. This result directly reflects the material's ablation resistance.

[0036] According to an embodiment of the present invention, parameter combination iterative optimization is performed based on a Bayesian optimization algorithm to obtain the optimal parameter combination matching the line ablation performance information, thereby generating the composite material processing technology and composition conditions, specifically including: Within a preset parameter value range, multiple sets of full characteristic parameter combinations of C / SiC composite materials are randomly generated; Each set of full feature parameters is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. The Bayesian optimization algorithm is used to iteratively optimize the combination of all feature parameters. When the iteration meets the convergence condition, the optimal parameter combination is output. The processing technology and component conditions of C / SiC composite materials are converted according to the optimal parameter combination.

[0037] It should be noted that, within the preset parameter value range, multiple sets of full feature parameter combinations of C / SiC composite materials are randomly generated. Each set of full feature parameter combinations is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. Based on the Bayesian optimization of the hyper_opt algorithm, the full feature parameter combinations are iteratively optimized. With the goal of "optimal line ablation performance (lowest line ablation rate)", the parameter values ​​are continuously adjusted and the prediction results are verified. When the iteration meets the convergence condition, the optimal parameter combination is output. This optimal parameter combination can be directly converted into the processing technology (such as preparation process and process parameters) and composition conditions (such as modified phase content and fiber-matrix ratio) of C / SiC composite materials for experimental verification.

[0038] SHAP value analysis was performed based on the optimal machine learning model. A positive SHAP value indicates that the feature enhances the predicted value, exhibiting a positive effect; conversely, a negative SHAP value indicates that the feature reduces the predicted value, exhibiting a negative effect. Based on the plotted SHAP graph, the influence of different features on the ablation resistance of C / SiC composite materials can be visually assessed. Figure 5 As shown, the feature importance ranking given by the XGBoost regression model and the SHAP analysis plot reveal that the melting point and enthalpy change of the modified phase, as well as the density and porosity of the composite material, have a significant impact on the ablation resistance of the composite material. Furthermore, based on the SHAP plot, it can be concluded that the higher the melting point of the modified phase, the lower the linear ablation rate of the composite material, i.e., the better the ablation resistance. In terms of process conditions, the PIP process is more effective than other processes in improving the ablation resistance of the composite material.

[0039] Secondly, embodiments of this application provide a C / SiC composite material design system, which includes: a memory and a processor. The memory includes a program for a C / SiC composite material design method. When the program for the C / SiC composite material design method is executed by the processor, it performs the following steps: Acquire literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; Extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; A prediction model is established by training the optimized features using multiple machine learning algorithms. Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on the required parameters for C / SiC composite materials, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite materials; The parameter combination is iteratively optimized based on the Bayesian optimization algorithm to obtain the optimal parameter combination that matches the line ablation performance information, and to generate the composite material processing technology and composition conditions.

[0040] It should be noted that by establishing a machine learning prediction model, the influence of the weaving structure, process parameters, modified components, and ablation conditions on the ablation resistance of C / SiC composite materials can be effectively analyzed, and an intuitive ranking of importance can be provided. By inputting specified parameters, the ablation resistance of the corresponding material can be predicted efficiently, which can serve as a reference and guide for the development of C / SiC composite materials with high ablation resistance. Iterative optimization of experimental parameter combinations based on sample datasets yields optimal design parameters, which are then used to better design experiments. Subsequent experimental data can be used to supplement and expand the sample dataset. This iterative method not only reduces the cost and time of trial-and-error experiments but also enables more accurate prediction of material properties and behavior, providing valuable reference and guidance for material design.

[0041] According to an embodiment of the present invention, literature data is acquired, an original dataset is established based on the literature data, the original dataset is preprocessed to obtain preprocessed data, and a sample dataset for machine learning is constructed based on the preprocessed data, specifically including: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

[0042] It should be noted that the normalization formula is as follows: in The normalized value obtained after MinMaxScaler (minimum-maximum normalization); It is the raw data from the dataset; and These are the maximum and minimum values ​​of the dataset, respectively. The sample dataset includes data on the braided structure of the C / SiC composite material, preparation process parameters, composite material performance parameters, physicochemical properties of the modified phase, and ablation conditions.

[0043] According to an embodiment of the present invention, data features are extracted from a sample dataset, and the data features are filtered based on filtering rules to obtain redundant features. These redundant features are then removed to generate optimized data features. Specifically, this includes: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph, such as... Figure 3 As shown; Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

[0044] It should be noted that the formula for calculating linear correlation is as follows:

[0045] in: It is the Pearson correlation coefficient; numerator: and The covariance, i.e. ; Denominator: Standard deviation and Standard deviation The product of. , : No. Observations of each sample point; , Sample mean, i.e. , ; Sample size.

[0046] Pearson correlation coefficient The value range is -1 to 1; negative values ​​indicate negative correlation, positive values ​​indicate positive correlation, and an absolute value close to 1 indicates a significant linear correlation. For highly correlated features... Generally, only one of them is kept.

[0047] According to embodiments of the present invention, a prediction model is established by training optimized features based on multiple machine learning algorithms, specifically including: The optimized data features are randomly divided into training and test sets according to a set ratio. Multiple machine learning models are obtained by iteratively training the model based on the training set. The prediction error of the machine learning model is analyzed based on the test set, and the performance information of the machine learning model is evaluated based on the prediction error. Machine learning models whose performance information is greater than the set performance information are selected as prediction models based on performance information.

[0048] It should be noted that the dataset after feature filtering is divided into training and test sets. The algorithm randomly divides the original dataset into training and test datasets according to an 80 / 20 ratio. Hyperparameter search was performed on each algorithm using a Bayesian optimization algorithm. The hyperparameter space was searched using Bayesian optimization based on the hyper_opt algorithm, and the prediction error of each model on the test set was calculated to evaluate the model's performance. The study selects the best-performing model from a variety of established machine learning models. Based on the machine learning models trained on the training set, the best-performing model is selected by comparing their errors on the test set. This study uses mean absolute error (MAE), mean squared error (MSE), and coefficient of determination (CPE). The performance of the model is evaluated using three indicators, and the formula is as follows: in For the sample size, For predicted values, For accurate values, The mean of the accurate values. MAE measures the average absolute difference between the predicted and actual values, while MSE measures the mean squared difference between the predicted and actual values. The smaller these two parameters are, the better the model performance. The value ranges from 0 to 1, and a higher value (closer to 1) indicates a better fit between the model and the input data; Figure 4The results of several superior machine learning prediction models are shown. The closer the scatter plot distribution is to the 45° diagonal, the better the model's fit. In this example, the XGBoost regression model has the best prediction performance. Feature importance analysis was performed based on the optimal machine learning model, and the contribution of different features to improving the ablation resistance of C / SiC composite materials was obtained based on the analysis results. SHAP analysis was performed based on the optimal machine learning model. A positive SHAP value indicates that the feature enhances the predicted value, exhibiting a positive effect; conversely, a negative SHAP value indicates that the feature reduces the predicted value, exhibiting a negative effect. Based on the plotted SHAP graph, the influence of different features on the ablation resistance of C / SiC composite materials can be intuitively determined.

[0049] According to an embodiment of the present invention, the required parameters for C / SiC composite materials are obtained, several parameter combinations are established based on the required parameters for C / SiC composite materials, and the several parameter combinations are input into a prediction model to analyze the linear ablation performance information of the composite material, specifically including: Obtain relevant parameters of C / SiC composite materials, including the braiding structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters of C / SiC composite materials; Based on the relevant parameters, an initial parameter combination is constructed. For the missing feature parameters in the initial parameter combination, a model is built to fill in the missing values ​​based on the sample dataset to obtain the full feature parameter combination. By inputting the full set of feature parameters into the prediction model, the predicted results of the line ablation performance corresponding to the full set of feature parameters are obtained.

[0050] It should be noted that the relevant parameters of C / SiC composite materials include the weaving structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters. Based on these parameters, an initial parameter combination is constructed. Then, for the missing feature parameters in the initial combination (such as the density and porosity of the composite material), a gap-filling model is established based on the sample dataset to complete the data, forming a full feature parameter combination. Finally, the full feature parameter combination is input into the trained prediction model, and the model outputs the predicted results of the linear ablation performance of the composite material under the corresponding full feature parameter combination. This result directly reflects the material's ablation resistance.

[0051] According to an embodiment of the present invention, parameter combination iterative optimization is performed based on a Bayesian optimization algorithm to obtain the optimal parameter combination matching the line ablation performance information, thereby generating the composite material processing technology and composition conditions, specifically including: Within a preset parameter value range, multiple sets of full characteristic parameter combinations of C / SiC composite materials are randomly generated; Each set of full feature parameters is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. The Bayesian optimization algorithm is used to iteratively optimize the combination of all feature parameters. When the iteration meets the convergence condition, the optimal parameter combination is output. The processing technology and component conditions of C / SiC composite materials are converted according to the optimal parameter combination.

[0052] It should be noted that, within the preset parameter value range, multiple sets of full feature parameter combinations of C / SiC composite materials are randomly generated. Each set of full feature parameter combinations is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. Based on the Bayesian optimization of the hyper_opt algorithm, the full feature parameter combinations are iteratively optimized. With the goal of "optimal line ablation performance (lowest line ablation rate)", the parameter values ​​are continuously adjusted and the prediction results are verified. When the iteration meets the convergence condition, the optimal parameter combination is output. This optimal parameter combination can be directly converted into the processing technology (such as preparation process and process parameters) and composition conditions (such as modified phase content and fiber-matrix ratio) of C / SiC composite materials for experimental verification.

[0053] SHAP value analysis was performed based on the optimal machine learning model. A positive SHAP value indicates that the feature enhances the predicted value, exhibiting a positive effect; conversely, a negative SHAP value indicates that the feature reduces the predicted value, exhibiting a negative effect. Based on the plotted SHAP graph, the influence of different features on the ablation resistance of C / SiC composite materials can be visually assessed. Figure 5 As shown, the feature importance ranking given by the XGBoost regression model and the SHAP analysis plot reveal that the melting point and enthalpy change of the modified phase, as well as the density and porosity of the composite material, have a significant impact on the ablation resistance of the composite material. Furthermore, based on the SHAP plot, it can be concluded that the higher the melting point of the modified phase, the lower the linear ablation rate of the composite material, i.e., the better the ablation resistance. In terms of process conditions, the PIP process is more effective than other processes in improving the ablation resistance of the composite material.

[0054] A third aspect of the present invention provides a computer-readable storage medium including a C / SiC composite material design method program, which, when executed by a processor, implements the steps of the C / SiC composite material design method as described in any of the above claims.

[0055] This invention discloses a design method, system, and medium for C / SiC composite materials. The method involves acquiring literature data, establishing an original dataset based on this data, preprocessing the original dataset to obtain preprocessed data, and constructing a sample dataset for machine learning based on the preprocessed data. Data features are extracted from the sample dataset, and redundant features are filtered according to selection rules. These redundant features are then removed to generate optimized data features. The optimized features are trained using multiple machine learning algorithms to establish a prediction model. The required parameters for C / SiC composite materials are obtained, and several parameter combinations are established based on these parameters. These parameter combinations are input into the prediction model to analyze the linear ablation performance information of the composite material. The parameter combinations are iteratively optimized using a Bayesian optimization algorithm to obtain the optimal parameter combination matching the linear ablation performance information, generating the composite material processing technology and composition conditions. By establishing a machine learning model, the ablation resistance performance of C / SiC composite materials with different parameters can be predicted, accelerating the auxiliary design of highly ablation-resistant C / SiC composite materials.

[0056] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0057] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0058] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0059] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] Alternatively, if the integrated units of the present invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A design method for C / SiC composite materials, characterized in that, include: Acquire literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; Extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; A prediction model is established by training the optimized features using multiple machine learning algorithms. Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on the required parameters for C / SiC composite materials, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite materials; The parameter combination is iteratively optimized based on the Bayesian optimization algorithm to obtain the optimal parameter combination that matches the line ablation performance information, and to generate the composite material processing technology and composition conditions.

2. The C / SiC composite material design method according to claim 1, characterized in that, The process involves acquiring literature data, building a raw dataset based on this data, preprocessing the raw dataset to obtain preprocessed data, and then constructing a sample dataset for machine learning based on the preprocessed data. Specifically, this includes: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

3. The C / SiC composite material design method according to claim 1, characterized in that, Extract data features from the sample dataset, filter these features based on selection rules to identify redundant features, remove these redundant features, and generate optimized data features. Specifically, this includes: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph. Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

4. The C / SiC composite material design method according to claim 3, characterized in that, A prediction model is built by training the optimized features using various machine learning algorithms, specifically including: The optimized data features are randomly divided into training and test sets according to a set ratio. Multiple machine learning models are obtained by iteratively training the model based on the training set. The prediction error of the machine learning model is analyzed based on the test set, and the performance information of the machine learning model is evaluated based on the prediction error. Machine learning models whose performance information is greater than the set performance information are selected as prediction models based on performance information.

5. The C / SiC composite material design method according to claim 1, characterized in that, Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on these parameters, and input these parameter combinations into a prediction model to analyze the linear ablation performance information of the composite materials. Specifically, this includes: Obtain relevant parameters of C / SiC composite materials, including the braiding structure, preparation process, carbon fiber and matrix content, type and content of modified phases, and ablation parameters of C / SiC composite materials; Based on the relevant parameters, an initial parameter combination is constructed. For the missing feature parameters in the initial parameter combination, a model is built to fill in the missing values ​​based on the sample dataset to obtain the full feature parameter combination. By inputting the full set of feature parameters into the prediction model, the predicted results of the line ablation performance corresponding to the full set of feature parameters are obtained.

6. The C / SiC composite material design method according to claim 5, characterized in that, Based on the Bayesian optimization algorithm, iterative optimization of parameter combinations is performed to obtain the optimal parameter combination that matches the line ablation performance information, thereby generating the composite material processing technology and composition conditions, specifically including: Within a preset parameter value range, multiple sets of full characteristic parameter combinations of C / SiC composite materials are randomly generated; Each set of full feature parameters is input into the established prediction model to obtain the corresponding predicted value of line ablation performance. The Bayesian optimization algorithm is used to iteratively optimize the combination of all feature parameters. When the iteration meets the convergence condition, the optimal parameter combination is output. The processing technology and component conditions of C / SiC composite materials are converted according to the optimal parameter combination.

7. A C / SiC composite material design system, characterized in that, The system includes a memory and a processor. The memory contains a program for a C / SiC composite material design method. When the program for the C / SiC composite material design method is executed by the processor, it performs the following steps: Acquire literature data, establish a raw dataset based on the literature data, preprocess the raw dataset to obtain preprocessed data, and construct a sample dataset for machine learning based on the preprocessed data; Extract data features from the sample dataset, filter the data features based on the filtering rules to obtain redundant features, remove the redundant features, and generate optimized data features; A prediction model is established by training the optimized features using multiple machine learning algorithms. Obtain the required parameters for C / SiC composite materials, establish several parameter combinations based on the required parameters for C / SiC composite materials, and input the several parameter combinations into the prediction model to analyze the linear ablation performance information of the composite materials; The parameter combination is iteratively optimized based on the Bayesian optimization algorithm to obtain the optimal parameter combination that matches the line ablation performance information, and to generate the composite material processing technology and composition conditions.

8. The C / SiC composite material design system according to claim 7, characterized in that, The process involves acquiring literature data, building a raw dataset based on this data, preprocessing the raw dataset to obtain preprocessed data, and then constructing a sample dataset for machine learning based on the preprocessed data. Specifically, this includes: Obtain literature data, and collect ablation performance data of modified C / SiC composites with different weaving structures, processes and components based on the literature data to establish an original dataset; Density and box plots of line ablation rate values ​​were plotted based on the original dataset. Outlier test analysis was then performed based on the plotted density and box plots of line ablation rate values ​​to obtain the test results. Based on the test results, analyze the sample difference values ​​of the original data, and analyze whether the sample difference values ​​are greater than or equal to the set difference threshold. If the difference is greater than or equal to the set difference threshold, the outliers and missing values ​​in the original data are analyzed, the outliers are deleted and the missing values ​​are filled. If the difference is less than the set difference threshold, the original data is cleaned. After data cleaning, outlier removal, and missing value imputation, the original data is normalized to obtain the sample dataset.

9. The C / SiC composite material design system according to claim 7, characterized in that, Extract data features from the sample dataset, filter these features based on selection rules to identify redundant features, remove these redundant features, and generate optimized data features. Specifically, this includes: Extract data features from the sample dataset and establish feature correlation thresholds based on screening rules; Calculate the Pearson correlation coefficient between any two data features, measure the linear correlation between the two data features based on the Pearson correlation coefficient, and plot a thermodynamic graph. Analyze whether the linear correlation value between any two data features is greater than or equal to the feature correlation threshold; If the correlation threshold is greater than or equal to the feature correlation threshold, then the two data features are considered to have redundant features. The data features are analyzed for feature importance to obtain a feature importance ranking. Based on the feature importance ranking, the data features with lower feature importance among two data features with redundant features are removed to generate optimized data features.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a C / SiC composite material design method program, which, when executed by a processor, implements the steps of the C / SiC composite material design method as described in any one of claims 1 to 6.

Images