A machine learning-based method and related apparatus for optimizing the mechanical properties of cyanate ester / quartz fibers

Abstract

This invention belongs to the field of materials science and technology, and discloses a machine learning-based method and related apparatus for optimizing the mechanical properties of cyanate ester / quartz fiber. Using a trained neural network prediction model, it recommends the tensile modulus and tensile strength targets of cyanate ester corresponding to the tensile strength target of the cyanate ester / quartz fiber composite material. Using the trained machine learning model, it recommends the composition and process parameters of the cyanate ester corresponding to the tensile modulus and tensile strength targets. The recommended cyanate ester composition and process parameters are then screened to obtain the cyanate ester composition and process parameters corresponding to the tensile strength target of the cyanate ester / quartz fiber composite material. This invention can utilize limited production line data, reduce the number of experiments and costs, rapidly optimize cyanate ester performance, and thus obtain high-performance composite materials.

Description

Technical Field

[0001] This invention belongs to the field of materials science and technology, specifically relating to a method and related apparatus for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning. Background Technology

[0002] Fiber-reinforced composite materials are composite materials made by combining reinforcing fibers and matrix materials through processes such as molding, pultrusion, and winding. Cyanate ester / quartz fiber composites, due to their advantages such as high strength, light weight, excellent wave transmission properties, and strong design flexibility, are widely used in radomes for aircraft, ground radar, satellites, and 5G communications. With the development of aerospace and defense equipment, higher requirements are being placed on the tensile strength of cyanate ester / quartz fiber composites. Only by mastering the laws governing these mechanical properties can experimental parameters be controlled to ensure excellent performance of the material during service.

[0003] Currently, in actual production, composite materials need to be processed into rolls using a composite coating machine and a composite impregnation machine. After the composite materials are rolled, before delivery, they need to be laid up and cured, and their tensile strength needs to be sampled and tested to ensure that their performance is up to standard. This product production and performance testing process is time-consuming and labor-intensive. Shortening the delivery cycle of composite materials and reducing sampling and testing costs is expected to be an important way for enterprises to reduce costs and increase efficiency. The performance of composite materials is closely related to the performance of resins. The mechanical properties of cyanate esters are determined by the ingredients and curing process. However, there are many combinations of catalyst and toughening agent content and curing process, which are difficult to verify one by one through experiments. How to utilize limited production line data, reduce the number of experiments and costs, quickly optimize the performance of cyanate esters, and thus obtain high-performance composite materials has become an urgent problem to be solved. Summary of the Invention

[0004] To address the shortcomings of the existing technologies, the present invention aims to provide a machine learning-based method and apparatus for optimizing the mechanical properties of cyanate ester / quartz fiber. This invention can utilize limited production line data, reduce the number of experiments and costs, rapidly optimize the properties of cyanate ester, and thus obtain high-performance composite materials.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A machine learning-based method for optimizing the mechanical properties of cyanate ester / quartz fibers includes the following steps:

[0007] Using a pre-trained neural network prediction model, the tensile modulus target and tensile strength target of cyanate ester corresponding to the tensile strength target of cyanate ester / quartz fiber composite material are recommended; the input of the neural network prediction model is the tensile modulus and tensile strength of cyanate ester, and the output is the tensile strength of cyanate ester / quartz fiber composite material.

[0008] Using a pre-trained machine learning model, the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester are recommended; the input of the machine learning model is the composition and process parameters of the cyanate ester, and the output is the tensile modulus and tensile strength of the cyanate ester.

[0009] The recommended cyanate composition and process parameters were screened to obtain the cyanate composition and process parameters corresponding to the tensile strength target of the cyanate / quartz fiber composite material.

[0010] Preferably, the process of recommending the target tensile modulus and target tensile strength of cyanate ester corresponding to the target tensile strength of cyanate ester / quartz fiber composite material using a pre-trained neural network prediction model includes:

[0011] Using the gradient descent algorithm, we first set the target tensile strength of the cyanate ester / quartz fiber composite material, the initial values ​​of the cyanate ester tensile modulus and tensile strength. Then, we use the mean squared error as the loss function to calculate the gradient between the loss function and the input of the neural network prediction model. We iteratively optimize the cyanate ester tensile modulus and tensile strength until the loss function is stable. Then, we input the neural network model corresponding to the tensile strength of the cyanate ester / quartz fiber composite material output by the neural network prediction model at this time into the tensile modulus and tensile strength of the cyanate ester, which are used as the target tensile modulus and tensile strength of the cyanate ester corresponding to the target tensile strength of the cyanate ester / quartz fiber composite material.

[0012] Preferably, the process of recommending the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester using a trained machine learning model includes:

[0013] By calculating the tensile modulus and tensile strength of the cyanate corresponding to the same component process combination output by the machine learning model, and the Euclidean distance between these values ​​and the target cyanate's tensile modulus and tensile strength, several cyanate component process combinations corresponding to smaller Euclidean distances are selected as the components and process parameters of the cyanate corresponding to the target tensile modulus and tensile strength of the cyanate. Among these, the components and process parameters of the cyanate corresponding to the target tensile modulus and tensile strength of the cyanate are the components and process parameters of the cyanate that ensure the tensile strength of the cyanate / quartz fiber composite material is not less than the target tensile strength of the cyanate / quartz fiber composite material.

[0014] Preferably, the training process of the neural network prediction model includes:

[0015] The neural network prediction model is pre-trained using the original virtual dataset to obtain the pre-trained neural network prediction model. The construction process of the original virtual dataset includes: using the Standard module in Abaqus software to perform finite element calculations to obtain the tensile strength of cyanate ester / quartz fiber composites corresponding to different cyanate ester tensile moduli and tensile strengths, and constructing the original virtual dataset with the tensile strength of the cyanate ester / quartz fiber composites corresponding to the different cyanate ester tensile moduli and tensile strengths.

[0016] The pre-trained neural network prediction model is fine-tuned using the original real dataset to obtain the trained neural network prediction model. The construction process of the original real dataset includes: obtaining the tensile strength of the cyanate / quartz fiber composite material corresponding to different cyanate tensile moduli and tensile strengths by conducting tensile mechanical property tests on cyanate ester / quartz fiber composite materials with different cyanate ester ingredients and curing process parameters, and constructing the original real dataset with the tensile strength of the cyanate ester / quartz fiber composite material corresponding to different cyanate ester tensile moduli and tensile strengths.

[0017] Preferably, the training process of the machine learning model includes:

[0018] The tensile modulus and tensile strength test data corresponding to different combinations of cyanate components are constructed into a dataset for training a machine learning model; the dataset for training the machine learning model includes: phenolphthalein polyaryletherketone content, polysulfone content, polyimide content, cobalt acetylacetone content, organotin content, nonylphenol content, curing temperature, curing time, curing pressure, tensile modulus and tensile strength.

[0019] The machine learning model is trained using the dataset used to train the machine learning model to obtain a trained machine learning model. During training, the input parameters of the machine learning model are first normalized, and then the input is processed by the machine learning model to obtain the tensile modulus and tensile strength of the cyanate ester.

[0020] Preferably, the machine learning model adopts a linear regression model, a multinomial regression model, a support vector machine regression model, a random forest regression model, a limit gradient boosting regression model, or a Gaussian process regression model.

[0021] The neural network prediction model uses an artificial neural network regression model.

[0022] Preferably, the tensile modulus and tensile strength of cyanate are normalized, and the normalization results are processed using a neural network prediction model to obtain the tensile strength of the cyanate / quartz fiber composite material.

[0023] This invention also provides a machine learning-based system for optimizing the mechanical properties of cyanate ester / quartz fibers, comprising:

[0024] The first screening unit is used to recommend the tensile modulus target and tensile strength target of cyanate ester corresponding to the tensile strength target of cyanate ester / quartz fiber composite material using a pre-trained neural network prediction model; the input of the neural network prediction model is the tensile modulus and tensile strength of cyanate ester, and the output is the tensile strength of cyanate ester / quartz fiber composite material.

[0025] The second screening unit is used to recommend the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester using a pre-trained machine learning model; the input of the machine learning model is the composition and process parameters of the cyanate ester, and the output is the tensile modulus and tensile strength of the cyanate ester.

[0026] The third screening unit is used to screen the recommended cyanate composition and process parameters to obtain the cyanate composition and process parameters corresponding to the tensile strength target of the cyanate / quartz fiber composite material.

[0027] The present invention also provides an electronic device, comprising:

[0028] One or more processors;

[0029] A storage device on which one or more programs are stored;

[0030] When the one or more programs are executed by the one or more processors, the one or more processors implement the machine learning-based mechanical property optimization method for cyanate ester / quartz fiber as described above.

[0031] The present invention also provides a storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the machine learning-based mechanical property optimization method for cyanate ester / quartz fiber described above.

[0032] The present invention has the following beneficial effects:

[0033] This invention addresses the composite material production process by employing artificial intelligence to mine the relationship between component material properties and composite material properties based on component material performance data. It establishes a digital analysis of performance evolution, enabling rapid prediction of composite material properties. After composite materials are rolled into rolls, using predicted values ​​instead of tested mechanical property values ​​significantly reduces the frequency of performance sampling, shortens delivery cycles, and lowers economic costs. Furthermore, combining this with big data analysis allows for product quality anomaly tracing and assists in optimizing product processes, accelerating product development. For the optimized mechanical property targets of cyanate esters, active learning can selectively test samples based on the current model's predictions, rapidly narrowing down the optimal solution space. The model obtained through active learning provides targeted guidance, helping to quickly find the optimal ingredient composition during material design and optimization, thereby improving the mechanical properties of cyanate esters. Active learning improves sample utilization efficiency, reduces experimental costs, rapidly explores the optimal solution space, and provides guidance for material design and optimization in achieving the optimal ingredient composition for cyanate esters. Attached Figure Description

[0034] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, do not constitute an undue limitation of the invention. In the drawings:

[0035] Figure 1 This is a schematic diagram of the process for predicting and optimizing the mechanical properties of cyanate ester / quartz fiber composites based on machine learning, as described in this invention.

[0036] Figure 2 This is a schematic diagram of the optical microscope structure of the cyanate ester / quartz fiber composite material in this invention.

[0037] Figure 3 This is a schematic diagram of the finite element periodic mesh division in this invention.

[0038] Figure 4 This is a schematic diagram of the tensile force-displacement curve of the cyanate ester / quartz fiber composite material in the finite element simulation of this invention.

[0039] Figure 5 This diagram illustrates the comparison between the predicted data and actual data of the neural network prediction model for fitting the tensile strength of cyanate ester / quartz fiber composite materials in this invention.

[0040] Figure 6 This is a schematic diagram comparing the performance of the tensile modulus model of the machine learning model used in this invention.

[0041] Figure 7 This is a schematic diagram comparing the tensile strength model performance of the machine learning model used in this invention.

[0042] Figure 8 This is a schematic diagram comparing the predicted data and actual data of the Gaussian process regression model fitting the tensile modulus of cyanate ester in this invention.

[0043] Figure 9 This is a schematic diagram comparing the predicted data and actual data of the Gaussian process regression model fitting the tensile strength of cyanate ester in this invention. Detailed Implementation

[0044] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention, but are not intended to limit the present invention.

[0045] See Figure 1 The present invention provides a method for optimizing the mechanical properties of cyanate ester / quartz fiber composite materials, comprising the following steps:

[0046] The original data of the composite material was obtained through tensile mechanical property tests and finite element simulation calculations. The original data consisted of the tensile modulus and tensile strength of the cyanate / quartz fiber composite material corresponding to different cyanate composition processes. The data obtained from the finite element simulation was the original virtual dataset, while the data obtained from the tensile mechanical tests was the original real dataset. The original virtual dataset obtained from the finite element simulation was used as the pre-training dataset for training a neural network prediction model, and the dataset was divided into a training set and a test set. The neural network prediction model was trained based on the training set, and its accuracy was evaluated using the test set. The neural network prediction model was then fine-tuned using the original real dataset obtained from the tensile mechanical property tests of the cyanate / quartz fiber composite material to obtain the trained neural network prediction model.

[0047] The trained neural network prediction model recommends the target cyanate tensile modulus and the target tensile strength of the cyanate / quartz fiber composite material, corresponding to the target tensile strength of the cyanate / quartz fiber composite material. The recommendation process uses a gradient descent algorithm. First, the initial values ​​of the target tensile strength, cyanate tensile modulus, and initial tensile strength of the cyanate / quartz fiber composite material are given. Mean square error is used as the loss function, and the gradient between the loss function and the input (i.e., the cyanate tensile modulus and tensile strength) is calculated. The cyanate tensile modulus and tensile strength are iteratively optimized until the loss function stabilizes. At this point, the predicted value of the neural network prediction model is close to the initially set target tensile strength of the cyanate / quartz fiber composite material. The cyanate tensile modulus and tensile strength corresponding to the predicted value of the cyanate / quartz fiber composite material at this point are taken as the recommended targets (i.e., the target tensile modulus and tensile strength of the cyanate). The mean square error of the loss function calculates the variance between the predicted value of the neural network prediction model and the target tensile strength of the composite material.

[0048] The original cyanate dataset for the machine learning model is obtained from cyanate ester test data. This dataset is divided into a training set and a test set. The original cyanate ester dataset contains the tensile modulus and tensile strength corresponding to different cyanate ester composition processes. The machine learning model is trained using the training set, and its accuracy is evaluated using the test set to obtain a trained model. Based on the trained machine learning model, the components and process parameters corresponding to the target cyanate ester tensile modulus and tensile strength are recommended. The recommendation process involves calculating the Euclidean distance between the predicted tensile modulus and tensile strength of the same component process combination and the target cyanate ester tensile modulus and tensile strength, ranking them from smallest to largest. Preface; Select several combinations of cyanate ester components and processes corresponding to smaller Euclidean distances as the cyanate ester components and process parameters corresponding to the target tensile modulus and tensile strength of the cyanate ester; wherein, the cyanate ester components and process parameters corresponding to the target tensile modulus and tensile strength of the cyanate ester are the cyanate ester components and process parameters that can ensure that the tensile strength of the cyanate ester / quartz fiber composite material is not less than the target tensile strength of the cyanate ester / quartz fiber composite material; the number of combinations of cyanate ester components and processes can be determined according to the actual situation and can be flexibly adjusted. This invention does not impose specific limitations, as long as the tensile strength of the cyanate ester / quartz fiber composite material is not less than the target tensile strength of the cyanate ester / quartz fiber composite material. In the above scheme, the typical composition and process combination of cyanate ester, by mass percentage, includes: toughening agent components of cyanate ester of 0-20% phenolphthalein polyarylether ketone (PEK-C), 0-20% polysulfone (PSF), and 0-20% polyimide (PI), with the total toughening agent components not exceeding 20%; catalyst components of 0-0.08% cobalt acetylacetone and 0-0.08% organotin, with the total catalyst components not exceeding 0.08%; co-catalyst component of 0-8% nonyl; curing temperature of 170-220℃, curing time of 120-480 min, and curing pressure of 0.2-0.8 MPa. It is recommended that the machine learning model traverse all composition and process combinations of cyanate ester and use the machine learning model to predict the tensile modulus and tensile strength of cyanate ester under different composition and process combinations.

[0049] Finally, the recommended cyanate ester can be synthesized and made into a cyanate ester / quartz fiber composite material. The mechanical properties are then tested to verify the recommended results and obtain the optimized composite material.

[0050] In the above-mentioned scheme of the present invention, the Standard module in Abaqus software can be used to perform finite element simulation calculations to obtain the relevant mechanical parameters of the cyanate ester / quartz fiber composite material. When performing finite element simulation calculations, the tensile modulus and tensile strength of different cyanate esters can be changed to obtain the corresponding tensile strength of the cyanate ester / quartz fiber composite material.

[0051] In the above-described scheme of the present invention, the mechanical property test includes: a cyanate ester tensile test to obtain the tensile strength and tensile modulus of the cyanate ester; and a cyanate ester / quartz fiber composite tensile test to obtain the tensile strength of the cyanate ester / quartz fiber composite.

[0052] In the above-described scheme of the present invention, the neural network dataset partitioning method is a 10-fold cross-validation method, including:

[0053] The training set is divided into 10 parts;

[0054] Nine of the samples are used as the training set to train the neural network prediction model, and the remaining sample is used as the test set. The optimal parameters are then used to calculate the prediction accuracy of the test set.

[0055] The average of the prediction accuracy is used as an evaluation of the model's prediction accuracy.

[0056] In the above-described scheme of the present invention, the dataset partitioning method for the machine learning model is leave-one-out cross-validation, including:

[0057] Divide the total number of training sets K into K parts;

[0058] The machine learning model is trained by taking K-1 parts as the training set in turn, and the remaining part is used as the test set. The optimal parameters are then used to calculate the prediction accuracy of the test set.

[0059] The average of the prediction accuracy is used as an evaluation of the model's prediction accuracy.

[0060] In the above-described scheme of the present invention, the machine learning model can be any one of the following: linear regression model, polynomial regression model, support vector machine regression model (SVR), random forest regression model (RF), extreme gradient boosting regression model (XGBR), and Gaussian process regression model (GPR).

[0061] In the above-described scheme of the present invention, regression coefficients and root mean square error are used to measure the prediction accuracy of the machine learning model on the test set.

[0062] In the above-described scheme of the present invention, detecting the predicted result by a tensile testing machine includes: obtaining the force and displacement curves during the experiment by a sensor, and obtaining the stress-strain curve; and calculating the elastic modulus and tensile strength based on the stress-strain curve.

[0063] Example

[0064] Please refer to Figure 1 The practical example of a machine learning-based method for optimizing the mechanical properties of cyanate ester / quartz fiber includes the following steps:

[0065] Step 1: Obtain the raw data obtained through mechanical property tests and finite element simulation calculations of cyanate ester / quartz fiber composite materials;

[0066] Specifically, mechanical property tests and finite element simulations were conducted on cyanate ester / quartz fiber under different cyanate ester ingredients and curing process parameters. The tensile strength of the cyanate ester / quartz fiber composite material was simulated using the finite element method.

[0067] This embodiment obtains the tensile modulus and tensile strength of cyanate by changing the composition of cyanate ester ingredients and curing process parameters, and obtains the corresponding tensile strength mechanical properties by synthesizing cyanate ester / quartz fiber composite materials.

[0068] Specifically, finite element analysis was performed in the Standard module of Abaqus software to obtain the tensile strength of cyanate ester / quartz fiber composites corresponding to different cyanate ester tensile moduli and tensile strengths.

[0069] The fiber size and porosity were determined using an optical microscope, and a unit cell model of the cyanate ester / quartz fiber composite material was built in UG software. Figure 2 A schematic diagram of the optical microscope structure of the cyanate ester / quartz fiber composite material. The structured mesh of the unit cell model was generated using Hypermesh software, ultimately producing a periodic mesh. Figure 3 This is a schematic diagram of a periodic mesh for a unit cell model. The elastic modulus parameters of the composite material are obtained by applying periodic boundary conditions to the unit cell using the Easypbc plugin in Abaqus software.

[0070] The tensile strength of the composite material was determined using an Abaqus custom field subroutine (USDFLD). Field variables were defined, and the relationship between the properties of the cyanate ester and quartz fiber components and these field variables was established. The failure criterion for the components was the maximum stress criterion. USDFLD was used as the interface for secondary development, and the constitutive model of the components was defined within the USDFLD subroutine. The tensile conditions were then simulated using the Abaqus main program. Finally, the tensile strength of the cyanate ester / quartz fiber composite material was obtained. Figure 4 This is a schematic diagram of the tensile force-displacement curve of cyanate ester / quartz fiber composite material from finite element simulation.

[0071] The tensile strength of the final composite material is obtained by adjusting the composition and process parameters of cyanate ester, a component of cyanate ester / quartz fiber composite material.

[0072] In this embodiment, the toughening agent components of cyanate ester are selected as follows: 0-20% phenolphthalein polyarylether ketone (PEK-C), 0-20% polysulfone (PSF), and 0-20% polyimide (PI), with the total toughening agent components not exceeding 20%; the catalyst components are 0-0.08% cobalt acetylacetone and 0-0.08% organotin, with the total catalyst components not exceeding 0.08%; the co-catalyst component is nonyl 0-8%; the curing temperature is 170-220℃, the curing time is 120-480 min, and the curing pressure is 0.2-0.8 MPa.

[0073] Step 2: Divide the dataset according to the original dataset and train the neural network prediction model;

[0074] Specifically, the original dataset for training the neural network prediction model is obtained from the data generated by the finite element simulation. The finite element method generates a dataset for the tensile strength of cyanate ester / quartz fiber, and the dataset is divided into a training set and a test set.

[0075] In this step, the dataset is partitioned using 10-fold cross-validation: the training set is divided into 10 parts, and 9 parts are used as the training set to train the neural network prediction model in turn. The remaining part is used as the test set. The prediction accuracy of the test set is calculated, and the average of the prediction accuracy is used as the evaluation of the model's prediction accuracy to determine the optimal hyperparameters of the neural network prediction model.

[0076] The dataset contains three features: the tensile modulus of cyanate, the tensile strength, and the tensile strength of the cyanate / quartz fiber composite.

[0077] Normalizing the dataset is necessary because the units of the input data for the neural network prediction model are different, and the range of some data may be particularly large, resulting in slow convergence and long training time for the neural network. Since the range of the activation function of the output layer of the neural network prediction model is limited, it is necessary to map the target data for network training to the range of the activation function.

[0078] The neural network prediction model is trained based on the training dataset, and the parameters of the model are adjusted and optimized. The accuracy of the neural network prediction model is evaluated on the test set.

[0079] For the tensile strength dataset, the optimal neural network structure after debugging is as follows: the neural network has a total of 6 layers, including 4 hidden layers. The input layer has 2 neurons, the hidden layer has 256, 128, 64, and 32 neurons respectively, and the output layer has 1 neuron. The hidden layers use the ReLU activation function, the training iterations are 100, and the learning rate is 0.0001. The optimal network structure parameters are saved as a pre-trained model for transfer learning.

[0080] The optimal model network structure parameters are fitted by importing limited metadata, and the neural network prediction model is fine-tuned using experimental data of cyanate ester / quartz fiber composite material to obtain the neural network prediction model trained after transfer learning.

[0081] The performance of the neural network prediction model in predicting the tensile strength of cyanate ester / quartz fiber composites was observed. Figure 5 This diagram illustrates the comparison between the predicted data and actual data of the neural network prediction model after transfer learning in this invention, which fits the tensile strength of cyanate ester / quartz fiber composite materials. The optimal network structure parameters are retained as the base model for the gradient descent optimization algorithm.

[0082] The current dataset shows a maximum tensile strength of cyanate ester / quartz fiber composites of 597 MPa. The target tensile strength of the cyanate ester / quartz fiber composites is set to 600 MPa, with the cyanate tensile modulus initialized to 3000 MPa and the tensile strength initialized to 60 MPa. There are 16 combinations of cyanate tensile modulus and tensile strength. Each initial cyanate tensile modulus and tensile strength is assigned a random value to ensure that the 16 initial values ​​are distinct. Gradient descent is used to recommend the target cyanate modulus and tensile strength, using mean squared error as the loss function. The gradient between the loss function and the input (i.e., cyanate tensile modulus and tensile strength) is calculated, and the cyanate tensile modulus and tensile strength are iteratively optimized until the loss function stabilizes. At this point, the predicted value of the neural network prediction model is close to the initially set target tensile strength of the composite material. The cyanate tensile modulus and tensile strength corresponding to this predicted value are then used as the recommended targets. The final recommended mechanical property combination of cyanate esters is as follows: Combination 1: tensile modulus of 4202.93 MPa, tensile strength of 66.46 MPa, corresponding to a composite tensile strength of 599.89 MPa; Combination 2: tensile modulus of cyanate esters of 4125.73 MPa, tensile strength of 73.26 MPa, corresponding to a composite tensile strength of 598.73 MPa.

[0083] The neural network prediction model is an artificial neural network regression model, using root mean square error (RMSE) and regression coefficients (R²). 2 This allows us to observe and measure the model's prediction accuracy on the test set.

[0084] Step 3, please refer to Figure 1 A machine learning model was established to recommend the components and process parameters corresponding to the mechanical properties of cyanate esters;

[0085] Specifically, the original dataset for the machine learning model is obtained based on the tensile modulus and tensile strength test data corresponding to different process combinations of cyanate ester components, and the original dataset is divided into a training set and a test set.

[0086] In this step, the dataset is partitioned using leave-one-out cross-validation: the total number of training sets is divided into K parts, and K-1 parts are used as the training set to train the machine learning model in turn, while the remaining part is used as the test set. The optimal parameters are found and the prediction accuracy of the machine learning model on the test set is calculated. The average of the prediction accuracy is used as an evaluation of the prediction accuracy of the machine learning model.

[0087] The dataset contains 11 features: phenolphthalein polyaryletherketone (PEK-C) content, polysulfone (PSF) content, polyimide (PI) content, cobalt acetylacetone content, organotin content, nonylphenol content, curing temperature, curing time, curing pressure, tensile modulus of cyanate ester, and tensile strength of cyanate ester.

[0088] Normalizing the dataset is crucial because the input data has different units, and some data ranges are particularly wide. When the numerical ranges of features vary greatly, the model might assign higher weights to features with larger numerical ranges, thus ignoring features with smaller numerical ranges, even if these features are more important for the prediction results. Therefore, normalization is needed to ensure that all features are treated fairly during model training, thereby improving the model's accuracy.

[0089] The machine learning model was trained using the training set, and its prediction accuracy was evaluated using the test set to obtain the trained machine learning model. A total of six models were used to fit the tensile modulus of cyanate esters: Linear Regression, Polynomial Regression, Support Vector Machine Regression (SVR), Random Forest Regression (RF), Extreme Gradient Boosting Regression (XGBR), and Gaussian Process Regression (GPR). Figure 6 This diagram illustrates the performance comparison of the machine learning models used in this embodiment for predicting the tensile modulus. It can be seen that the Gaussian process regression model outperforms other models in both regression coefficients and root mean square error, demonstrating the highest accuracy in predicting the tensile modulus of cyanate esters.

[0090] The machine learning model was trained using the training set, and its prediction accuracy was evaluated using the test set to obtain the trained machine learning model. A total of six models were used to fit the tensile strength of cyanate esters: Linear Regression, Polynomial Regression, Support Vector Machine Regression (SVR), Random Forest Regression (RF), Extreme Gradient Boosting Regression (XGBR), and Gaussian Process Regression (GPR). Figure 7This diagram illustrates the performance comparison of the tensile strength models using the machine learning models employed in this embodiment. It can be seen that the Gaussian process regression model outperforms other models in both regression coefficients and root mean square error, demonstrating the highest accuracy in predicting the tensile strength of cyanate esters.

[0091] The performance of the tensile modulus and tensile strength of cyanate esters in a Gaussian process regression model was observed. Figure 8 This is a comparison chart of the predicted data and actual data of the tensile modulus of cyanate ester by the Gaussian process regression model in this invention; Figure 9 This is a comparison chart of the predicted and actual data of the tensile strength of cyanate ester by the Gaussian process regression model in this invention. The Gaussian process regression model has excellent performance in the tensile modulus and tensile strength properties of cyanate ester. Therefore, it is recommended to use the Gaussian process model as the base model for cyanate ester experiments.

[0092] The process space for cyanate ester composition is defined, with the following combinations: toughening agent composition of cyanate ester: phenolphthalein polyaryletherketone (PEK-C) 0-20%, step size 1%; polysulfone (PSF) 0-20%, step size 1%; polyimide (PI) 0-20%, step size 1%; total toughening agent composition not exceeding 20%; catalyst composition: cobalt acetylacetonate 0-0.08%, step size 0.01%; organotin 0-0.08%, step size 0.01%; total catalyst composition not exceeding 0.08%; co-catalyst composition: nonyl 0-8%, step size 1%; curing temperature: 170-220℃, step size 10℃; curing time: 120-480 min, step size 60 min; curing pressure: 0.2-0.8 MPa, step size 0.1 MPa. All composition and process combinations are explored, and a machine learning model is used to predict the tensile modulus and tensile strength of cyanate ester under different composition and process combinations.

[0093] By calculating the Euclidean distance between the tensile modulus and tensile strength of the cyanate ester corresponding to the same composition and process combination predicted by the Gaussian process regression model and the target tensile modulus and tensile strength of the cyanate ester, and sorting the Euclidean distances from smallest to largest, the composition and process parameters corresponding to the target mechanical properties of the cyanate ester are recommended. The formula for calculating the Euclidean distance is shown below:

[0094]

[0095] in , These are the mean values ​​of the predicted tensile modulus and tensile strength of cyanate ester by the Gaussian process regression model, respectively. This is the currently calculated Euclidean distance; and These are the target cyanate ester tensile modulus and tensile strength, respectively. The composition and process parameters of the cyanate ester with the smaller Euler-Stokes distance are selected.

[0096] The recommended cyanate ester was synthesized and subjected to tensile tests to verify the model optimization effect.

[0097] For the target combination of tensile modulus and tensile strength of cyanate ester 1, the recommended composition and process parameters of cyanate ester 1 are shown in Table 1:

[0098] Table 1

[0099]

[0100] In Example 1 of the present invention, a recommended cyanate ester was selected for verification. Tensile tests were conducted according to GB / T2567-2021 "Test Methods for Performance of Resin Castings" using a universal testing machine at room temperature. The force and displacement curves during the experiment were measured by sensors, and the stress-strain curves were obtained. The tensile strength was obtained by curve analysis.

[0101] In Example 1 of the present invention, the total length of the tensile test specimen was 200 mm, the working section length was 60 mm, the width was 10 mm, the thickness was 4 mm, and the tensile rate was 2 mm / min.

[0102] Stress-strain curves of cyanate ester under tensile conditions were plotted based on the force-displacement curves measured by the sensors. Theoretical calculations yielded an experimentally measured tensile modulus of 3.8 GPa and a tensile strength of 68 MPa. These values ​​are close to the combined mechanical properties of cyanate ester recommended by the neural network, which have a modulus of 4202.93 MPa and a strength of 66.46 MPa.

[0103] For the second combination of cyanate ester tensile modulus and tensile strength, the recommended composition and process parameters of cyanate ester 2 are shown in Table 2:

[0104] Table 2

[0105]

[0106] In Example 2 of the present invention, a recommended cyanate ester was selected for verification. Tensile tests were conducted according to GB / T2567-2021 "Test Methods for Performance of Resin Castings" using a universal testing machine at room temperature. The force and displacement curves during the experiment were measured by sensors, and the stress-strain curves were obtained. The tensile strength was obtained by curve analysis.

[0107] In Example 2 of the present invention, the total length of the tensile test specimen was 200 mm, the working section length was 60 mm, the width was 10 mm, the thickness was 4 mm, and the tensile rate was 2 mm / min.

[0108] Stress-strain curves of cyanate ester under tensile conditions were plotted based on the force-displacement curves measured by the sensors. Theoretical calculations yielded an experimentally measured tensile modulus of 3.6 GPa and a tensile strength of 70 MPa. These values ​​are close to the second recommended mechanical property combination for cyanate ester from the neural network, which has a modulus of 4125.73 MPa and a strength of 73.26 MPa.

[0109] Step 4: Impregnate the cyanate synthesized in Step 3 with quartz fiber to obtain a composite material roll;

[0110] Specifically, tensile tests were conducted on the synthesized cyanate ester / quartz fiber composite material to verify the model's predictive performance.

[0111] In Example 3 of the present invention, a composite material impregnated with resin and fiber was selected for verification. The tensile test was conducted according to ASTM D3039 "Standard Test Method for Tensile Properties of Polymer-Based Composite Materials" using a universal testing machine at room temperature. The force and displacement curves during the experiment were measured by sensors, and the stress-strain curve was obtained to guide the calculation of tensile strength.

[0112] In Example 3 of the present invention, the total length of the tensile test specimen was 250 mm, the working section length was 138 mm, the width was 25 mm, the thickness was 2.25 mm, and the tensile rate was 2 mm / min.

[0113] Based on the force and displacement curves measured by the sensor, the stress-strain curve of the cyanate ester / quartz fiber composite material under tensile conditions was plotted. According to the theoretical calculation formula, the experimentally measured tensile strength is 607 MPa, which is greater than 600 MPa, and meets the optimization requirements for the tensile strength of the composite material.

[0114] In Example 4 of this invention, a composite material impregnated with resin and fiber was used for verification. The tensile test was conducted according to ASTM D3039, "Standard Test Method for Tensile Properties of Polymer-Based Composite Materials", using a universal testing machine at room temperature. The force and displacement curves during the experiment were measured by sensors, and the stress-strain curve was obtained to guide the calculation of tensile strength.

[0115] In Example 4 of the present invention, the total length of the tensile test specimen was 250 mm, the working section length was 138 mm, the width was 25 mm, the thickness was 2.25 mm, and the tensile rate was 2 mm / min.

[0116] Based on the force and displacement curves measured by the sensor, the stress-strain curve of the cyanate ester / quartz fiber composite material under tensile conditions was plotted. According to the theoretical calculation formula, the experimentally measured tensile strength is 615 MPa, which meets the requirements for optimizing the tensile strength of the composite material.

[0117] As can be seen from the above embodiments, this invention provides a machine learning-based method for optimizing the mechanical properties of cyanate ester / quartz fiber composites. By combining finite element analysis and experimental data, machine learning can quickly optimize the cyanate ester composition and process parameters corresponding to the desired tensile strength of the composite material. This effectively solves the problem of time-consuming and labor-intensive traditional "trial and error" methods for testing composite material performance, providing significant guidance and assistance for process production. It can offer targeted guidance, helping to quickly find the optimal composition during material design and optimization, and greatly reducing the R&D and experimental costs required for new material development.

[0118] This invention also provides a machine learning-based system for optimizing the mechanical properties of cyanate ester / quartz fibers, used to implement the optimization methods provided in any of the above embodiments. The system includes:

[0119] The first screening unit is used to recommend the tensile modulus target and tensile strength target of cyanate ester corresponding to the tensile strength target of cyanate ester / quartz fiber composite material using a pre-trained neural network prediction model; the input of the neural network prediction model is the tensile modulus and tensile strength of cyanate ester, and the output is the tensile strength of cyanate ester / quartz fiber composite material.

[0120] The second screening unit is used to recommend the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester using a pre-trained machine learning model; the input of the machine learning model is the composition and process parameters of the cyanate ester, and the output is the tensile modulus and tensile strength of the cyanate ester.

[0121] The third screening unit is used to screen the recommended cyanate composition and process parameters to obtain the cyanate composition and process parameters corresponding to the tensile strength target of the cyanate / quartz fiber composite material.

[0122] The embodiments of the present invention also provide corresponding electronic devices and computer-readable storage media for implementing the solutions provided in the embodiments of the present invention.

[0123] The device includes a memory and a processor. The memory is used to store instructions or code, and the processor is used to execute the instructions or code to enable the device to perform the machine learning-based mechanical property optimization method for cyanate ester / quartz fiber described in any embodiment of this application.

[0124] The storage medium stores a computer program, which, when executed by a processor, implements the machine learning-based mechanical property optimization method for cyanate ester / quartz fiber described in any embodiment of this application.

[0125] Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort should fall within the scope of protection of the present invention.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning, characterized in that, The process includes the following: Using a pre-trained neural network prediction model, the tensile modulus and tensile strength targets of cyanate ester / quartz fiber composites are recommended, corresponding to the tensile strength targets. The neural network prediction model takes the tensile modulus and tensile strength of the cyanate ester as input and outputs the tensile strength of the cyanate ester / quartz fiber composite as output. The training process of the neural network prediction model includes: pre-training the neural network prediction model using an original virtual dataset to obtain a pre-trained neural network prediction model. The construction process of the original virtual dataset includes: performing finite element calculations using the Standard module in Abaqus software to obtain the cyanate ester / quartz fiber composites corresponding to different tensile moduli and tensile strengths. The tensile strength of the material is determined by constructing an original virtual dataset of tensile strengths of cyanate / quartz fiber composites corresponding to different cyanate ester tensile moduli and tensile strengths. The pre-trained neural network prediction model is then fine-tuned using the original real dataset to obtain the trained neural network prediction model. The construction process of the original real dataset includes: obtaining the tensile mechanical properties of cyanate / quartz fiber composites with different cyanate ester tensile moduli and tensile strengths through tensile mechanical property tests under different cyanate ester composition and curing process parameters; and constructing the original real dataset using the tensile strengths of the cyanate / quartz fiber composites corresponding to different cyanate ester tensile moduli and tensile strengths. Using a pre-trained machine learning model, the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester are recommended. The input of the machine learning model is the composition and process parameters of the cyanate ester, and the output is the tensile modulus and tensile strength of the cyanate ester. The training process of the machine learning model includes: constructing a dataset for training the machine learning model from the experimental data of tensile modulus and tensile strength corresponding to different combinations of cyanate ester composition and process. The dataset for training the machine learning model includes: phenolphthalein polyaryletherketone content, polysulfone content, polyimide content, cobalt acetylacetonate content, organotin content, nonylphenol content, curing temperature, curing time, curing pressure, tensile modulus, and tensile strength. The machine learning model is trained using the dataset for training the machine learning model to obtain a trained machine learning model. During training, the input parameters of the machine learning model are first normalized, and then the machine learning model processes the input to obtain the tensile modulus and tensile strength of the cyanate ester. The recommended cyanate composition and process parameters were screened to obtain the cyanate composition and process parameters corresponding to the tensile strength target of the cyanate / quartz fiber composite material.

2. The method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning according to claim 1, characterized in that, The process of recommending the target tensile modulus and tensile strength of cyanate ester corresponding to the target tensile strength of cyanate ester / quartz fiber composite material using a pre-trained neural network prediction model includes: Using the gradient descent algorithm, we first set the target tensile strength of the cyanate ester / quartz fiber composite material, the initial values ​​of the cyanate ester tensile modulus and tensile strength. Then, we use the mean squared error as the loss function to calculate the gradient between the loss function and the input of the neural network prediction model. We iteratively optimize the cyanate ester tensile modulus and tensile strength until the loss function is stable. Then, we input the neural network model corresponding to the tensile strength of the cyanate ester / quartz fiber composite material output by the neural network prediction model at this time into the tensile modulus and tensile strength of the cyanate ester, which are used as the target tensile modulus and tensile strength of the cyanate ester corresponding to the target tensile strength of the cyanate ester / quartz fiber composite material.

3. The method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning according to claim 1, characterized in that, The process of recommending the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester using a pre-trained machine learning model includes: By calculating the tensile modulus and tensile strength of the cyanate corresponding to the same component process combination output by the machine learning model, and the Euclidean distance between these values ​​and the target cyanate's tensile modulus and tensile strength, several cyanate component process combinations corresponding to smaller Euclidean distances are selected as the components and process parameters of the cyanate corresponding to the target tensile modulus and tensile strength of the cyanate. Among these, the components and process parameters of the cyanate corresponding to the target tensile modulus and tensile strength of the cyanate are the components and process parameters of the cyanate that ensure the tensile strength of the cyanate / quartz fiber composite material is not less than the target tensile strength of the cyanate / quartz fiber composite material.

4. The method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning according to claim 1, characterized in that, The machine learning model used is a linear regression model, a multinomial regression model, a support vector machine regression model, a random forest regression model, a limit gradient boosting regression model, or a Gaussian process regression model. The neural network prediction model uses an artificial neural network regression model.

5. The method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning according to claim 1, characterized in that, The tensile modulus and tensile strength of cyanate ester were normalized, and the normalization results were processed using a neural network prediction model to obtain the tensile strength of cyanate ester / quartz fiber composite material.

6. A machine learning-based system for optimizing the mechanical properties of cyanate ester / quartz fiber, characterized in that, A method for optimizing the mechanical properties of cyanate ester / quartz fiber based on machine learning as described in any one of claims 1-5 includes: The first screening unit is used to recommend the tensile modulus target and tensile strength target of cyanate ester corresponding to the tensile strength target of cyanate ester / quartz fiber composite material using a pre-trained neural network prediction model; the input of the neural network prediction model is the tensile modulus and tensile strength of cyanate ester, and the output is the tensile strength of cyanate ester / quartz fiber composite material. The second screening unit is used to recommend the composition and process parameters of the cyanate ester corresponding to the target tensile modulus and tensile strength of the cyanate ester using a pre-trained machine learning model; the input of the machine learning model is the composition and process parameters of the cyanate ester, and the output is the tensile modulus and tensile strength of the cyanate ester. The third screening unit is used to screen the recommended cyanate composition and process parameters to obtain the cyanate composition and process parameters corresponding to the tensile strength target of the cyanate / quartz fiber composite material.

7. An electronic device, characterized in that, include: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the machine learning-based mechanical property optimization method for cyanate ester / quartz fiber as described in any one of claims 1 to 5.

8. A storage medium, characterized in that, It stores a computer program, wherein the computer program, when executed by a processor, implements the machine learning-based method for optimizing the mechanical properties of cyanate ester / quartz fiber as described in any one of claims 1 to 5.

Images