Composite material mechanical property prediction method based on data mining

Abstract

The invention provides a composite material mechanical property prediction method based on data mining, and belongs to the technical field of composite material mechanical property tests.The method specifically comprises the steps that parameter information and mechanical properties of a composite material are collected to form a data set, an initial prediction model is constructed, and the initial prediction model comprises a conversion module and an initial prediction network module; the conversion module is used for correcting the thermal conductivity of the composite material sample and calculating the bearing capacity of the composite material sample, and training and optimizing the initial prediction model by using the data set to obtain a final prediction model; and collecting parameter information of the to-be-predicted composite material, preprocessing the parameter information, inputting the preprocessed parameter information into the final prediction model, and outputting the mechanical properties of the to-be-predicted composite material by the final prediction model. Through the treatment scheme, the accuracy of predicting the mechanical property of the composite material is improved.

Description

Technical Field

[0001] This application relates to the field of mechanical property testing of composite materials, and in particular to a method for predicting the mechanical properties of composite materials based on data mining. Background Technology

[0002] Composite materials have been widely used in many fields due to their excellent mechanical properties. However, the mechanical properties of composite materials are influenced by a variety of complex factors, such as the characteristics of the fiber and matrix materials, the fiber arrangement, and the manufacturing process. Traditional mechanical property testing methods often only obtain a limited number of data points, making it difficult to comprehensively and deeply reveal the intrinsic laws governing the mechanical properties of composite materials. Summary of the Invention

[0003] In view of this, this application provides a data mining-based method for predicting the mechanical properties of composite materials, which solves the problems in the prior art and improves the accuracy of predicting the mechanical properties of composite materials.

[0004] The method for predicting the mechanical properties of composite materials based on data mining provided in this application adopts the following technical solution: A method for predicting the mechanical properties of composite materials based on data mining includes the following steps: Collect parameter information of composite materials, including material parameters and processing parameters, process the collected parameter information, and label the processed parameter information with corresponding mechanical property labels to form a dataset; An initial prediction model is constructed, which includes a transformation module and an initial prediction network module. The transformation module is used to correct the thermal conductivity of the composite material sample and calculate the load-bearing capacity of the composite material sample. The initial prediction model is trained and optimized using the dataset to obtain the final prediction model; The parameter information of the composite material to be predicted is collected, the parameter information is preprocessed, the preprocessed parameter information is input into the final prediction model, and the final prediction model outputs the mechanical properties of the composite material to be tested.

[0005] Optionally, the conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the corrected thermal conductivity of the composite material sample. ; ; ; In the formula, The relative structural density of the composite material sample. The thermal conductivity of the composite material is given. As a structural influencing factor, The angle of the concave corner of the composite material sample. The thickness of the composite material sample, The length of the crossbar in the microstructure of the composite material sample. The length of the longitudinal bar in the microstructure of the composite material sample is given by , and T is the ambient temperature.

[0006] Optionally, the conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the load-bearing capacity of the composite material sample. ; ; ; In the formula, The overall stiffness matrix of the composite material sample. Let be the nodal displacement vector of the composite material sample. For the nodal load vector of the composite material sample, Here is the stiffness matrix of the composite material sample element. The elastic modulus in the axial direction of the rod structure in the microstructure of the composite material sample is given. The cross-sectional area of ​​the rod structure in the microstructure of the composite material sample. The length of the rod structure in the microstructure of the composite material sample. The x-component of the axial direction of the rod structure in the microstructure of the composite material sample is given. The y-component of the axial direction of the rod structure in the microstructure of the composite material sample is represented by its cosine. It is the in-plane shear modulus. Let be the torsional constant, ( , ) represents the coordinates of the composite material sample before the nodal displacement. , () represents the coordinates of the composite material sample before the node displacement.

[0007] Optionally, the initial prediction network module is constructed based on an artificial neural network, and the initial prediction network module includes an input layer, three fully connected hidden layers, an output layer, and a loss function in sequence; The input layer receives a feature vector composed of parameter information, corrected thermal conductivity, and load-bearing capacity of the composite material sample. Three fully connected hidden layers are used for non-linear transformation and mapping of features. The three fully connected hidden layers use the GELU activation function to enhance the non-linear expressive power. The output layer is used to output the predicted mechanical properties of the composite material. An absolute value transformation is added after the activation function to the output layer to ensure the non-negativity of the modulus. The loss function is used to calculate the difference between the model's predicted value and the true label. The loss function integrates the mean squared error and the physical constraint term to ensure thermodynamic consistency.

[0008] Optionally, during the training of the initial prediction model using the dataset, the AdamW optimizer is used in conjunction with gradient pruning and ReduceLROnPlateau for dynamic adjustment, and the best prediction model is saved based on the early stopping mechanism.

[0009] Optional, the initial learning rate of the AdamW optimizer is 10⁻ 4 .

[0010] Optionally, the steps to optimize the trained prediction model include: optimizing the hyperparameters of the trained prediction model through K-fold cross-validation, selecting the optimal model using the validation set performance, quantifying the prediction confidence interval using Monte Carlo Dropout, and resolving feature sensitivity using Jacobian matrix analysis and SHAP values.

[0011] In summary, this application includes the following beneficial technical effects: This application explores the complex relationships between the mechanical properties of composite materials and factors such as material parameters and manufacturing processes. These relationships help us understand how different factors individually and synergistically influence the mechanical properties of composite materials. It can uncover potential patterns and regularities from large amounts of seemingly disorganized data, and based on the established relationships and discovered patterns, predict the potential impact of new formulations or new processing conditions on composite materials. This provides strong support for early-stage evaluation and scheme selection in the R&D process, reducing redundant testing, unnecessary experiments and cost inputs, and improving R&D efficiency. Finally, it can help identify key factors affecting mechanical properties, thus pointing the way for further optimization of composite material performance. It contributes to the discovery of potential patterns and trends, and promotes the development of composite material mechanical property testing technology. Attached Figure Description To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a data mining-based method for predicting the mechanical properties of composite materials. Detailed Implementation

[0013] The embodiments of this application will now be described in detail with reference to the accompanying drawings.

[0014] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this application, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0016] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of this application. The illustrations only show the components related to this application and are not drawn according to the number, shape and size of the components in actual implementation. In actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0017] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0018] This application provides a method for predicting the mechanical properties of composite materials based on data mining.

[0019] like Figure 1 As shown, a method for predicting the mechanical properties of composite materials based on data mining includes the following steps: Step 1: Collect parameter information of composite materials, including material parameters and processing parameters. Process the collected parameter information and label the processed parameter information with corresponding mechanical property labels to form a dataset.

[0020] Material parameters include fiber properties, resin matrix properties, interfacial properties, prepreg / fabric form, and layup design. Fiber properties specifically include material type, tensile modulus and strength, elongation at break, fiber diameter, fiber surface condition, and fiber volume fraction; material types include carbon fiber, glass fiber, aramid fiber, and basalt fiber, etc., and fiber surface condition affects the interface. Resin matrix properties specifically include material type, modulus, strength, fracture toughness, viscoelastic behavior, heat resistance, curing shrinkage, and viscosity; material types include epoxy, bismaleimide, polyimide, and thermoplastic resins, etc., and viscosity affects wetting. Prepreg / fabric form includes unidirectional tape, woven fabric, non-buckling fabric, etc. Different reinforcement configurations directly affect the anisotropy and interlaminar properties of the composite material. Layup design specifically includes layup sequence and the combination of anisotropic layup angles, such as [0° / 45° / 90° / -45°] S-type structures, which determine the overall stiffness tensor and strength distribution of the composite material.

[0021] Processing parameters include curing / molding process parameters, pressure history, vacuum level and holding pressure, layup process, fiber tension, environmental conditions, and mold design. Curing / molding process parameters include curing cycles; the heating rate, peak curing temperature, holding time, and cooling rate of the temperature history are among the most critical parameters, directly affecting the degree of cross-linking reaction / degree of curing of the resin. Pressure history includes the application method, magnitude, timing and rate of pressurization / depressurization, controlling the degree of material compaction, fiber distribution, porosity, and geometric accuracy; application methods include vacuum bags, autoclaves, and molding. Vacuum level and holding pressure parameters are crucial for removing gas, ensuring sufficient wetting, and reducing porosity. Laying up process includes layup accuracy: fiber orientation deviation, layup gaps, or overlaps. The tension applied to the fiber bundles during automatic fiber / tape laying affects interlayer stress and compaction. Temperature, humidity, and cleanliness in the molding environment significantly affect the reaction process and final properties of hygroscopic materials such as epoxy resins. The thermal conductivity, heat capacity, and surface condition of the mold affect local heat transfer and curing uniformity.

[0022] Mechanical properties include: macroscopic stiffness and strength: tensile modulus / strength, compressive modulus / strength, flexural modulus / strength, in-plane shear modulus / strength. Key interface-related properties: interlaminar shear strength, interlaminar fracture toughness, damage tolerance, and impact resistance. Fatigue properties. Creep and stress relaxation behavior. Degradation of hygrothermal properties.

[0023] The preprocessing process includes filtering and analyzing the collected data. Specifically, it includes: Data cleaning: Identify and handle missing and outlier values.

[0024] Integration and conversion: Extract features from process parameter curves, such as temperature peaks, heating / cooling rates, duration of specific temperature ranges, predicted viscosity values ​​when pressure is applied, pressure point, and magnitude of temperature gradients, etc.

[0025] Normalization: Transform data of different dimensions to a similar scale, such as [0,1] or with a mean of 0 and a variance of 1, to avoid the model being affected by the dimensions.

[0026] Step 2: Construct an initial prediction model, which includes a conversion module and an initial prediction network module. The conversion module is used to correct the thermal conductivity of the composite material sample and calculate the load-bearing capacity of the composite material sample based on the microstructure of the material sample.

[0027] Step 3: Use the dataset to train and optimize the initial prediction model to obtain the final prediction model.

[0028] Step 4: Collect parameter information of the composite material to be predicted, preprocess the parameter information, input the preprocessed parameter information into the final prediction model, and the final prediction model outputs the mechanical properties of the composite material to be tested.

[0029] In the prediction process using the predictive model, this application considers the influence of the microstructure of the composite material sample on the thermal conductivity and load-bearing capacity of the material structure. This improves the accuracy of the prediction model's results.

[0030] The molded composite material sample is a truss-like structural material composed of porous materials to support the formation of larger-scale porous materials. This application considers the microscopic truss structure of the sample and corrects the thermal conductivity of the composite material. The conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the corrected thermal conductivity of the composite material sample. ; ; ; In the formula, The relative structural density of the composite material sample. The thermal conductivity of the composite material is given. As a structural influencing factor, This indicates structural deformation, and together with temperature T, it determines the change in the concave angle of the structure. Let be the angle of the concave angle of the composite material sample, which is a function of temperature T. The thickness of the composite material sample, The length of the crossbar in the microstructure of the composite material sample. denoted as the length of the longitudinal bar in the microstructure of the composite material sample, and T as the ambient temperature.

[0031] The conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the load-bearing capacity of the composite material sample. ; ; ; In the formula, The overall stiffness matrix of the composite material sample. Let be the nodal displacement vector of the composite material sample. For the nodal load vector of the composite material sample, Here is the stiffness matrix of the composite material sample element. The elastic modulus in the axial direction of the rod structure in the microstructure of the composite material sample is given. The cross-sectional area of ​​the rod structure in the microstructure of the composite material sample. The length of the rod structure in the microstructure of the composite material sample. The x-component of the axial direction of the rod structure in the microstructure of the composite material sample is given. The y-component of the axial direction of the rod structure in the microstructure of the composite material sample is represented by its cosine. It is the in-plane shear modulus. Let be the torsional constant, ( , ) represents the coordinates of the composite material sample before the nodal displacement. , () represents the coordinates of the composite material sample before the node displacement.

[0032] The initial prediction network module is built based on an artificial neural network and consists of an input layer, three fully connected hidden layers, an output layer, and a loss function. The input layer receives a feature vector composed of parameter information, corrected thermal conductivity, and load-bearing capacity of the composite material sample.

[0033] The three fully connected hidden layers are used for nonlinear transformation and mapping of features. The three fully connected hidden layers use the GELU activation function to enhance the nonlinear expressive power, and physical constraint modules are embedded between the hidden layers to enforce the monotonicity of the stress-strain relationship.

[0034] The output layer is used to output the predicted mechanical properties of the composite material. An absolute value transformation is added after the activation function to ensure the non-negativity of the modulus.

[0035] The loss function is used to calculate the difference between the model's predicted value and the true label. The loss function integrates the mean squared error and the physical constraint term to ensure thermodynamic consistency.

[0036] During the training of the initial prediction model using the dataset, the AdamW optimizer, combined with gradient clipping and ReduceLROnPlateau dynamic adjustment, is employed, and the optimal prediction model is saved based on an early stopping mechanism. The initial learning rate of the QAdamW optimizer is 10⁻⁻⁶. 4 .

[0037] The steps for optimizing the trained prediction model include: optimizing the hyperparameters of the trained prediction model through K-fold cross-validation, selecting the optimal model using the validation set performance, quantifying the prediction confidence interval using Monte Carlo Dropout, and resolving feature sensitivity using Jacobian matrix analysis and SHAP values. Finally, the ONNX format is deployed for engineering applications, supporting three main functions: forward performance prediction, inverse microstructure design, and virtual experiments, with a computational speed three orders of magnitude faster than traditional finite element methods.

[0038] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting the mechanical properties of composite materials based on data mining, characterized in that: Includes the following steps: Collect parameter information of composite materials, including material parameters and processing parameters, process the collected parameter information, and label the processed parameter information with corresponding mechanical property labels to form a dataset; An initial prediction model is constructed, which includes a transformation module and an initial prediction network module. The transformation module is used to correct the thermal conductivity of the composite material sample and calculate the load-bearing capacity of the composite material sample. The initial prediction model is trained and optimized using the dataset to obtain the final prediction model; The parameter information of the composite material to be predicted is collected, the parameter information is preprocessed, the preprocessed parameter information is input into the final prediction model, and the final prediction model outputs the mechanical properties of the composite material to be tested.

2. The method for predicting the mechanical properties of composite materials based on data mining according to claim 1, characterized in that, The conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the corrected thermal conductivity of the composite material sample. ; ； ； In the formula, The relative structural density of the composite material sample. The thermal conductivity of the composite material is given. As a structural influencing factor, The angle of the concave corner of the composite material sample. The thickness of the composite material sample, The length of the crossbar in the microstructure of the composite material sample. The length of the longitudinal bar in the microstructure of the composite material sample is given by , and T is the ambient temperature.

3. The method for predicting the mechanical properties of composite materials based on data mining according to claim 1, characterized in that, The conversion module combines the microstructure, material parameters, and processing parameters of the composite material sample to obtain the load-bearing capacity of the composite material sample. ； ； ； In the formula, The overall stiffness matrix of the composite material sample. Let be the nodal displacement vector of the composite material sample. For the nodal load vector of the composite material sample, Here is the stiffness matrix of the composite material sample element. The elastic modulus in the axial direction of the rod structure in the microstructure of the composite material sample is given. The cross-sectional area of ​​the rod structure in the microstructure of the composite material sample. The length of the rod structure in the microstructure of the composite material sample. The x-component of the axial direction of the rod structure in the microstructure of the composite material sample is given. The y-component of the axial direction of the rod structure in the microstructure of the composite material sample is represented by its cosine. It is the in-plane shear modulus. Let be the torsional constant, ( , ) represents the coordinates of the composite material sample before the nodal displacement. , () represents the coordinates of the composite material sample before the node displacement.

4. The method for predicting the mechanical properties of composite materials based on data mining according to claim 1, characterized in that, The initial prediction network module is built on an artificial neural network and consists of an input layer, three fully connected hidden layers, an output layer, and a loss function. The input layer receives a feature vector composed of parameter information, corrected thermal conductivity, and load-bearing capacity of the composite material sample. Three fully connected hidden layers are used to perform non-linear transformations and mappings of features. The three fully connected hidden layers use the GELU activation function to enhance the non-linear expressive power. The output layer is used to output the predicted mechanical properties of the composite material. An absolute value transformation is added after the activation function to the output layer to ensure the non-negativity of the modulus. The loss function is used to calculate the difference between the model's predicted value and the true label. The loss function integrates the mean squared error and the physical constraint term to ensure thermodynamic consistency.

5. The method for predicting the mechanical properties of composite materials based on data mining according to claim 4, characterized in that, During the training of the initial prediction model using the dataset, the AdamW optimizer is used in conjunction with gradient pruning and ReduceLROnPlateau for dynamic adjustment, and the best prediction model is saved based on the early stopping mechanism.

6. The method for predicting the mechanical properties of composite materials based on data mining according to claim 5, characterized in that, The AdamW optimizer has an initial learning rate of 10⁻ 4 .

7. The method for predicting the mechanical properties of composite materials based on data mining according to claim 5, characterized in that, The steps to optimize the trained prediction model include: optimizing the hyperparameters of the trained prediction model through K-fold cross-validation, selecting the optimal model using the validation set performance, quantifying the prediction confidence interval using Monte Carlo Dropout, and resolving feature sensitivity using Jacobian matrix analysis and SHAP values.

Images