Method for predicting flame retardant property of fiber reinforced composite material based on data driving

Abstract

The invention relates to a method for predicting the flame retardant property of a fiber reinforced composite material based on data driving. The prediction method comprises the following steps: S1) constructing and preprocessing a data set; s2) constructing a deep neural network model for predicting the flame retardant property of the fiber reinforced composite material; the deep neural network model takes a fiber parameter, a flame retardant parameter and a resin parameter as input variables and takes the flame retardant property as an output variable; s3) verifying and testing the model; and S4) flame retardant performance prediction application. According to the method, global optimization is carried out by comprehensively considering characteristic variables such as fiber types, flame retardant types, resin types and raw material proportions, the accuracy of predicting the flame retardant property of the fiber reinforced composite material analyzed by a single flame retardant can be improved through multi-variable cooperation and unification, and batch construction, calculation and processing of simulation models and results are achieved.

Description

Technical Field

[0001] The present invention belongs to the field of fiber-reinforced composite materials, and particularly relates to a method for predicting the flame retardancy of fiber-reinforced composite materials based on data-driven approach. Background Art

[0002] Fiber-reinforced composite materials are widely used in aerospace, energy and power, electronics and electrical appliances, biomedicine and other fields due to their excellent properties, and have become the key basic materials for lightweight and functional components. In order to meet the actual application requirements, flame retardant technologies are usually applied to the raw material formula or production process of fiber-reinforced composite materials to improve their flame retardancy. In actual production, the flame retardancy of fiber-reinforced composite materials is affected by various factors such as fiber type, flame retardant type, resin type, raw material ratio, and production process. Therefore, predicting the flame retardancy of fiber-reinforced composite materials has important guiding significance for industrial production.

[0003] Chinese invention patent CN103678875A discloses a method for predicting the dispersion behavior of flame retardants and flame retardant synergists in a polymer matrix, including establishing a molecular model of a flame-retardant polymer material and a layer structure model of a polymer / inorganic flame retardant system or a polymer / inorganic flame retardant / flame retardant synergist system, using the Flory-Huggins model to obtain the compatibility behavior of the polymer / inorganic flame retardant system or the polymer / inorganic flame retardant / flame retardant synergist system, and calculating the binding energy and radial distribution function of the layer structure of the polymer / inorganic flame retardant system or the polymer / inorganic flame retardant / flame retardant synergist system by molecular dynamics method, so as to determine the influence of flame retardants and flame retardant synergists on the microscopic interaction of polymer flame-retardant materials. Analyze the phase distribution and dispersion behavior of the polymer / inorganic flame retardant system or the polymer / inorganic flame retardant / flame retardant synergist system by dissipative particle dynamics method.

[0004] Chinese invention patent CN116312885A discloses a method and device for predicting the flame retardancy of wood-based composite materials. According to existing conditions such as the structural data and production process data of wood veneers, the actual value of its flame retardancy can be quickly predicted through a trained machine learning model.

[0005] Japanese invention patent JP2024035451A discloses a method and device for predicting the physical properties of a resin composition, providing a method for predicting the elongation and tensile strength of a resin composition that can accurately predict the physical properties of a resin composition, and a prediction device for predicting the physical properties of a resin composition manufactured using materials such as a base polymer, a flame retardant, and a flame retardant aid.

[0006] The prior art has not solved the problem of predicting the flame retardancy of fiber-reinforced composites, and more attention has been focused on predicting the dispersion behavior of flame retardants in polymer matrices. Summary of the Invention

[0007] Aiming at the problems existing in the prior art, the purpose of the present invention is to propose a method for predicting the flame retardancy of fiber-reinforced composites.

[0008] One aspect of the present invention provides a data-driven method for predicting the flame retardancy of fiber-reinforced composites, and the prediction method includes the following steps: S1) Dataset construction and preprocessing Collect experimental data of glass fiber-reinforced epoxy resin composites and material database information, and construct a sample dataset with standardized fiber parameters, flame retardant parameters, and resin parameters as input variables and flame retardancy as the output variable; S2) Construct a deep neural network model for predicting the flame retardancy of fiber-reinforced composites; S201) Neural network structure design: Design a deep neural network model with a fully connected structure or a deep neural network model with a multi-channel hybrid structure; The deep neural network model with a fully connected structure includes an input layer, multiple fully connected hidden layers, and an output layer; The deep neural network model with a multi-channel hybrid structure includes constructing a numerical feature channel and a categorical feature channel, and fusing the two-channel features through an attention mechanism and then outputting through the output layer; S202) Model training: Use the constructed sample dataset to train the deep neural network model. Initialize the weights and biases before training, define the loss function and introduce a thermodynamic constraint term to make the prediction results physically consistent with the material stability parameters. Use the Adam optimization algorithm or introduce the gradient penalty of Wasserstein GAN for optimization. Monitor the change of the loss value during the training process and use the Early Stopping strategy to determine convergence; S3) Model verification and testing, and evaluate the prediction error and generalization ability of the model through the validation set and the test set respectively; S4) Application of flame retardancy prediction, input the parameters of the material to be tested into the trained deep neural network model, and output the prediction results of its flame retardancy.

[0009] Furthermore, the fiber parameters include fiber diameter and volume fraction, the flame retardant parameters include flame retardant type, decomposition temperature, and volume fraction, the resin parameters include resin type, resin content, viscosity, and decomposition temperature; the flame retardancy includes limiting oxygen index, 12s vertical burning length, and extinguishing time.

[0010] Further, the fiber volume fraction is 30 - 47 vol% of the total volume fraction of the fiber in the composite material; the flame retardant volume fraction is 3% - 10 vol% of the total volume fraction of the flame retardant in the composite material; the resin content is 50 - 67 vol% of the total volume fraction of the resin in the composite material. Further, in the flame retardant performance, the value range of the limiting oxygen index is 28% - 40%, the vertical burning length in 12 s is 1 cm - 15 cm, and the extinguishing time is 1 s - 15 s.

[0011] Further, step S1) includes the following steps: S101) Collect experimental data of the fiber-reinforced composite material and information in the material database. The data includes fiber parameters, flame retardant parameters, resin parameters, and flame retardant performance data, and generate virtual samples based on the thermodynamic model to enhance the feature dimension; S102) Conduct standardized preprocessing on the collected data, including normalizing numerical data, performing one-hot encoding or embedding layer mapping on text type data, and processing missing values and outliers. When the data dimension exceeds the preset condition, perform dimensionality reduction processing; S103) Use the standardized glass fiber, flame retardant, and resin parameters as input variables and the flame retardant performance as the output variable to construct a sample data set.

[0012] Further, the thermodynamic model introduced in S101) is the Kissinger equation, which is used to simulate the decomposition behavior of the flame retardant.

[0013] Further, the normalization method includes the Z-score standardization method or the Min-Max normalization method.

[0014] Further, in S201), the numerical feature channels extract features through fully connected layers, the categorical feature channels are processed by a bidirectional long short-term memory network (Bi-LSTM), and an attention mechanism is introduced to fuse the outputs of each channel; Further, in the categorical feature channels of S201), the categorical features are mapped into low-dimensional dense vectors through the embedding layer and input into the Bi-LSTM network for sequence modeling.

[0015] Further, in step S202), the loss function used in the model training process includes the mean square error (MSE) and at least one thermodynamic constraint term, so that the predicted limiting oxygen index is positively correlated with the thermal decomposition temperature of the fiber or resin.

[0016] Further, in step S202), the optimization algorithm in the model training process adopts the gradient penalty strategy of Adam or Wasserstein GAN.

[0017] Further, in step S202), the model training process optimization algorithm adopts the Early Stopping strategy to control the training process.

[0018] Further, the training data is divided into a training set, a validation set, and a test set according to the ratios of 80%, 10%, and 10% respectively.

[0019] Further, the number of neurons in the hidden layer of the DNN neural network model is 64 to 256 for each layer, and the number of hidden layers is 3 to 5 layers.

[0020] Further, in step S3), the errors include the mean square error (MSE), the root mean square error (RMSE), and the coefficient of determination (R²).

[0021] Further, in step S3), the generalization ability of the model is evaluated using the test set, and the model is optimized according to the results.

[0022] Further, the type of fiber in the fiber reinforced composite material is one of glass fiber, basalt fiber, carbon fiber, aramid fiber, polyimide fiber, and polyarylate fiber.

[0023] Further, the type of resin in the fiber reinforced composite material is one of polyethylene, polypropylene, polymethylpentene, polystyrene, polyamide, and polycarbonate.

[0024] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the above prediction method are implemented.

[0025] The present invention also provides a computer device, including a memory and a processor. A computer program capable of running on the processor is stored on the memory. When the processor executes the computer program, the steps of the above prediction method are implemented.

[0026] Beneficial Effects Compared with the prior art, the present invention analyzes the flame retardant performance of fiber reinforced composite materials through parametric modeling, constructs a database of analysis results of the flame retardant performance of fiber reinforced composite materials, combines with data driving and introduces a neural network method to build a model, realizing the rapid and intelligent prediction of the flame retardant performance of fiber reinforced composite materials. The present invention comprehensively considers characteristic variables such as fiber type, flame retardant type, resin type, and raw material ratio for global optimization. The coordinated unity of multiple variables can improve the accuracy of predicting the flame retardant performance of fiber reinforced composite materials analyzed by a single flame retardant, and realize the batch construction, calculation, and processing of simulation models and results. Description of the Drawings

[0027] Figure 1This is the topological structure diagram of the DNN neural network according to the embodiments of the present invention.

[0028] Figure 2 This is a schematic diagram of the model construction and application process of the method for predicting the flame retardancy performance of fiber-reinforced composite materials according to the present invention. Detailed implementation manners

[0029] Now, various exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. It should be noted that: unless otherwise specifically stated, the relative arrangements, numerical expressions, and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present invention. The following description of at least one exemplary embodiment is merely illustrative in nature and in no way serves as a limitation on the present invention or its application or use.

[0030] For technologies, methods, and devices known to those of ordinary skill in the relevant art, detailed discussions may not be made. However, where appropriate, the present invention aims to provide a method for predicting the flame retardancy performance of fiber-reinforced composite materials, which uses a data-driven method to quickly and accurately predict the flame retardancy performance of fiber-reinforced composite materials, helps optimize the material ratio and production process, and improves production efficiency and product quality.

[0031] The said technologies, methods, and devices should be regarded as part of the specification. In all the examples shown and discussed here, any specific value should be construed as merely exemplary and not as a limitation. Therefore, other examples of the exemplary embodiments may have different values.

[0032] The present invention aims to provide a method for predicting the flame retardancy performance of fiber-reinforced composite materials, which uses a data-driven method to quickly and accurately predict the flame retardancy performance of fiber-reinforced composite materials, helps optimize the material ratio and production process, and improves production efficiency and product quality.

[0033] Embodiment 1 Specifically, in combination with Figure 1 As shown, a method for predicting the flame retardancy performance of a glass fiber-reinforced polypropylene composite material provided by the present invention includes the following steps: Step S1) Dataset construction and preprocessing S101) Data collection Collect experimental data of glass fiber-reinforced polypropylene composite materials, including parameters of glass fibers, flame retardant parameters, resin parameters, and flame retardancy performance.

[0034] The parameters of glass fibers include fiber diameter and volume fraction; the flame retardant parameters include flame retardant type, decomposition temperature, and volume fraction; the resin parameters include polypropylene grade, viscosity, and decomposition temperature; the flame retardancy performance includes limiting oxygen index, 12s vertical burning length, and extinguishing time.

[0035] In some specific embodiments, among the parameters of the glass fiber reinforced polypropylene composite material, the fiber diameter is 11 μm, and the volume fraction is 30 - 47 vol%; among the parameters of the flame retardant, the type of flame retardant is a phosphorus-based flame retardant, the decomposition temperature is 250 - 350 °C, and the volume fraction is 5% - 10 vol%; the resin parameters are polypropylene, the viscosity is 1.05 - 2.00 dL / g, and the decomposition temperature is 350 - 370 °C; in terms of the flame retardant performance, the value range of the limiting oxygen index is 28% - 35%, the vertical burning length at 12 s is 5 cm - 15 cm, and the extinguishing time is 5 s - 15 s.

[0036] In addition to the experimental data, it is necessary to collect existing literature, patent data, reference books, and publicly available material databases (such as NIST, Polymer Database), introduce material property data such as fiber properties, thermal conductivity, and flame retardant decomposition temperature, and enhance the feature dimensions. Use thermodynamic models (such as the Kissinger equation) to simulate the decomposition behavior of different flame retardants and generate virtual samples.

[0037] S102) Data preprocessing Perform standardized preprocessing on the data collected in S101). The standardized preprocessing includes normalizing the numerical type data, performing one-hot encoding on the text type data to convert the text type data into digital form, or using an embedding layer to map it into a low-dimensional dense vector instead of the traditional one-hot encoding to capture the potential correlations between materials, and comprehensively checking for missing values and outliers in the processed data. After processing, it is also possible to judge according to the dimension of the standardized data. When the amount of standardized data exceeds 1000 and the dimension exceeds 10, dimensionality reduction processing is performed.

[0038] For numerical data such as raw material ratios and flame retardant properties, use the normalization method to convert data with different dimensions to the same scale for the neural network model to effectively learn. In some embodiments, the normalization method uses the Z-score normalization method or the Min-Max normalization method.

[0039] Comprehensively check the fiber type, flame retardant type, resin type, raw material ratio, and flame retardant property data, and identify and process missing values and outliers. For the case of missing values in the collected data, use the mean filling or interpolation method to ensure data integrity. And remove outliers through data denoising. In some embodiments, data with a deviation from the mean exceeding 3 times the standard deviation is used as an outlier for removal. The dimensionality reduction processing uses the principal component analysis (PCA) or t-SNE dimensionality reduction method to reduce the data dimension.

[0040] S103) Construct a dataset Taking the parameters of the standardized glass fiber, flame retardant parameters, and resin parameters as variable parameters, and the flame retardant properties under different values of these variable parameters, a sample data set is jointly constructed. The sample data set contains at least 200 groups of sample data.

[0041] S2) Construction of DNN neural network model S201) Design of DNN neural network structure The structure of the deep neural network (DNN) includes an input layer, a hidden layer, and an output layer.

[0042] Among them, the feature variables received by the input layer are the variable parameters in the data set obtained in step S1). The hidden layer consists of 3 - 5 fully connected layers, and each layer contains 64 - 256 neurons. In some embodiments, the hidden layer uses the ReLU (Rectified Linear Unit) activation function to improve the nonlinear fitting ability of the model; in some embodiments, Batch Normalization is introduced to accelerate convergence and prevent gradient disappearance. The output layer is used to predict the flame retardant properties of the glass fiber reinforced polypropylene composite material, and the flame retardant property indexes are the limiting oxygen index, the vertical burning length at 12 s, and the extinguishing time.

[0043] Based on the above DNN neural network structure, a multi-channel hybrid DNN structure is further constructed, including a numerical feature processing channel, a categorical feature processing channel, and feature fusion and output.

[0044] Numerical feature processing channel: The numerical features in the data set obtained in step S1), including the fiber diameter and volume fraction of the glass fiber, the decomposition temperature and volume fraction of the flame retardant, the viscosity and decomposition temperature of the resin, etc., such as raw material ratio and physical property data, are input into the fully connected layer (FC). This fully connected layer consists of 3 layers, and the number of neurons in each layer is 128, 64, and 32 respectively. Through the fully connected layer, deep feature extraction and nonlinear transformation of the numerical features are realized.

[0045] Categorical feature processing channel: For the categorical features in the data set, including the type of glass fiber, the type of flame retardant, the type of polypropylene resin, etc., they are mapped into low-dimensional dense vectors through the embedding layer, and the bidirectional long short-term memory network (Bi-LSTM) is used for feature extraction processing. A network structure of Bi-LSTM with 2 hidden layers is constructed, and the number of hidden units in each layer is 64. Through the Bi-LSTM network, the sequential dependencies between different categorical features are captured, so as to explore the synergistic effects between the flame retardant and the glass fiber and resin.

[0046] Feature fusion and output: Based on the outputs of the numerical feature processing channel and the categorical feature processing channel, the attention mechanism is introduced.

[0047] The outputs of the fully connected layer and the Bi-LSTM are used as the inputs of the attention mechanism. Through the attention mechanism, the outputs of different channels are dynamically weighted, automatically focusing on the feature combinations that play a key role in predicting the flame retardant properties (limiting oxygen index, 12s vertical burning length, and extinguishing time) of glass fiber-reinforced epoxy composites. Finally, the weighted and fused features are input into the output layer for flame retardant performance prediction.

[0048] S202) Model training 80% of the sample dataset is used as the training set, 10% as the validation set, and 10% as the test set. The parameters of glass fiber, flame retardant, and resin are used as input data, and the flame retardant performance is used as output data. The DNN neural network is systematically trained with the training set.

[0049] Before the training work is carried out, the weights and biases of the DNN neural network are initialized. The loss function and optimization algorithm are defined. Among them, the mean square error (MSE) is used as the loss function to measure the error between the predicted flame retardant performance of the model and the actual optimal flame retardant performance. To improve the physical rationality of the model, a thermodynamic constraint term can be introduced into the loss function, such as making the predicted limiting oxygen index (LOI) of the model positively correlated with stability indicators such as the thermal decomposition temperature of fibers or resins.

[0050] The random gradient descent (SGD) algorithm and its variants are used as the optimization algorithm during the optimization process. For example, using the Adam optimization algorithm can adaptively adjust the learning rate and accelerate the model convergence speed; or introducing the gradient penalty matrix of Wasserstein GAN can force the model to generate a prediction distribution that conforms to the flame retardant mechanism, improving the physical consistency of the prediction results generated by the model with respect to the flame retardant mechanism.

[0051] The training set is divided into several batches, and each batch contains a certain number of samples. For example, each batch contains 32 samples. In each training step, the gradient is calculated from the current batch of data, and the weights and biases of the neural network are updated. During the training process, monitoring is carried out, and the loss values of the training set and the validation set are calculated regularly, and the loss curves are plotted. Model convergence judgment: When it is considered that the model has reached the convergence state, the training is stopped. The model convergence judgment can adopt the Early Stopping strategy. For example, if the loss function of the validation set has not decreased for 10 consecutive rounds, the training is stopped.

[0052] S3) Model validation and testing S301) Model performance evaluation Use the validation set to verify the effectiveness of the trained DNN neural network, and calculate the prediction error of the model on the validation set. If the error exceeds the preset range, check whether there are abnormalities in the data, adjust the model structure and retrain.

[0053] Calculate indicators such as mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination (R²).

[0054] Use the test set to predict and verify the DNN neural network model that has passed the verification. Compare the model prediction results with the actual test results to evaluate the accuracy and generalization ability of the model. If the model's performance on the prediction set is less than expected, further optimize the model. The methods for optimizing the model include increasing the sample data volume, adjusting the model structure, or optimizing the training algorithm.

[0055] S4) Flame retardancy performance prediction application S401) Data input Obtain information such as the fiber type, flame retardant type, resin type, and raw material ratio of the fiber-reinforced composite material to be tested, and input it into the trained DNN neural network model.

[0056] S402) Result output: The model outputs the prediction result of the limiting oxygen index of the glass fiber-reinforced polypropylene composite material to be tested, providing a decision-making basis for material R & D personnel and production enterprises in terms of material ratio design and production process optimization. Enterprises can select the optimal material formula and production process according to the prediction results, reduce production costs, and improve the flame retardancy and quality of products.

[0057] In summary, the prediction method of the present invention has at least the following advantages compared with the prior art: 1) By building a DNN neural network model, the present invention takes parameters such as fiber type and flame retardant type as inputs. After the model is trained and learned, it can quickly give the flame retardancy performance prediction result, greatly shortening the R & D cycle. In actual production, the material formula and process may need to be adjusted in real time due to various factors. The model of the present invention can quickly process the newly input parameters and timely predict the flame retardancy performance after adjustment, providing immediate support for production decision-making.

[0058] 2) When analyzing the influence of a single flame retardant on the flame retardancy performance of fiber-reinforced composite materials in the prior art, it is difficult to comprehensively consider the interaction of various factors, resulting in poor prediction accuracy. By parametric modeling, the present invention systematically considers the synergistic effect of factors such as fiber type, resin type, and raw material ratio with the flame retardant, enabling the model to capture the complex physical and chemical relationships inside the material.

[0059] The method of the present invention can be used for prediction of other different fiber-reinforced thermoplastic composites in addition to glass fiber-reinforced polypropylene composites. Fiber-reinforced composites are a type of high-performance materials composed of fibers and matrix materials through specific processes. Among them, the reinforcing fibers bear the main load-bearing role, endowing the materials with high strength, high modulus and other properties; the matrix materials bond the reinforcing fibers together, making the composites have a certain shape, protecting the fibers from the external environment, and transmitting stress when the materials are stressed. The fibers include different fibers such as carbon fibers, glass fibers, natural fibers, ceramic fibers, etc. And the matrix materials include resin matrices, etc.

[0060] The present invention can be a system, a method, and / or a computer program product. The computer program product can include a computer-readable storage medium having thereon computer-readable program instructions for causing a processor to implement various aspects of the present invention.

[0061] The computer-readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (non-exhaustive list) of the computer-readable storage medium include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disc (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punched card or raised structures in grooves having instructions stored thereon, and any suitable combination of the foregoing. The computer-readable storage medium used herein is not construed as an instantaneous signal itself, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., an optical pulse through an optical fiber cable), or an electrical signal transmitted through a wire.

[0062] The computer-readable program instructions described herein can be downloaded from the computer-readable storage medium to various computing / processing devices, or downloaded to an external computer or external storage device through a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, fiber optic transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0063] The computer program instructions for performing the operations of the present invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine - related instructions, microcode, firmware instructions, state - setting data, or source code or object code written in any combination of one or more programming languages, including object - oriented programming languages such as Smalltalk, C++, Python, etc., and conventional procedural programming languages such as the "C" language or similar programming languages. The computer - readable program instructions may be executed entirely on the user's computer, partially on the user's computer, executed as a stand - alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or, alternatively, may be connected to an external computer (e.g., through the Internet using an Internet service provider). In some embodiments, by using the state information of the computer - readable program instructions to customize an electronic circuit, such as a programmable logic circuit, a field - programmable gate array (FPGA), or a programmable logic array (PLA), the electronic circuit can execute the computer - readable program instructions to implement various aspects of the present invention.

[0064] Aspects of the present invention are described herein with reference to the flowchart and / or block diagram of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each block of the flowchart and / or block diagram, and the combinations of blocks in the flowchart and / or block diagram, can be implemented by computer - readable program instructions.

[0065] These computer - readable program instructions can be provided to a processor of a general - purpose computer, a special - purpose computer, or other programmable data - processing device, thereby producing a machine such that when these instructions are executed by the processor of the computer or other programmable data - processing device, a device is produced that implements the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer - readable program instructions can also be stored in a computer - readable storage medium, which causes a computer, a programmable data - processing device, and / or other devices to operate in a particular manner. Thus, the computer - readable medium storing the instructions includes a manufacture that includes instructions for implementing various aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0066] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device, causing a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process such that the instructions executed on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in one or more boxes of the flowchart and / or block diagram.

[0067] The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, a segment of code, or a portion of an instruction, which contains one or more executable instructions for implementing the specified logical function. In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two consecutive blocks may in fact be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or acts, or by a combination of dedicated hardware and computer instructions. As is well known to those skilled in the art, implementations by hardware, by software, and by a combination of software and hardware are equivalent.

[0068] The embodiments of the present invention have been described above. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the market, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein. The scope of the present invention is defined by the appended claims.

Claims

1. A prediction method for the flame retardancy of fiber-reinforced composites based on data-driven, characterized in that, The prediction method includes the following steps: S1) Dataset construction and preprocessing; Collect experimental data of fiber-reinforced composites and information from the material database, and construct a sample dataset with the standardized fiber parameters, flame retardant parameters, and resin parameters as input variables and the flame retardancy performance as the output variable; S2) Construct a deep neural network model for predicting the flame retardancy performance of fiber-reinforced composites; S201) Neural network structure design: Design a deep neural network model with a fully connected structure or a deep neural network model with a multi-channel hybrid structure; The deep neural network model with a fully connected structure includes an input layer, multiple fully connected hidden layers, and an output layer; The deep neural network model with a multi-channel hybrid structure includes constructing a numerical feature channel and a categorical feature channel, and outputs through the output layer after weighted fusion of the two-channel features by an attention mechanism; S202) Model training: Use the constructed sample dataset to train the deep neural network model. Before training, initialize the weights and biases, define the loss function and introduce a thermodynamic constraint term to make the prediction results physically consistent with the material stability parameters. Use the Adam optimization algorithm or introduce the gradient penalty of Wasserstein GAN for optimization. Monitor the change of the loss value during the training process and use the Early Stopping strategy to determine convergence; S3) Model validation and testing, and evaluate the prediction error and generalization ability of the model through the validation set and the test set respectively; S4) Application of flame retardancy performance prediction, input the parameters of the material to be tested into the trained deep neural network model, and output the prediction results of its flame retardancy performance.

2. The prediction method according to claim 1, wherein The fiber parameters include fiber diameter and volume fraction, the flame retardant parameters include flame retardant type, decomposition temperature, and volume fraction, and the resin parameters include resin type, resin content, viscosity, and decomposition temperature; the flame retardancy performance includes limiting oxygen index, 12s vertical burning length, and extinguishing time.

3. The prediction method according to claim 1, characterized in that Step S1) includes the following steps: S101) Collect experimental data of fiber-reinforced composites and information from the material database. The data includes fiber parameters, flame retardant parameters, resin parameters, and flame retardancy performance data, and generate virtual samples based on the thermodynamic model to enhance the feature dimension; S102) Perform standardized preprocessing on the collected data, including normalizing numerical data, one-hot encoding or embedding layer mapping for text type data, and handling missing values and outliers. Perform dimensionality reduction when the data dimension exceeds the preset conditions; S103) Construct a sample dataset with the standardized fiber, flame retardant, and resin parameters as input variables and the flame retardancy performance as the output variable.

4. The prediction method according to claim 1, wherein In S201), the numerical feature channel extracts features through a fully connected layer.

5. The prediction method according to claim 4, characterized in that In S201), the categorical feature channel is processed through a bidirectional long short-term memory network.

6. The prediction method according to claim 5, wherein In step S202), the loss function used in the model training process includes mean square error and at least one thermodynamic constraint term, so that the predicted limiting oxygen index is positively correlated with the thermal decomposition temperature of the fiber or resin.

7. The prediction method according to claim 1, wherein The type of fiber in the fiber-reinforced composite is one of glass fiber, basalt fiber, carbon fiber, aramid fiber, polyimide fiber, and polyarylate fiber.

8. The prediction method according to claim 1, wherein The type of resin in the fiber-reinforced composite material is one of polyethylene, polypropylene, polymethylpentene, polystyrene, polyamide, and polycarbonate.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, the steps of the prediction method according to any one of claims 1-8 are implemented.

10. A computer device, comprising a memory and a processor, wherein a computer program capable of running on the processor is stored on the memory, and is characterized in that When the processor executes the computer program, the steps of the prediction method according to any one of claims 1-8 are implemented.

Images