Method, device, medium and product for predicting mechanical properties of thermoplastic composites

Abstract

The application discloses a thermoplastic composite mechanical property prediction method, device, medium and product, and the method comprises the following steps: obtaining component parameter sets matched with a thermoplastic composite to be simulated and a plurality of target parameter value sets matched with the thermoplastic composite to be simulated, constructing a matched target simulation representative volume element to perform a simulation experiment and obtaining a simulation experiment result. A matched target physical representative volume element is constructed according to each target parameter value set to perform a real physical experiment and obtain a physical experiment result, a plurality of training samples are obtained according to each target parameter value set, a corresponding simulation experiment result and a physical experiment result, a target loss function is set, a neural network prediction model is trained to obtain a mechanical property prediction model, and a parameter value set to be measured is input into the mechanical property prediction model to obtain a mechanical property prediction result. The technical scheme can improve the model prediction precision and generalization ability, and efficiently complete performance prediction of the thermoplastic composite in different scenes.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, device, medium, and product for predicting the mechanical properties of thermoplastic composite materials. Background Technology

[0002] Thermoplastic composites, with their excellent toughness, re-molding capability, and recyclability, are widely used in engineering fields such as aerospace, rail transportation, and new energy equipment. The equivalent elastic modulus, as a core mechanical parameter characterizing the macroscopic stiffness and mechanical load-bearing properties of composite materials, is an important basis for structural strength verification, molding process design, and service performance evaluation.

[0003] Currently, the prediction of mechanical properties of composite materials is mainly divided into two technical systems: pure mechanism modeling and pure data-driven prediction. Pure mechanism modeling has strong physical logic, good interpretability, and high long-term stability. However, when faced with complex microscopic conditions such as pore defects, incomplete interface bonding, and temperature coupling effects in composite materials, it suffers from problems such as high theoretical modeling difficulty, many boundary assumptions, complex solutions, and poor generalization ability. On the other hand, pure data-driven prediction methods rely on mining data patterns from massive samples, have wide adaptability and strong engineering applicability, but generally suffer from defects such as lack of physical constraints, weak interpretability, insufficient extrapolation prediction accuracy, and poor model timeliness.

[0004] Most existing fusion models adopt fixed mechanism constraints and do not take into account the anisotropic mechanical properties of thermoplastic composite fibers and matrix, as well as the coupling and variation laws of multiple variables such as temperature and volume fraction. They cannot accurately constrain the overall variation trend of the longitudinal and transverse equivalent modulus of composite materials, resulting in insufficient prediction stability and generalization ability of the models under small sample, multi-condition, and temperature perturbation scenarios. They are difficult to meet the actual engineering needs of rapid and high-precision prediction of the mechanical properties of thermoplastic composite materials under multiple conditions. Summary of the Invention

[0005] This invention provides a method, device, medium, and product for predicting the mechanical properties of thermoplastic composites, in order to solve the problems that existing composite mechanical property prediction technologies are not adapted to the calculation of the equivalent elastic modulus of thermoplastic composites under temperature influence, and are difficult to balance physical rationality, prediction accuracy, and prediction stability under small sample conditions.

[0006] According to one aspect of the present invention, a method for predicting the mechanical properties of thermoplastic composite materials is provided, comprising: Obtain a set of component parameters that match the thermoplastic composite material to be simulated, and obtain a set of multiple target parameter values ​​that match the set of component parameters. The component parameters in the set of component parameters include fiber content, temperature, and interface stiffness. A target simulation representative volume element matching thermoplastic composite material is constructed, and simulation experiments are conducted on the target simulation representative volume element after setting according to each target parameter value set to obtain the mechanical property simulation experimental results corresponding to each target parameter value set. According to the target parameter value set, multiple target physical representative volume units that match the thermoplastic composite material are constructed respectively. After conducting real physical experiments on each target physical representative volume unit, the mechanical property physical experimental results corresponding to the target parameter value set are obtained respectively. The sets of target parameter values, along with the corresponding mechanical performance simulation results and mechanical performance physical experiment results, are combined to obtain multiple training samples. Obtain a preset neural network prediction model and set a target loss function for the neural network prediction model. The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function takes the model prediction results and the mechanical performance simulation experiment results as independent variables, and the data loss function takes the model prediction results and the mechanical performance physical experiment results as independent variables. Each training sample is input into the neural network prediction model, and the neural network prediction model is trained based on the target loss function to obtain the mechanical performance prediction model; Obtain the set of parameter values ​​to be tested, and input the set of parameter values ​​to be tested into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite materials for the set of parameter values ​​to be tested.

[0007] According to another aspect of the present invention, a device for predicting the mechanical properties of thermoplastic composite materials is provided, comprising: The target parameter value set construction module is used to obtain the set of component parameters that match the thermoplastic composite material to be simulated, and to obtain multiple target parameter value sets that match the set of component parameters. The component parameters in the set of component parameters include fiber content, temperature and interface stiffness. The simulation experiment execution module is used to construct a target simulation representative volume element that matches the thermoplastic composite material, and to conduct simulation experiments on the target simulation representative volume element after setting according to each target parameter value set, so as to obtain the mechanical property simulation experiment results corresponding to each target parameter value set. The physical experiment execution module is used to construct multiple target physical representative volume units that match the thermoplastic composite material according to each target parameter value set, and to conduct real physical experiments on each target physical representative volume unit to obtain the mechanical property physical experiment results corresponding to the target parameter value sets respectively. The training sample construction module is used to combine the sets of target parameter values, as well as the mechanical performance simulation experimental results and mechanical performance physical experimental results corresponding to each set of target parameter values, to obtain multiple training samples; The target loss function construction module is used to obtain a preset neural network prediction model and set a target loss function for the neural network prediction model. The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function uses the model prediction results and the mechanical performance simulation experiment results as independent variables, and the data loss function uses the model prediction results and the mechanical performance physical experiment results as independent variables. The mechanical performance prediction model acquisition module is used to input each training sample into the neural network prediction model, train the neural network prediction model based on the target loss function, and obtain the mechanical performance prediction model. The mechanical property prediction module is used to obtain the set of test parameter values ​​and input the set of test parameter values ​​into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite materials for the set of test parameter values.

[0008] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, such that the at least one processor can perform the method for predicting the mechanical properties of thermoplastic composite materials according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the method for predicting the mechanical properties of thermoplastic composite materials according to any embodiment of the present invention.

[0010] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps of the method as described in any embodiment of the present invention.

[0011] The technical solution of this invention obtains a set of component parameters matching the thermoplastic composite material to be simulated, as well as multiple sets of matching target parameter values. It then constructs a target simulation representative volumetric unit matching the thermoplastic composite material, sets corresponding target parameter values, and conducts simulation experiments to obtain the corresponding mechanical property simulation results. Based on each set of target parameter values, it constructs multiple matching target physical representative volumetric units for the thermoplastic composite material and conducts real physical experiments to obtain the corresponding mechanical property physical experiment results. It combines each set of target parameter values ​​with the corresponding mechanical property simulation and physical experiment results to obtain multiple corresponding training samples. After obtaining a preset neural network prediction model, it sets a target loss function including a mechanism loss function and a data loss function. The training samples are input into the neural network prediction model for training, and the model is iteratively optimized based on the target loss function to obtain a mechanical property prediction model. Inputting the set of parameters to be tested into the mechanical property prediction model allows for the prediction of the mechanical properties of the thermoplastic composite material. This technical solution obtains sample data through simulation experiments and real physical experiments, and completes model training based on a small number of samples, effectively reducing experimental costs and R&D cycles. The training phase integrates dual constraints of mechanism loss function and data loss function, which conforms to the inherent laws of material mechanics and the characteristics of real experimental data, greatly improving the model's prediction accuracy and generalization ability. It can efficiently complete the performance prediction of thermoplastic composite materials in different scenarios, and provide convenient and reliable data support for the optimization of thermoplastic composite material formulations and practical engineering applications.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart of a method for predicting the mechanical properties of thermoplastic composite materials according to Embodiment 1 of the present invention; Figure 2 This is a flowchart of a method for predicting the mechanical properties of thermoplastic composite materials according to Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of a representative volume unit of a thermoplastic composite material applicable to an embodiment of the present invention; Figure 4 This is a schematic diagram of a neural network prediction model applicable to an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating a model prediction accuracy error applicable to an embodiment of the present invention; Figure 6 This is a schematic diagram of a device for predicting the mechanical properties of thermoplastic composite materials according to Embodiment 3 of the present invention; Figure 7 This is a schematic diagram of the structure of an electronic device that implements the method for predicting the mechanical properties of thermoplastic composite materials according to embodiments of the present invention. Detailed Implementation

[0015] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0016] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Figure 1 This is a flowchart of a method for predicting the mechanical properties of thermoplastic composites according to Embodiment 1 of the present invention. This embodiment is applicable to the prediction of the mechanical properties of thermoplastic composites whose equivalent mechanical properties change with temperature. The method can be executed by a thermoplastic composite mechanical property prediction device, which can be implemented in hardware and / or software and is generally configured in equipment for performing thermoplastic composite mechanical property prediction. Figure 1 As shown, the method includes: S110. Obtain the set of component parameters that match the thermoplastic composite material to be simulated, and obtain a set of multiple target parameter values ​​that match the set of component parameters.

[0018] The component parameters in the component parameter set include fiber content, temperature, and interface stiffness.

[0019] The component parameter set refers to a collection of characteristic parameters representing thermoplastic composite materials. This set may include three categories of component-related parameters: fiber content, temperature, and interfacial stiffness. These parameters characterize the component properties and performance influencing conditions of the composite material. The target parameter value set refers to the combination of experimental values ​​selected for each parameter in the corresponding component parameter set. It consists of the specific values ​​of the currently selected fiber content, temperature, and interfacial stiffness.

[0020] Understandably, the mechanical and structural properties of thermoplastic composites can be represented by the parameters of each component in the component parameter set. This set can include fiber content, temperature, and interfacial stiffness. Fiber content represents the percentage of fiber filling within the thermoplastic composite, temperature is the ambient temperature, and interfacial stiffness indicates the strength of the bond between the fibers and the matrix. The overall performance characteristics of the thermoplastic composite can be directly observed through the values ​​of each parameter in the component parameter set. During experiments, multiple target parameter sets matching the component parameter set can be selected based on the experimental scenario to set specific values ​​for fiber content, temperature, and interfacial stiffness for the thermoplastic composite under various experimental conditions.

[0021] S120. Construct a target simulation representative volume element that matches the thermoplastic composite material, and conduct simulation experiments on the target simulation representative volume element after setting according to each target parameter value set to obtain the mechanical property simulation experiment results corresponding to each target parameter value set.

[0022] The target simulation representative volume element refers to the microscale simulation model element constructed when conducting simulation experiments on thermoplastic composite materials. It is the smallest characteristic volume structure that can realistically reflect the distribution of fibers and matrix and the interfacial bonding state inside the material, and can accurately reflect the actual microscopic composition and stress deformation law of thermoplastic composite materials. The mechanical property simulation experiment results refer to the various mechanical data and performance of thermoplastic composite materials under the current experimental scenario obtained after conducting simulation experiments based on the target simulation representative volume element. It can reflect the actual mechanical performance of thermoplastic composite materials under different sets of target parameter values.

[0023] Understandably, during simulation experiments, a representative volumetric element can be constructed within the simulation environment to reflect the true microstructure, material composition characteristics, and internal bonding state of thermoplastic composites. Based on the experimental scenario, the fiber content, ambient temperature, and interface stiffness values ​​corresponding to the selected set of target parameter values ​​are set as the experimental parameters for the representative volumetric element. Mechanical simulation calculations are then performed on this element to obtain the corresponding mechanical performance simulation results. By iterating through each set of target parameter values ​​and performing the simulation calculations, the mechanical performance data of the thermoplastic composite under different parameter combinations are finally obtained, providing reliable simulation data support for subsequent processing.

[0024] S130. Based on the set of target parameter values, construct multiple target physical representative volume units that match the thermoplastic composite material. After conducting real physical experiments on each target physical representative volume unit, obtain the physical experimental results of mechanical properties corresponding to the set of target parameter values.

[0025] The target physical representative volume unit refers to a physical experimental unit prepared according to the actual component ratio and molding process of the thermoplastic composite material during real physical experiments. Its microstructure, fiber arrangement, and interfacial bonding state can truly reflect the intrinsic structural characteristics of the material, serving as a standard test carrier for conducting real physical experiments on mechanical properties. The mechanical property physical experiment results refer to the actual performance data obtained after conducting real physical experiments on the target physical representative volume unit. These data represent the measured values ​​of various mechanical indices such as strength, deformation, and stiffness of the thermoplastic composite material under different sets of target parameter values.

[0026] Understandably, in conducting real-world physical experiments, multiple representative volumetric units capable of realistically replicating the microstructure, actual component ratios, and internal bonding states of the thermoplastic composite material can be prepared based on the actual molding process and preparation standards of the thermoplastic composite. According to the target parameter value set corresponding to each experimental scenario, the corresponding fiber content, experimental temperature, and interface stiffness parameters are matched to complete the sample preparation and experimental condition settings for each representative volumetric unit. Standardized real-world mechanical and physical experiments are then conducted on each representative volumetric unit. By traversing all target parameter value sets and completing the corresponding real-world physical experiment testing process, the actual mechanical properties of the thermoplastic composite material under different target parameter value sets are obtained. This data serves as verification data for subsequent data comparison and model calibration processes.

[0027] S140. Combine the sets of target parameter values, as well as the mechanical performance simulation results and mechanical performance physical test results corresponding to each set of target parameter values, to obtain multiple training samples.

[0028] Understandably, after completing all testing procedures for both simulation and real physical experiments, the experimental results can be integrated and processed to construct model training data samples from the mechanical performance simulation results and physical experimental results corresponding to each set of target parameter values. Component-related parameters such as fiber content, temperature, and interface stiffness included in each set of target parameter values ​​can be used as model training input samples, while the physical experimental results corresponding to the same set of target parameter values ​​can be used as standard measured reference data. By traversing each set of target parameter values ​​and sequentially integrating them to form corresponding training sample data, the model can be trained and learned using the constructed training sample dataset.

[0029] S150. Obtain the preset neural network prediction model and set the target loss function for the neural network prediction model.

[0030] The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function uses the model prediction results and the mechanical performance simulation experiment results as independent variables, while the data loss function uses the model prediction results and the mechanical performance physical experiment results as independent variables.

[0031] The mechanistic loss function can be a function constructed based on the intrinsic mechanisms of material mechanics, using the model's predicted values ​​and the results of mechanical performance simulation experiments as independent variables to calculate relevant data errors and constrain the model to conform to the intrinsic laws of material mechanics reflected in the simulation. The data loss function can be a function constructed based on actual physical experimental data, using the model's predicted values ​​and the results of physical experiments on mechanical performance as independent variables to measure numerical deviations and correct model prediction deviations based on actual physical experimental data. The target loss function can be a comprehensive loss function formed by fusing the mechanistic loss function and the data loss function, taking into account both mechanistic constraints and measured data constraints during model training, and is used to optimize the model in iterative training rounds.

[0032] Understandably, after obtaining a pre-defined neural network prediction model for predicting the mechanical properties of thermoplastic composites, in order to improve the model's prediction accuracy and generalization ability, and to make the prediction results more consistent with the mechanical mechanism of thermoplastic composites and the actual experimental laws, a target loss function can be set for this neural network prediction model. This target loss function is a composite loss function, including a mechanism loss function and a data loss function. The mechanism loss function uses the model prediction results and the mechanical property simulation experimental results as independent variables, quantifying the error between the predicted value and the simulation data, and constraining the model to conform to the simulated mechanical mechanism of thermoplastic composites. The data loss function uses the model prediction results and the mechanical property physical experimental results as independent variables, quantifying the deviation between the predicted value and the measured data, and correcting the model error based on the actual experimental data. Through the synergistic constraint of these two types of loss functions, model mechanism fitting and measured data calibration can be completed, providing reliable constraint support for model iterative training and parameter optimization.

[0033] S160. Input each training sample into the neural network prediction model, and train the neural network prediction model based on the target loss function to obtain the mechanical performance prediction model.

[0034] Among them, the mechanical performance prediction model can refer to the neural network prediction model obtained by iterative training based on the dual constraints of mechanism and data. With component parameters such as fiber content, temperature, and interface stiffness as input, it can quickly calculate and output various mechanical performance indicators corresponding to thermoplastic composite materials.

[0035] Understandably, the integrated training samples are sequentially input into a pre-defined neural network prediction model. Through a pre-constructed target loss function that includes both mechanistic and data loss functions, the weight parameters within the neural network prediction model are continuously iteratively adjusted and optimized. This process continuously reduces the error between the model output and the verification data provided by real physical experiments. After multiple rounds of model iteration, learning, and optimization, a mechanical performance prediction model that can stably provide prediction functions is finally obtained.

[0036] S170. Obtain the set of parameter values ​​to be tested, and input the set of parameter values ​​to be tested into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite material for the set of parameter values ​​to be tested.

[0037] The set of parameters to be measured can refer to the set of component parameters in the actual application scenario to be predicted, which may include the actual set parameters such as fiber content, ambient temperature, and interface stiffness of the thermoplastic composite material in the actual prediction scenario.

[0038] Understandably, when using a mechanical property prediction model to predict mechanical properties, a set of test parameter values ​​required in the actual application scenario can be obtained. This set includes the set of component parameters corresponding to the thermoplastic composite material in the scenario to be predicted. The prepared set of test parameter values ​​is input into the trained mechanical property prediction model. After model calculation, analysis and feature inference, the various mechanical indices corresponding to the thermoplastic composite material under the set of test parameter values ​​can be output, that is, the mechanical property prediction results of the thermoplastic composite material for the set of test parameter values.

[0039] The technical solution of this invention obtains a set of component parameters matching the thermoplastic composite material to be simulated, as well as multiple sets of matching target parameter values. It then constructs a target simulation representative volumetric unit matching the thermoplastic composite material, sets corresponding target parameter values, and conducts simulation experiments to obtain the corresponding mechanical property simulation results. Based on each set of target parameter values, it constructs multiple matching target physical representative volumetric units for the thermoplastic composite material and conducts real physical experiments to obtain the corresponding mechanical property physical experiment results. It combines each set of target parameter values ​​with the corresponding mechanical property simulation and physical experiment results to obtain multiple corresponding training samples. After obtaining a preset neural network prediction model, it sets a target loss function including a mechanism loss function and a data loss function. The training samples are input into the neural network prediction model for training, and the model is iteratively optimized based on the target loss function to obtain a mechanical property prediction model. Inputting the set of parameters to be tested into the mechanical property prediction model allows for the prediction of the mechanical properties of the thermoplastic composite material. This technical solution obtains sample data through simulation experiments and real physical experiments, and completes model training based on a small number of samples, effectively reducing experimental costs and R&D cycles. The training phase integrates dual constraints of mechanism loss function and data loss function, which conforms to the inherent laws of material mechanics and the characteristics of real experimental data, greatly improving the model's prediction accuracy and generalization ability. It can efficiently complete the performance prediction of thermoplastic composite materials in different scenarios, and provide convenient and reliable data support for the optimization of thermoplastic composite material formulations and practical engineering applications.

[0040] Example 2 Figure 2 This is a flowchart illustrating a method for predicting the mechanical properties of thermoplastic composite materials according to Embodiment 2 of the present invention. This embodiment is a further specification based on the above embodiments, including: specific methods for correcting coefficients in simulation experimental results, and specific methods for training a neural network prediction model. Figure 2 As shown, the method includes: S210. Obtain the set of component parameters that match the thermoplastic composite material to be simulated, and obtain a set of multiple target parameter values ​​that match the set of component parameters.

[0041] The component parameters in the component parameter set include fiber content, temperature, and interface stiffness.

[0042] S220. Construct a target simulation representative volume element that matches the thermoplastic composite material, and conduct simulation experiments on the target simulation representative volume element after setting according to each target parameter value set to obtain the mechanical property simulation experiment results corresponding to each target parameter value set.

[0043] Optionally, simulation experiments are performed on representative volumetric elements of the target simulation after setting each set of target parameter values, to obtain simulation results of mechanical properties corresponding to each set of target parameter values, including: Select a target reference point and set periodic boundary conditions in the representative volume element of the target simulation; The current target parameter value set is obtained sequentially from each target parameter value set, and the representative volume element of the current target simulation is obtained after being set according to the current target parameter value set; By gradually applying displacement loads and / or force loads to the target reference point of the current target simulation representative volume element, simulation operations such as fiber direction stretching, perpendicular fiber direction stretching, in-plane shearing, and out-of-plane shearing are performed to obtain the current mechanical response that matches the current target simulation representative volume element. Extract the current target matrix elastic modulus and the current target fiber elastic modulus that match the target reference point from the current mechanical response, and use them as the mechanical performance simulation results corresponding to the representative volume element of the current target simulation. Return to the previous state and perform the operation of retrieving the current target parameter value set in each target parameter value set in turn, until the processing of all target parameter value sets is completed.

[0044] The target reference point refers to a selected feature point on the surface of the target simulated representative volume element. It is the location where the simulated load is applied, and tensile, shear, or displacement constraints can be applied at this point to simulate mechanical loading operations such as tension and shear. It can serve as a benchmark observation point for extracting mechanical response data such as stress and strain, and is used to uniformly collect mechanical signals under various loading conditions. Periodic boundary conditions refer to constraints used to limit the periodic changes in the relative surface displacement of the target simulated representative volume element, eliminate boundary interference, and ensure the continuous distribution of internal stress. The target matrix elastic modulus refers to the mechanical parameter of the target simulated representative volume element of the thermoplastic composite material that resists elastic deformation. It reflects the ease with which the target simulated representative volume element deforms under stress and is a core performance indicator characterizing rigidity. The target fiber elastic modulus refers to the mechanical parameter characterizing the fiber component of the target simulated representative volume element of the thermoplastic composite material that resists elastic deformation. It reflects the magnitude of the fiber's own stiffness and is a key performance indicator for measuring the fiber's resistance to deformation. Mechanical response refers to the various mechanical parameters output by the target simulated representative volume element of thermoplastic composite material after being subjected to load during the simulation experiment. These parameters include the matrix elastic modulus and the fiber elastic modulus. They can characterize the overall mechanical changes such as stress, strain, and deformation.

[0045] Specifically, when conducting simulation experiments, one can do as follows: Figure 3 The target reference point O is selected on the target simulation representative volume element of the thermoplastic composite material. A three-dimensional coordinate system is established with the target reference point O as the origin and Z, T and R as three of the coordinate axes, which serve as the reference for the relevant experimental data. Periodic boundary conditions are set for the target simulation representative volume element so that the target simulation representative volume element can equivalently represent the mechanical behavior of an infinite composite material, constrain the period of the relative surface displacement field of the model to be consistent, eliminate boundary effects, and ensure the continuity of the micro-stress distribution. From a pre-set set of all target parameter values, one target parameter value set is sequentially selected as the current target parameter value set. A representative volume element for the current target simulation is then set based on the fiber content, temperature, and interface stiffness parameters in the current target parameter value set. Based on the selected target reference point, four types of mechanical loading simulations are sequentially performed: fiber-directed tension, perpendicular fiber-directed tension, in-plane shear, and out-of-plane shear. Displacement and force loads are applied step-by-step to simulate the mechanical response process of thermoplastic composites under different working conditions. Finally, the stress-strain data of the mechanical response at the target reference point are extracted, and the elastic modulus of the current target matrix and the current target fiber, as well as the in-plane shear modulus and corresponding shear modulus in the fiber direction and perpendicular fiber direction, are calculated to generate the corresponding simulation dataset. After traversing and completing the mechanical performance simulation experiments corresponding to all target parameter value sets, the corresponding simulation datasets for each target parameter value set constitute the mechanical performance simulation experiment results.

[0046] S230. The target matrix elastic modulus and target fiber elastic modulus corresponding to each target parameter value set are corrected by coefficients to obtain the equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction after coefficient correction, which are used as the new mechanical property simulation experimental results.

[0047] Understandably, to make the simulation results more consistent with the mechanical response of thermoplastic composites and reduce the deviation between theoretical parameters and actual working conditions, coefficient corrections can be applied to the corresponding target matrix elastic modulus and target fiber elastic modulus for each set of target parameter values. Based on the coefficient-corrected target matrix elastic modulus and target fiber elastic modulus, the equivalent elastic modulus along the fiber axis and perpendicular to the fiber axis is obtained, and this is used as the updated mechanical property simulation result to ensure that subsequent data accurately reflects the mechanical properties of thermoplastic composites.

[0048] Optionally, coefficient corrections are applied to the target matrix elastic modulus and target fiber elastic modulus corresponding to each set of target parameter values ​​to obtain the equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction after coefficient correction, which serves as the new mechanical property simulation experimental results, including: Given the current set of target parameter values, obtain the representative volume element of the current target simulation after setting according to the current set of target parameter values; Obtain the matrix volume fraction corresponding to the representative volume element of the current target simulation. Target matrix elastic modulus and fiber volume fraction ; According to the formula: The fiber-direction matrix elastic modulus of the representative volume element of the current target simulation is calculated. And according to the formula The matrix elastic modulus perpendicular to the fiber direction of the representative volume element of the current target simulation is calculated. ; According to the formula The equivalent elastic modulus in the fiber direction corresponding to the representative volume element of the current target simulation is calculated. ; According to the formula The equivalent elastic modulus in the vertical fiber direction corresponding to the representative volume element of the current target simulation is calculated. .

[0049] Specifically, by performing mechanical property simulation experiments on a representative volume element of the current target simulation based on the current target parameter value set, the output elastic modulus of the target matrix can be obtained. With the target fiber elastic modulus Furthermore, the matrix volume fraction corresponding to the representative volume element of the current target simulation can be calculated based on the component parameter values ​​in the current target parameter value set. and fiber volume fraction And then according to the formula The fiber-direction matrix elastic modulus of the representative volume element of the current target simulation was calculated. According to the formula The matrix elastic modulus perpendicular to the fiber direction of the representative volume element of the current target simulation was calculated. Substitute the relevant parameters into the equivalent elastic modulus in the fiber direction. Calculation formula Later obtained This allows us to calculate the equivalent elastic modulus in the current fiber direction corresponding to the representative volume element of the current target simulation. Then substitute the relevant parameters into the equivalent elastic modulus in the direction perpendicular to the fiber. Calculation formula Later obtained This allows us to calculate the equivalent elastic modulus in the current vertical fiber direction corresponding to the representative volume element of the current target simulation. .

[0050] S240. Based on the set of target parameter values, construct multiple target physical representative volume units that match the thermoplastic composite material. After conducting real physical experiments on each target physical representative volume unit, obtain the physical experimental results of mechanical properties corresponding to the set of target parameter values.

[0051] The physical test results of mechanical properties include: equivalent elastic modulus in the fiber direction and equivalent elastic modulus perpendicular to the fiber direction. The physical test results of mechanical properties also include the true equivalent elastic modulus.

[0052] Specifically, one target parameter value set is selected sequentially from each target parameter value set as the current target parameter value set. Based on the target reference point selected in the simulation experiment and the set periodic boundary conditions corresponding to the current target parameter value set, a corresponding thermoplastic composite material target physical representative volume element is constructed. A real physical and mechanical property experiment under the same working conditions is performed. The target physical representative volume element is physically measured during the experiment using a special instrument to obtain the fiber-direction equivalent elastic modulus and the perpendicular fiber-direction equivalent elastic modulus corresponding to each target parameter value set, thereby obtaining the corresponding real equivalent elastic modulus.

[0053] S250. Combine the sets of target parameter values, as well as the mechanical performance simulation results and mechanical performance physical experiment results corresponding to each set of target parameter values, to obtain multiple training samples.

[0054] Specifically, when constructing training samples for training neural network prediction models, one of each target parameter value set can be selected sequentially as the current target parameter value set. The component parameter values ​​in the current target parameter value set, the equivalent elastic modulus and shear modulus in the corresponding mechanical performance simulation experimental results are used as input data in the training samples. The real equivalent elastic modulus in the mechanical performance physical experimental results corresponding to the current target parameter value set is used as verification data, thereby obtaining multiple training samples corresponding to all target parameter value sets.

[0055] S260. Obtain the preset neural network prediction model and set the target loss function for the neural network prediction model.

[0056] The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function uses the model prediction results and the mechanical performance simulation experiment results as independent variables, while the data loss function uses the model prediction results and the mechanical performance physical experiment results as independent variables.

[0057] Specifically, in order to both accurately reflect the intrinsic mechanical mechanisms of thermoplastic composites and obtain suitable simulation data, thereby effectively improving the model's prediction accuracy and reliability, the target loss function for constructing the training neural network prediction model can include both a mechanistic loss function and a data loss function. The mechanistic loss function for the fiber-direction equivalent elastic modulus can be designed as follows: The mechanism loss function for the equivalent elastic modulus perpendicular to the fiber direction can be designed as follows: Therefore, a comprehensive mechanism loss function can be designed as follows: ,in, This refers to the predicted value of the equivalent elastic modulus in the fiber direction from the model prediction results. This represents the predicted value of the equivalent elastic modulus perpendicular to the fiber direction in the model prediction results. The equivalent elastic modulus in the fiber direction is the result of mechanical property simulation experiments. The mechanical property simulation results show the equivalent elastic modulus perpendicular to the fiber direction, where c and d are preset empirical coefficients, and nor is a preset normalization operation. The data loss function can be designed as follows: ,in, Let y be the predicted equivalent elastic modulus from the model prediction results, and y be the actual equivalent elastic modulus measured in real physical experiments. To correspond to the number of input samples in each training round, a target loss function that includes both a mechanistic loss function and a data loss function can be constructed. ,in, Weights for data loss The weights for the mechanistic loss can be designed as follows: Less than By assigning higher weights to the mechanistic loss term, the neural network prediction model can prioritize following the modified constitutive equation during the initial training phase, aligning with the physical evolution of thermoplastic composites. This ensures stable physical monotonicity and good predictive performance even with a limited number of experimental samples. Conversely, setting lower weights for the data loss term allows for calibration and compensation of the physical model framework based on real-world experimental results. This balances the logical coherence of the physical mechanism with the authenticity of measured data, ultimately enabling high-precision prediction of the mechanical properties of thermoplastic composites across the entire temperature range.

[0058] S270. Input each training sample into the neural network prediction model, and train the neural network prediction model based on the target loss function to obtain the mechanical performance prediction model.

[0059] Optionally, each training sample is input into the neural network prediction model, and the neural network prediction model is trained based on the target loss function to obtain a mechanical performance prediction model, including: Obtain multiple current input samples corresponding to the current training round. Each current input sample includes: a set of current target parameter values, and the equivalent elastic modulus of the current fiber direction corresponding to the set of current target parameter values. Current equivalent elastic modulus in the vertical fiber direction And the current true equivalent elastic modulus y, the current number of input samples is ; The current training samples are input into the neural network prediction model to obtain the predicted value of the equivalent elastic modulus in the current fiber direction that matches the current target parameter value set. Current predicted value of equivalent elastic modulus in the vertical fiber direction and the predicted value of the equivalent elastic modulus. ; According to the formula: The function value of the target loss function in the current training round is calculated. in, For data loss function, For the mechanism loss function, Weights for data loss The weights for the mechanism loss are c and d, which are preset empirical coefficients, and nor is a preset normalization operation. After optimizing the model parameters of the neural network prediction model based on the function value of the target loss function, a new training round is started, and the operation of obtaining multiple current input samples corresponding to the current training round is returned until the end of training conditions are met, and the mechanical performance prediction model is obtained.

[0060] Specifically, when training a neural network prediction model, each training epoch requires multiple input samples corresponding to each set of target parameter values. The input samples in a single model training process may include the current set of target parameter values ​​and the equivalent elastic modulus in the current fiber direction. Current equivalent elastic modulus in the vertical fiber direction And the current true equivalent elastic modulus y, and the number of current input samples can be obtained. By inputting the current training samples into the neural network prediction model, the corresponding predicted value of the equivalent elastic modulus in the current fiber direction can be output. Current predicted value of equivalent elastic modulus in the vertical fiber direction and the predicted value of the equivalent elastic modulus. Based on the current round of model training sample data, using the formula The function value of the target loss function in the current training round is calculated, and the parameters of the neural network prediction model are optimized based on the function value of the target loss function. Then, a new round of training is started, and the samples are input into the neural network prediction model after optimization for iterative operation. After multiple rounds of training, the training termination condition is reached, and the finally optimized neural network prediction model is used as the mechanical performance prediction model.

[0061] Optionally, after optimizing the model parameters of the neural network prediction model based on the function value of the target loss function, a new training epoch is started, and the operation of obtaining multiple current input samples corresponding to the current training epoch is returned until the end of training condition is met, and a mechanical performance prediction model is obtained, including: Based on the function value of the target loss function, the network weights, biases, and all other model parameters of the neural network prediction model are iteratively optimized in reverse to correct the model parameter deviations and complete the model update operation for the current training round. After each round of parameter optimization, it is determined whether the preset training termination condition is met. If the training termination condition is met, the iterative training is terminated, and a mechanical property prediction model that can accurately predict the mechanical properties of thermoplastic composite materials is obtained. If the training termination condition is not met, a new training round is executed until the preset training termination condition is met.

[0062] Specifically, based on the function value calculated from the target loss function, it can be used for, for example... Figure 4 The weight parameters in the neural network prediction model shown (i.e., a1 to a1 in the figure) n ), bias parameters (i.e., b1 to b in the figure) n Backpropagation iterative optimization is performed on all core parameters, including temperature T and fiber volume fraction. The model's predicted output data can include the predicted value of the equivalent elastic modulus in the fiber direction, along with other parameters. Predicted equivalent elastic modulus perpendicular to fiber direction In-plane shear modulus and out-of-plane shear modulus And from this, the predicted value of the equivalent elastic modulus can be calculated. Update data loss function L data And predicted values ​​through the equivalent elastic modulus in the fiber direction. Predicted equivalent elastic modulus perpendicular to the fiber direction Update the incentive loss function L LP This process gradually corrects the deviations and errors in the parameters, thereby completing the model structure and parameter updates in a single round of training. After each round of parameter optimization, the system automatically determines whether the current state has reached the pre-set training termination condition. If the termination condition is met, the iterative training process is stopped immediately, resulting in a predictive model that can accurately predict the mechanical properties of thermoplastic composite materials. If the termination condition is not met, a new round of training iteration continues until the training termination index data converges.

[0063] Optionally, it is determined whether a preset training termination condition is met. If the training termination condition is met, the iterative training is terminated, resulting in a mechanical property prediction model that can accurately predict the mechanical properties of thermoplastic composites, including: The function value of the target loss function is statistically analyzed in each training round, and the slope of the curve of the change of the function value of the target loss function in each training round is used as the error index of the model prediction accuracy. If the model prediction accuracy error index changes and stabilizes within the preset accuracy error range within the preset detection time, then the iterative optimization of the neural network prediction model will stop, and the current neural network prediction model will be used as the mechanical performance prediction model.

[0064] Among them, the model prediction accuracy error index can refer to [the specific indicator]. The preset accuracy error range can refer to [the specific range].

[0065] Specifically, the function value of the target loss function is statistically analyzed in real time during each training round of the model (i.e., Figure 5 The loss function value is recorded, and the changes in the target loss function value are recorded for different iterations. The slope of the loss function value change curve is used as the model prediction accuracy error index to evaluate the deviation of the model prediction accuracy. Figure 5 The accuracy of the model prediction accuracy error index is continuously monitored to visually reflect the model's convergence and error fluctuation status. If, within a preset detection period, the fluctuation range of this error index remains within a preset accuracy error range (i.e., ...), the model's prediction accuracy error index is considered stable. Figure 5When the accuracy curve in the model stabilizes, it indicates that the model parameters have stabilized and the prediction error no longer decreases significantly. At this point, the iterative optimization process of the neural network prediction model can be stopped, and the neural network prediction model under the current parameter state can be determined as the final usable mechanical performance prediction model.

[0066] S280. Obtain the set of parameter values ​​to be tested, and input the set of parameter values ​​to be tested into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite material for the set of parameter values ​​to be tested.

[0067] The technical solution of this invention obtains a set of component parameters matching the thermoplastic composite material to be simulated and a set of multiple target parameter values. It then constructs a target simulation representative volume element matching the thermoplastic composite material, sets the target parameter values ​​accordingly, and conducts simulation experiments to obtain corresponding mechanical property simulation results. After coefficient correction of the target matrix elastic modulus and target fiber elastic modulus in the mechanical property simulation results, it obtains the corresponding coefficient-corrected equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction, which serves as a new mechanical property simulation result. Based on each target parameter value set, multiple matching thermoplastic composite target physical representative volume elements are constructed, and real physical experiments are conducted to obtain corresponding mechanical property physical experiment results. The target parameter value sets, along with the corresponding mechanical property simulation and physical experiment results, are combined to obtain multiple corresponding training samples. After obtaining a preset neural network prediction model, a target loss function including a mechanism loss function and a data loss function is set. The training samples are input into the neural network prediction model for training, and the neural network prediction model is iteratively optimized based on the target loss function to obtain the mechanical property prediction model. By inputting the set of parameters to be measured into the mechanical property prediction model, the predicted mechanical properties of thermoplastic composites can be obtained. This technical solution takes into account both the physical mechanism constraints of thermoplastic composites and the correction of experimental data, making the mechanical property prediction model highly generalizable and extrapolable, and capable of predicting the mechanical properties of thermoplastic composites under actual working conditions with high accuracy.

[0068] Example 3 Figure 6 This is a schematic diagram of a device for predicting the mechanical properties of thermoplastic composite materials according to Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a target parameter value set construction module 610, a simulation experiment execution module 620, a physical experiment execution module 630, a training sample construction module 640, a target loss function construction module 650, a mechanical performance prediction model acquisition module 660, and a mechanical performance prediction module 670.

[0069] The target parameter value set construction module 610 is used to obtain a set of component parameters that match the thermoplastic composite material to be simulated, and to obtain multiple sets of target parameter values ​​that match the set of component parameters. The component parameters in the set of component parameters include fiber content, temperature and interface stiffness. The simulation experiment execution module 620 is used to construct a target simulation representative volume element that matches the thermoplastic composite material, and to conduct simulation experiments on the target simulation representative volume element after setting according to each target parameter value set, so as to obtain the mechanical property simulation experiment results corresponding to each target parameter value set. The physical experiment execution module 630 is used to construct multiple target physical representative volume units that match the thermoplastic composite material according to each target parameter value set, and to conduct real physical experiments on each target physical representative volume unit to obtain the mechanical property physical experiment results corresponding to the target parameter value sets respectively. The training sample construction module 640 is used to combine the sets of target parameter values, as well as the mechanical performance simulation experimental results and mechanical performance physical experimental results corresponding to each set of target parameter values, to obtain multiple training samples. The target loss function construction module 650 is used to obtain a preset neural network prediction model and set a target loss function for the neural network prediction model. The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function takes the model prediction results and the mechanical performance simulation experiment results as independent variables, and the data loss function takes the model prediction results and the mechanical performance physical experiment results as independent variables. The mechanical performance prediction model acquisition module 660 is used to input each training sample into the neural network prediction model, train the neural network prediction model based on the target loss function, and obtain the mechanical performance prediction model. The mechanical property prediction module 670 is used to acquire the set of test parameter values ​​and input the set of test parameter values ​​into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite materials for the set of test parameter values.

[0070] The technical solution of this invention obtains a set of component parameters matching the thermoplastic composite material to be simulated, as well as multiple sets of matching target parameter values. It then constructs a target simulation representative volumetric unit matching the thermoplastic composite material, sets corresponding target parameter values, and conducts simulation experiments to obtain the corresponding mechanical property simulation results. Based on each set of target parameter values, it constructs multiple matching target physical representative volumetric units for the thermoplastic composite material and conducts real physical experiments to obtain the corresponding mechanical property physical experiment results. It combines each set of target parameter values ​​with the corresponding mechanical property simulation and physical experiment results to obtain multiple corresponding training samples. After obtaining a preset neural network prediction model, it sets a target loss function including a mechanism loss function and a data loss function. The training samples are input into the neural network prediction model for training, and the model is iteratively optimized based on the target loss function to obtain a mechanical property prediction model. Inputting the set of parameters to be tested into the mechanical property prediction model allows for the prediction of the mechanical properties of the thermoplastic composite material. This technical solution obtains sample data through simulation experiments and real physical experiments, and completes model training based on a small number of samples, effectively reducing experimental costs and R&D cycles. The training phase integrates dual constraints of mechanism loss function and data loss function, which conforms to the inherent laws of material mechanics and the characteristics of real experimental data, greatly improving the model's prediction accuracy and generalization ability. It can efficiently complete the performance prediction of thermoplastic composite materials in different scenarios, and provide convenient and reliable data support for the optimization of thermoplastic composite material formulations and practical engineering applications.

[0071] Optionally, the simulation experiment execution module 620 can be specifically used to select a target reference point and set periodic boundary conditions in the representative volume element of the target simulation. It sequentially obtains the current target parameter value set from each target parameter value set and then obtains the current representative volume element of the target simulation after setting according to the current target parameter value set. By gradually applying displacement loads and / or force loads to the target reference point of the current representative volume element of the target simulation, simulation operations of fiber-direction tension, perpendicular fiber-direction tension, in-plane shear, and out-of-plane shear are performed to obtain the current mechanical response matching the current representative volume element of the target simulation. The current target matrix elastic modulus and the current target fiber elastic modulus matching the target reference point in the current mechanical response are extracted as the simulation results of the mechanical properties corresponding to the current representative volume element of the target simulation. The process returns to re-executing the operation of sequentially obtaining the current target parameter value set from each target parameter value set until the processing of all target parameter value sets is completed.

[0072] Optionally, it may also include a coefficient correction module, which is used to correct the coefficients of the target matrix elastic modulus and the target fiber elastic modulus corresponding to each set of target parameter values, and obtain the equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction after coefficient correction, which is used as the new mechanical property simulation experimental result.

[0073] Optionally, the coefficient correction module can be specifically used to obtain the representative volume element of the current target simulation after setting according to the current target parameter value set. It can also obtain the matrix volume fraction corresponding to the representative volume element of the current target simulation. Target matrix elastic modulus and fiber volume fraction According to the formula: The fiber-direction matrix elastic modulus of the representative volume element of the current target simulation is calculated. And according to the formula The matrix elastic modulus perpendicular to the fiber direction of the representative volume element of the current target simulation is calculated. According to the formula The equivalent elastic modulus in the fiber direction corresponding to the representative volume element of the current target simulation is calculated. According to the formula The equivalent elastic modulus in the vertical fiber direction corresponding to the representative volume element of the current target simulation is calculated. .

[0074] Optionally, the mechanical performance prediction model acquisition module 660 can be specifically used to acquire multiple current input samples corresponding to the current training round, wherein each current input sample includes: a set of current target parameter values, and the equivalent elastic modulus of the current fiber direction corresponding to the set of current target parameter values. Current equivalent elastic modulus in the vertical fiber direction And the current true equivalent elastic modulus y, the current number of input samples is The current training samples are input into the neural network prediction model to obtain the predicted value of the equivalent elastic modulus in the current fiber direction that matches the current target parameter value set. Current predicted value of equivalent elastic modulus in the vertical fiber direction and the predicted value of the equivalent elastic modulus. According to the formula: The function value of the target loss function in the current training round is calculated. For data loss function, For the mechanism loss function, Weights for data loss Here, c and d are the weights for the mechanistic loss, c and d are preset empirical coefficients, and nor is a preset normalization operation. After optimizing the model parameters of the neural network prediction model based on the function value of the target loss function, a new training epoch is started, and the process returns to obtain multiple current input samples corresponding to the current training epoch until the end of training conditions are met, and the mechanical performance prediction model is obtained.

[0075] Optionally, the mechanical property prediction model acquisition module 660 can also be specifically used to perform reverse iterative optimization of all model parameters, such as network weights and biases, of the neural network prediction model based on the function value of the target loss function, correct model parameter deviations, and complete the model update operation for the current training round. After each round of parameter optimization, it is determined whether the preset training termination condition is met. If the training termination condition is met, the iterative training is terminated, and a mechanical property prediction model that can accurately predict the mechanical properties of thermoplastic composite materials is obtained. If the training termination condition is not met, a new training round is executed until the preset training termination condition is met.

[0076] Optionally, the mechanical performance prediction model acquisition module 660 can also be used to statistically analyze the function value of the target loss function in each training round in real time, and use the slope of the curve of the change in the function value of the target loss function in each training round as the model prediction accuracy error index. When the change error of the model prediction accuracy index stabilizes within the preset accuracy error range within the preset detection time, the iterative optimization of the neural network prediction model is stopped, and the current neural network prediction model is used as the mechanical performance prediction model.

[0077] The thermoplastic composite mechanical property prediction device provided in this embodiment of the invention can execute the thermoplastic composite mechanical property prediction method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0078] Example 4 Figure 7 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0079] like Figure 7As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0080] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0081] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as methods for predicting the mechanical properties of thermoplastic composite materials.

[0082] In some embodiments, the method for predicting the mechanical properties of thermoplastic composites can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for predicting the mechanical properties of thermoplastic composites described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the method for predicting the mechanical properties of thermoplastic composites by any other suitable means (e.g., by means of firmware).

[0083] That is, obtain the set of component parameters that match the thermoplastic composite material to be simulated, and obtain a set of multiple target parameter values ​​that match the set of component parameters. The component parameters in the set of component parameters include fiber content, temperature and interface stiffness. A target simulation representative volume element matching thermoplastic composite material is constructed, and simulation experiments are conducted on the target simulation representative volume element after setting according to each target parameter value set to obtain the mechanical property simulation experimental results corresponding to each target parameter value set. According to the target parameter value set, multiple target physical representative volume units that match the thermoplastic composite material are constructed respectively. After conducting real physical experiments on each target physical representative volume unit, the mechanical property physical experimental results corresponding to the target parameter value set are obtained respectively. The sets of target parameter values, along with the corresponding mechanical performance simulation results and mechanical performance physical experiment results, are combined to obtain multiple training samples. Obtain a preset neural network prediction model and set a target loss function for the neural network prediction model. The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function takes the model prediction results and the mechanical performance simulation experiment results as independent variables, and the data loss function takes the model prediction results and the mechanical performance physical experiment results as independent variables. Each training sample is input into the neural network prediction model, and the neural network prediction model is trained based on the target loss function to obtain the mechanical performance prediction model; Obtain the set of parameter values ​​to be tested, and input the set of parameter values ​​to be tested into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite materials for the set of parameter values ​​to be tested.

[0084] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0085] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0086] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0087] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0088] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0089] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0090] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0091] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for predicting the mechanical properties of thermoplastic composite materials, characterized in that, include: Obtain a set of component parameters that match the thermoplastic composite material to be simulated, and obtain a set of multiple target parameter values ​​that match the set of component parameters. The component parameters in the set of component parameters include fiber content, temperature, and interface stiffness. A target simulation representative volume element matching thermoplastic composite material is constructed, and simulation experiments are conducted on the target simulation representative volume element after setting according to each target parameter value set to obtain the mechanical property simulation experimental results corresponding to each target parameter value set. According to the target parameter value set, multiple target physical representative volume units that match the thermoplastic composite material are constructed respectively. After conducting real physical experiments on each target physical representative volume unit, the mechanical property physical experimental results corresponding to the target parameter value set are obtained respectively. The sets of target parameter values, along with the corresponding mechanical performance simulation results and mechanical performance physical experiment results, are combined to obtain multiple training samples. Obtain a preset neural network prediction model and set a target loss function for the neural network prediction model. The target loss function includes a mechanism loss function and a data loss function. The mechanism loss function takes the model prediction results and the mechanical performance simulation experiment results as independent variables, and the data loss function takes the model prediction results and the mechanical performance physical experiment results as independent variables. Each training sample is input into the neural network prediction model, and the neural network prediction model is trained based on the target loss function to obtain the mechanical performance prediction model; Obtain the set of parameter values ​​to be tested, and input the set of parameter values ​​to be tested into the mechanical property prediction model to obtain the mechanical property prediction results of thermoplastic composite materials for the set of parameter values ​​to be tested.

2. The method according to claim 1, characterized in that, Simulation experiments were conducted on representative volumetric elements of the target simulation after setting each set of target parameter values. The simulation results of mechanical properties corresponding to each set of target parameter values ​​were obtained, including: Select a target reference point and set periodic boundary conditions in the representative volume element of the target simulation; The current target parameter value set is obtained sequentially from each target parameter value set, and the representative volume element of the current target simulation is obtained after being set according to the current target parameter value set; By gradually applying displacement loads and / or force loads to the target reference point of the current target simulation representative volume element, simulation operations such as fiber direction stretching, perpendicular fiber direction stretching, in-plane shearing, and out-of-plane shearing are performed to obtain the current mechanical response that matches the current target simulation representative volume element. Extract the current target matrix elastic modulus and the current target fiber elastic modulus that match the target reference point from the current mechanical response, and use them as the mechanical performance simulation results corresponding to the representative volume element of the current target simulation. Return to the previous state and perform the operation of retrieving the current target parameter value set in each target parameter value set in turn, until the processing of all target parameter value sets is completed.

3. The method according to claim 2, characterized in that, After obtaining the mechanical performance simulation results corresponding to each set of target parameter values, the method further includes: The target matrix elastic modulus and target fiber elastic modulus corresponding to each set of target parameter values ​​are corrected by coefficients to obtain the equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction after coefficient correction, which are used as the new mechanical property simulation results.

4. The method according to claim 3, characterized in that, The target matrix elastic modulus and target fiber elastic modulus corresponding to each set of target parameter values ​​are corrected by coefficients to obtain the equivalent elastic modulus in the fiber direction and perpendicular to the fiber direction corresponding to each set of target parameter values. These are used as new simulation results of mechanical properties, including: Given the current set of target parameter values, obtain the representative volume element of the current target simulation after setting according to the current set of target parameter values; Obtain the matrix volume fraction corresponding to the representative volume element of the current target simulation. Target matrix elastic modulus and fiber volume fraction ; According to the formula: The fiber-direction matrix elastic modulus of the representative volume element of the current target simulation is calculated. And according to the formula The matrix elastic modulus perpendicular to the fiber direction of the representative volume element of the current target simulation is calculated. ; According to the formula The equivalent elastic modulus in the fiber direction corresponding to the representative volume element of the current target simulation is calculated. ; According to the formula The equivalent elastic modulus in the vertical fiber direction corresponding to the representative volume element of the current target simulation is calculated. .

5. The method according to any one of claims 1-4, characterized in that, The physical test results of the mechanical properties include: the equivalent elastic modulus in the fiber direction and the equivalent elastic modulus perpendicular to the fiber direction, and the physical test results of the mechanical properties include the true equivalent elastic modulus; Accordingly, each training sample is input into the neural network prediction model, and the neural network prediction model is trained based on the target loss function to obtain the mechanical performance prediction model, including: Obtain multiple current input samples corresponding to the current training round. Each current input sample includes: a set of current target parameter values, and the equivalent elastic modulus of the current fiber direction corresponding to the set of current target parameter values. Current equivalent elastic modulus in the vertical fiber direction And the current true equivalent elastic modulus y, the current number of input samples is ; The current training samples are input into the neural network prediction model to obtain the predicted value of the equivalent elastic modulus in the current fiber direction that matches the current target parameter value set. Current predicted value of equivalent elastic modulus in the vertical fiber direction and the predicted value of the equivalent elastic modulus. ; According to the formula: The function value of the target loss function in the current training round is calculated. in, For data loss function, For the mechanism loss function, Weights for data loss The weights for the mechanism loss are c and d, which are preset empirical coefficients, and nor is a preset normalization operation. After optimizing the model parameters of the neural network prediction model based on the function value of the target loss function, a new training round is started, and the operation of obtaining multiple current input samples corresponding to the current training round is returned until the end of training conditions are met, and the mechanical performance prediction model is obtained.

6. The method according to claim 5, characterized in that, After optimizing the model parameters of the neural network prediction model based on the function value of the target loss function, a new training epoch is started, and the process returns to obtain multiple current input samples corresponding to the current training epoch until the end-of-training condition is met, resulting in a mechanical performance prediction model, including: Based on the function value of the target loss function, the network weights, biases, and all other model parameters of the neural network prediction model are iteratively optimized in reverse to correct the model parameter deviations and complete the model update operation for the current training round. After each round of parameter optimization, it is determined whether the preset training termination condition is met. If the training termination condition is met, the iterative training is terminated, and a mechanical property prediction model that can accurately predict the mechanical properties of thermoplastic composite materials is obtained. If the training termination condition is not met, a new training round is executed until the preset training termination condition is met.

7. The method according to claim 6, characterized in that, Determine whether the preset training termination condition is met. If the training termination condition is met, terminate the iterative training to obtain a mechanical property prediction model that can accurately predict the mechanical properties of thermoplastic composite materials, including: The function value of the target loss function is statistically analyzed in each training round, and the slope of the curve of the change of the function value of the target loss function in each training round is used as the error index of the model prediction accuracy. If the model prediction accuracy error index changes and stabilizes within the preset accuracy error range within the preset detection time, then the iterative optimization of the neural network prediction model will stop, and the current neural network prediction model will be used as the mechanical performance prediction model.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method for predicting the mechanical properties of thermoplastic composite materials according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the method for predicting the mechanical properties of thermoplastic composite materials according to any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method for predicting the mechanical properties of thermoplastic composite materials according to any one of claims 1-7.

Images