Microscopic damage evolution-based composite gas cylinder mechanical property prediction method and system
By establishing a microscopic representative volumetric unit model and training a machine learning proxy model, the problem of large dispersion in the prediction of mechanical properties of composite gas cylinders in traditional methods is solved, and high-precision mechanical property prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES)
- Filing Date
- 2026-05-08
- Publication Date
- 2026-07-07
Smart Images
Figure CN122135857B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mechanical property analysis technology for composite materials, and in particular to a method and system for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution. Background Technology
[0002] Composite material gas cylinders, with their high specific strength, lightweight, and excellent fatigue resistance, have been widely used in key pressure-bearing equipment such as aerospace propulsion systems and high-pressure hydrogen storage devices for new energy vehicles. Traditional methods for predicting the mechanical properties of composite material gas cylinders mainly rely on macroscopic empirical formulas and fatigue life models. Empirical parameters are obtained by fitting a large amount of experimental data, and then the burst pressure and fatigue life of the gas cylinder are predicted.
[0003] While macroscopic empirical methods are widely used in engineering, empirical formulas cannot reflect the actual damage evolution process of fibers, resin matrices, and their interfaces at the microscale. They simplify composite materials as homogeneous continuous media, ignoring the influence of microstructural inhomogeneities on macroscopic properties. Fatigue life models are usually based on data fitting under specific experimental conditions. When material composition, winding process, or load conditions change, the model needs to be recalibrated, resulting in limited generalization ability. Regardless of whether it is a macroscopic empirical formula or a fatigue life model, the prediction results suffer from large dispersion, making it difficult to meet the accuracy requirements of high-reliability applications. Summary of the Invention
[0004] To address the aforementioned technical problems in the existing technology, this invention provides a method and system for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution.
[0005] The first aspect of this invention provides a method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, comprising:
[0006] The microstructure parameters of the composite gas cylinder are obtained, and a microscopic representative volumetric unit model containing randomly distributed fibers, resin matrix and fiber-matrix cohesive interface is established based on the microstructure parameters.
[0007] Periodic boundary conditions are applied to the microscopic representative volume element model and multiaxial progressive loading finite element analysis is performed to extract damage initialization conditions and damage evolution laws, and homogenization processing is performed to obtain homogenized output parameters.
[0008] Using the microstructure parameters as input and the homogenized output parameters as output, a machine learning proxy model is trained to obtain a parameterized mapping model. The damage parameters output by the parameterized mapping model are then implanted into a macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model.
[0009] The macroscopic damage constitutive model is used to output the predicted failure pressure and predicted fatigue life of the composite gas cylinder.
[0010] Furthermore, the periodic boundary conditions are achieved through displacement constraint equations:
[0011] ;
[0012] in, For the positive boundary (e.g., in the microscopic representative volume element model) The corresponding node on the surface Displacement components in the direction; For the negative boundary in the microscopic representative volume element model (e.g.) The nodes on the surface that correspond one-to-one with the positive boundary nodes are in The displacement component in the direction; the corresponding nodes mentioned above refer to the node pairs in the periodic boundary that are spatially different by one RVE side length and have the same topological position; For the applied macroscopic strain tensor components; For the microscopic representative volume element model in the first... Geometric dimensions (side length) in the direction.
[0013] Furthermore, the damage initialization conditions include:
[0014] When the matrix stress satisfies the maximum principal stress criterion When, or when the interface stress satisfies the secondary stress criterion At that time, the corresponding macroscopic strain state is recorded as the damage initiation point;
[0015] in, For the principal stress of the matrix, The matrix strength limit, For the interface normal stress, For the interfacial tangential stress, For the interface normal intensity, The tangential strength of the interface.
[0016] Furthermore, the damage evolution law is expressed through damage variables. Quantify and fit the evolution equation Extract damage threshold and evolution index ;
[0017] in, To unload the modulus, For the initial modulus, This is an equivalent change.
[0018] Furthermore, the machine learning proxy model employs a fully connected neural network, wherein the number of input layer nodes of the fully connected neural network is consistent with the dimension of the microstructure parameters, the hidden layer is at least one layer, and the number of output layer nodes is consistent with the number of damage parameters to be predicted; the homogenized output parameters include equivalent elastic modulus, strength parameters, damage threshold, evolution exponent, and critical energy release rate.
[0019] Furthermore, it also includes:
[0020] Obtain actual damage data of real gas cylinders during destructive testing;
[0021] The measured damage data is compared with the simulation prediction results of the macroscopic damage constitutive model, and the parameters of the machine learning proxy model are corrected according to the comparison results to obtain a self-calibrated mechanical performance prediction model.
[0022] The predicted failure pressure and predicted fatigue life are updated using the self-calibrated mechanical property prediction model.
[0023] A second aspect of the present invention provides a system for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, comprising:
[0024] The modeling unit is used to obtain the microstructure parameters of the composite gas cylinder and to establish a microscopic representative volume element model containing randomly distributed fibers, resin matrix and fiber-matrix cohesive interface based on the microstructure parameters.
[0025] The analysis unit is used to apply periodic boundary conditions to the microscopic representative volume element model and perform multi-axis progressive loading finite element analysis to extract damage initialization conditions and damage evolution laws, and perform homogenization processing to obtain homogenized output parameters.
[0026] The surrogate model construction unit is used to train a machine learning surrogate model with the microstructure parameters as input and the homogenized output parameters as output to obtain a parameterized mapping model, and to implant the damage parameters output by the model into the macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model.
[0027] The prediction unit is used to output the predicted failure pressure and predicted fatigue life of the composite gas cylinder using the macroscopic damage constitutive model.
[0028] A third aspect of the present invention provides an electronic device including a memory, a processor, and a program stored in the memory and running on the processor, wherein the processor executes the program to implement the steps in the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in the first aspect of the present invention.
[0029] A fourth aspect of the present invention provides a computer-readable storage medium having a program stored thereon that, when executed by a processor, implements the steps in the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in the first aspect of the present invention.
[0030] A fifth aspect of the present invention provides a computer program product comprising software code, wherein the program in the software code performs the steps of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in the first aspect of the present invention.
[0031] Compared with existing technologies, the present invention provides a method and system for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, which realizes high-precision prediction of the mechanical properties of composite gas cylinders from the perspective of micro-mechanism, and solves the problem of large dispersion in prediction by traditional macro-empirical methods and fatigue life models. Attached Figure Description
[0032] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.
[0033] Figure 1 A flowchart of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution provided in Embodiment 1 of the present invention;
[0034] Figure 2 This is a schematic diagram of a representative microscopic volumetric unit model of composite materials provided in Embodiment 1 of the present invention;
[0035] Figure 3 This is a schematic diagram of the periodic boundary conditions of a representative volume element provided in Embodiment 1 of the present invention;
[0036] Figure 4 This is a three-dimensional schematic diagram of the damage initialization surface provided in Embodiment 1 of the present invention;
[0037] Figure 5 This is a comparison diagram of damage evolution curves provided in Embodiment 1 of the present invention;
[0038] Figure 6 This is a block diagram of the composite material gas cylinder mechanical property prediction system based on micro-damage evolution provided in Embodiment 2 of the present invention. Detailed Implementation
[0039] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0040] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form as well. Furthermore, it should be understood that 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.
[0041] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.
[0042] All data acquisition in this embodiment is carried out in accordance with laws and regulations and with user consent, and the data is used legally.
[0043] Example 1
[0044] like Figure 1 This embodiment provides a method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, including:
[0045] S1. Obtain the microstructure parameters of the composite material gas cylinder, and establish a system based on the microstructure parameters as follows: Figure 2 The model shown is a microscopic representative volumetric unit containing randomly distributed fibers, resin matrix, and fiber-matrix cohesive interfaces.
[0046] Specifically, this step includes the following sub-steps:
[0047] S1.1 Determine the geometric parameters of the representative volume element (hereinafter referred to as RVE): A three-dimensional model is adopted, and the RVE size is 40μm×40μm×40μm. Carbon fibers are randomly distributed in the epoxy resin matrix. The fiber volume fraction Vf is set to three groups: 40%, 50%, and 60%, and each group is achieved by adjusting the number and radius of fibers. The fiber radius is uniformly set to r=2.85μm (diameter≈5.7μm), and the cross-sectional area of a single fiber is Af=πr²≈25.5μm². When Vf=40%, the total fiber area Attotal=0.4×40×40=640μm², and the number of fibers N=640 / 25.5≈25 fibers; when Vf=50%, Attotal=800μm², and the number of fibers N≈31 fibers; when Vf=60%, Attotal=960μm², and the number of fibers N≈38 fibers. The fiber orientation angle is mainly circumferential according to the gas cylinder winding process, and the basic orientation angle is set to (This can be extended to the helical direction ±θ, with 90° as an example here). An intermolecular interface is introduced between the fiber and the matrix, using the linear traction-separation law.
[0048] S1.2 Setting Material Properties: The carbon fiber uses the HTS type, with the following elastic parameters: E11=238GPa, E22=E33=28GPa, ν12=0.28, ν23=0.33, ν31=0.02, G12=G13=24GPa, G23=7.2GPa, linear elasticity, and transverse isotropic. The epoxy resin matrix uses the RTM6 type, with elastic parameters E=3GPa and ν=0.34. Cohesive interface parameters: including interfacial shear stiffness. interface normal stiffness And the value shear strength normal intensity Type I critical energy release rate Type II .in, Shear stiffness represents the interface's ability to resist relative slip deformation under tangential stress. Normal stiffness represents the interface's ability to resist opening (separation) deformation under normal forces.
[0049] S1.3 Generating Random Fiber Distribution: In Digimat software, input the fiber volume fraction (or fiber quantity), fiber diameter, fiber orientation (typically 90°), and the spatial range of the fiber distribution. A random location generation algorithm is used, which can simulate random fiber locations using a Poisson distribution or a normal distribution. Assuming there are n fibers, their locations... and Through formula and Generate, where and For the length and width of the region, Indicates from the interval A number is randomly selected from the data. To simulate the spacing between fibers in real materials, a distance constraint is introduced to avoid fiber overlap, typically achieved using minimum spacing or contact detection algorithms. The output is the distance between each fiber. coordinate.
[0050] S1.4 Execution of modeling tools and processes: Generate random fiber distribution RVE in Digimat 2023, import it into Abaqus 2022 to assign material properties: use C3D8 elements for fibers and matrix, and use COH2D4 cohesive elements for interface, and complete mesh generation (element size 1μm to ensure accuracy).
[0051] Microstructural parameters include fiber volume fraction, fiber orientation angle, fiber diameter, and material properties of the matrix and interface. These parameters determine the macroscopic mechanical properties of the composite material. This step transforms the microstructure of the actual gas cylinder material into a digital model in a computer by establishing a microscopic RVE model. This model provides the foundation for subsequent damage analysis, solves the problem of traditional methods ignoring the randomness of microstructure, and achieves an accurate description of the microscopic geometric characteristics of the composite material.
[0052] S2. Apply periodic boundary conditions to the microscopic representative volume element model and perform finite element analysis with multi-axis progressive loading to extract damage initialization conditions and damage evolution laws, and perform homogenization processing to obtain homogenized output parameters.
[0053] Specifically, this step includes the following sub-steps:
[0054] S2.1, Apply periodic boundary conditions: such as Figure 3 As shown, this is achieved through displacement constraint equations: ,in Let RVE be the displacement of the node opposite the edge. For macroscopic strain tensor, Let be the side length of the RVE. In Abaqus, this is specifically applied through displacement constraint equations.
[0055] S2.2 Setting the loading path: Select 6 typical multiaxial loads (uniaxial tension / compression, pure shear, biaxial tension, tension-shear composite), and unify the strain rate to . .
[0056] S2.3, Define the damage initialization criterion:
[0057] Matrix damage: Maximum principal stress criterion ,in .
[0058] Interface damage: secondary stress criterion ,in For normal stress, For tangential stress, , .
[0059] S2.4 Extracting the damage evolution pattern: such as Figure 5 As shown, through damage variables Quantification, among which To unload the modulus, The initial modulus. Fitting the evolution equation. Extract damage threshold and evolution index The debonding area evolution rate is described by a power-law evolution equation based on the strain energy release rate, with the specific functional form as follows:
[0060]
[0061] in, For macroscopic stress; For interfacial mixing critical energy release rate (via and (obtained through combined calculations) This is the equivalent elastic modulus under the current condition; The characteristic length of the microscopic representative volume element (RVE); The debonding evolution coefficient; This is the debonding evolution index.
[0062] Interface hybrid equivalent critical energy release rate The calculation formula follows the Benzeggaggh-Kenane (BK) criterion:
[0063]
[0064] Parameter definition: The critical energy release rate for Type I (opening type) characterizes the interfacial tear resistance (0.002 N / mm in this embodiment). The critical energy release rate is the Type II (shear-type) value, which characterizes the interface's shear resistance (0.006 N / mm in this example). , These represent the type I and type II strain energy release rates at the interface during the current microscopic analysis step;
[0065] is the BK criterion constant, which characterizes the sensitivity of a material to mixed-mode loads (for epoxy resin systems, the value is typically in the range of 1.0 to 3.0).
[0066] S2.5 Perform homogenization and output parameters: Record the equivalent elastic modulus (E≈55-75GPa, increases with increasing Vf), strength parameters (tensile strength≈1200-1500MPa), and damage threshold. (0.008-0.012), evolutionary index (2.5-3.5) Critical energy release rate (0.002-0.006 N / mm), forming a "microscopic parameter-macroscopic response" database. Radial basis function interpolation was performed on the RVE analysis results under at least six different loading paths to construct a database as follows: Figure 4 The damage initialization three-dimensional surface is shown.
[0067] Periodic boundary conditions ensure that the deformation of the RVE model remains consistent with the overall material properties. Multiaxial progressive loading simulates the complex stress conditions of the gas cylinder in actual use. This step extracts the damage initialization conditions and damage evolution laws from the analysis results, indicating when the material begins to fail, and the evolution laws reveal how the stiffness decreases after failure. Homogenization transforms the microscopic non-uniform response into macroscopic equivalent parameters. This converts microscopic damage information into parameters usable by the macroscopic constitutive model.
[0068] S3. Using the microstructure parameters as input and the homogenized output parameters as output, train a machine learning proxy model to obtain a parameterized mapping model, and then implant the damage parameters output by the parameterized mapping model into the macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model.
[0069] Specifically, this step includes the following sub-steps:
[0070] S3.1 Constructing the dataset: Input parameters include fiber volume fraction (40%, 50%, 60%), fiber orientation angle θ (0°-180° in 5° increments, 37 angles in total), interfacial shear stiffness interface normal stiffness Output parameters include the damage initiation threshold. Damage evolution index Critical energy release rate 1000 sets of samples were generated based on microscopic RVE simulation: Take 3 values, We selected 37 values, resulting in 111 combinations. Each combination was simulated 9 times, and the average was taken, resulting in 999 sets. One boundary value was then added. The dataset was divided as follows: 70% training set (700 sets), 20% validation set (200 sets), and 10% test set (100 sets), ensuring a uniform data distribution (each...). The percentages in the training set were 40%, 30%, and 30%, respectively.
[0071] S3.2 Design the proxy model architecture: Use a fully connected neural network, with the number of input layer nodes matching the dimension of the input parameters (scalar input). Two hidden layers are set up with 20 and 10 nodes respectively, using ReLU activation function. The output layer has 3 nodes, corresponding to the damage initiation threshold, damage evolution index, and critical energy release rate. The mean squared error function is used as the loss function. Where N is the number of samples, and k=1,2,3 corresponds to 3 output parameters. For predicted values, This is the actual value.
[0072] S3.3, Evaluate the performance of the surrogate model: Evaluate the model accuracy based on the test set (100 samples): Damage threshold R² = 0.97, RMSE = 2.5 MPa; Evolutionary index R² = 0.95, RMSE = 0.08; Critical Energy Release Rate R² = 0.96, RMSE = 0.12 N / mm. Overall performance average R² = 0.96, average RMSE = 3%.
[0073] S3.4, Implanting the Macroscopic Constitutive Model: The damage parameters output by the trained surrogate model are used as input variables and passed to the user material subroutine (UMAT for static analysis, VUMAT for dynamic analysis). In the Abaqus software, UMAT / VUMAT is called to map the microscopic damage model parameters to the corresponding nodes or elements in the macroscopic structure. Specifically, the algorithm formulas and principles in the UMAT / VUMAT subroutines are as follows:
[0074] (1) Damage initialization conditions (matrix and interface)
[0075] Matrix damage initialization criterion: When matrix stress Damage initialization when the maximum principal stress criterion is satisfied: ;
[0076] Interface damage initialization criterion: when the interface stress When the secondary stress criterion is satisfied, the interface damage is initialized as follows: .
[0077] (2) Damage evolution law (cohesion model)
[0078] Damage evolution is described by the cohesive force model, assuming damage variables The degree of damage to a material is represented by the following formula, and its evolution is described by:
[0079]
[0080] in, For effective stress, In order to respond effectively, The critical energy release rate is defined as the critical energy release rate. In each analysis step, the rate of damage evolution is proportional to stress, strain, and energy release rate; the specific evolution model depends on the material's fracture mechanism.
[0081] (3) Stress-strain update (constitutive relation)
[0082] Based on the damage evolution results, the constitutive relation of the material is updated, and the new stress and strain states are calculated using the following formulas:
[0083] Updated stress: ,in It is the updated effective elastic modulus;
[0084] Updated Response: ,in It is the incremental strain calculated based on the current loading conditions.
[0085] Microscopic RVE analysis is computationally intensive and time-consuming. Machine learning surrogate models, by learning from a large amount of RVE simulation data, have established a mapping relationship between microscopic parameters and macroscopic damage parameters. After training, given any microscopic parameters, the surrogate model can quickly output the corresponding damage parameters, thus significantly accelerating the calculation. The damage parameters are then fed into the user material subroutine of the macroscopic finite element model. This subroutine determines whether damage has occurred based on the current stress and updates the material parameters accordingly, achieving rapid transfer of microscopic to macroscopic parameters.
[0086] S4. Using the macroscopic damage constitutive model, output the predicted failure pressure and predicted fatigue life of the composite gas cylinder.
[0087] Specifically, this step includes the following sub-steps:
[0088] S4.1 Determine the macroscopic structural analysis object: a high-pressure composite gas cylinder for aerospace applications, 300mm in diameter, 800mm in length, circumferential and helical winding, fiber volume fraction... =50%, Orientation Angle The material system is HTS carbon fiber + RTM6 epoxy resin.
[0089] S4.2 Simulated hydrostatic burst test: Simulated in Abaqus by hydrostatic pressure loading (0→150MPa, step size 5MPa), and the optimized macroscopic damage constitutive model (UMAT subroutine, containing damage parameters mapped by proxy model) is implanted to predict burst pressure, failure location (such as debonding of circumferential fiber bundles in the cylinder) and damage evolution cloud map.
[0090] S4.3, Perform self-calibration (optional step): Obtain measured damage data of a real gas cylinder during destructive testing (in-situ monitoring using acoustic emission and digital image correlation techniques; acoustic emission is used to record the initial pressure and energy release rate of the damage event, and digital image correlation is used to obtain the full-field strain distribution and crack propagation path on the cylinder surface). Compare the measured damage data with the simulation prediction results of the macroscopic damage constitutive model, and correct the parameters of the machine learning surrogate model based on the comparison results. Specifically, the self-calibration process includes:
[0091] (1) Proxy model output and macroscopic simulation prediction
[0092] Based on the fully connected neural network proxy model , The predicted damage parameters are: damage threshold. Evolutionary index Critical energy release rate .
[0093] After incorporating the above parameters into the Abaqus UMAT subroutine, the damage threshold predicted by the macroscopic simulation is: .
[0094] The experimental measured value is .
[0095] The initial error is: .
[0096] (2) Calculation of loss function and optimization objective
[0097] Define the loss function (mean squared error, MSE) as a measure of the error between the model's predicted values and the actual values:
[0098]
[0099] For a single trial: .
[0100] The goal is to reduce the value of the loss function by optimizing the algorithm, so that the prediction results of the surrogate model are closer to the actual experimental data.
[0101] (3) Parameter optimization algorithm: gradient descent method
[0102] To optimize the damage threshold parameters, gradient descent is chosen. The loss function is minimized by continuously calculating its gradient and adjusting the parameters. The update formula for the damage threshold is:
[0103]
[0104] in The learning rate determines the step size for each update; This is the partial derivative of the loss function with respect to the damage threshold.
[0105] (4) Numerical optimization case: Correcting the damage threshold
[0106] Let the learning rate be... Calculate the gradient of the loss function with respect to the damage threshold:
[0107]
[0108] Update damage threshold:
[0109] Recalculate the loss: The loss was significantly reduced compared to the initial loss of 900.
[0110] The gradient descent method was continued iteratively multiple times, gradually updating the damage threshold until the loss function converged. After 5 iterations, the damage threshold converged to approximately 180 MPa, and the final loss function value converged to a small constant. The performance validation data of the corrected model are as follows:
[0111]
[0112] (5) Updated proxy model performance metrics
[0113] R² (coefficient of determination) = 0.99, RMSE (root mean square error) = 1.2 MPa.
[0114] These results demonstrate that the optimized surrogate model has very high prediction accuracy and can accurately predict the damage threshold of composite gas cylinders.
[0115] After obtaining the self-calibrating mechanical property prediction model, the model is used to update the predicted failure pressure and predicted fatigue life.
[0116] Example 2
[0117] like Figure 6 This embodiment provides a composite material gas cylinder mechanical property prediction system based on micro-damage evolution, including:
[0118] The modeling unit is used to obtain the microstructure parameters of the composite gas cylinder and to establish a microscopic representative volume element model containing randomly distributed fibers, resin matrix and fiber-matrix cohesive interface based on the microstructure parameters.
[0119] The analysis unit is used to apply periodic boundary conditions to the microscopic representative volume element model and perform multi-axis progressive loading finite element analysis to extract damage initialization conditions and damage evolution laws, and perform homogenization processing to obtain homogenized output parameters.
[0120] The surrogate model construction unit is used to train a machine learning surrogate model with the microstructure parameters as input and the homogenized output parameters as output to obtain a parameterized mapping model, and to implant the damage parameters output by the model into the macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model.
[0121] The prediction unit is used to output the predicted failure pressure and predicted fatigue life of the composite gas cylinder using the macroscopic damage constitutive model.
[0122] Example 3
[0123] Embodiment 3 of the present invention provides an electronic device.
[0124] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. The processor includes, but is not limited to, at least one of a central processing unit (CPU), a graphics processing unit (GPU), a neural network processor (NPU), a tensor processor (TPU), or an artificial intelligence acceleration chip. The program is used to execute the steps in the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in Embodiment 1 of the present invention.
[0125] The detailed steps are the same as those provided in Example 1 for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, and will not be repeated here.
[0126] Example 4
[0127] Embodiment 4 of the present invention provides a computer-readable storage medium.
[0128] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps in the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in Embodiment 1 of the present invention.
[0129] The detailed steps are the same as those provided in Example 1 for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, and will not be repeated here.
[0130] Example 5
[0131] Embodiment 5 of the present invention provides a computer program product.
[0132] A computer program product includes software code, wherein the program in the software code performs the steps of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution as described in Embodiment 1 of the present invention.
[0133] The detailed steps are the same as those provided in Example 1 for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, and will not be repeated here.
[0134] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages. For example, in one implementation, the methods and systems can be developed based on deep learning frameworks (such as TensorFlow, PyTorch, etc.) and using the Python language. Those skilled in the art will understand that other suitable programming languages or tools can also be used for implementation without departing from the core ideas of the present invention.
[0135] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0136] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0137] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0138] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.
Claims
1. A method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, characterized in that, include: The microstructure parameters of the composite gas cylinder are obtained, and a microscopic representative volumetric unit model containing randomly distributed fibers, resin matrix and fiber-matrix cohesive interface is established based on the microstructure parameters. A finite element analysis with multiaxial progressive loading is performed on the microscopic representative volume element model under periodic boundary conditions to extract damage initialization conditions and damage evolution laws, and then homogenized to obtain homogenized output parameters; the periodic boundary conditions are achieved through displacement constraint equations. ； in, For the nodes on the positive boundary in the microscopic representative volume element model, Displacement components in the direction, For the nodes corresponding to the negative boundary in the microscopic representative volume element model, Displacement components in the direction, For macroscopic strain tensor, Let be the side length of the microscopic representative volume element model; The damage evolution law is expressed through damage variables. Quantify and fit the evolution equation Extract damage threshold and evolution index ; in, To unload the modulus, For the initial modulus, For equivalent change; Using the microstructure parameters as input and the homogenized output parameters as output, a machine learning proxy model is trained to obtain a parameterized mapping model. The damage parameters output by the parameterized mapping model are then implanted into a macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model. The macroscopic damage constitutive model is used to output the predicted failure pressure and predicted fatigue life of the composite gas cylinder.
2. The method according to claim 1, characterized in that, The damage initialization conditions include: When the matrix stress satisfies the maximum principal stress criterion When, or when the interface stress satisfies the secondary stress criterion At that time, the corresponding macroscopic strain state is recorded as the damage initiation point; in, For the principal stresses of the matrix, The matrix strength limit, For the interface normal stress, For the interfacial tangential stress, For the interface normal intensity, The tangential strength of the interface.
3. The method according to claim 1, characterized in that, The machine learning proxy model employs a fully connected neural network. The number of nodes in the input layer of the fully connected neural network is consistent with the dimension of the microstructure parameters, the hidden layer is at least one layer, and the number of nodes in the output layer is consistent with the number of damage parameters to be predicted. The homogenized output parameters include the equivalent elastic modulus, strength parameters, damage threshold, evolution exponent, and critical energy release rate.
4. The method according to claim 1, characterized in that, Also includes: Obtain actual damage data of real gas cylinders during destructive testing; The measured damage data is compared with the simulation prediction results of the macroscopic damage constitutive model, and the parameters of the machine learning proxy model are corrected according to the comparison results to obtain a self-calibrated mechanical performance prediction model. The predicted failure pressure and predicted fatigue life are updated using the self-calibrated mechanical property prediction model.
5. A system for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution, characterized in that, include: The modeling unit is used to obtain the microstructure parameters of the composite gas cylinder and to establish a microscopic representative volume element model containing randomly distributed fibers, resin matrix and fiber-matrix cohesive interface based on the microstructure parameters. The analysis unit is used to apply periodic boundary conditions to the microscopic representative volume element model and perform multiaxial asymptotic loading finite element analysis, extract damage initialization conditions and damage evolution laws, and perform homogenization processing to obtain homogenized output parameters; the periodic boundary conditions are implemented through displacement constraint equations: ； in, For the nodes on the positive boundary in the microscopic representative volume element model, Displacement components in the direction, For the nodes corresponding to the negative boundary in the microscopic representative volume element model, Displacement components in the direction, For macroscopic strain tensor, Let be the side length of the microscopic representative volume element model; The damage evolution law is expressed through damage variables. Quantify and fit the evolution equation Extract damage threshold and evolution index ; in, To unload the modulus, For the initial modulus, For equivalent change; The surrogate model construction unit is used to train a machine learning surrogate model with the microstructure parameters as input and the homogenized output parameters as output to obtain a parameterized mapping model, and to implant the damage parameters output by the model into the macroscopic gas cylinder finite element model to obtain a macroscopic damage constitutive model. The prediction unit is used to output the predicted failure pressure and predicted fatigue life of the composite gas cylinder using the macroscopic damage constitutive model.
6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution according to any one of claims 1 to 4.
7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution according to any one of claims 1 to 4.
8. A computer program product, comprising software code, characterized in that, The program in the software code executes the steps of the method for predicting the mechanical properties of composite gas cylinders based on micro-damage evolution according to any one of claims 1 to 4.