Macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation

Abstract

The invention discloses a macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation, and the method comprises the steps: selecting a plurality of representative materials to carry out a microscopic mechanical test, and observing the damage evolution process under different loading conditions through MicroCT to obtain a microscopic mechanical test result; simulating a crack state based on a micro-mechanical model to reproduce a micro-mechanical test result, and performing material parameter correction of simulation calculation by taking test data as a constraint to obtain a micro-numerical modeling calculation result; and determining a representative volume element (RVE) scale based on a microscopic numerical modeling calculation result, and describing a macroscopic damage behavior by utilizing a continuous damage mechanical model so as to obtain evolution relation data of a macroscopic damage variable, stress and strain. According to the method, the damage behaviors of metal and composite material structures under complex working conditions can be accurately predicted, and the safety and reliability of aircraft structures can be improved.

Description

Technical Field

[0001] The present invention relates to the field of computational mechanics and material damage technology, and in particular to a macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation. Background Art

[0002] In the destruction process of metals and composite materials, the initiation and expansion of microscopic cracks in materials and their association with macroscopic failure behavior are important issues in current engineering analysis. At present, the finite element method has advantages in simulating macroscopic material behavior, but it is difficult to accurately capture the dynamic evolution of microscopic damage; on the contrary, the microscopic detection technology of fracture mechanics based on MicroCT and the micromechanical model can reveal the internal mechanism of the material, but the computational cost is high when dealing with complex macroscopic structures. Existing technologies often find it difficult to take into account both macroscopic efficiency and microscopic accuracy at the same time.

[0003] Although the commonly used damage analysis methods such as finite element analysis based on continuum damage mechanics (CDM) and fracture mechanics model have been widely used, they still have the following shortcomings:

[0004] The mapping accuracy of micro cracks to macro damage is insufficient. The influence of micro crack expansion on the evolution of macro properties of materials is analyzed by statistical damage distribution method, and a macro model is obtained based on continuous damage mechanics. However, the influence of micro cracks is shielded in the statistical process, and the direct effect of micro cracks on progressive damage cannot be directly evaluated; the crack expansion path cannot be dynamically updated. It is precisely because the generalized model shields the initiation and expansion process of micro cracks, and the crack expansion path cannot be directly calculated in the damage model, resulting in inaccurate and unstable prediction results of crack expansion; the calculation efficiency is insufficient. The use of damage models still requires a lot of computing power support. When facing complex industrial structures, there is still the problem of low calculation efficiency. Summary of the invention

[0005] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0006] In view of the problems existing in current technologies, this paper proposes a macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation. The mechanical behavior of the macroscopic structure is dynamically updated using microscopic damage parameters, thereby achieving high-precision damage prediction and path evaluation, and improving the efficiency and stability of failure analysis of complex structures.

[0007] Another object of the present invention is to propose a macroscopic progressive damage intelligent evaluation system driven by multi-scale microscopic damage observation.

[0008] To achieve the above objectives, the present invention proposes, on one hand, a macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation, comprising:

[0009] Select a variety of representative materials for micro-mechanical tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain the micro-mechanical test results;

[0010] Based on the micro-mechanical model, simulate the crack state to reproduce the micro-mechanical test results, and correct the material parameters of the simulation calculation with the test data as a constraint to obtain the micro-numerical modeling calculation results;

[0011] Based on the micro-numerical modeling calculation results, determine the representative volume element (RVE) scale, and use the continuous damage mechanics model to describe the macroscopic damage behavior to obtain the evolution relationship data of the macroscopic damage variable with stress and strain.

[0012] The macro-progressive damage intelligent evaluation method driven by multi-scale micro-damage observation in the embodiments of the present invention may further have the following additional technical features:

[0013] In an embodiment of the present invention, selecting a variety of representative materials for micro-mechanical tests, and using MicroCT to observe the damage evolution process under different loading conditions to obtain the micro-mechanical test results, including:

[0014] Select a variety of representative materials, including one or more of metal materials, composite materials, ceramic materials, and polymer materials;

[0015] Use composite materials to prepare cubic samples, and use MicroCT equipment for in-situ loading scans under different conditions, and observe and record the three-dimensional geometric characteristics of the initiation and propagation of test cracks;

[0016] Gradually increase the load according to the preset loading steps, and pause the loading between each loading step or at the key loading stage to perform in-situ MicroCT scans to obtain MicroCT images and acquire the corresponding key crack features;

[0017] Based on the image segmentation algorithm, perform threshold segmentation and morphological processing on the MicroCT images during the loading process to extract the key damage parameters.

[0018] In an embodiment of the present invention, the different conditions include constraint forms of full fixation, partial constraint, and free boundary; the in-situ loading scan includes quasi-static loading and dynamic loading; the loading steps include equal displacement loading and equal load loading; the key crack features include crack density, orientation distribution, and pore volume fraction; the key damage parameters include crack length density and local strain distribution.

[0019] In an embodiment of the present invention, based on the micro-mechanical model, simulate the crack state to reproduce the micro-mechanical test results, and correct the material parameters of the simulation calculation with the test data as a constraint to obtain the micro-numerical modeling calculation results, including:

[0020] Modeling is carried out using the AT2 phase field model, and a degradation function is adopted to describe the damage process during the entire loading stage to determine the quantity functional formula;

[0021] Calibrate the phase field opening parameter, length parameter and fracture toughness in the energy functional formula through MicroCT observation data to obtain the microscopic numerical modeling calculation results.

[0022] In an embodiment of the present invention, based on the microscopic numerical modeling calculation results, determine the representative volume element (RVE) scale, and use the continuous damage mechanics model to describe the macroscopic damage behavior to obtain the evolution relationship data of macroscopic damage variables with stress and strain, including:

[0023] Based on the microscopic numerical modeling calculation results, and using the damage factor as the judgment basis to determine the scale of the representative volume element (RVE);

[0024] Based on the representative volume element, perform multi-condition finite element-phase field coupling calculations to generate a training data set containing strain-stress sequences and damage variables;

[0025] Construct a multi-layer recurrent neural network model using gated recurrent units, and use the training data set to train the multi-layer recurrent neural network model to obtain a trained multi-layer recurrent neural network model; input the macroscopic strain tensor and historical damage state into the trained multi-layer recurrent neural network model, and the output is the equivalent damage variable and the updated stress;

[0026] According to the calculated equivalent damage variable and updated stress, use the continuous damage mechanics model to describe the damage behavior of the macroscopic structure to define the macroscopic damage variable and its evolution law.

[0027] To achieve the above object, on the other hand, the present invention proposes a macroscopic progressive damage intelligent evaluation system driven by multi-scale microscopic damage observation, including:

[0028] A microscopic mechanics test module, which is used to select a variety of representative materials for microscopic mechanics tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain microscopic mechanics test results;

[0029] A microscopic evolution modeling module, which is used to simulate the crack state based on the microscopic mechanics model to reproduce the microscopic mechanics test results, and correct the material parameters of the simulation calculation with the test data as a constraint to obtain the microscopic numerical modeling calculation results;

[0030] A macroscopic damage equivalence module, which is used to determine the representative volume element (RVE) scale based on the microscopic numerical modeling calculation results, and use the continuous damage mechanics model to describe the macroscopic damage behavior to obtain the evolution relationship data of macroscopic damage variables with stress and strain.

[0031] The macro-progressive damage intelligent evaluation method and system driven by multi-scale microscopic damage observation according to the embodiments of the present invention, aiming at the deficiencies of existing damage analysis methods, selects a variety of representative materials for microscopic mechanical tests, and uses MicroCT to observe the damage evolution process under different loading conditions. The phase field fracture theory is used for modeling, and the material parameters are corrected according to the test data. The RVE scale is determined by macroscopically equivalent calculation of microscopic parameters. Based on this, the equivalent stress, strain and damage factor are calculated, and the continuous damage mechanics model is used to describe the macroscopic damage behavior, obtaining the evolution relationship data of macroscopic damage variables with stress and strain, realizing high-precision damage prediction and path evaluation, and improving the efficiency and stability of complex structure failure analysis.

[0032] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. Brief Description of the Drawings

[0033] The above and / or additional aspects and advantages of the present invention will become apparent and be readily understood from the following description of the embodiments in conjunction with the drawings, wherein:

[0034] Figure 1 is a flowchart of the macro-progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation according to the embodiments of the present invention;

[0035] Figure 2 is a schematic diagram of calculating the RVE scale according to the evolution process of the volume-averaged damage factor with size according to the embodiments of the present invention;

[0036] Figure 3 is a schematic diagram of the realization of upscaling modeling and RNN surrogate model based on microscopic damage observation according to the embodiments of the present invention;

[0037] Figure 4 is a schematic diagram of the unit structure of the recurrent neural network RNN according to the embodiments of the present invention;

[0038] Figure 5 is a schematic diagram of the stress state distribution of the microscopic cracking damage process according to the embodiments of the present invention;

[0039] Figure 6 is a schematic diagram of the macroscopic equivalent stress-strain curve according to the embodiments of the present invention;

[0040] Figure 7 is a schematic diagram of the equivalent value of the macroscopic damage factor and the predicted value of the RNN according to the embodiments of the present invention;

[0041] Figure 8 is a structural diagram of the macro-progressive damage intelligent evaluation system driven by multi-scale microscopic damage observation according to the embodiments of the present invention. Detailed implementation manners

[0042] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0043] In order to enable those skilled in the art of the present technology to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0044] The multi-scale microscopic damage observation-driven macroscopic progressive damage intelligent evaluation method and system proposed according to the embodiments of the present invention will be described below with reference to the accompanying drawings.

[0045] Figure 1 is a flowchart of the multi-scale microscopic damage observation-driven macroscopic progressive damage intelligent evaluation method according to the embodiments of the present invention. As Figure 1 shown, the method includes:

[0046] S1, Select a variety of representative materials for microscopic mechanical tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain the microscopic mechanical test results.

[0047] Specifically, select a variety of representative materials, including metal materials (such as aluminum alloy, steel, etc.), composite materials (such as carbon fiber reinforced composite materials, glass fiber reinforced composite materials, etc.), ceramic materials (such as alumina ceramics, zirconia ceramics, etc.), and polymer materials (such as polyethylene, polypropylene, etc.). These materials cover different crystal structures, chemical bonding types, and mechanical properties, which helps to comprehensively study the microscopic mechanism of progressive failure of materials.

[0048] Use composite materials to prepare cubic samples with a size of 10 mm, and perform in-situ loading scans (including quasi-static loading: 0.001 - 0.1 mm / s; dynamic loading: 1 - 1000 mm / s) under different conditions (including constraint forms such as full fixation, partial constraint, free boundary, etc.) using a MicroCT device with a resolution ≤ 5 μm (such as Bruker SkyScan 1174), and observe and record the three-dimensional geometric characteristics of crack initiation and propagation during the test.

[0049] Increase the load step by step according to a predetermined loading procedure (such as equal displacement loading, equal load loading, etc.), and pause the loading between each loading step or at key loading stages (such as crack initiation, accelerating crack propagation, etc.) to perform in-situ MicroCT scanning to obtain key crack features, such as crack density, orientation distribution, pore volume fraction, etc. During the scanning process, the loading state should be kept stable to ensure that the acquired images can accurately reflect the internal microstructural state of the material at this loading stage. The entire loading process should continue until obvious fracture or failure of the material occurs, and the complete progressive failure process should be recorded.

[0050] Quantification of damage parameters: Based on the image segmentation algorithm, perform threshold segmentation and morphological processing on the MicroCT images during the loading process to extract key damage parameters:

[0051] Crack length density: (L i is the crack length, V RVE is the volume of the representative volume element)

[0052] Local strain field: Through Digital Volume Correlation (DVC) technology, by tracking the voxel (three-dimensional pixel) displacement field of the internal microstructure of the material before and after deformation, the local strain distribution is deduced and calculated.

[0053] S2, based on the microscopic mechanical model to simulate the crack state and reproduce the microscopic mechanical test results, and correct the material parameters of the simulation calculation with the test data as the constraint to obtain the microscopic numerical modeling calculation results.

[0054] Using the phase field fracture theory, based on the microscopic mechanical model to simulate the initiation and propagation of cracks, reproduce the microscopic mechanical test results, and with the test data as the constraint, correct the material parameters of the simulation calculation. In this case, the AT2 phase field model is used for modeling. Different from the AT1 model and the PFM-CZM model, the AT2 model uses a degradation function to describe the damage process throughout the loading stage, so the stress-strain is non-linear throughout the loading process, that is, damage occurs immediately when the load is applied. The energy functional is expressed as the following formula:

[0055]

[0056] where g(d)=(1 - d) 2 +k is the degradation function, k = 1×10 -6 to avoid numerical singularities. G c is the Griffith-type critical fracture energy release rate, that is, the material fracture toughness. The energy storage function ψ describes the volume energy stored in the object per unit volume. (In an elastic body, where \(C\) is the elastic stiffness tensor, \(\varepsilon\) is the strain tensor, and \(\Omega\) is the continuum. The damage variable of the d-phase field (\(d = 0\) for the intact material, \(d = 1\) for complete fracture), and \(l\) is the crack length parameter in the phase field theory.

[0057] Furthermore, parameter correction is carried out: calibrate the phase field opening parameter \(d\) and length parameter \(l\) with the fracture toughness \(G\) through the MicroCT observation data c to ensure that the simulated crack morphology is consistent with the experiment.

[0058] S3, determine the scale of the Representative Volume Element (RVE) based on the results of the microscopic numerical modeling calculation, and use the continuum damage mechanics model to describe the macroscopic damage behavior to obtain the evolution relationship data of the macroscopic damage variable with stress and strain.

[0059] In one embodiment of the present invention, based on the results of the microscopic numerical modeling calculation, use the damage factor as the judgment basis to determine the scale of the Representative Volume Element (RVE).

[0060] In one embodiment of the present invention, based on the representative volume element, perform multi-condition finite element-phase field coupling calculations to generate a training data set containing strain-stress sequences and damage variables.

[0061] In one embodiment of the present invention, a gated recurrent unit (GRU) is used to construct a multi-layer recurrent neural network (RNN), and the multi-layer recurrent neural network (RNN) is trained using the training data set to obtain a trained multi-layer recurrent neural network. The input of the trained multi-layer recurrent neural network is the macroscopic strain tensor \(\varepsilon\) t and the historical damage state \(D\) t-1 , and the output is the equivalent damage variable \(D\) t and the updated stress \(\sigma\) t .

[0062] In one embodiment of the present invention, according to the calculated equivalent stress and strain evolution, use the continuum damage mechanics (CDM) model to describe the damage behavior of the macroscopic structure, and define the macroscopic damage variable \(D\) and its evolution law as:

[0063] \(D = f\) NN \((\sigma,\varepsilon,t)\ (2)\)

[0064] where \(\sigma\) is the equivalent stress tensor, \(\varepsilon\) is the equivalent strain tensor, and \(t\) is the time. \(f\) NN is a neural network. That is, a rapid evaluation model of the evolution relationship between the damage variable \(D\) and the stress \(\sigma\) and strain \(\varepsilon\) in the macroscopic state is obtained.

[0065] The multi-scale finite element and microscopic evolution coupling method for progressive damage evaluation of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings:

[0066] Data Preparation

[0067] a) MicroCT Scanning and Crack Parameter Extraction

[0068] Specimen Preparation: Select a material sample to prepare a standard tensile specimen (dimensions: 10 mm × 5 mm × 2 mm), and spray graphite speckles (particle size ≤ 2 μm) on the surface for DVC analysis.

[0069] Scanning Parameters:

[0070] Use a Bruker SkyScan 1272 MicroCT device with a resolution of 4 μm, a voltage of 80 kV, a current of 125 μA, a rotation step of 0.3°, and 1200 projection numbers.

[0071] Perform in-situ tension on an Instron 5848 micro-force testing machine, and pause and scan every 0.5% engineering strain.

[0072] Extract crack geometry through threshold segmentation (Otsu algorithm) and morphological processing, and calculate the crack length density:

[0073]

[0074] where L i is the length of the i-th crack, and V is the statistical space volume, which is generally not less than 0.5 mm 3 .

[0075] b) Digital Volume Correlation (DVC) Strain Field Calculation

[0076] Displacement Field Calculation: Divide the MicroCT images in the reference state (undeformed) and the deformed state into sub-volumes of 32×32×32 voxels, and maximize the matching displacement vector u = (ux, uy, uz) through normalized cross-correlation (NCC).

[0077] Strain Field Deduction: Calculate the Green-Lagrange strain tensor based on the displacement gradient tensor F:

[0078]

[0079] Extract the maximum principal strain E max for calibrating the phase field model.

[0080] c. Phase Field Fracture Model Construction and Parameter Calibration

[0081] The crack evolution simulation scheme is established using the AT2 method phase field model, and the total energy functional is expressed by Formula 1. Using ABAQUS software, a representative volume element (RVE) model at the microscale is established. Different regions are selected, and the volume-weighted average damage value within the region is used as the average damage value. The range of the selected region is changed from small to large. When the average damage value does not change as the selection range increases, it is the RVE scale D of the macroscopic damage.

[0082] After determining the RVE scale, the spatial integration of stress and strain is performed respectively to calculate the resultant external force and deformation. Under the RVE control volume, the average stress and strain on the boundary surface are calculated as the equivalent stress and strain at the RVE scale. Figure 2 In the present invention, the RVE scale is calculated according to the evolution process of the volume average damage factor with size, as Figure 2 shown.

[0083] The crack length density d c is used as the morphology constraint parameter, and the maximum principal strain E max is used as the mechanical constraint parameter to optimize the values of the fracture toughness and the phase field crack length, so as to minimize the following comprehensive residual:

[0084]

[0085] where λ = 0.5 is the weight coefficient and M = 10 is the number of loading steps.

[0086] The overall solution is iteratively solved using the Levenberg - Marquardt algorithm, and the convergence condition is that the relative error ≤ 5%.

[0087] Figure 3 It is a flowchart for upscaling modeling based on micro - damage observation and the implementation of the RNN surrogate model.

[0088] Neural network equivalent model

[0089] a. Dataset generation

[0090] Six loading paths (uniaxial tension, plane shear, cyclic loading, etc.) are applied to the RVE, and strain - stress - damage data are calculated and generated through the phase field - finite element coupling model established and verified previously.

[0091] b. Data pre - processing

[0092] The experimental observation data of micro - damage evolution and the numerical modeling data of macro - damage are integrated and cleaned. Outliers and incorrect records in the data are removed to ensure the accuracy and reliability of the data. For example, in the experimental observation data of micro - damage, if it is found that the crack length recorded at a certain moment significantly does not conform to the physical law or is too different from the data at adjacent moments, it should be verified and corrected.

[0093] Normalize the data from different sources to make the data have a unified dimension and numerical range, which is convenient for the training of the neural network. For data such as stress and strain, the min-max method is used to normalize them to [0, 1].

[0094]

[0095] c. Neural network architecture design

[0096] Figure 4 It is the unit structure of the recurrent neural network RNN, as Figure 4 shown.

[0097] Use the recurrent neural network (RNN) to extract, train and predict the evolution data of stress, strain and damage factor. A 3-layer network is used, with 64 neurons in each layer, and the ReLU activation function is used.

[0098] Design the input layer, hidden layer and output layer of the network. The number of nodes in the input layer is consistent with the sequence length of the strain, and the number of nodes in the output layer is consistent with the sequence length of the stress plus the damage factor.

[0099] The loss function is obtained by using the mean square error weighted loss, which is expressed as follows:

[0100]

[0101] where α = 1.0 and β = 0.5 are the weight coefficients.

[0102] d. Training process

[0103] Divide the preprocessed data into a training set, a validation set and a test set. The proportions of the training set, the validation set and the test set are 70%, 20% and 10% respectively.

[0104] Use the Adam optimizer to optimize the parameters of the neural network. The learning rate is set to ε = 0.001, and the number of training epochs is set to 1000.

[0105] Use regularization to process the output data to prevent overfitting.

[0106] e. Model evaluation and optimization

[0107] 1. Finite element implementation:

[0108] Material subroutine: Embed the trained neural network into the Abaqus UMAT subroutine, and the calculation process is as follows:

[0109] Input the macroscopic strain increment Δε.

[0110] Call the neural network to predict the current stress σ and damage D.

[0111] Updated Jacobian matrix (implemented by automatic differentiation).

[0112] Failure criterion: When the damage variable D ≥ 0.95, the element is determined to have failed and is deleted.

[0113] Such as Figure 5 、 Figure 6 、 Figure 7 shown, Figure 5 is the stress state distribution of the microscopic cracking damage process, Figure 6 is the macroscopic equivalent stress-strain curve, Figure 7 is the equivalent value of the macroscopic damage factor and the RNN prediction value.

[0114] 2. Verification case (carbon fiber composite laminate)

[0115] Model parameters:

[0116] Macroscopic dimensions: 200mm × 100mm × 2mm, ply sequence [0° / 90°]6.

[0117] Boundary conditions: One end is fixed, and an in-plane tensile displacement is applied to the other end until failure.

[0118] Result comparison:

[0119] The final failure mode predicted by the neural network is consistent with the experimental crack propagation path (error ≤ 8%).

[0120] Load-displacement curve correlation coefficient.

[0121] In summary, the present invention uses a multi-scale computational model to take the propagation behavior of microscopic cracks as the macroscopic calculation basis, improving the accuracy of damage prediction. The prediction error of the microscopic crack propagation path is ≤ 8% (compared with 25% of the traditional CDM), and the dynamic update mechanism improves the coincidence degree of the macroscopic stress-strain curve with the experiment to 95%. Using a neural network to give an equivalent model of macroscopic progressive damage evolution reduces expensive high-precision calculations. The recurrent neural network reduces the multi-scale calculation time from the traditional FE 2 method's 10 4 seconds to 10 2 seconds. The model can adjust the macroscopic element stiffness and damage variable in real time to adapt to the progressive failure analysis of complex structures. It is applicable to the research on damage and failure of metals, composites and other multi-scale complex structures.

[0122] The present invention can be widely applied to many fields such as aerospace, automotive manufacturing, and ship engineering. In the aerospace field, it can accurately predict the damage behavior of metal and composite material structures under complex working conditions, improve the safety and reliability of aircraft structures, optimize structural design, reduce material waste, and lower maintenance costs. In automotive manufacturing, it helps to evaluate the durability of components under different loads and environmental conditions, guide material selection and process improvement, and extend the service life of automobiles. In ship engineering, it can analyze the damage evolution of hull structures under complex loads such as wave impacts to ensure the navigation safety of ships.

[0123] The intelligent evaluation method for macroscopic progressive damage driven by multi-scale microscopic damage observation in the embodiment of the present invention uses the phase-field fracture theory for modeling and modifies material parameters based on experimental data. The representative volume element (RVE) scale is determined through the macroscopic equivalent calculation of microscopic parameters. Based on this, the equivalent stress, strain, and damage factor are calculated, and the continuous damage mechanics model is used to describe the macroscopic damage behavior, obtaining the evolution relationship data of macroscopic damage variables with stress and strain, realizing high-precision damage prediction and path evaluation, and improving the efficiency and stability of failure analysis of complex structures.

[0124] To implement the above embodiment, as Figure 8 shown, the intelligent evaluation system 10 for macroscopic progressive damage driven by multi-scale microscopic damage observation is further provided in this embodiment, including:

[0125] The micro-mechanics test module 100 is used to select a variety of representative materials for micro-mechanics tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain micro-mechanics test results;

[0126] The micro-evolution modeling module 200 is used to simulate the crack state based on the micro-mechanics model to reproduce the micro-mechanics test results, and correct the material parameters through simulation calculations with experimental data as constraints to obtain micro-numerical modeling calculation results;

[0127] The macroscopic damage equivalence module 300 is used to determine the representative volume element (RVE) scale based on the micro-numerical modeling calculation results, and use the continuous damage mechanics model to describe the macroscopic damage behavior to obtain the evolution relationship data of macroscopic damage variables with stress and strain.

[0128] Furthermore, the micro-mechanics test module 100 is also used for:

[0129] Select a variety of representative materials, including one or more of metal materials, composite materials, ceramic materials, and polymer materials;

[0130] Select composite materials to prepare cubic samples, and use MicroCT equipment for in-situ loading scanning under different conditions to observe and record the three-dimensional geometric characteristics of the initiation and propagation of test cracks;

[0131] Gradually increase the load according to the preset loading steps, and pause the loading between each loading step or at the key loading stages to perform in-situ MicroCT scanning to obtain MicroCT images and acquire the corresponding key crack features.

[0132] Based on the image segmentation algorithm, perform threshold segmentation and morphological processing on the MicroCT images during the loading process to extract the key damage parameters.

[0133] Furthermore, different conditions include constraint forms of full fixation, partial constraint, and free boundary; in-situ loading scans include quasi-static loading and dynamic loading; loading steps include equal-displacement loading and equal-load loading; key crack features include crack density, orientation distribution, and pore volume fraction; key damage parameters include crack length density and local strain distribution.

[0134] Furthermore, the micro-evolution modeling module 200 is also used for:

[0135] Use the AT2 phase field model for modeling, and adopt a degradation function to describe the damage process throughout the loading stage to determine the quantity functional formula.

[0136] Calibrate the phase field opening parameter, length parameter, and fracture toughness in the energy functional formula through MicroCT observation data to obtain the micro-numerical modeling calculation results.

[0137] Furthermore, the macro damage equivalence module 300 is also used for:

[0138] Based on the micro-numerical modeling calculation results, and use the damage factor as the judgment basis to determine the scale of the representative volume element (RVE).

[0139] Perform multi-condition finite element-phase field coupling calculations based on the representative volume element to generate a training dataset containing strain-stress sequences and damage variables.

[0140] Use a gated recurrent unit to construct a multi-layer recurrent neural network model, and use the training dataset to train the multi-layer recurrent neural network model to obtain a trained multi-layer recurrent neural network model; input the macro strain tensor and historical damage state into the trained multi-layer recurrent neural network model, and the output is the equivalent damage variable and the updated stress.

[0141] According to the calculated equivalent damage variable and updated stress, use the continuous damage mechanics model to describe the damage behavior of the macro structure to define the macro damage variable and its evolution law.

[0142] The macroscopic progressive damage intelligent evaluation system driven by multi-scale microscopic damage observation in the embodiments of the present invention uses the phase field fracture theory for modeling and modifies the material parameters according to the test data. The RVE scale is determined by the macroscopic equivalent calculation of microscopic parameters. Based on this, the equivalent stress, strain, and damage factor are calculated. The continuous damage mechanics model is used to describe the macroscopic damage behavior, and the evolution relationship data of macroscopic damage variables with stress and strain are obtained, realizing high-precision damage prediction and path evaluation, and improving the efficiency and stability of the failure analysis of complex structures.

[0143] In the description of this specification, the description referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0144] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, "a plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

Claims

1. A macroscopic progressive damage intelligent evaluation method driven by multi-scale microscopic damage observation, characterized in that: include: Select a variety of representative materials for micromechanical tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain micromechanical test results; Simulating the crack state based on the micromechanics model to reproduce the micromechanics test results, and correcting the material parameters of the simulation calculation based on the test data to obtain the micro numerical modeling calculation results; The representative volume unit RVE scale is determined based on the microscopic numerical modeling calculation results, and the macroscopic damage behavior is described using a continuous damage mechanics model to obtain the evolution relationship data between the macroscopic damage variable and stress and strain.

2. The method according to claim 1, characterized in that A variety of representative materials were selected for micromechanical tests, and MicroCT was used to observe the damage evolution process under different loading conditions to obtain micromechanical test results, including: Select a variety of representative materials, including one or more of metal materials, composite materials, ceramic materials and polymer materials; Cube samples were prepared from composite materials, and in-situ loading scans were performed using MicroCT equipment under different conditions to observe and record the three-dimensional geometric features of crack initiation and propagation. The load is gradually increased according to the preset loading steps, and the loading is suspended between each loading step or at the critical loading stage. In-situ MicroCT scanning is performed to obtain MicroCT images and the corresponding key features of the cracks; Based on the image segmentation algorithm, the MicroCT images of the loading process were subjected to threshold segmentation and morphological processing to extract the key damage parameters.

3. The method according to claim 2, characterized in that The different conditions include fully fixed, partially constrained and free boundary constraint forms; in-situ loading scanning includes quasi-static loading and dynamic loading; loading steps include equal displacement loading and equal load loading; key crack characteristics include crack density, orientation distribution, and pore volume fraction; key damage parameters include crack length density and local strain distribution.

4. The method according to claim 1, characterized in that The micromechanical test results are reproduced by simulating the crack state based on the micromechanical model, and the material parameters of the simulation calculation are modified based on the test data to obtain the micro-numerical modeling calculation results, including: The AT2 phase field model is used for modeling, and the degradation function is used to describe the damage process in the whole loading stage to determine the quantity functional formula; The phase field opening parameter, length parameter and fracture toughness in the energy functional formula are calibrated by MicroCT observation data to obtain the microscopic numerical modeling calculation results.

5. The method according to claim 1, characterized in that Based on the calculation results of the microscopic numerical modeling, the scale of the representative volume unit RVE is determined, and the macroscopic damage behavior is described using a continuous damage mechanics model to obtain the evolution relationship data of the macroscopic damage variable and stress and strain, including: Based on the calculation results of microscopic numerical modeling, the scale of the representative volume unit RVE is determined by taking the damage factor as the judgment basis; Perform multi-case finite element-phase field coupled calculations based on representative volume elements to generate a training data set containing strain-stress series and damage variables; A multi-layer recurrent neural network model is constructed by using a gated recurrent unit, and the multi-layer recurrent neural network model is trained by using the training data set to obtain a trained multi-layer recurrent neural network model; the macroscopic strain tensor and the historical damage state are input into the trained multi-layer recurrent neural network model and output as an equivalent damage variable and an updated stress; According to the calculated equivalent damage variables and updated stress, the continuum damage mechanics model is used to describe the damage behavior of the macroscopic structure in order to define the macroscopic damage variables and their evolution law.

6. A macroscopic progressive damage intelligent evaluation system driven by multi-scale microscopic damage observation, characterized in that: include: Micromechanics test module, which is used to select a variety of representative materials for micromechanics tests, and use MicroCT to observe the damage evolution process under different loading conditions to obtain micromechanics test results; A micro-evolution modeling module is used to simulate the crack state based on a micro-mechanics model to reproduce the micro-mechanics test results, and to modify the material parameters of the simulation calculation based on the test data to obtain the micro-numerical modeling calculation results; The macro damage equivalent module is used to determine the scale of the representative volume unit RVE based on the calculation results of the micro numerical modeling, and use the continuous damage mechanics model to describe the macro damage behavior to obtain the evolution relationship data of the macro damage variable and stress and strain.

7. The system according to claim 6, characterized in that The Micromechanics Testing Module is also used for: Select a variety of representative materials, including one or more of metal materials, composite materials, ceramic materials and polymer materials; Cube samples were prepared from composite materials, and in-situ loading scans were performed using MicroCT equipment under different conditions to observe and record the three-dimensional geometric features of crack initiation and propagation. The load is gradually increased according to the preset loading steps, and the loading is suspended between each loading step or at the critical loading stage. In-situ MicroCT scanning is performed to obtain MicroCT images and the corresponding key features of the cracks; Based on the image segmentation algorithm, the MicroCT images of the loading process were subjected to threshold segmentation and morphological processing to extract the key damage parameters.

8. The system according to claim 7, characterized in that The different conditions include fully fixed, partially constrained and free boundary constraint forms; in-situ loading scanning includes quasi-static loading and dynamic loading; loading steps include equal displacement loading and equal load loading; key crack characteristics include crack density, orientation distribution, and pore volume fraction; key damage parameters include crack length density and local strain distribution.

9. The system according to claim 6, characterized in that The Microevolution Modeling Module is also used to: The AT2 phase field model is used for modeling, and the degradation function is used to describe the damage process in the whole loading stage to determine the quantity functional formula; The phase field opening parameter, length parameter and fracture toughness in the energy functional formula are calibrated by MicroCT observation data to obtain the microscopic numerical modeling calculation results.

10. The system according to claim 6, characterized in that The macro damage equivalence module is also used for: Based on the calculation results of microscopic numerical modeling, the scale of the representative volume unit RVE is determined by taking the damage factor as the judgment basis; Perform multi-case finite element-phase field coupled calculations based on representative volume elements to generate a training data set containing strain-stress series and damage variables; A multi-layer recurrent neural network model is constructed by using a gated recurrent unit, and the multi-layer recurrent neural network model is trained by using the training data set to obtain a trained multi-layer recurrent neural network model; the macroscopic strain tensor and the historical damage state are input into the trained multi-layer recurrent neural network model and output as an equivalent damage variable and an updated stress; According to the calculated equivalent damage variables and updated stress, the continuum damage mechanics model is used to describe the damage behavior of the macroscopic structure in order to define the macroscopic damage variables and their evolution law.

Images