Method for establishing and optimizing constitutive parameter multi-scale prediction model of fiber reinforced composite material

Abstract

The invention relates to a method for establishing and optimizing a multi-scale prediction model for constitutive parameters of a fiber reinforced composite material. Comprising the following steps: constructing a fiber-matrix microscopic RVE model, applying a boundary condition and a load, determining a failure criterion, and calculating equivalent mechanical parameters of a fiber bundle; a fiber bundle-matrix mesoscopic RVE model is generated, fiber bundle equivalent mechanical parameters serve as material attributes of fiber bundles in the mesoscopic RVE, a viscoelastic model is adopted for a matrix, and fiber reinforced material unit cell equivalent mechanical parameters are calculated; constructing a macroscopic fiber reinforced composite material finite element model, endowing a macroscopic model with unit cell equivalent mechanical parameters, defining an orthotropic material card, simulating an explosion impact process by adopting an explicit dynamics algorithm, outputting structural displacement, a damage area and stress distribution, and comparing calculation and test data. And correcting the microscopic RVE model and the mesoscopic RVE model through reverse optimization iteration to obtain a final macroscopic prediction model. The explosion-proof performance of the fiber composite structure is remarkably improved.

Description

Technical Field

[0001] The present invention belongs to the field of optimization of fiber - reinforced composites, and particularly relates to a method for establishing and optimizing a multi - scale prediction model of the constitutive parameters of fiber - reinforced composites. Background Art

[0002] Due to their high strength, high stiffness and lightweight characteristics, fiber - reinforced composites are widely used in the fields of aerospace, automotive manufacturing, military protection, etc. However, the mechanical properties of fiber materials are complexly affected by their micro - structures (such as fiber arrangement, fiber volume fraction) and macro - properties (such as elastic modulus, shear strength). Traditional test methods are difficult to efficiently and accurately predict their constitutive parameters, especially in the optimization of explosion - proof performance, and there are the following challenges:

[0003] 1. High test cost and long cycle: Determining the range of constitutive parameters of fiber materials through tests requires a large number of samples and complex tests, which are costly and time - consuming.

[0004] 2. Complex correlation of multi - scale properties: The macro - mechanical properties of fiber materials are closely related to their micro - structures (such as fiber - matrix interface) and meso - structures (such as weaving patterns). Traditional single - scale modeling methods are difficult to comprehensively reflect their mechanical behaviors.

[0005] 3. Difficulty in optimizing explosion - proof performance: Factors such as fiber type, fiber arrangement, fiber volume fraction and matrix type can be combined and designed according to actual needs, but their mechanical properties vary significantly, which in turn affects the explosion - proof performance of the structure under explosion loads. The application of fiber materials in explosion - proof structures requires comprehensive consideration of various mechanical parameters (such as elastic modulus, shear strength) and their effects on explosion - proof performance (such as plastic displacement of steel plates, peak velocity). The existing technologies lack systematic optimization methods. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for establishing and optimizing a multi - scale prediction model of the constitutive parameters of fiber - reinforced composites.

[0007] The technical solution for achieving the purpose of the present invention is: A method for establishing a multi - scale prediction model of the constitutive parameters of fiber - reinforced composites, comprising the following steps:

[0008] Step (1): Based on the fiber volume fraction and the fiber random distribution algorithm, construct a fiber - matrix micro - RVE model, apply boundary conditions and loads, determine the failure criterion, obtain the stress - strain curves of the micro - RVE under each applied load, and calculate the equivalent mechanical parameters of the fiber bundle;

[0009] Step (2): Generate a fiber bundle-matrix mesoscopic RVE model using the parametric modeling method. Generate finite element cube elements for the mesoscopic RVE geometric model. Use the equivalent mechanical parameters of the fiber bundle obtained in step (1) as the material properties of the fiber bundle in the mesoscopic RVE. The matrix adopts a viscoelastic model, and introduce the cohesive zone model CZM of the fiber bundle-matrix interface to describe the interface debonding behavior. Apply loads, obtain the equivalent stress-strain curve of the mesoscopic RVE, and calculate the equivalent mechanical parameters of the fiber-reinforced material unit cell;

[0010] Step (3): According to the fiber-reinforced composite structure form and the explosion boundary conditions, construct a macroscopic fiber-reinforced composite finite element model using the homogenization method. Assign the equivalent mechanical parameters of the unit cell obtained in step (2) to the macroscopic model, define the orthotropic material card, and use the explicit dynamics algorithm to simulate the explosion shock process. Output the structural displacement, damage area, and stress distribution. Compare the calculated data with the experimental data, and correct the microscopic RVE model in step (1) and the mesoscopic RVE model in step (2) through reverse optimization iteration to obtain the final macroscopic prediction model.

[0011] Further, step (1) specifically includes the following steps:

[0012] Step (11): Construct a fiber-matrix microscopic RVE model based on the fiber volume fraction and the fiber random distribution algorithm. The geometric model of the fiber-matrix microscopic RVE is a cuboid with a length-width-height ratio of 1:1:√3;

[0013] Step (12): Use structured hexahedral meshing for the fiber-matrix interface. The mesh size is determined through mesh sensitivity analysis, and the mesh quality is optimized by the symmetric cutting method;

[0014] Step (13): Automatically apply periodic boundary conditions through a Python script. The loads are longitudinal, transverse tensile and compressive, in-plane and out-of-plane shear displacement loads;

[0015] Step (14): Based on the multi-level failure criterion and the damage evolution criterion, use the finite element solver to obtain the stress-strain curve of the mesoscopic RVE under each applied load, and calculate the equivalent mechanical parameters of the fiber bundle.

[0016] Further, the fiber volume fraction in step (11) is determined by the following formula:

[0017] Use thermogravimetric analysis to obtain the overall fiber mass fraction;

[0018] Use the mixture law to correct and calculate the overall fiber volume fraction:

[0019]

[0020] where, W ffis the fiber mass fraction measured by thermogravimetry, ρ mm is the matrix density, ρ ff is the fiber density;

[0021] Calculate the fiber volume fraction of the fiber bundle:

[0022]

[0023] where is the fiber volume fraction of the mesoscopic unit cell structure.

[0024] Furthermore, in step (13), the periodic boundary conditions apply the load through the high-order periodic displacement boundary conditions, specifically as follows:

[0025] Establish a displacement relationship equation between the parallel plane node pairs (x + , x - ):

[0026]

[0027] where u mm (x + ), u mm (x - ) are the corresponding boundary node displacements, ∈ mmii is the applied strain field, η mmiiii is the high-order strain gradient term. When considering the size effect, L is the characteristic length of the RVE;

[0028] Apply the master-slave constraint to the free edges and vertex nodes, and only apply the independent constraint equations to the master nodes.

[0029] Furthermore, the multi-level failure criterion in step (14) is specifically as follows:

[0030] The fiber failure uses the maximum stress criterion to judge the longitudinal tensile failure and compressive failure, that is:

[0031] σ 11 ≥X tt ;

[0032] σ 11 ≤-X mm

[0033] X tt 、X mm are obtained through the single fiber axial tension / compression test;

[0034] The fiber bundle failure uses the three-dimensional Hashin criterion, including the tensile and compressive failures of the fiber, the tensile and compressive failures of the matrix in the transverse and normal directions, and the shear failure of the fiber bundle, specifically as:

[0035] Fiber tensile or compressive failure:

[0036]

[0037] X is the longitudinal tensile or compressive strength. When compressed, S 12 is the longitudinal shear strength;

[0038] Matrix tensile or compressive failure:

[0039]

[0040] Y is the transverse or normal tensile or compressive strength, and S is the transverse or normal shear strength;

[0041] Normal shear failure of fiber bundle:

[0042]

[0043] S N is the normal shear strength;

[0044] The damage evolution criterion is specifically as follows: The fiber adopts an instantaneous stiffness degradation model, and the matrix adopts a modified von Mises criterion, combining shear stress and hydrostatic stress:

[0045]

[0046] σ vvvv is the von Mises stress, I1 is the first stress invariant, and k1, k2 are material constants;

[0047] The equivalent mechanical parameters of the fiber bundle calculated in step (14) include: longitudinal elastic modulus E 11 、transverse elastic modulus E 22 、shear modulus G 12 、Poisson's ratio v 12 and strength parameters.

[0048] Furthermore, step (2) is specifically as follows:

[0049] Use a microscope to obtain the mesoscopic structure of plain woven fibers. Based on the geometric configuration of plain woven fibers, use a parametric modeling method to generate a fiber bundle-matrix mesoscopic RVE and calculate the mesoscopic RVE volume fraction;

[0050] Use a voxel grid division method to generate finite element cube elements with uniform shape and size from the mesoscopic RVE geometric model. Use the equivalent mechanical parameters of the fiber bundle obtained in step (1) as the material properties of the fiber bundle in the mesoscopic RVE. The matrix adopts a viscoelastic model, and a cohesive zone model CZM of the fiber bundle-matrix interface is introduced to describe the interface debonding behavior;

[0051] Using the periodic loading and damage evolution criterion in step (1), apply five displacement loads, namely longitudinal tension, longitudinal compression, out-of-plane compression, in-plane and out-of-plane shear, to the mesoscopic RVE, obtain the equivalent stress-strain curve of the mesoscopic RVE, and calculate the equivalent parameters of the fiber-reinforced material unit cell.

[0052] An optimization method for the constitutive parameters of a fiber-reinforced composite material, comprising the following steps:

[0053] Step S1: Determine the value range of the constitutive parameters of the fiber-reinforced composite material:

[0054] According to the volume fraction determination method, obtain the volume fraction range of the fiber-reinforced material;

[0055] Through the above macroscopic prediction model, obtain the value range of each constitutive parameter of the fiber-reinforced composite material at different volume fractions;

[0056] Step S2: Design an orthogonal experiment, use the above macroscopic prediction model to generate constitutive parameter combinations and explosion-proof performance data; adopt lasso regression analysis to screen key constitutive parameters;

[0057] Step S3: Construct a spatial interpolation Kriging surrogate model, use the optimal Latin hypercube to generate training samples within the value range of the key constitutive parameters, combine the above macroscopic prediction model, obtain the response data space, and fit the surrogate model with the Gaussian spatial correlation function;

[0058] Step S4: Adopt a multi-objective genetic optimization algorithm, with the minimum structural deformation and the minimum peak velocity as the objectives, to obtain the optimal combination of constitutive parameters.

[0059] Further, in step S2, the specific process of using lasso regression analysis to screen key constitutive parameters is as follows:

[0060]

[0061] Where: β is the parameter to be regressed, λ is the regularization coefficient, N is the number of samples, and p is the parameter dimension.

[0062] Further, the expression of the spatial interpolation Kriging surrogate model in step S3 is as follows:

[0063]

[0064] Compared with the prior art, the significant advantages of the present invention are:

[0065] The present invention realizes the efficient prediction from the micro-mechanical behavior of fibers and matrix to the macroscopic properties of composites by constructing multi-scale models at the micro, meso, and macro scales, combining periodic boundary conditions and failure criteria, and further establishes the connection between the constitutive parameters of fiber materials and the protective performance of composite structures by using the prediction model. Combining the multi-objective optimization method, the effective improvement from the optimal constitutive parameter combination of fiber materials to the explosion-proof performance of composite structures is achieved.

[0066] The present invention can effectively solve the problems of high cost and long cycle faced by the experimental research on the mechanical properties of fiber materials, and also solves the problem that it is difficult to effectively optimize in the explosion-proof design process of fiber composite structures because the macroscopic mechanical properties of materials are closely related to their microstructures (such as fiber-matrix interfaces) and mesostructures (such as weaving methods). Brief Description of the Drawings

[0067] Figure 1 It is the establishment process of the multi-scale prediction model of the constitutive parameters of the fiber-reinforced composite material of the present invention.

[0068] Figure 2 It is the micro and meso RVE finite element models of the fiber material of the present invention.

[0069] Figure 3 It is the schematic diagram of the application method of the periodic boundary condition of the present invention; among them, (a) local coordinate system, (b) node constraint, (c) load application. Detailed Embodiment

[0070] The present invention will be further described in detail below with reference to the drawings.

[0071] The present invention provides a method for establishing and optimizing a multi-scale prediction model of the constitutive parameters of fiber-reinforced composite materials, including a multi-scale prediction method for the constitutive parameters of fiber-reinforced composite materials and an optimization method for the constitutive parameters of fiber-reinforced composite materials.

[0072] The method for establishing a multi-scale prediction model of the constitutive parameters of fiber-reinforced composite materials includes the following steps:

[0073] S1: Predicting the equivalent mechanical parameters of fiber bundles at the microscale of fiber-reinforced composite materials, specifically including:

[0074] S11: Constructing a fiber-matrix micro RVE based on the fiber volume fraction and the fiber random distribution algorithm, and its geometric model is a cuboid with a length-width-height ratio of 1:1:√3, simulating the periodic hexagonal close-packed structure of fibers in the matrix;

[0075] The fiber volume fraction is determined by the following method:

[0076] The overall fiber mass fraction is obtained by using thermogravimetric analysis;

[0077] The overall fiber volume fraction is calculated by modifying the rule of mixtures: (W ff is the fiber mass fraction measured by thermogravimetry, ρ mm is the matrix density, ρ ff is the fiber density)

[0078] Calculate the fiber volume fraction of the fiber bundle: ( is the fiber volume fraction of the mesoscopic unit cell structure, determined by the mesoscopic RVE)

[0079] S12: Structured hexahedral meshes are used for the fiber and matrix interface, and the mesh size is determined by mesh sensitivity analysis and the mesh quality is optimized by the symmetric cutting method;

[0080] S13: Periodic boundary conditions are automatically applied through Python scripts, and the loads are longitudinal, transverse tensile and compressive, in-plane and out-of-plane shear displacement loads;

[0081] The periodic boundary conditions apply the load through high-order periodic displacement boundary conditions, specifically including:

[0082] Establish a displacement relationship equation between parallel plane node pairs (x + , x - ):

[0083] where u mm (x + ), u mm (x - ) are the corresponding boundary node displacements, ∈ mmii is the applied strain field, η mmiiii is the high-order strain gradient term (when considering the size effect, L is the characteristic length of the RVE);

[0084] The master-slave constraints are adopted for the free edges and vertex nodes, and only the master nodes are applied with independent constraint equations to eliminate the repeated constraint conflicts;

[0085] S14: Based on the multi-level failure criterion and damage evolution criterion, using the finite element solver, obtain the stress-strain curves of the mesoscopic RVE under each applied load, and calculate the equivalent mechanical parameters of the fiber bundle including the longitudinal elastic modulus E 11 , transverse elastic modulus E 22 , shear modulus G 12 , Poisson's ratio v 12 , strength parameters and other material constitutive parameters.

[0086] The multi-level failure criterion specifically includes:

[0087] The failure of fiber filaments is judged by the maximum stress criterion for longitudinal tensile failure (σ 11 ≥X tt ) and compressive failure (σ 11 ≤-X mm ), where X tt and X mm are obtained through single fiber axial tensile / compressive tests;

[0088] The failure of fiber bundles is judged by the three-dimensional Hashin criterion, including fiber tensile and compressive failures, matrix transverse and normal tensile and compressive failures, and fiber bundle shear failure. Specifically:

[0089] Fiber tensile or compressive failure:

[0090]

[0091] Matrix tensile or compressive failure:

[0092]

[0093] ≥1 (Y is the transverse or normal tensile or compressive strength, and S is the transverse or normal shear strength)

[0094] Fiber bundle normal shear failure: (S NN is the normal shear strength)

[0095] The damage evolution criterion specifically includes: the fiber adopts an instantaneous stiffness degradation model, the matrix adopts a modified von Mises criterion, combined with shear stress and hydrostatic stress:

[0096] (σ vvvv is the von Mises stress, I1 is the first stress invariant, and k1, k2 are material constants)

[0097] S2: Prediction of the equivalent mechanical parameters of the mesoscale unit cell of fiber-reinforced composites, specifically including:

[0098] S21: Observe the mesostructure of plain woven fibers using a microscope. Based on the geometric configuration of plain woven fibers, generate a fiber bundle-matrix mesoscale RVE using a parametric modeling method. The parameters include the warp and weft yarn densities, the cross-sectional shape of the fiber bundle (elliptical or rectangular), and the interlacing angle (0° to 90°), and calculate the volume fraction of the mesoscale RVE;

[0099] S22: Use the voxel grid division method to generate finite element cube elements with uniform shape and size from the mesoscale RVE geometric model. Take the equivalent mechanical parameters of the fiber bundle obtained in S1 as the material properties of the fiber bundle in the mesoscale RVE. The matrix adopts a viscoelastic model, and a cohesive zone model (CZM) of the fiber bundle-matrix interface is introduced to describe the interface debonding behavior;

[0100] S23: Apply five types of displacement loads, namely longitudinal tension, longitudinal compression, out-of-plane compression, in-plane and out-of-plane shear, to the mesoscopic RVE using the periodic loading and damage evolution criterion of S1, obtain the equivalent stress-strain curve of the mesoscopic RVE, and calculate the equivalent parameters of the fiber-reinforced material unit cell.

[0101] S3: Macroscopic scale verification of the constitutive parameters of fiber-reinforced composites, specifically including:

[0102] S31: Construct a macroscopic scale finite element model of the structure using the homogenization method according to the fiber-reinforced composite structure form and explosion boundary conditions;

[0103] S32: Assign the unit cell equivalent parameters obtained in S2 to the macroscopic model and define the orthotropic material card;

[0104] S33: Simulate the explosion shock process using the explicit dynamics algorithm and output the structure displacement, damage area and stress distribution;

[0105] S34: Compare the calculation results with the explosion test data. If the error exceeds 10%, correct the models in S1 and S2 through reverse optimization iteration.

[0106] An optimization method for the constitutive parameters of fiber-reinforced composites, comprising the following steps:

[0107] Step (1): Determine the optimization range of the constitutive parameters of fiber-reinforced composites, specifically including:

[0108] Obtain the volume fraction range of the fiber-reinforced material according to the volume fraction determination method described;

[0109] Obtain the value range of each constitutive parameter of the fiber-reinforced composite under different volume fraction ranges according to the multi-scale prediction method;

[0110] Step (2): Determine the key constitutive parameters affecting the explosion-proof performance of the fiber composite structure, specifically including:

[0111] Design an orthogonal experiment and use the macroscopic scale model described to generate constitutive parameter combinations and explosion-proof performance data;

[0112] Adopt lasso regression analysis to screen the key constitutive parameters affecting the explosion-proof performance of the composite structure:

[0113]

[0114] Where: β is the parameter to be regressed, λ is the regularization coefficient, N is the number of samples, and p is the parameter dimension

[0115] Step (3): Construct a spatial interpolation Kriging surrogate model Specifically include:

[0116] Adopt the optimal Latin hypercube to generate training samples within the range of key constitutive parameters, and combine with the macroscopic scale model described in claim 2 to obtain the response data space;

[0117] Use the Gaussian spatial correlation function to fit the surrogate model.

[0118] Step (4): Adopt the multi-objective genetic optimization algorithm, with the minimum deformation and minimum peak velocity of the structure as the objectives, to obtain the optimal combination of constitutive parameters.

[0119] Embodiment

[0120] The present invention provides a multi-scale prediction method for the constitutive parameters of fiber-reinforced composite materials. Taking the fiber type T800 as an implementation case, the selected matrix epoxy resin has a density of 1.2 g / cm 3 , and the carbon fiber has a density of 1.79 g / cm 3 . It includes the following steps:

[0121] S1: Prediction of the equivalent parameters of fiber bundles at the microscale of fiber-reinforced composite materials, specifically including:

[0122] (a) Construct a fiber-matrix micro RVE based on the fiber volume fraction and the fiber random distribution algorithm. As Figure 2 described, the ratio of the radius R of the fiber to the length of the cuboid in the obtained RVE is 0.47:1;

[0123] The fiber volume fraction: The overall fiber mass fraction is obtained by thermogravimetric analysis as 0.629, and the overall fiber volume fraction is calculated and corrected by the mixing law to be 53.2%. The fiber volume fraction of the fiber bundle is calculated to be 80%;

[0124] (b) Use structured hexahedral meshing for the fiber and matrix interface, and analyze the mesh size as 0.0005 mm through mesh sensitivity;

[0125] (c) Automatically apply periodic boundary conditions through Python scripts. The loads are longitudinal, transverse tensile and compressive (strain rate 0.001 / s, strain range 0-2%), in-plane and out-of-plane shear shear angle (shear angle 0.1 rad) displacement loads, as Figure 3 described.

[0126] (d) Based on the multi-level failure criterion and damage evolution criterion, use the finite element solver to obtain the stress-strain curves of the mesoscopic RVE under each applied load, and calculate the longitudinal elastic modulus E 11 of the fiber bundle, the transverse elastic modulus E 22 , the shear modulus G 12 , and the Poisson's ratio v 12The equivalent mechanical parameters of material constitutive parameters such as strength parameters are shown in Table 1.

[0127] Table 1 Prediction results of fiber bundle constitutive parameters

[0128]

[0129] S2: Prediction of equivalent parameters of the mesoscale unit cell of fiber-reinforced composites, specifically including:

[0130] (a) Observe the mesostructure of plain woven fibers using a microscope. Based on the geometric configuration of plain woven fibers, generate a fiber bundle-matrix mesoscale RVE using a parametric modeling method. As Figure 2 described, the warp and weft yarn density is 1.79 g / cm 3 , the cross-sectional shape of the fiber bundle (elliptical) and the interlacing angle (90°), and calculate the volume fraction of the mesoscale RVE to be 66.5%;

[0131] (b) Use the voxel grid division method to generate finite element cube elements (0.2 mm × 0.2 mm × 0.02 mm) with uniform shape and size from the mesoscale RVE geometric model. Use the equivalent mechanical parameters of the fiber bundle obtained in S1 as the material properties of the fiber bundle in the mesoscale RVE. The matrix uses a viscoelastic model, and a fiber bundle-matrix interface cohesive force model (CZM) is introduced to describe the interface debonding behavior;

[0132] (c) Apply five displacement loads of longitudinal tension, longitudinal compression, out-of-plane compression, in-plane and out-of-plane shear to the mesoscale RVE using the periodic loading and damage evolution criteria described in S1, obtain the equivalent stress-strain curve of the mesoscale RVE, and calculate the equivalent parameters of the unit cell of the fiber-reinforced material, as shown in Table 2.

[0133] Table 2 Prediction results of CFRP macroscopic constitutive parameters

[0134]

[0135]

[0136] S3: Macroscopic scale verification of the constitutive parameters of fiber-reinforced composites. Use TNT with an equivalent of 200 g as the explosion boundary condition for the implementation case, specifically including:

[0137] (a) Construct a macroscopic scale finite element model of the structure using the homogenization method according to the fiber-reinforced composite structure form and the explosion boundary condition;

[0138] (b) Assign the unit cell equivalent stiffness matrix parameters obtained in S2 to the macroscopic model and define the orthotropic material card;

[0139] (c) The explicit dynamics algorithm is used to simulate the explosion shock process, and the structural displacement, damage area and stress distribution are output;

[0140] (d) Compare the calculated results with the explosion test data, and iteratively correct until the error is less than 10%, as shown in Table 3.

[0141] Table 3 Comparison of errors between multi-scale prediction results and test results

[0142]

[0143] An optimization method for the constitutive parameters of a fiber-reinforced composite material, selecting fiber types T300 and T800 as implementation cases, with the density of the matrix epoxy resin being 1.2 g / cm 3 , and the density of carbon fiber being 1.79 g / cm 3 , including the following steps:

[0144] Step (1): Determine the optimization range of the constitutive parameters of the fiber-reinforced composite material, specifically including:

[0145] (a) According to the volume fraction determination method described above, determine the fiber-reinforced material volume fraction range of 20% - 65%;

[0146] (b): According to the multi-scale prediction method, obtain the value ranges of the constitutive parameters of the fiber-reinforced composite material under different volume fraction ranges, as shown in Table 4;

[0147] Table 4 Fiber constitutive parameter range

[0148]

[0149] Step (2): Determine the key parameters affecting the explosion-proof performance of the fiber composite structure, specifically including:

[0150] (a) Design an orthogonal experiment (11 experimental factors, each factor divided into 3 levels), and use the macroscopic scale model described above to generate constitutive parameter combinations and explosion-proof performance data;

[0151] (b) Use lasso regression analysis to screen the key constitutive parameters affecting the explosion-proof performance of the composite structure as E 11 / E 22 、G 13 、X 1T / Y 1T and S 12 :

[0152] Step (3): Construct a spatial interpolation Kriging surrogate model Specifically including:

[0153] Adopt the optimal Latin hypercube to generate 30 training samples within the range of the key constitutive parameter values described above, and combine with the above macroscopic scale model to obtain the response data space;

[0154] Use the Gaussian spatial correlation function to fit the surrogate model.

[0155] Step (4): Adopt the multi-objective genetic optimization algorithm, with the minimum structural deformation and the minimum peak velocity as the objectives, and establish an optimization model:

[0156] Min f(x)=[f D (X i ),f V (X i )]

[0157]

[0158] Obtain the optimal combination of constitutive parameters as E 11 / E 22 is 49.73 GPa, G 13 is 7.04 GPa, X 1T / Y 2T is 875.82 MPa, S 12 is 108.78 MPa.

[0159] In the present invention, the failure criterion and damage evolution criterion of the material are completed by a self-written subroutine, which effectively guarantees the prediction accuracy in establishing the constitutive parameter prediction model.

[0160] The above description of the present invention with reference to the accompanying drawings is exemplary. Obviously, the specific implementation of the present invention is not limited by the above methods. As long as this non-substantial improvement is made by adopting the method concept and technical solution of the present invention, or the concept and technical solution of the present invention are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A method for establishing a multi-scale prediction model of constitutive parameters of fiber-reinforced composite materials, characterized in that: The steps include: Step (1): construct a fiber-matrix micro RVE model based on the fiber volume fraction and fiber random distribution algorithm, apply boundary conditions and loads, determine the failure criterion, obtain the stress-strain curve of the micro RVE under each applied load, and calculate the equivalent mechanical parameters of the fiber bundle; Step (2): Generate a fiber bundle-matrix mesoscopic RVE model using a parametric modeling method, generate finite element cube elements from the mesoscopic RVE geometric model, use the fiber bundle equivalent mechanical parameters obtained in step (1) as the material properties of the fiber bundle in the mesoscopic RVE, adopt a viscoelastic model for the matrix, introduce the fiber bundle-matrix interface cohesion model CZM to describe the interface debonding behavior, apply loads, obtain the mesoscopic RVE equivalent stress-strain curves, and calculate the equivalent mechanical parameters of the fiber reinforced material unit cell; Step (3): According to the fiber-reinforced composite structure and explosion boundary conditions, a macroscopic fiber-reinforced composite material finite element model is constructed by homogenization method, the unit cell equivalent mechanical parameters obtained in step (2) are assigned to the macroscopic model, an orthotropic material card is defined, an explicit dynamics algorithm is used to simulate the explosion impact process, the structural displacement, damage area and stress distribution are output, the calculated and experimental data are compared, and the microscopic RVE model in step (1) and the mesoscopic RVE model in step (2) are iteratively corrected by inverse optimization to obtain the final macroscopic prediction model.

2. The method according to claim 1, characterized in that Step (1) specifically includes the following steps: Step (11): A fiber-matrix micro RVE model is constructed based on the fiber volume fraction and the fiber random distribution algorithm. The geometric model of the fiber-matrix micro RVE is a length-width-height ratio of A rectangular parallelepiped; Step (12): using structured hexahedral meshing for the fiber-matrix interface, the mesh size is determined by mesh sensitivity analysis, and the mesh quality is optimized by symmetric cutting method; Step (13): Periodic boundary conditions are automatically applied through Python scripts, and the loads are longitudinal, transverse tension and compression, in-plane and out-of-plane shear displacement loads; Step (14): Based on the multi-level failure criterion and the damage evolution criterion, the finite element solver is used to obtain the stress-strain curve of the mesoscopic RVE under each applied load and calculate the equivalent mechanical parameters of the fiber bundle.

3. The method according to claim 2, characterized in that The fiber volume fraction in step (11) is determined by the following formula: Thermogravimetric analysis was used to obtain the overall fiber mass fraction; The overall fiber volume fraction is calculated using the mixed law correction: Where W f is the fiber mass fraction measured by thermogravimetric method, ρ m is the matrix density, ρ f is the fiber density; Calculate the fiber volume fraction of the fiber bundle: in, is the fiber volume fraction of the mesoscopic unit cell structure.

4. The method according to claim 3, characterized in that In step (13), the periodic boundary conditions apply loads through high-order periodic displacement boundary conditions, as follows: On the parallel plane node pair (x + , x - ) to establish the displacement relationship equation: Among them, u i (x + ),u i (x - ) is the corresponding boundary node displacement, ∈ ik is the applied strain field, η ikl is a high-order strain gradient term. When the size effect is considered, L is the characteristic length of RVE; Master-slave constraints are used for free edges and vertex nodes, and independent constraint equations are only applied to the master nodes.

5. The method according to claim 4, characterized in that The multi-level failure criteria in step (14) are as follows: Fiber failure uses the maximum stress criterion to judge longitudinal tensile failure and compression failure, namely: s 11 ≥X t ; s 11 ≤-X c X t , X c Obtained through single fiber axial tension / compression test; The fiber bundle failure adopts the three-dimensional Hashin criterion, including the tensile and compressive failure of the fiber and the tensile and compressive failure of the matrix in the transverse and normal directions, as well as the shear failure of the fiber bundle, specifically: Fiber tension or compression failure: X is the longitudinal tensile or compressive strength. S 12 is the longitudinal shear strength; Matrix tensile or compressive failure: Y is the transverse or normal tensile or compressive strength, and S is the transverse or normal shear strength; Fiber bundle normal shear failure: S N is the normal shear strength; The damage evolution criterion is as follows: the fiber adopts the instantaneous stiffness degradation model, and the matrix adopts the modified von Mises criterion, combining shear stress and hydrostatic stress: σ vM is the von Mises stress, I1 is the first stress invariant, k1, k2 are material constants; The fiber bundle equivalent mechanical parameters calculated in step (14) include: longitudinal elastic modulus E 11 , transverse elastic modulus E 22 , shear modulus G 12 , Poisson's ratio v 12 and intensity parameters.

6. The method according to claim 5, characterized in that Step (2) is as follows: The mesostructure of plain woven fibers was obtained by microscope. Based on the geometric configuration of plain woven fibers, the fiber bundle-matrix mesoscopic RVE was generated by parametric modeling method, and the mesoscopic RVE volume fraction was calculated. The voxel meshing method is used to generate finite element cubic elements of uniform shape and size from the mesoscopic RVE geometric model. The fiber bundle equivalent mechanical parameters obtained in step (1) are used as the material properties of the fiber bundle in the mesoscopic RVE. The matrix adopts a viscoelastic model, and the fiber bundle-matrix interface cohesion model CZM is introduced to describe the interface debonding behavior. Using the cyclic loading and damage evolution criteria of step (1), five displacement loads, including longitudinal tension, longitudinal compression, out-of-plane compression, in-plane shear and out-of-plane shear, are applied to the mesoscopic RVE to obtain the mesoscopic RVE equivalent stress-strain curve and calculate the equivalent parameters of the fiber reinforced material unit cell.

7. A method for optimizing constitutive parameters of a fiber-reinforced composite material, characterized in that: The steps include: Step S1: Determine the value range of the constitutive parameters of the fiber reinforced composite material: According to the volume fraction determination method, the volume fraction range of the fiber reinforcement material is obtained; Obtaining the value range of each constitutive parameter of the fiber reinforced composite material at different volume fractions by using the macro prediction model described in any one of claims 1 to 6; Step S2: designing an orthogonal experiment, using the macro prediction model described in any one of claims 1 to 6 to generate constitutive parameter combinations and explosion-proof performance data; using lasso regression analysis to screen key constitutive parameters; Step S3: construct a spatial interpolation Kriging proxy model, generate training samples using the optimal Latin hypercube within the range of key constitutive parameters, obtain the response data space in combination with the macro prediction model described in any one of claims 1 to 6, and fit the proxy model using a Gaussian spatial correlation function; Step S4: Using a multi-objective genetic optimization algorithm, with the minimum structural deformation and minimum peak velocity as the goals, the optimal constitutive parameter combination is obtained.

8. The method according to claim 7, characterized in that In step S2, lasso regression analysis is used to screen the key constitutive parameters: Where: β is the parameter to be regressed, λ is the regularization coefficient, N is the number of samples, and p is the parameter dimension.

9. The method according to claim 7, characterized in that: The spatial interpolation Kriging proxy model expression in step S3 is as follows:

Images