A method for predicting multi-axial yield surface of additively manufactured metal material and reverse design of microstructure

Abstract

The application belongs to computational materials science, and discloses a kind of multi-axial yield surface prediction and structure reverse design method of additive manufacturing metal material, comprising the following steps: step 1: construct structure representative volume element model;Step 2: carry out crystal plasticity simulation, obtain stress-strain response under loading path;Step 3: based on the preset yield criterion, extract the yield point data corresponding to the loading path;Step 4: construct the multi-axial initial yield surface of additive manufacturing metal material, extract the characteristic parameters of yield surface;Step 5: construct the mapping relationship between additive manufacturing structure parameters and yield surface characteristic parameters;Step 6: reverse the target additive manufacturing structure parameters, realize the reverse design of additive manufacturing metal material structure;The application can reduce the cost of complex multi-axial experiment, realize the prediction from real structure to multi-axial yield surface of additive manufacturing metal material, and the reverse design from target yield performance to structure parameter.

Description

Technical Field

[0001] This invention relates to the field of materials computation, specifically to a method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials. Background Technology

[0002] Additive manufacturing technology can directly fabricate complex metal components through layer-by-layer deposition, offering advantages such as near-net-shape forming, high design freedom, and high material utilization. Laser powder bed melting is an important technical route in metal additive manufacturing and has been widely applied in aerospace, gas turbines, energy equipment, medical implants, and high-end equipment manufacturing.

[0003] Compared to traditionally cast, forged, or rolled metal materials, additively manufactured metal materials undergo rapid solidification, high temperature gradients, periodic molten pool overlap, and complex thermal cycling during the forming process. This readily leads to the formation of columnar crystals, strong textures, molten pool boundaries, lamellar structures, and periodic microstructures. These microstructural characteristics cause additively manufactured metal materials to exhibit significant anisotropy under different loading directions and complex stress states.

[0004] For aero-engines, gas turbines, and other high-reliability components, materials typically endure complex multiaxial loads during actual service. Obtaining yield strength solely through uniaxial tensile tests is insufficient to comprehensively evaluate the yield behavior of materials under complex stress states. The yield surface is a crucial tool for describing the critical state at which a material transitions from the elastic to the plastic stage. It reflects the yield characteristics of a material in a multidimensional stress space and is of great significance for the structural design, performance evaluation, and service safety of additively manufactured metal components.

[0005] Current methods for obtaining the yield surface of materials mainly rely on experimental techniques such as uniaxial tensile testing, biaxial tensile testing, shear testing, bulging testing, and plane strain testing. These methods suffer from drawbacks, including complex multiaxial experimental setups, high testing costs, limited stress paths, and difficulty in obtaining a complete yield surface. Furthermore, the experimental results primarily reflect the macroscopic response of the material and are insufficient to reveal the influence of additive manufacturing-specific microstructure parameters on the yield surface morphology.

[0006] In recent years, crystal plasticity simulation methods have been increasingly used to predict the anisotropic plastic deformation behavior of metallic materials. This method can link grain orientation, slip system activity, and microstructure with macroscopic mechanical responses. However, existing methods still have shortcomings in predicting the yield surface of additively manufactured metallic materials: First, some models use statistically generated or idealized microstructures, making it difficult to accurately reflect the melt pool boundaries, texture alternation, and periodic characteristics of additively manufactured materials; second, most studies focus on predicting uniaxial stress-strain curves, with few establishing quantitative mappings between microstructure parameters and complete multiaxial yield surfaces; third, existing methods typically remain at the forward analysis stage of "predicting performance given a microstructure," and cannot yet design additive manufacturing microstructure parameters and process control directions based on the target yield surface. Summary of the Invention

[0007] This invention addresses the problems existing in the prior art by providing a method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials.

[0008] The technical solution adopted in this invention is: a method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials, comprising the following steps: Step 1: Obtain additive manufacturing microstructure parameters and construct a representative volumetric unit model of the microstructure; Step 2: Perform crystal plasticity simulation on the microstructure representative volume element model obtained in Step 1 to obtain the stress-strain response under the loading path; Step 3: Based on the preset yield criterion, extract the yield point data corresponding to the loading path according to the stress-strain response obtained in Step 2; Step 4: Construct the multiaxial initial yield surface of the additive manufacturing metal material based on the yield point data, and extract the yield surface feature parameters based on the multiaxial initial yield surface of the additive manufacturing metal material; Step 5: Establish the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters; Step 6: Based on the mapping relationship, reverse the calculation of the target additive manufacturing microstructure parameters to achieve reverse design of the microstructure of additive manufacturing metal materials.

[0009] Furthermore, the additive manufacturing microstructure parameters in step 1 include grain size, grain morphology, texture type, texture volume fraction, and microstructure period parameters; the additive manufacturing microstructure parameters are obtained based on real microstructure data.

[0010] Furthermore, the tissue representative volume unit model in step 1 is obtained through electron backscatter diffraction tissue reconstruction.

[0011] Furthermore, in step 2, the crystal plasticity simulation employs either the crystal plasticity finite element method or the crystal plasticity fast Fourier transform method.

[0012] Furthermore, the loading path in step 2 includes uniaxial tensile loading, biaxial tensile loading, and shear loading.

[0013] Furthermore, the process of constructing the multiaxial initial yield surface of the additive manufacturing metal material based on the yield point data in step 4 is as follows: Based on the yield point data, the anisotropic yield function parameters are identified, and a multiaxial initial yield surface for additive manufacturing metallic materials is constructed based on the anisotropic yield function parameters.

[0014] Furthermore, the mapping relationship between the additive manufacturing microstructure parameters and the yield surface characteristic parameters in step 5 is obtained using a nonlinear fitting method.

[0015] Furthermore, the process of inversely calculating the target additive manufacturing microstructure parameters based on the target yield surface characteristic parameters in step 6 is as follows: Bayesian optimization is used to obtain the target additive manufacturing microstructure parameter vector based on the target yield surface characteristic parameters; Generative adversarial networks are used to obtain a microscopic image of the target microstructure based on the target additive manufacturing microstructure parameter vector; and the target additive manufacturing microstructure parameters are obtained based on the target microstructure image.

[0016] Furthermore, the texture type includes <001> / / BD texture <011> / / BD texture or a combination of the two, where BD is the direction of additive manufacturing construction; the organizational cycle parameters include the melt pool cycle, lamellar cycle, and texture alternation cycle formed during the additive manufacturing process.

[0017] Furthermore, the directions of the uniaxial tensile loading, biaxial tensile loading, and shear loading all include the following loading directions: 0°, 15°, 30°, 45°, 60°, 75° and 90°.

[0018] The beneficial effects of this invention are: This invention is based on additive manufacturing of metallic materials. It establishes a representative volumetric element model using real microstructure data, which can more accurately reflect the actual microstructure of additively manufactured materials. It obtains the yield response under different uniaxial, biaxial, and shear loading paths through crystal plasticity simulation, reducing the need for complex multiaxial experiments. It constructs a continuous multiaxial initial yield surface using anisotropic yield functions, which can characterize the yield behavior of additively manufactured metallic materials under complex stress states. By establishing a mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters, it achieves quantitative prediction from microstructure to negative values. By inversely calculating the target microstructure parameters based on the target yield surface, it can realize reverse design from target performance to microstructure. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow used in the embodiment.

[0020] Figure 2 The example illustrates the process of constructing a representative volumetric unit model.

[0021] Figure 3 This example illustrates the crystal plasticity simulation and the construction process of the multiaxial initial yield surface.

[0022] Figure 4 This example illustrates the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters, as well as the reverse design process. Detailed Implementation

[0023] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0024] A method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials includes the following steps: Step 1: Obtain additive manufacturing microstructure parameters and construct a representative volumetric unit model of the microstructure; Additive manufacturing microstructure parameters include grain size, grain morphology, texture type, texture volume fraction, and microstructure periodicity parameters; these parameters are obtained from actual microstructure data. Texture type includes... <001> / / BD texture <011> / / BD texture or a combination of the two, where BD is the direction of additive manufacturing construction; the organizational cycle parameters include the melt pool cycle, lamellar cycle, and texture alternation cycle formed during the additive manufacturing process.

[0025] The actual microstructure data is derived from electron backscatter diffraction (EBSD) data, which includes information on grain boundaries, grain orientation, grain size, texture distribution, and microstructure periodicity. The microstructure representative volumetric unit model is obtained through electron backscatter diffraction microstructure reconstruction.

[0026] Step 2: Perform crystal plasticity simulation on the representative volumetric element model obtained in Step 1 to obtain the stress-strain response under the loading path; the crystal plasticity simulation uses either the crystal plasticity finite element method or the crystal plasticity fast Fourier transform method. The loading path includes uniaxial tensile loading, biaxial tensile loading, and shear loading. The directions of uniaxial tensile loading, biaxial tensile loading, and shear loading all include the following loading directions: 0°, 15°, 30°, 45°, 60°, 75° and 90°.

[0027] Step 3: Based on the preset yield criterion, extract the yield point data corresponding to the loading path according to the stress-strain response obtained in Step 2; the preset yield criterion is the 0.2% plastic strain offset criterion.

[0028] Step 4: Construct the multiaxial initial yield surface of the additive manufacturing metal material based on the yield point data, and extract the yield surface feature parameters based on the multiaxial initial yield surface of the additive manufacturing metal material; Anisotropic yield function parameters were identified based on yield point data, and a multiaxial initial yield surface for additive manufacturing metallic materials was constructed based on these parameters. The anisotropic yield function is Yld2004-18p.

[0029] Step 5: Establish the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters; The mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters is obtained using a nonlinear fitting method.

[0030] Step 6: Based on the mapping relationship, reverse the calculation of the target additive manufacturing microstructure parameters to achieve reverse design of the microstructure of additive manufacturing metal materials.

[0031] Bayesian optimization is used to obtain the target additive manufacturing microstructure parameter vector based on the target yield surface characteristic parameters; Generative adversarial networks are used to obtain a microscopic image of the target microstructure based on the target additive manufacturing microstructure parameter vector; and the target additive manufacturing microstructure parameters are obtained based on the target microstructure image.

[0032] Example 1 A method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain additive manufacturing microstructure parameters and construct a representative volumetric unit model of the microstructure. In this embodiment, the additively manufactured metal is a nickel-based superalloy prepared by laser powder bed melting. The chemical composition of the nickel-based superalloy powder is Ni-10Co-7Cr-7W-4.6Al-1.2Ti-5.8Ta-1.5Mo-0.019B-0.08C, in wt.%. A crystal layered metal (CLM) sample was prepared using laser powder bed melting. The process parameters for the CLM sample were: laser power 260 W; scanning speed 1100 mm / s; scanning spacing 80 μm; powder layer thickness 30 μm; spot diameter 180 μm; and parallel reciprocating scanning strategy.

[0033] In comparison, a laser powder bed melting process was used to prepare a directional solidified microstructure (DSM) specimen. The main process parameters for the DSM specimen were: laser power 160 W; scanning interval 50 μm.

[0034] The TD-BD sections of the CLM and DSM specimens were mechanically ground, polished, and electrolytically etched. EBSD technology was used to obtain information on grain morphology, grain orientation, grain boundary distribution, and texture distribution.

[0035] Through EBSD data processing, the following additive manufacturing microstructure parameters were extracted: average grain size. D Grain aspect ratio A Texture type T Texture volume fraction V Tissue cycle parameters λ This is represented as: (1) in, This is a vector of additive manufacturing organization parameters.

[0036] For CLM organization, its main organizational characteristic is alternating arrangement. <001> / / BD texture region and <011> / / BD texture region; for DSM tissue, its main organizational characteristics are as follows: <001> / / BD-based unidirectional texture.

[0037] The process of establishing a volumetric element model representing the actual microstructure based on EBSD microstructure data includes: reading EBSD orientation data; identifying grain boundaries; establishing grain numbers; extracting grain geometric contours and crystal orientations; mapping grain geometric and orientation information to a finite element mesh; and applying periodic boundary conditions.

[0038] This yields the CLM (Constant Volumetric Model) and DSM (Distributed Volumetric Model) models corresponding to the real tissue, such as... Figure 2 As shown.

[0039] Step 2: Perform crystal plasticity simulation on the representative volumetric element model obtained in Step 1 using the crystal plasticity fast Fourier transform method to obtain the stress-strain response under the loading path; the loading path includes uniaxial tensile loading, biaxial tensile loading, and shear loading in different directions; the corresponding loading directions can be set to 0°, 15°, 30°, 45°, 60°, 75°, and 90°. Of course, the loading direction can also be selected according to the actual situation.

[0040] Step 3: Based on the preset yield criterion, using the 0.2% plastic strain offset criterion as the yield criterion, extract the yield point data corresponding to the loading path according to the stress-strain response obtained in Step 2; Step 4: Construct a multiaxial initial yield surface for the additively manufactured metallic material based on the yield point data, and extract the yield surface feature parameters based on the multiaxial initial yield surface; the yield surface feature parameters are expressed as follows: (2) In the formula: The characteristic parameter vector of the yield surface. For reference yield strength, R The ratio of the major and minor axes of the yield surface. CFor yield surface curvature parameters, S is the shear sensitivity coefficient.

[0041] Step 5: Construct the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters to obtain a quantitative correlation between the microstructure parameters and yield surface characteristic parameters; the mapping relationship can be obtained directly from the sample set using conventional nonlinear fitting methods. For example, texture type T and texture volume fraction V Combined into texture influence factor (This processing method is a standard practice in the field, used to couple the combined effects of texture type and content): (3) In the formula: , The strengthening weighting coefficients for different textures can be pre-calibrated through crystal plasticity simulation of a single texture system; for <001> / / BD texture volume fraction for <011> / / BD texture volume fraction. ( <001> / / The enhancement effect of BD texture is significantly higher than <011> / / BD texture, > ) For each component of the yield surface eigenvector, a well-known quadratic polynomial form can be used for fitting, and the coefficients are solved by the least squares method.

[0042] The specific process is as follows: Figure 3 As shown: Based on the CLM organization model, different <001> / / BD texture and <011> / / BD texture volume fraction organization model.

[0043] Set different texture proportions: : =1:1, 2:1, 3:1, 6:1 Maintain grain size D Grain aspect ratio A and tissue cycle parameters λ The basic structure remains unchanged, with only the volume fraction of the two types of textures being altered.

[0044] Crystal plasticity simulation and yield surface construction were performed on models with different texture ratios, and characteristic parameters of the yield surface were extracted: (4) In the formula: For the first k The yield surface characteristic parameters of an organization model For the first kThe reference yield surface strength of the tissue model. For the first k The ratio of the major and minor axes of the yield surface of each tissue model. For the first k The yield surface curvature parameters of the organization model For the first k Shear sensitivity coefficient of an individual tissue model.

[0045] Establish texture influencing factors: (5) In the formula: For the first k Texture influencing factors of an organizational model For the first k In an organizational model <001> / / BD texture volume fraction For the first k In an organizational model <011> / / BD texture volume fraction.

[0046] Further establish mapping relationships: (6) (7) (8) (9) In the formula: , , , This is the corresponding mapping function.

[0047] Based on the above, the mapping relationship can be obtained as follows: (10) along with <001> / / The change in BD texture volume fraction leads to regular changes in the multi-angle yield strength and yield surface morphology of additively manufactured nickel-based alloys; therefore, the texture ratio and microstructure periodic parameters can be used as key microstructure parameters for controlling the multiaxial yield surface.

[0048] Based on the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters, one-to-one corresponding data can be obtained: the internal microstructure parameters of each part, and the corresponding yield performance of this structure, such as: average grain size (average equivalent grain diameter, CLM approximately 118.34 μm, DSM approximately 84.98 μm; the finer the grain, the more dislocations are hindered and the higher the yield strength), grain aspect ratio (long axis / short axis of columnar crystals, characterizing the degree of grain elongation; the larger the ratio, the stronger the unidirectional anisotropy), texture type (divided into three categories: pure...). <001> / / BD, Pure <011> / / BD, both are alternately mixed textures; CLM is an alternating mixture, DSM is a single texture. <001> / / BD), texture volume fraction (V 001 ( <001> / / BD (grain area ratio), V 011 ( <011> / / BD grain area ratio), microstructure period parameters (layer width period of alternating melt pool, lamellar, and texture, directly controlling the density of CLM layered arrangement, with multiple ratios of 1:1 / 2:1 / 3:1 / 6:1 that can be set); by inputting the additive manufacturing microstructure parameters according to the mapping relationship, the corresponding yield surface characteristic parameters can be obtained, thus obtaining the corresponding yield effect; based on the yield surface characteristic parameters, the reasonable range that the additive manufacturing microstructure parameters cannot exceed can also be calculated; in this step, the additive manufacturing microstructure parameters can also be assigned to the metallographic image closest to the parameter in the material database. Input the obtained mapping relationship, parameter range, and metallographic image into step 6.

[0049] Step 6: Based on the mapping relationship, reverse the calculation of the target additive manufacturing microstructure parameters to achieve reverse design of the microstructure of additive manufacturing metal materials.

[0050] The direction of laser power, scanning speed, scanning spacing, powder layer thickness, or scanning strategy can also be determined based on the target additive manufacturing microstructure parameters.

[0051] Bayesian optimization and GAN were used to inversely determine the target additive manufacturing microstructure parameters: Bayesian optimization (BO) is responsible for finding the target additive manufacturing microstructure parameters in the parameter space. ,make GAN found This can be transformed into a microscopic image, or a record, that resembles the actual additive structure, for verification or use in the downstream process loop: Bayesian optimization (BO) is used to obtain the target additive manufacturing microstructure parameter vector based on the target yield surface characteristic parameters; (11) In the formula: As a search variable, each step requires a surrogate model (GP) to predict yield surface features; the BP acquisition function automatically balances search and utilization, making it much more efficient than grid search / genetic algorithms.

[0052] Generative adversarial networks are used to obtain a microscopic image of the target microstructure from the target additive manufacturing microstructure parameter vector; then, the target additive manufacturing microstructure parameters are obtained from the microscopic image. This step involves... Visualize as a tissue image / RVE; BO only provides one set of statistical parameters. Here, the GAN trains a conditional generator G(z,c): →Synthetic microstructure images / orientation fields During training, real EBSD images (metallographic images corresponding to the additive manufacturing parameters in step 5) and corresponding labeled parameters are used for conditional supervision. During inference, the input is BO and the output is... It outputs an image that "approximates the real microstructure" and converts the orientation field of the image into a readable mesh file.

[0053] The specific reverse design process is as follows: Figure 4 As shown: Target yield surface characteristic parameters for: (12) In the formula: The target parameter is yield strength. The ratio of the major and minor axes of the target yield surface. The target yield surface curvature parameter, The target shear sensitivity coefficient.

[0054] Solve the inverse problem according to equation (10): (13) Obtain the target additive manufacturing parameters: (14) In the formula: For the target grain size, For the target grain aspect ratio, For the target texture type, For the target texture volume fraction, The target organization's cycle parameters.

[0055] When the target yield surface is required to have low in-plane anisotropy and relatively balanced biaxial bearing capacity in the TD-BD plane, the target yield surface is set as a near-rectangular yield surface, and the aspect ratio of the yield surface is constrained to be close to 1. The target microstructure parameters are obtained through inverse solving, exhibiting alternating... <001> / / BD and <011> / / The CLM structure of BD texture was determined, and the corresponding texture volume fraction and structure period parameters were further determined.

[0056] After obtaining the target microstructure parameters, a representative volumetric element model is rebuilt and crystal plasticity simulation is performed for verification. If the error between the verified yield surface and the target yield surface is less than a preset threshold, the target microstructure design scheme is output; if the error is greater than the preset threshold, the mapping relationship is updated and the inverse solution is performed again.

[0057] The direction of laser powder bed melting process control can also be determined based on the target microstructure parameters. For example, if the target microstructure parameters require the formation of alternating... <001> / / BD and <011> / / The crystal lamellar structure of BD texture is controlled by adjusting the laser power, scanning speed, scanning spacing, powder layer thickness and scanning strategy to control the direction of heat flow in the molten pool, the structure period and the volume fraction of the texture, so that the formed structure approaches the target structure parameters.

[0058] This invention first acquires real microstructure data of additively manufactured metallic materials, extracting grain size, grain morphology, texture type, texture volume fraction, and microstructure periodic parameters to establish an additive manufacturing microstructure parameter vector. Then, based on the real microstructure data, a representative volume element model is built, and stress-strain responses under different loading paths are obtained through crystal plasticity simulation. Yield point data is extracted according to a preset yield criterion, and yield function parameters are identified using anisotropic yield functions to construct a multiaxial initial yield surface. Subsequently, yield surface characteristic parameters are extracted, establishing a mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters. Finally, the target microstructure parameters are derived from the target yield surface characteristic parameters. This invention solves the problems in existing technologies, such as high cost of obtaining multiaxial yield surfaces of additively manufactured metallic materials, insufficient utilization of real microstructure information, lack of quantitative mapping between microstructure parameters and yield surface characteristics, and difficulty in reverse-engineering microstructure and process parameters based on the target yield surface.

[0059] This invention enables a closed-loop design of "target yield surface → target microstructure parameters → additive manufacturing process parameter control → microstructure verification → yield surface verification," allowing for proactive design of the yield performance of additively manufactured metallic materials under complex stress states. The method is not limited to laser powder bed fusion nickel-based superalloys, but can also be applied to other additively manufactured metallic materials with significant microstructure anisotropy and periodicity, including additively manufactured titanium alloys, aluminum alloys, stainless steel, superalloys, and heterogeneous metallic materials. For materials with different crystal structures, the crystal plasticity model can be adjusted according to their slip system type; for different additive manufacturing processes, the microstructure parameter vector and mapping function form can be adjusted according to their microstructure formation characteristics. It has a wide range of applications and strong applicability.

Claims

1. A method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials, characterized in that, Includes the following steps: Step 1: Obtain additive manufacturing microstructure parameters and construct a representative volumetric unit model of the microstructure; Step 2: Perform crystal plasticity simulation on the microstructure representative volume element model obtained in Step 1 to obtain the stress-strain response under the loading path; Step 3: Based on the preset yield criterion, extract the yield point data corresponding to the loading path according to the stress-strain response obtained in Step 2; Step 4: Construct the multiaxial initial yield surface of the additive manufacturing metal material based on the yield point data, and extract the yield surface feature parameters based on the multiaxial initial yield surface of the additive manufacturing metal material; Step 5: Establish the mapping relationship between additive manufacturing microstructure parameters and yield surface characteristic parameters Step 6: Based on the mapping relationship, reverse the calculation of the target additive manufacturing microstructure parameters to achieve reverse design of the microstructure of additive manufacturing metal materials.

2. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, In step 1, the additive manufacturing microstructure parameters include grain size, grain morphology, texture type, texture volume fraction, and microstructure period parameters; these parameters are obtained based on actual microstructure data.

3. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, In step 1, the tissue representative volumetric unit model is obtained through electron backscatter diffraction tissue reconstruction.

4. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, In step 2, the crystal plasticity simulation employs either the crystal plasticity finite element method or the crystal plasticity fast Fourier transform method.

5. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, The loading paths in step 2 include uniaxial tensile loading, biaxial tensile loading, and shear loading.

6. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, The process of constructing the multiaxial initial yield surface of the additive manufacturing metallic material based on the yield point data in step 4 is as follows: Based on the yield point data, the anisotropic yield function parameters are identified, and a multiaxial initial yield surface for additive manufacturing metallic materials is constructed based on the anisotropic yield function parameters.

7. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, In step 5, the mapping relationship between the additive manufacturing microstructure parameters and the yield surface characteristic parameters is obtained using a nonlinear fitting method.

8. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 1, characterized in that, The process of inversely calculating the target additive manufacturing microstructure parameters based on the target yield surface characteristic parameters in step 6 is as follows: Bayesian optimization is used to obtain the target additive manufacturing microstructure parameter vector based on the target yield surface characteristic parameters; Generative adversarial networks are used to obtain a microscopic image of the target microstructure based on the target additive manufacturing microstructure parameter vector; and the target additive manufacturing microstructure parameters are obtained based on the target microstructure image.

9. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 2, characterized in that, The texture type includes <001> / / BD texture <011> / / BD texture or a combination of the two, where BD is the direction of additive manufacturing construction; the organizational cycle parameters include the melt pool cycle, lamellar cycle, and texture alternation cycle formed during the additive manufacturing process.

10. The method for predicting the multiaxial yield surface and reverse designing the microstructure of additively manufactured metallic materials according to claim 5, characterized in that, The directions of uniaxial tensile loading, biaxial tensile loading, and shear loading all include the following loading directions: 0°, 15°, 30°, 45°, 60°, 75° and 90°.

Images