Zero sample prediction and numerical phase diagram generation method and system for creep behavior of metal material
By establishing a mapping relationship library between macroscopic properties of materials and creep model parameters, a dedicated creep prediction model was constructed, solving the problem of zero-sample prediction of creep behavior of new materials. This enabled rapid and accurate evaluation of creep behavior and generation of numerical phase diagrams, reducing R&D costs and time.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-17
- Publication Date
- 2026-03-31
AI Technical Summary
Existing technologies rely on time-consuming and costly experimental tests to predict the creep behavior of new materials, and data-driven models lack the ability to generalize with zero or few samples, making it difficult to achieve rapid and accurate creep behavior assessment.
By establishing a mapping relationship library between macroscopic properties of materials and creep model parameters, and utilizing the crystal structure and melting point information of the target metallic material, a dedicated creep prediction model is constructed to generate a quantitative creep numerical phase diagram, enabling rapid prediction without experimental data.
It enables rapid and accurate prediction of creep behavior of new materials, reducing R&D costs and time. The generated numerical phase diagrams can be used for service safety assessment and design of materials.
Smart Images
Figure CN121768544A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the interdisciplinary field of materials science and engineering computation, and in particular to a zero-sample prediction method and system for generating numerical phase diagrams of creep behavior in metallic materials. This is a technique for predicting creep behavior in metallic materials, specifically a zero-sample prediction method and system that integrates macroscopic material properties and physical mechanism models. Background Technology
[0002] Creep behavior of metallic materials under high temperature and sustained stress is a decisive factor in the long-term service safety of key components in aerospace, energy, and power industries. Traditional creep behavior assessment relies heavily on time-consuming and costly experimental testing. While physics-based numerical simulations (such as the finite element method) can partially replace experiments, they consume enormous computational resources, making rapid design and screening difficult.
[0003] In recent years, data-driven machine learning methods have provided new insights into creep prediction. However, these methods are often considered "black boxes," as their predictions heavily rely on training with large amounts of high-quality experimental data specific to the material. For new materials lacking experimental data or materials in the early design stages, their predictive ability is almost zero. In other words, existing data-driven models lack the ability to generalize predictions with "zero samples" or "few samples."
[0004] Therefore, there is an urgent need in this field for a new paradigm that can break through the "one material, one model" approach and achieve accurate prediction without relying on specific material creep experimental data, so as to enable rapid performance evaluation and design of new materials. Summary of the Invention
[0005] (a) Purpose of the invention
[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a zero-sample prediction method and system for creep behavior of metallic materials and generation of numerical phase diagrams. This method can achieve rapid and accurate prediction of creep behavior over a wide stress-temperature range using only readily available macroscopic properties of the material, without requiring any experimental data on the creep behavior of the material, and generate a quantitative creep numerical phase diagram.
[0007] (II) Technical Solution
[0008] A method for zero-sample prediction and numerical phase diagram generation of creep behavior in metallic materials includes the following steps:
[0009] Step S100: Establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types.
[0010] Step S200: Obtain the macroscopic property parameters of the target metallic material, wherein the macroscopic property parameters include at least its material microstructure characteristics, thermodynamic properties, and macroscopic mechanical parameters;
[0011] Step S300: Based on the crystal structure type of the target metal material, call the corresponding preset weight distribution model from the mapping relationship library; determine the creep physical model parameters based on the phase transition temperature or melting point of the target metal material; construct a dedicated creep prediction model for the target metal material based on the preset weight distribution model and the creep physical model parameters.
[0012] Step S400: Input the target stress-temperature conditions into the dedicated creep prediction model and output the predicted creep behavior of the target metallic material under the conditions;
[0013] Step S500: Within the specified full stress-temperature load space, the dedicated creep prediction model is used for continuous calculation, and the creep rate prediction results of all spatial coordinate points are integrated to generate a quantitative and continuous creep numerical phase diagram.
[0014] A zero-sample prediction and numerical phase diagram generation system for implementing the method of creep behavior of metallic materials includes:
[0015] The mapping relationship library module is used to establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types.
[0016] The parameter acquisition module is used to acquire the macroscopic property parameters of the target metallic material.
[0017] The model building module is used to call the corresponding preset weight distribution model from the mapping relationship library according to the crystal structure type of the target metal material; determine its creep physical model parameters according to the phase transition temperature or melting point of the target metal material; and construct a special creep prediction model for the target metal material based on the preset weight distribution model and creep physical model parameters.
[0018] The calculation and prediction module is used to input the target stress-temperature conditions into the dedicated creep prediction model and output the prediction results of the creep behavior of the target metallic material under the conditions.
[0019] The phase diagram generation module is used to perform continuous calculations within a specified full stress-temperature load space using a dedicated creep prediction model, integrate the creep rate prediction results of all spatial coordinate points, and then generate a quantitative and continuous creep numerical phase diagram.
[0020] A computing device includes: one or more processors; a memory for storing one or more programs; and, when the one or more programs are executed by the one or more processors, causing the one or more processors to implement the method.
[0021] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method.
[0022] (III) Beneficial Effects
[0023] Compared with the prior art, the significant advantages of the present invention include:
[0024] 1. Breakthrough zero-sample prediction capability: It can quantitatively predict the creep behavior of materials using only two easily obtainable macroscopic properties, namely the crystal structure and melting point, which greatly reduces the cost and cycle of new material research and development and material selection.
[0025] 2. Generate new data assets – numerical phase diagrams: The generated phase diagrams are continuous, quantitative, and interpretable. They not only replace traditional qualitative Ashby diagrams, but can also be directly used as digital design tools to guide the service safety assessment and microstructure design of materials under complex working conditions.
[0026] 3. Efficient model fine-tuning interface: When a small amount of new data is obtained, a flexible fine-tuning strategy is provided, which can quickly adapt a general zero-sample model into a high-precision special model for specific materials at extremely low cost. Attached Figure Description
[0027] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.
[0028] Figure 1 Flowchart of the zero-sample prediction and numerical phase diagram generation method for creep behavior of metallic materials provided in this embodiment of the invention.
[0029] Figure 2 : A schematic diagram of the average weight distribution curve of cubic crystal materials provided in the embodiments of the present invention.
[0030] Figure 3 : A schematic diagram of the linear relationship between the creep physical model parameters and the melting point of the material provided in this embodiment of the invention.
[0031] Figure 4A schematic diagram comparing the quantitative creep numerical phase diagram generated in this embodiment of the invention with the traditional qualitative Ashby diagram. Detailed Implementation
[0032] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
[0033] This invention provides a zero-sample prediction method and system for creep behavior of metallic materials and numerical phase diagram generation. The core idea is to decouple the creep behavior of materials into "common laws" and "individual parameters." The common laws (i.e., the competitive relationships between different physical mechanisms) are determined by the crystal structure type of the material; the individual parameters (i.e., the strength of each mechanism) are determined by the thermodynamic properties of the material (such as phase transition temperature or melting point). By learning these mapping relationships in advance from a known material database, a high-precision prediction model can be "assembled" for any new material.
[0034] Figure 1 This is a schematic flowchart illustrating the zero-sample prediction and numerical phase diagram generation method for the creep behavior of metallic materials according to the present invention. Figure 1 As shown, the method includes the following steps:
[0035] Step S100: Establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types.
[0036] Step S200: Obtain the macroscopic property parameters of the target metallic material. The macroscopic property parameters include at least its material microstructure characteristics (including crystal structure type, grain size, etc.), thermodynamic properties (phase transition temperature or melting point, etc.), and macroscopic mechanical parameters (shear modulus, etc.).
[0037] Step S300: Based on the crystal structure type of the target metal material, call the corresponding preset weight distribution model from the mapping relationship library; determine the creep physical model parameters based on the phase transition temperature or melting point of the target metal material; construct a dedicated creep prediction model for the target metal material based on the preset weight distribution model and the creep physical model parameters.
[0038] Step S400: Input the target stress-temperature conditions into the dedicated creep prediction model and output the predicted creep behavior of the target metal material under the conditions. The entire process does not require any creep experimental data of the target metal material.
[0039] Step S500: Within the specified full stress-temperature load space, continuous calculations are performed using a dedicated creep prediction model, integrating the creep rate prediction results of all spatial coordinate points, and then generating a quantitative and continuous creep numerical phase diagram.
[0040] In step S100, the preset weight distribution model in the mapping relationship library is an average weight distribution function obtained by statistical learning of creep experimental data of homocrystalline material groups.
[0041] In step S100, the crystal structure types include cubic, hexagonal, tetragonal, trigonal, orthorhombic, monoclinic, triclinic, and other structures; the preset weight distribution model is a Gaussian mixture normal model, and its model parameters are set according to the crystal structure type.
[0042] In step S300, the creep physical model parameters are determined based on the phase transition temperature or melting point of the target metallic material. Specifically, the creep physical model parameters are linearly related to the phase transition temperature or melting point of the material, and are obtained by linear regression of creep experimental data of various materials.
[0043] In step S300, the dedicated creep prediction model is a physical mechanism weighted network model, and its network structure is initialized based on the preset weight distribution model and creep physical model parameters.
[0044] In step S400, the creep behavior prediction result includes the steady-state creep rate. In step S500, the creep numerical phase diagram can simultaneously display the numerical distribution of the steady-state creep rate and the spatial distribution of the weights of each physical mechanism in the form of a contour map or a cloud map.
[0045] The above method may also include step S600: for cases with a small number of creep test data samples, the parameters of the dedicated creep prediction model can be fine-tuned and corrected by least squares fitting.
[0046] This invention also provides a zero-sample prediction and numerical phase diagram generation system for realizing the above methods of creep behavior of metallic materials, comprising:
[0047] The mapping relationship library module is used to establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types.
[0048] The parameter acquisition module is used to acquire the macroscopic property parameters of the target metallic material.
[0049] The model building module is used to call the corresponding preset weight distribution model from the mapping relationship library according to the crystal structure type of the target metal material; determine its creep physical model parameters according to the phase transition temperature or melting point of the target metal material; and construct a special creep prediction model for the target metal material based on the preset weight distribution model and creep physical model parameters.
[0050] The calculation and prediction module is used to input the target stress-temperature conditions into the dedicated creep prediction model and output the prediction results of the creep behavior of the target metallic material under the conditions.
[0051] The phase diagram generation module is used to perform continuous calculations within a specified full stress-temperature load space using a dedicated creep prediction model, integrate the creep rate prediction results of all spatial coordinate points, and then generate a quantitative and continuous creep numerical phase diagram.
[0052] The present invention also provides a computing device, comprising: one or more processors; a memory for storing one or more programs; wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-described method.
[0053] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.
[0054] The following is a specific example to illustrate the process of the above method:
[0055] 1. Construction of the mapping relationship library: From a pre-existing creep experiment database containing various cubic crystal system metallic materials, a Gaussian mixture model with corresponding average weight distribution is fitted through statistical learning, and a linear relationship library between creep physical model parameters and melting point is established. For example... Figure 2 As shown, five typical cubic crystal system metals—lead, aluminum, copper, iridium, and platinum—are studied at different temperatures. Based on steady-state creep rate test data under stress conditions, three stress-related physical mechanism models were selected: Stress Mechanism Model 1— Stress mechanism model 2— Stress mechanism model 3— Two physical mechanism models for temperature correlation: Temperature mechanism model 1— Temperature mechanism model 2— .in and This indicates the temperature environment and stress load that the material endures. and These represent the material's intrinsic melting point and shear modulus values, respectively, and are represented by... , , The magnitude of the steady-state creep rate caused by stress is represented by stress mechanism model 1, stress mechanism model 2 and stress mechanism model 3, respectively. and These represent two temperature mechanism models for the effect of temperature on the steady-state creep rate, namely temperature mechanism model 1 and temperature mechanism model 2. , , and These are the model parameters to be estimated. The model weight distribution function uses a Gaussian mixture distribution function, which can calculate the proportion (i.e., weight) of each stress and temperature location.
[0056] 2. Macroscopic Attribute Acquisition: By fitting the temperature-stress-creep rate test data from the experiment, the four parameters in the above model were obtained (parameter 1: Parameter 2: Parameter 3: Parameter 4: The linear correlation between the melting point of the target material (e.g., aluminum) and the melting point of the target material (e.g., aluminum metal). Figure 3 (As shown).
[0057] 3. Dedicated Model Construction: Based on the cubic crystal structure, the corresponding average weight distribution model is invoked; based on its melting point, its physical model parameters are calculated through linear relationships. These parameters are used to initialize a physical mechanism weighted network as a dedicated prediction model for aluminum.
[0058] 4. Zero-sample prediction: Input the stress-temperature conditions of interest (e.g., stress 150 MPa, temperature 500℃), and the model will instantly output its steady-state creep rate.
[0059] 5. Numerical Phase Diagram Generation: Dense mesh calculations are performed within a defined stress-temperature space (e.g., stress: 1-300 MPa, temperature: 400-600℃) to generate the creep numerical phase diagram for the aluminum material. This phase diagram, presented as a contour plot, simultaneously displays the distribution of the creep rate and the dominant regions of mechanisms such as dislocation creep and diffusion creep (e.g., ...). Figure 4 (As shown).
[0060] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.
Claims
1. A method for zero-sample prediction and numerical phase diagram generation of creep behavior in metallic materials, characterized in that, Includes the following steps: Step S100: Establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types. Step S200: Obtain the macroscopic property parameters of the target metallic material, wherein the macroscopic property parameters include at least its material microstructure characteristics, thermodynamic properties, and macroscopic mechanical parameters; Step S300: Based on the crystal structure type of the target metal material, call the corresponding preset weight distribution model from the mapping relationship library; determine the creep physical model parameters based on the phase transition temperature or melting point of the target metal material; construct a dedicated creep prediction model for the target metal material based on the preset weight distribution model and the creep physical model parameters. Step S400: Input the target stress-temperature conditions into the dedicated creep prediction model and output the predicted creep behavior of the target metallic material under the conditions; Step S500: Within the specified full stress-temperature load space, the dedicated creep prediction model is used for continuous calculation, and the creep rate prediction results of all spatial coordinate points are integrated to generate a quantitative and continuous creep numerical phase diagram.
2. The method according to claim 1, characterized in that, In step S100, the preset weight distribution model in the mapping relationship library is an average weight distribution function obtained by statistical learning of creep experimental data of homocrystalline material groups.
3. The method according to claim 2, characterized in that, In step S100, the crystal structure types include cubic, hexagonal, tetragonal, trigonal, orthorhombic, monoclinic, and triclinic crystal systems; the preset weight distribution model is a Gaussian mixture normal model, and its model parameters are set according to the crystal structure type.
4. The method according to claim 1, characterized in that, In step S300, determining the creep physical model parameters based on the phase transition temperature or melting point of the target metallic material specifically involves: the creep physical model parameters having a linear relationship with the material's phase transition temperature or melting point, obtained through linear regression of creep experimental data from various materials.
5. The method according to claim 1, characterized in that, The dedicated creep prediction model is a physical mechanism weighted network model, and its network structure is initialized based on the preset weight distribution model and the creep physical model parameters.
6. The method according to claim 1, characterized in that, In step S400, the creep behavior prediction result includes the steady-state creep rate; in step S500, the generated creep numerical phase diagram can simultaneously display the numerical distribution of the steady-state creep rate and the spatial distribution of the weights of each physical mechanism in the form of a cloud map or contour map.
7. The method according to claim 1, characterized in that, It also includes step S600: for cases with creep test data samples, the parameters of the dedicated creep prediction model are fine-tuned and corrected by least squares fitting.
8. A zero-sample prediction and numerical phase diagram generation system for implementing the method described in any one of claims 1-7 of metallic materials, characterized in that, include: The mapping relationship library module is used to establish and store a mapping relationship library between macroscopic properties of materials and creep model parameters. The mapping relationship library includes at least a preset weight distribution model corresponding to different crystal structure types. The parameter acquisition module is used to acquire the macroscopic property parameters of the target metallic material. The model building module is used to call the corresponding preset weight distribution model from the mapping relationship library according to the crystal structure type of the target metal material; determine its creep physical model parameters according to the phase transition temperature or melting point of the target metal material; and construct a special creep prediction model for the target metal material based on the preset weight distribution model and creep physical model parameters. The calculation and prediction module is used to input the target stress-temperature conditions into the dedicated creep prediction model and output the prediction results of the creep behavior of the target metallic material under the conditions. The phase diagram generation module is used to perform continuous calculations within a specified full stress-temperature load space using a dedicated creep prediction model, integrate the creep rate prediction results of all spatial coordinate points, and then generate a quantitative and continuous creep numerical phase diagram.
9. A computing device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-7.