Prediction method, device and equipment for material manufacturability and medium

Abstract

The invention relates to a method, a device and equipment for predicting the manufacturability of a material, and a medium. The method comprises the following steps: acquiring material preparation information of a to-be-predicted material; the material preparation information at least comprises component information and process information; through a plurality of material calculation modules, the microstructure parameters of the to-be-predicted material are predicted in parallel and in the forward direction, and a plurality of prediction results are obtained; analyzing the difference between different prediction results to obtain a consistency analysis result; according to a consistency analysis result, determining predicted microstructure parameters of the to-be-predicted material; according to the consistency analysis result and the material preparation information, determining a target confidence coefficient for representing the preparation degree of the to-be-predicted material; and when the target confidence coefficient is greater than a first threshold value, outputting a predicted microstructure parameter. Therefore, through mixed parallel calculation of the material calculation module and mutual verification of a plurality of prediction results, the microstructure parameter prediction precision is improved, and through confidence calculation, predicted microstructure parameters meeting user expectations are obtained.

Description

Technical Field

[0001] This application relates to the field of materials science and technology, and in particular to a method, apparatus, device and medium for predicting the fabricatability of materials. Background Technology

[0002] Multiphase metallic materials (such as steel and titanium alloys) are widely used in aerospace, energy equipment, transportation, and high-end manufacturing due to their excellent comprehensive mechanical properties and functional characteristics. The macroscopic properties of multiphase metallic materials are determined by their microstructure, including but not limited to key structural parameters such as phase composition, volume fraction of each phase, grain size, orientation distribution, and interface characteristics. Therefore, precise control of the microstructure is the core of achieving customized design and optimization of material properties.

[0003] Currently, the research and development of multiphase metallic materials mainly relies on the "trial and error method," which involves repeatedly conducting numerous experiments to predict the microstructure parameters and fabrication feasibility of materials within the parameter space of composition design and thermo-mechanical processing. This method is not only time-consuming, labor-intensive, and costly, but also limited by experimental conditions and experience accumulation. It is difficult to systematically reveal the intrinsic correlation mechanism between composition, process, microstructure, and properties, and cannot guarantee prediction accuracy, severely restricting the efficiency and innovation potential of new material development.

[0004] Therefore, how to improve the accuracy and efficiency of predicting the fabrication feasibility of multiphase metallic materials is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, one aspect of this application provides a method for predicting the prepareability of materials, the method comprising: Obtain material preparation information of the material to be predicted; the material preparation information includes at least composition information and process information. The material preparation information is analyzed in parallel using multiple material calculation modules to positively predict the microstructure parameters of the material to be predicted, and multiple prediction results are obtained. The differences between the different prediction results are analyzed to obtain the consistency analysis results; and the predicted microstructure parameters of the material to be predicted are determined based on the consistency analysis results. Based on the consistency analysis results and the material preparation information, a target confidence level is determined to characterize the fabricatability of the predicted material; the target confidence level is positively correlated with the fabricatability. When the target confidence level is greater than the first threshold, the predicted microstructure parameters are output.

[0006] Optionally, the plurality of material calculation modules include at least a thermodynamic calculation module, a kinetic calculation module, and a machine learning prediction module; The machine learning prediction module is a module that obtains prediction results by training on historical material data. The predicted microstructure parameters include at least the target phase composition and the target phase volume fraction.

[0007] Optionally, determining the predicted microstructure parameters of the material to be predicted based on the consistency analysis results includes: Determine whether the consistency analysis results indicate that the multiple prediction results meet preset difference conditions; the preset difference conditions include at least the same phase composition and the difference in volume fraction of the same phase within a preset difference range; If so, the average volume fraction of the same phase is taken as the target phase volume fraction; and the phase composition in the prediction results is integrated to obtain the target phase composition, so as to obtain the predicted microstructure parameters; If not, proceed with the following steps: Based on the process information, determine the thermodynamic state of the microstructure evolution process of the material to be predicted; If the thermodynamic state is in equilibrium, the first prediction result of the thermodynamic calculation module shall be used as the predicted microstructure parameter; If the thermodynamic state is non-equilibrium, determine whether the quality control of the second prediction results of the kinetic calculation module and the machine learning prediction module meets the preset difference condition; if it does, integrate the second prediction results to obtain the predicted microstructure parameters.

[0008] Optionally, determining the target confidence level for characterizing the fabricatability of the predicted material based on the consistency analysis results and the material preparation information includes: Based on the consistency analysis results, a target consistency score is determined; the target consistency score is used to characterize the degree of consistency among the multiple prediction results. Based on the material preparation information, an input clarity score and a reliability score are determined; the input clarity score characterizes the clarity of the material preparation information; the reliability score characterizes the reliability of the positive predictions made by the material calculation module under the conditions of the material preparation information. Based on the multiple prediction results, a conflict penalty item score is determined; the conflict penalty item score is used to characterize the interpretability of the conflict between the multiple prediction results. The target confidence level is determined based on the target consistency score, the input clarity score, the reliability score, and the conflict penalty item score.

[0009] Optionally, the prediction results include at least one of qualitative and quantitative indicators; The step of determining the target consistency score based on the consistency analysis results includes: Obtain a pre-set mapping relationship for the multiple prediction results; the mapping relationship includes the correspondence between the number of identical qualitative indicators and the first score, and the correspondence between the number of differences of identical quantitative indicators within a specified range and the second score; Based on the mapping relationship and the consistency analysis results, the target first score and the target second score are determined; Assign weights to the qualitative and quantitative indicators respectively; Based on the weights, the first target score and the second target score are weighted and summed to obtain the target consistency score.

[0010] Optionally, a method for predicting material prepareability, the method further comprising: Obtain the target microstructure parameters to be designed; Historical materials with an organization similarity greater than a similarity threshold are obtained from a pre-built material database; the organization similarity is used to characterize the degree of similarity between the historical microstructure parameters of the historical materials and the target microstructure parameters. The historical material is used as the material to be predicted; and the process proceeds to the step of obtaining the material preparation information of the material to be predicted, so as to obtain the predicted microstructure parameters. With the goal of approximating the target microstructure parameters with the predicted microstructure parameters, the historical microstructure parameters are iteratively optimized using a genetic algorithm to obtain the target material preparation information; Determine the reverse design confidence level used to characterize the reliability of the predictions for the preparation information of the target material; When the confidence level of the reverse design is greater than the second threshold, the reverse design result, which includes the target material preparation information and the predicted microstructure parameters, is output.

[0011] Optionally, determining the reverse design confidence level used to characterize the predictive reliability of the target material preparation information includes: The step of determining the target confidence level to characterize the fabrication feasibility of the material to be predicted based on the consistency analysis results and the material preparation information is performed to obtain the target confidence level; Determine the quality parameters used to characterize the iterative search quality of the genetic algorithm; Determine the safety parameters used to characterize the feasibility of the preparation information for the target material; The reverse design confidence level is determined based on the target confidence level, the quality parameter, and the safety parameter; the reverse design confidence level is positively correlated with the target confidence level, the quality parameter, and the safety parameter.

[0012] Another aspect of this application provides a device for predicting the prepareability of materials, the device comprising: A preparation information acquisition module is used to acquire material preparation information of the material to be predicted; the material preparation information includes at least composition information and process information. The forward prediction module is used to analyze the material preparation information in parallel through multiple material calculation modules to forward predict the microstructure parameters of the material to be predicted and obtain multiple prediction results. The consistency analysis module is used to analyze the differences between different prediction results to obtain consistency analysis results; and to determine the predicted microstructure parameters of the material to be predicted based on the consistency analysis results. The confidence level determination module is used to determine a target confidence level to characterize the fabricatability of the material to be predicted, based on the consistency analysis results and the material preparation information; the target confidence level is positively correlated with the fabricatability. The prediction result output module is used to output the predicted microstructure parameters when the target confidence level is greater than the first threshold.

[0013] Another aspect of this application provides an electronic device including a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the computer program to implement the steps of the method for predicting the material prepareability.

[0014] Another aspect of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for predicting the material prepareability.

[0015] The method, apparatus, equipment, and medium for predicting material prepareability provided in this application offer the following advantages: By performing hybrid parallel computation through a material calculation module, multiple prediction results are cross-validated and complementary, i.e., consistency analysis is conducted on multiple prediction results, improving the accuracy of microstructure parameter prediction. Furthermore, through confidence level calculation, high-confidence predicted microstructure parameters that meet user expectations are obtained. In addition, this method can rapidly complete the screening and prediction of a large number of materials, avoiding numerous repetitive experiments, shortening the R&D cycle, and reducing R&D costs. Attached Figure Description

[0016] Figure 1 A schematic flowchart illustrating a method for predicting material prepareability provided in an embodiment of this application; Figure 2 A schematic diagram illustrating the principle of a method for predicting the prepareability of materials provided in an embodiment of this application; Figure 3A schematic diagram illustrating the principle of a method for predicting material prepareability provided in another embodiment of this application; Figure 4 A schematic diagram illustrating the principle of a method for predicting the prepareability of materials provided in another embodiment of this application; Figure 5 This is a schematic diagram of the structure of a material prepareability prediction device provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0017] The reference numerals in the attached figures are as follows: 50 is the information acquisition module, 51 is the forward prediction module, 52 is the consistency analysis module, 53 is the first confidence level determination module, 54 is the prediction result output module, 60 is the memory, 61 is the processor, 62 is the display screen, 63 is the input / output interface, 64 is the communication interface, 65 is the power supply, 66 is the communication bus, 601 is the computer program, 602 is the operating system, and 603 is the data. Detailed Implementation

[0018] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0019] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0020] Figure 1 This is a schematic flowchart of a method for predicting the prepareability of materials provided in an embodiment of this application, as shown below. Figure 1 As shown, the method includes: S10: Obtain material preparation information for the material to be predicted; the material preparation information shall include at least composition information and process information; Figure 2 This is a schematic diagram illustrating the principle of a method for predicting material prepareability provided in an embodiment of this application. In an optional embodiment, the prediction method provided in this application can be applied to a material prepareability prediction system, see [link to relevant documentation]. Figure 2 In a specific embodiment, the system includes a user input layer, a prediction unit, and a result output layer. An analysis engine is deployed in the prediction unit to analyze and predict the user-input data.

[0021] In a specific embodiment, the user inputs material preparation information of the material to be predicted through the user input layer, such as... Figure 2 As shown, the material preparation information is transmitted to the analysis engine for analysis via the input interface module. This material preparation information may include, but is not limited to, composition information and process information.

[0022] For example, when the material to be predicted is TC4 titanium alloy (Ti-6Al-4V), the composition information entered by the user in the material preparation information can include Al (aluminum) 6%, V (vanadium) 4%, and Ti (titanium) 90%. The process information includes a solution treatment temperature of 950 degrees Celsius (°C), a holding time of 1 hour, and a cooling method of water quenching (cooling rate of approximately >400°C / s).

[0023] S11: Through multiple material calculation modules, material preparation information is analyzed in parallel to positively predict the microstructure parameters of the material to be predicted and obtain multiple prediction results; Understandably, knowing the material preparation information of the material to be predicted aims to determine the microstructure parameters of the material and whether the corresponding material can be successfully prepared using this material preparation information; this process is called forward prediction. In a specific embodiment, to ensure prediction accuracy, the acquired material preparation information is analyzed and processed in parallel by multiple different material calculation modules.

[0024] See Figure 2 The analysis engine calls multiple different material calculation modules to perform parallel calculations and obtain multiple prediction results. It can be understood that each material calculation module can output a prediction result, so the number of prediction results is the same as the number of material calculation modules.

[0025] It should be noted that, in specific embodiments, the materials calculation module may include, but is not limited to, a thermodynamic calculation module, a kinetic calculation module, a machine learning prediction module, and a principle calculation module. This application does not limit the number of materials calculation modules.

[0026] S12: Analyze the differences between different prediction results to obtain consistency analysis results; and determine the predicted microstructure parameters of the material to be predicted based on the consistency analysis results. It is understandable that multiple different material calculation modules perform calculations in parallel, and the resulting predictions may be completely consistent or different. To further ensure prediction accuracy, in one optional embodiment, verification is performed between the multiple prediction results. Specifically, through... Figure 2 The analysis engine shown analyzes the differences between the prediction results output by different material calculation modules to obtain consistency analysis results.

[0027] If all predictions are consistent, highly reliable predicted microorganism parameters can be output. If multiple predictions differ, arbitration is required. This can be achieved by using a larger number of predictions as the predicted microorganism parameters, or by analyzing multiple predictions using a large model to obtain the predicted microorganism parameters. This application does not limit the specific arbitration and verification methods.

[0028] In specific embodiments, the predicted microstructure parameters may include, but are not limited to, phase composition, phase volume fraction, grain size, and morphology of precipitated phases.

[0029] S13: Based on the consistency analysis results and material preparation information, determine the target confidence level used to characterize the fabricatability of the material to be predicted; the target confidence level is positively correlated with the fabricatability. S14: When the target confidence level is greater than the first threshold, output the predicted microstructure parameters.

[0030] To further ensure the reliability of the output predicted microstructure parameters, in an optional embodiment, the target confidence level of the material to be predicted is calculated based on the above-described consistency analysis structure and the initially acquired material preparation information. The target confidence level reflects the degree of fabrication feasibility of the material to be predicted using the input material preparation information. The higher the target confidence level, the higher the degree of fabrication feasibility.

[0031] When the target confidence level is higher than the first threshold, it indicates that the material to be predicted can be prepared using the material preparation information input by the user, and the expected results can be achieved. See also Figure 2 At this point, the predicted microstructure parameters are output to the result output layer via the output interface module, allowing users to view the positive prediction results. In an optional embodiment, the positive prediction results may include, in addition to the predicted microstructure parameters, the result of whether the material to be predicted is fabricable, and the target confidence level.

[0032] Therefore, the material fabrication prediction method provided in this application embodiment uses a material calculation module for hybrid parallel computation, where multiple prediction results are mutually verified and complementary, i.e., consistency analysis is performed on multiple prediction results to improve the prediction accuracy of microstructure parameters. Furthermore, through confidence level calculation, high-confidence predicted microstructure parameters that meet user expectations are obtained. In addition, this method can quickly complete the screening and prediction of a large number of materials, avoiding numerous repeated experiments, shortening the R&D cycle, and reducing R&D costs.

[0033] In one optional embodiment, the multiple material calculation modules include at least a thermodynamic calculation module, a kinetic calculation module, and a machine learning prediction module; wherein the machine learning prediction module is a module that obtains prediction results by training on historical material data. Furthermore, in one optional embodiment, the predicted microstructure parameters include at least the target phase composition and the target phase volume fraction. In addition, information such as grain size and precipitated phase morphology may also be included, which is not limited in this application.

[0034] In a specific embodiment, the thermodynamic calculation module can calculate the phase composition and phase volume fraction at equilibrium based on the composition information. The thermodynamic calculation module can use the CALPHAD method for calculation, and can be configured to call thermodynamic data for equilibrium phase calculation. Figure 2 The database shown includes thermodynamic data.

[0035] The kinetic calculation module can calculate the phase transition process and microstructure evolution under non-equilibrium conditions based on composition and process information. In one optional embodiment, the kinetic calculation module can perform calculations based on at least one of the following: TTT diagram, CCT diagram, phase-field method, or JMAK equations. Figure 2 The database also includes dynamic data.

[0036] In a specific embodiment, the machine learning prediction module is trained based on historical material data and is used to perform data-driven predictions of microstructure parameters. Figure 2 The database also includes a large amount of historical material data, including but not limited to historical composition information, historical process information, and historical microscopic parameter organization.

[0037] Apart from Figure 2 In addition to the three modules shown, multiple material calculation modules may also include a principle calculation module, which can provide key thermodynamic data (e.g., formation energy) for thermodynamic calculations and parameters such as migration barriers and interface energies for kinetic calculation modules (e.g., phase field method). This can improve the prediction accuracy and reliability of thermodynamic and kinetic calculation modules, which is especially crucial for prediction accuracy in unknown systems lacking experimental data.

[0038] Figure 3 This is a schematic diagram illustrating the principle of a method for predicting the prepareability of a material, provided as another embodiment of this application. Based on the above embodiment, as an optional embodiment, the predicted microstructure parameters of the material to be predicted are determined according to the consistency analysis results, including: Determine whether the consistency analysis results indicate that multiple prediction results meet the preset difference conditions; the preset difference conditions include at least the same phase composition and the difference in volume fraction of the same phase within the preset difference range. If so, the average volume fraction of the same phase is taken as the target phase volume fraction; and the phase composition in the prediction results is integrated to obtain the target phase composition, so as to obtain the predicted microstructure parameters. If not, proceed with the following steps: Based on the process information, determine the thermodynamic state of the microstructure evolution process of the material to be predicted; If the thermodynamic state is in equilibrium, the first prediction result of the thermodynamic calculation module will be used as the predicted microstructure parameters. If the thermodynamic state is non-equilibrium, determine whether the second prediction results of the kinetic calculation module and the machine learning prediction module meet the preset difference conditions; if they do, integrate the second prediction results to obtain the predicted microstructure parameters.

[0039] See Figure 3 The material preparation information is calculated in parallel by multiple material calculation modules, and each module can output a prediction result. Specifically, the prediction result output by the thermodynamic calculation module includes at least phase composition and phase volume fraction, the result output by the kinetic calculation module includes at least a descriptive text of the tissue evolution process, and the prediction result output by the machine learning prediction module includes at least phase composition and phase volume fraction.

[0040] Furthermore, the analysis engine performs verification on multiple prediction results. See details below. Figure 3 In one optional embodiment, the analysis engine's verification includes three parts: consistency analysis, physical plausibility analysis, and target confidence analysis.

[0041] In a specific embodiment of the consistency analysis, it is determined that multiple prediction results all include the same phase composition, and the difference in volume fraction between the same phases is within a preset difference range. If so, it indicates that the prediction results of multiple material calculation modules are highly consistent, and at this time, the multiple prediction results can be directly combined to obtain a positive prediction result.

[0042] Specifically, in one optional embodiment, the average volume fraction of the same phase can be used as the target volume fraction, and the target phase composition can be obtained by combining the phase composition of the entire multiple prediction results, thereby obtaining the predicted microstructure parameters.

[0043] In another optional embodiment, if there are conflicts among multiple prediction results, i.e., the phase compositions of different prediction results are different, or the difference in the volume fraction of the same phase is not within the preset difference range, then further analysis, verification, and arbitration of the multiple prediction results are required, i.e., proceeding to... Figure 3 The physical rationality analysis is shown.

[0044] Specifically, based on the process information input by the user, the thermodynamic state of the material to be predicted during the microstructure evolution process is determined. The thermodynamic state includes equilibrium and non-equilibrium states. When the thermodynamic state is in equilibrium, the prediction results of the thermodynamic calculation module are more reliable, and in this case, the prediction results of the thermodynamic calculation module can be directly used as the final output predicted microstructure parameters.

[0045] When the thermodynamic state is non-equilibrium, the reliability of the kinetic calculation module and the machine learning prediction module is higher. In this case, it is further determined whether the two prediction results from these two modules meet a preset difference condition. If they do, the two prediction results are integrated to obtain the output predicted microstructure parameters. Of course, if the two prediction results still do not meet the preset difference condition, a prompt signal indicating that preparation is impossible or prediction is impossible is output to the terminal, along with directions for modifying the material preparation information so that the user can adjust the input material preparation information. For ease of understanding, an example will be given below.

[0046] For example, as illustrated above, when the material to be predicted is TC4 titanium alloy (Ti-6Al-4V), the user-input composition information includes 6% Al (aluminum), 4% V (vanadium), and 90% Ti (titanium). The process information includes a solution treatment temperature of 950 degrees Celsius (°C), a holding time of 1 hour, and a water quenching method (cooling rate approximately >400°C / s).

[0047] The above material preparation information is transmitted through Figure 2 The output interface module shown transmits data to the analysis engine, which then calls... Figure 2 and Figure 3 The three modules shown perform parallel calculations. Among them, the thermodynamic calculation module retrieves thermodynamic data of titanium alloy from the database during the calculation process and obtains the equilibrium state result: "At room temperature, the equilibrium phases of TC4 alloy are α phase (HCP structure) and β phase (BCC structure), of which the equilibrium volume fraction of α phase is about 89%".

[0048] Based on the composition information, the kinetic calculation module retrieves the CCT diagram data of the TC4 alloy from the database. Based on the high cooling rate of water quenching, it determines that the diffusion-type phase transformation is completely suppressed, and the β phase (high-temperature phase) will undergo a diffusionless martensitic transformation, transforming into the α' martensite phase (hcp structure, supersaturated solid solution). Therefore, the microstructure evolution process can be described as "a small amount of untransformed primary α phase and a large amount of α' martensite phase."

[0049] The machine learning prediction module extracts more than a specified number of sets (e.g., 500 sets) of titanium alloy heat treatment data from a knowledge base as a training set. Input features include elemental contents such as Al and V, solution temperature, and cooling rate. In an optional embodiment, a gradient boosting tree (GBDT) model is trained using the training set so that user-inputted material preparation information can be directly fed into the model for prediction. For the above example, the output prediction result could be "α phase volume fraction is 15%, α' phase volume fraction is 85%".

[0050] Regarding the three prediction results mentioned above, through Figure 2 , Figure 3 The analysis engine shown compares and analyzes the three results. First, a consistency analysis reveals that the equilibrium prediction (α+β) from the thermodynamic calculation module differs fundamentally from the non-equilibrium prediction (α+α') from the kinetic calculation module and the machine learning prediction module, i.e., the phase compositions are different.

[0051] At this point, a physical rationality analysis, namely a thermodynamic state analysis, is required. Specifically, based on the cooling rate in the process information, water quenching is an extremely non-equilibrium process. Therefore, the prediction results from the kinetic calculation module and the machine learning prediction module are more reliable. In this case, the output results of these two modules can be integrated to obtain a positive prediction result.

[0052] Specifically, since both the kinetic principle and the data-driven prediction results point to α+α' microstructure and corroborate each other, in one optional embodiment, the integrated predicted microstructure parameters include target phase composition comprising "primary α phase and α' martensite phase," target phase volume fraction comprising "α phase 15%, α' phase 85%," and target grain size comprising "due to rapid cooling, the primary α phase grain size is small (approximately 5 micrometers), and the α' phase is lath-shaped." In a specific embodiment, the above positive prediction results are processed through... Figure 2 The output layer outputs the results.

[0053] In one optional embodiment, based on the consistency analysis results and material preparation information, a target confidence level is determined to characterize the fabricatability of the material to be predicted, including: Based on the consistency analysis results, the target consistency score is determined; the target consistency score is used to characterize the degree of consistency among multiple prediction results. Based on the material preparation information, an input clarity score and a reliability score are determined; the input clarity score characterizes the clarity of the material preparation information; the reliability score characterizes the reliability of the positive predictions made by the material calculation module under the conditions of the material preparation information. Based on multiple prediction results, a conflict penalty score is determined; the conflict penalty score is used to characterize the interpretability of conflicts between multiple prediction results. The target confidence level is determined based on the target consistency score, input clarity score, reliability score, and conflict penalty item score.

[0054] Based on the above embodiments, in order to further improve the reliability of the positive prediction results, the feasibility of preparing the material to be predicted using the material preparation information input by the user is analyzed to obtain the target confidence level, and the feasibility of preparing the material to be predicted is determined based on the target confidence level.

[0055] Specifically, in one alternative embodiment, the target confidence level can be calculated using formula (1): (1) in, Target confidence level, target confidence level It is a quantitative assessment of the reliability of the judgment that "given composition information and process information, predict microstructure parameters". It integrates cross-validation of multiple material calculation modules and physical rationality analysis.

[0056] Score for consistency with objectives. Scoring for consistency with objectives The weight of the objective consistency score. The target consistency score is used to reflect the degree of consistency between the prediction results of different material calculation modules. It is positively correlated with the degree of consistency.

[0057] For reliability score, For reliability score The weighting of reliability score. The reliability score reflects the reliability of a material calculation model's positive predictions under user-inputted material preparation information. It is positively correlated with the degree of reliability.

[0058] To input a specificity score, Input clarity score The weight of the input explicitness score The input clarity score is used to reflect the clarity of the material preparation information entered by the user. It is positively correlated with the degree of clarity.

[0059] Scoring for conflict penalty items, conflict penalty item score The degree of interpretability of conflicts between predictions output by different material calculation modules, and the conflict penalty term score. It is positively correlated with the degree of explainability.

[0060] It should be noted that, in one alternative embodiment, wherein, All scores were between 0 and 1.

[0061] Based on the above embodiments, as an optional embodiment, the prediction results include at least one of qualitative indicators and quantitative indicators; Based on the consistency analysis results, the target consistency score is determined, including: Obtain a pre-set mapping relationship for multiple prediction results; the mapping relationship includes the correspondence between the number of identical qualitative indicators and the first score, and the correspondence between the number of differences of identical quantitative indicators within a specified range and the second score; Based on the mapping relationship and the consistency analysis results, the first target score and the second target score are determined; Weights are assigned to qualitative and quantitative indicators respectively; Based on the weights, the first target score and the second target score are weighted and summed to obtain the target consistency score.

[0062] Understandably, the goal consistency score It can be used to measure the consistency of the thermodynamic calculation module, the kinetic calculation module, and the machine learning prediction module on key prediction metrics. In an optional embodiment, a mapping relationship for multiple prediction results can be pre-built.

[0063] For qualitative indicators (e.g., phase composition), the mapping relationship can be as follows: if the prediction results of all modules are consistent, the first score is 1 point; if the prediction results of two modules are consistent, the first score is 0.5 points; and if all prediction results are different from each other, the first score is 0 points.

[0064] For quantitative indicators (e.g., phase volume fraction), calculate the differences between the same quantitative indicators and determine whether they are within a specified range. If the differences between all predictions are within the specified range, the second score is 1 point; if some are within the specified range, the second score is 0.5 points; if none are within the specified range, the second score is 0 points.

[0065] Furthermore, weights are assigned to quantitative and qualitative indicators, and a weighted calculation is performed to obtain the target consistency score. In one optional embodiment, the weight assigned to the qualitative index of phase composition is 0.4, the weight assigned to the quantitative index of phase volume fraction is 0.4, and the weight assigned to the quantitative index of grain size is 0.2.

[0066] In an alternative embodiment, the reliability score can be calculated using formula (2). : (2) in, , For the first The reliability of the positive predictions made by each material calculation module under the condition of user-input material preparation information. For the first Each material calculation module is based on a historically validated baseline reliability score. For example, in one optional embodiment, the baseline reliability score for the thermodynamic calculation module is 0.9, for the kinetic calculation module it is 0.85, and for the machine learning prediction module it is 0.88.

[0067] For the first The scenario applicability parameters for each material calculation module, for the thermodynamic calculation module. ,in, The cooling rate is approximately 1 for furnace cooling and approximately 0.2 for water quenching. In a specific embodiment, the scene applicability parameter ranges from 0 to 1. For the dynamics calculation module, the scene applicability parameter... The degree of matching is equal to the first parameter, which characterizes the completeness of the data, and the second parameter, which characterizes the matching degree of the module. The module matching degree refers to the degree to which the selected dynamic calculation model (TTT plot, CCT plot, phase-field method, JMAK equations) matches the current physical process. For machine learning prediction modules, the scenario applicability parameter... This is equal to the probability that the input falls within the training distribution, i.e., it measures whether the current input parameters are within the statistical distribution of the machine learning model's training data. Machine learning models make reliable predictions within the training data distribution, but may produce catastrophic extrapolation errors outside the distribution. It is an indicator that quantifies the similarity between the current input point and the training data.

[0068] In one alternative embodiment, an input explicitness score is determined. The score can be determined based on a pre-built target mapping relationship, which is the correspondence between quantitative information in the material preparation information and the score. For example, if the quantitative information is a specific numerical value, the score is 1 point; if a smaller value range is given, the score is 0.8 points; and if the given value range is too large, the score is 0.5 points. Larger and smaller value ranges can be set according to different quantitative information. Examples will be given below.

[0069] For example, in the above case, the material to be predicted is TC4 titanium alloy (Ti-6Al-4V). If the user inputs Al as 6%, the score is 1 point. If the input Al is 5.8%-6.2%, the score is 0.8 points. If the input Al is 5%-7%, the score is 0.5 points.

[0070] In another optional embodiment, for descriptive text such as qualitative information (e.g., phase composition), the completeness of the information description can be scored by specifying a large model, with the score ranging from 0 to 1. The specified large model can be a large language model such as Deepseek or Tongyiqianwen, and this application does not limit this. Further, after obtaining the scores corresponding to all material preparation information, they are summed to obtain the final input clarity score. .

[0071] In one alternative embodiment, a conflict penalty item score is determined. First, identify the differences between multiple prediction results output by different modules and determine whether they are interpretable or uninterpretable conflicts. For interpretable conflicts (e.g., equilibrium and non-equilibrium states in thermodynamics), the conflict penalty score can be calculated using formula (3). : (3) in, To explain the intensity, the value ranges from 0 to 1. This is the penalty coefficient, which can be set according to actual needs; for example, it can be set to 0.1.

[0072] For unexplainable conflicts, the conflict penalty score can be calculated using formula (4). : (4) in, To characterize the maximum difference between two prediction results, The difference threshold can be set according to actual business needs.

[0073] For example, in the case of the material to be predicted as TC4 titanium alloy (Ti-6Al-4V), if the equilibrium state predicted by the thermodynamic calculation module differs from the non-equilibrium state predicted by the kinetic calculation module, and the high cooling rate leads to a non-equilibrium martensitic phase transformation, which is different from the equilibrium state prediction, then the conflict can be explained. Therefore, the explanation strength is 0.95, and the conflict penalty score can be calculated using the above calculation formula (3). It is 0.005.

[0074] Figure 4 The schematic diagram illustrates the principle of a method for predicting material prepareability according to another embodiment of this application. In an optional embodiment, the method for predicting material prepareability further includes: Obtain the target microstructure parameters to be designed; Historical materials with a similarity greater than a similarity threshold are obtained from a pre-built material database; the similarity is used to characterize the degree of similarity between the historical microstructure parameters of the historical materials and the target microstructure parameters. Historical materials are used as the materials to be predicted; and the process proceeds to the step of obtaining the material preparation information of the materials to be predicted in order to obtain the predicted microstructure parameters. With the goal of predicting microstructure parameters to approximate target microstructure parameters, a genetic algorithm is used to iteratively optimize historical microstructure parameters to obtain information on the preparation of the target material. Determine the confidence level of the inverse design used to characterize the reliability of the predictions for the preparation of the target material; When the confidence level of the reverse design is greater than the second threshold, the output includes the reverse design results, which include information on the preparation of the target material and the predicted microstructure parameters.

[0075] The prediction method provided in this application can not only perform forward prediction of materials, but also reverse design. Specifically, it obtains the target microstructure parameters to be designed. In one optional embodiment, see [link to example]. Figure 2 and Figure 4 The target microstructure parameters input by the user may include information such as phase composition, phase volume fraction and grain size, which are not limited in this application.

[0076] For example, taking high-strength steel as an example for reverse verification and design, the target microstructure parameters input by the user include phase composition including bainite and retained austenite, phase volume fraction including bainite greater than 80%, retained austenite 10% to 15%, and grain size including original austenite grain size less than 10 micrometers (μm).

[0077] like Figure 4As shown, after obtaining the target microstructure parameters, the physical feasibility of the target microstructure parameters can first be judged based on a pre-built material database, that is, to confirm whether the microstructure parameter is physically feasible. For example, in the above example, the microstructure can be obtained in high-carbon steel through a specific process, therefore the target microstructure parameter is physically feasible.

[0078] In an alternative embodiment, if it is physically infeasible, a reason for the undesignability is output or an adjustment is provided. Figure 2 The output layer shown allows users to modify the input target microstructure parameters.

[0079] In another alternative embodiment, such as Figure 4 As shown, if physical feasibility is feasible, further steps include screening historical materials from the materials database that have high similarity between historical microstructure parameters and target microstructure parameters. Specifically, in one optional embodiment, for identical quantitative information, difference calculations can be performed, and historical materials with differences within a specified range can be selected. Simultaneously, for identical qualitative information, the descriptive text can be converted into a semantic vector, and semantic similarity calculations can be performed, filtering out historical materials with semantic similarity greater than a threshold. Alternatively, a large language model can be used to calculate the similarity between historical and target microstructure parameters, obtaining historical materials with similarity greater than a similarity threshold.

[0080] It should be noted that, in specific embodiments, the selected historical materials may include multiple materials. When the number of selected materials is very large, in order to save computing resources, the first specified number of historical materials (e.g., the first 3) in the descending similarity sort can be selected for subsequent calculations.

[0081] For example, in the case of the high-strength steel mentioned above, the selected historical microstructure parameters include historical composition information and historical process information. The historical composition information includes "C is 0.25%, Si is 1.5%, and Mn is 2.0%", and the historical process information includes "rapid cooling after annealing in the two-phase region".

[0082] After determining the historical materials, the historical materials are used as the materials to be predicted in the above embodiments, and steps S10 to S12 are executed to obtain the predicted microstructure parameters of the historical materials. That is, the screened historical composition information and historical process information are used as the material preparation information in the above embodiments for positive prediction, thereby obtaining the predicted microstructure parameters.

[0083] Furthermore, Figure 2 The analysis engine shown starts the built-in genetic algorithm, that is, it starts... Figure 4The genetic algorithm shown uses material preparation information such as historical composition information and historical process information from the aforementioned historical materials as variables. At the same time, it constructs an objective function with the goal of minimizing the difference between the predicted microstructure parameters and the target microstructure parameters. The genetic algorithm iteratively searches the variable search space to obtain the target material preparation information.

[0084] Specifically, see Figure 4 Using historical material preparation information as optimization variables, the population is initialized within the solution space defined by the variables. Simultaneously, an objective function is constructed to minimize the difference between the predicted microstructure parameters and the target microstructure parameters. In each iteration of the genetic algorithm, candidate schemes in the current population are evaluated, the difference between the predicted microorganism parameters and the target microorganism parameters corresponding to each candidate scheme is calculated, and the next generation of candidate schemes is generated based on the difference comparison results.

[0085] Repeat the above iterative process until the termination condition is met, and output the target scheme that minimizes the difference (i.e., the predicted microorganism parameters generated in the last iteration).

[0086] It should be noted that, in the specific embodiment, each iteration of the search will execute steps S10 to S12 once to obtain the predicted microorganism parameters corresponding to the current iteration, and compare the difference between the predicted microorganism parameters corresponding to the current iteration and the target microorganism parameters in order to guide the next generation of search.

[0087] For example, taking the aforementioned high-strength steel as an example, the genetic algorithm eventually converges to a target material preparation information, and the recommended composition information includes "C is 0.28%, Si is 1.8%, Mn is 2.2%, Cr is 0.5% (balance Fe)", and the recommended process information is "austenitization temperature is 850℃ / 5min, then cooling to 350℃ isothermally for 10 minutes at a cooling rate of 15℃ / s". The predicted microstructure parameters obtained by the positive prediction include "bainite volume fraction is 82%, retained austenite is 12%, and the original austenite grain size is 9.5μm".

[0088] Furthermore, in an optional embodiment, to improve the reliability of reverse design, the predictive credibility of the target material preparation information is evaluated to obtain the reverse design confidence level. In an optional embodiment, this can be calculated using formula (5): (5) in, The confidence level for reverse engineering is used to characterize the predictive credibility of information on the preparation of the target material, reflecting the multi-level requirements for the reliability of the scheme. The quality parameters characterize the iterative search quality of the genetic algorithm. Safety parameters used to characterize the feasibility of information on the preparation of target materials.

[0089] In one optional embodiment, when the confidence level of the reverse design is greater than the second threshold, it indicates that the reliability of the current reverse design result is high. At this time, the reverse design result including the target material preparation information and the predicted microstructure parameters can be output.

[0090] Based on the above embodiments, as an optional embodiment, determining the reverse design confidence level used to characterize the predictive reliability of target material preparation information includes: The steps are performed to determine the target confidence level used to characterize the fabricatability of the material to be predicted, based on the consistency analysis results and material preparation information, and the target confidence level is obtained. Determine the quality parameters used to characterize the quality of iterative search in the genetic algorithm; Determine the safety parameters used to characterize the feasibility of information on the preparation of the target material; The reverse design confidence level is determined based on the target confidence level, quality parameters, and safety parameters; the reverse design confidence level is positively correlated with the target confidence level, quality parameters, and safety parameters.

[0091] For target confidence level The quality parameters are obtained by performing step S13 of the above embodiment, and are calculated once in each round of iterative search of the genetic algorithm. This is used to measure the quality of the search process by which the optimization algorithm finds the current solution, ensuring that it is not a local optimum or a random result. In an optional embodiment, the quality parameter is calculated according to formula (6). : (6) in, The convergence score ranges from 0 to 1. Analyze the optimal objective value sequence for the last m generations of the population. The formula for calculating relative change is: Therefore, the convergence score is... The calculation formula is: ,in, The convergence threshold can be set according to the size of the optimization problem; for example, it can be set to 0.01.

[0092] The uniqueness score of the solution ranges from 0 to 1. In a specific embodiment, the top K solutions are selected from the final population (for example, the top 10% of solutions can be selected), and the clustering degree in the normalized parameter space is calculated using formula (7) to obtain the uniqueness score of the solution. : (7) in, For the first The solution and the first The Euclidean distance between the solutions.

[0093] For safety parameters This is used to evaluate the practical feasibility and robustness of the recommended scheme. In an optional embodiment, it can be calculated according to formula (8): (8) in, The robustness score ranges from 0 to 1. In a specific embodiment, Apply perturbation to the scheme parameters as follows And use a fast surrogate model to predict the perturbed organization Therefore, the rate of change of key targets can be calculated, and the calculation formula is shown in formula (9): (9) Furthermore, based on the rate of change of key targets The robustness score can be calculated using formula (10). (10) in, The sensitivity parameter is determined through sensitivity analysis.

[0094] The prepareability score ranges from 0 to 1. In a specific embodiment, based on the rules of a pre-set knowledge base, process feasibility is evaluated, including whether the cooling rate is within the equipment's capacity, whether the temperature control accuracy meets the standard, and the matching degree between the isothermal time and the production cycle. Cost and safety include the amount of rare elements used, heat treatment energy consumption, and environmental friendliness. Based on the above rules, a prepareability score can be obtained, as shown in formula (11): (11) in, To characterize the first The parameter indicates the degree to which a rule is satisfied, and its value ranges from 0 to 1. For rule weights.

[0095] In the above embodiments, the method for predicting material prepareability has been described in detail. This application also provides an embodiment of a device for predicting material prepareability.

[0096] Figure 5 This is a schematic diagram of the structure of a material prepareability prediction device provided in an embodiment of this application, as shown below. Figure 5As shown, the device includes: The preparation information acquisition module 50 is used to acquire the material preparation information of the material to be predicted; the material preparation information includes at least composition information and process information. The forward prediction module 51 is used to analyze material preparation information in parallel through multiple material calculation modules to forward predict the microstructure parameters of the material to be predicted and obtain multiple prediction results. The consistency analysis module 52 is used to analyze the differences between different prediction results to obtain consistency analysis results; and to determine the predicted microstructure parameters of the material to be predicted based on the consistency analysis results. The first confidence level determination module 53 is used to determine the target confidence level to characterize the fabricatability of the material to be predicted based on the consistency analysis results and material preparation information; the target confidence level is positively correlated with the fabricatability. The prediction result output module 54 is used to output the predicted microstructure parameters when the target confidence level is greater than the first threshold.

[0097] Furthermore, the material prepareability prediction device provided in this application embodiment also includes: The difference judgment module is used to determine whether the consistency analysis results indicate that multiple prediction results meet preset difference conditions. The preset difference conditions include at least the same phase composition and the difference in volume fraction of the same phase within a preset difference range. Among them, the multiple material calculation modules include at least a thermodynamic calculation module, a kinetic calculation module, and a machine learning prediction module. The machine learning prediction module is a module that obtains prediction results by training on historical material data. The predicted microstructure parameters include at least the target phase composition and the target phase volume fraction.

[0098] If so, the first processing module is invoked. The first processing module is used to take the average volume fraction of the same phase as the target phase volume fraction; and to integrate the phase composition in the prediction results to obtain the target phase composition, so as to obtain the predicted microstructure parameters. If not, call the following module: Thermodynamic state determination module is used to determine the thermodynamic state of the microstructure evolution process of the material to be predicted based on process information; The second processing module is used to use the first prediction result of the thermodynamic calculation module as the predicted microstructure parameters when the thermodynamic state is in equilibrium. The third processing module is used to determine whether the second prediction results of the kinetic calculation module and the machine learning prediction module meet the preset difference conditions when the thermodynamic state is non-equilibrium. If they do, the second prediction results are integrated to obtain the predicted microstructure parameters.

[0099] The first score determination module is used to determine the target consistency score based on the consistency analysis results; the target consistency score is used to characterize the degree of consistency between multiple prediction results. The second scoring module is used to determine the input clarity score and the reliability score based on the material preparation information. The input clarity score is used to characterize the clarity of the material preparation information; the reliability score is used to characterize the reliability of the positive predictions made by the material calculation module under the conditions of the material preparation information. The penalty score determination module is used to determine the conflict penalty item score based on multiple prediction results; the conflict penalty item score is used to characterize the interpretability of the conflict between multiple prediction results. The fourth processing module is used to determine the target confidence level based on the target consistency score, input clarity score, reliability score, and conflict penalty item score.

[0100] The mapping relationship acquisition module is used to acquire pre-set mapping relationships for multiple prediction results; the mapping relationship includes the correspondence between the number of identical qualitative indicators and the first score, and the correspondence between the number of differences of identical quantitative indicators within a specified range and the second score; wherein, the prediction result includes at least one of qualitative indicators and quantitative indicators; The first determination module is used to determine the first target score and the second target score based on the mapping relationship and the consistency analysis results. The weight allocation module is used to assign weights to qualitative and quantitative indicators respectively. The weighted summation module is used to perform a weighted summation of the first target score and the second target score based on weights to obtain the target consistency score.

[0101] The organization parameter acquisition module is used to acquire the target microstructure parameters to be designed; The second determining module is used to obtain historical materials with an organization similarity greater than a similarity threshold from a pre-built material database; organization similarity is used to characterize the degree of similarity between the historical microstructure parameters of the historical materials and the target microstructure parameters. The third processing module is used to treat historical materials as materials to be predicted and proceeds to the step of obtaining material preparation information of the materials to be predicted in order to obtain the predicted microstructure parameters. The iterative search module is used to obtain the target material preparation information by iteratively optimizing historical microstructure parameters through a genetic algorithm, with the goal of approximating the target microstructure parameters with the predicted microstructure parameters. The second confidence determination module is used to determine the reverse design confidence level used to characterize the reliability of the prediction information for the preparation of the target material; The reverse design result output module is used to output reverse design results, including target material preparation information and predicted microstructure parameters, when the reverse design confidence level is greater than the second threshold.

[0102] After calling the first confidence level determination module to obtain the target confidence level, the third determination module and the target determination module are called. The third determination module is used to determine the quality parameters used to characterize the quality of the iterative search of the genetic algorithm; and to determine the safety parameters used to characterize the feasibility of the target material preparation information. The target determination module is used to determine the reverse design confidence level based on the target confidence level, quality parameters, and safety parameters; the reverse design confidence level is positively correlated with the target confidence level, quality parameters, and safety parameters.

[0103] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 6 As shown, the electronic device includes: a memory 60 for storing computer programs; The processor 61 is configured to execute a computer program to implement the steps of the method for predicting the material prepareability as described in the above embodiments.

[0104] The electronic devices provided in this embodiment may include, but are not limited to, laptops or desktop computers.

[0105] The processor 61 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 61 may be implemented using at least one of the following hardware forms: Digital Signal Processor (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 61 may also include a main processor and a coprocessor. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 61 may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 61 may also include an Artificial Intelligence (AI) processor, which is used to handle computational operations related to machine learning.

[0106] The memory 60 may include one or more computer-readable storage media, which may be non-transitory. The memory 60 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 60 is used to store at least the following computer program 601, which, after being loaded and executed by the processor 61, is capable of implementing the relevant steps of the material prepareability prediction method disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 60 may also include an operating system 602 and data 603, etc., and the storage method may be temporary storage or permanent storage. The operating system 602 may include Windows, Unix, Linux, etc. The data 603 may include, but is not limited to, relevant data involved in the material prepareability prediction method.

[0107] In some embodiments, the electronic device may further include a display screen 62, an input / output interface 63, a communication interface 64, a power supply 65, and a communication bus 66.

[0108] Those skilled in the art will understand that Figure 6 The structures shown do not constitute a limitation on electronic devices and may include more or fewer components than those shown.

[0109] The electronic device provided in this application includes a memory and a processor. When the processor executes the program stored in the memory, it can implement the material prepareability prediction method described in the above embodiments.

[0110] It should be noted that although the operations are depicted in a specific order in the accompanying drawings, this should not be construed as requiring these operations to be performed in the specific order shown or sequentially, or requiring all illustrated operations to be performed to achieve the desired result. In some cases, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system modules and components in the above embodiments should not be construed as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

Claims

1. A method for predicting the prepareability of a material, characterized in that, The method includes: Obtain material preparation information of the material to be predicted; the material preparation information includes at least composition information and process information. The material preparation information is analyzed in parallel using multiple material calculation modules to positively predict the microstructure parameters of the material to be predicted, and multiple prediction results are obtained. The differences between the different prediction results are analyzed to obtain the consistency analysis results; and the predicted microstructure parameters of the material to be predicted are determined based on the consistency analysis results. Based on the consistency analysis results and the material preparation information, a target confidence level is determined to characterize the fabricatability of the predicted material; the target confidence level is positively correlated with the fabricatability. When the target confidence level is greater than the first threshold, the predicted microstructure parameters are output.

2. The method for predicting material prepareability as described in claim 1, characterized in that, The multiple material calculation modules include at least a thermodynamic calculation module, a kinetic calculation module, and a machine learning prediction module; The machine learning prediction module is a module that obtains prediction results by training on historical material data. The predicted microstructure parameters include at least the target phase composition and the target phase volume fraction.

3. The method for predicting material prepareability as described in claim 2, characterized in that, The step of determining the predicted microstructure parameters of the material to be predicted based on the consistency analysis results includes: Determine whether the consistency analysis results indicate that the multiple prediction results meet preset difference conditions; the preset difference conditions include at least the same phase composition and the difference in volume fraction of the same phase within a preset difference range; If so, the average volume fraction of the same phase is taken as the target phase volume fraction; and the phase composition in the prediction results is integrated to obtain the target phase composition, so as to obtain the predicted microstructure parameters; If not, proceed with the following steps: Based on the process information, determine the thermodynamic state of the microstructure evolution process of the material to be predicted; If the thermodynamic state is in equilibrium, the first prediction result of the thermodynamic calculation module shall be used as the predicted microstructure parameter; If the thermodynamic state is non-equilibrium, determine whether the quality control of the second prediction results of the kinetic calculation module and the machine learning prediction module meets the preset difference condition; if it does, integrate the second prediction results to obtain the predicted microstructure parameters.

4. The method for predicting material prepareability as described in claim 1, characterized in that, The step of determining the target confidence level for characterizing the fabricatability of the predicted material based on the consistency analysis results and the material preparation information includes: Based on the consistency analysis results, a target consistency score is determined; the target consistency score is used to characterize the degree of consistency among the multiple prediction results. Based on the material preparation information, an input clarity score and a reliability score are determined; the input clarity score characterizes the clarity of the material preparation information; the reliability score characterizes the reliability of the positive predictions made by the material calculation module under the conditions of the material preparation information. Based on the multiple prediction results, a conflict penalty item score is determined; the conflict penalty item score is used to characterize the interpretability of the conflict between the multiple prediction results. The target confidence level is determined based on the target consistency score, the input clarity score, the reliability score, and the conflict penalty item score.

5. The method for predicting material prepareability as described in claim 4, characterized in that, The prediction results include at least one of qualitative and quantitative indicators; The step of determining the target consistency score based on the consistency analysis results includes: Obtain a pre-set mapping relationship for the multiple prediction results; the mapping relationship includes the correspondence between the number of identical qualitative indicators and the first score, and the correspondence between the number of differences of identical quantitative indicators within a specified range and the second score; Based on the mapping relationship and the consistency analysis results, the target first score and the target second score are determined; Assign weights to the qualitative and quantitative indicators respectively; Based on the weights, the first target score and the second target score are weighted and summed to obtain the target consistency score.

6. The method for predicting material prepareability as described in claim 1, characterized in that, The method further includes: Obtain the target microstructure parameters to be designed; Historical materials with an organization similarity greater than a similarity threshold are obtained from a pre-built material database; the organization similarity is used to characterize the degree of similarity between the historical microstructure parameters of the historical materials and the target microstructure parameters. The historical material is used as the material to be predicted; and the process proceeds to the step of obtaining the material preparation information of the material to be predicted, so as to obtain the predicted microstructure parameters. With the goal of approximating the target microstructure parameters with the predicted microstructure parameters, the historical microstructure parameters are iteratively optimized using a genetic algorithm to obtain the target material preparation information; Determine the reverse design confidence level used to characterize the reliability of the predictions for the preparation information of the target material; When the confidence level of the reverse design is greater than the second threshold, the reverse design result, which includes the target material preparation information and the predicted microstructure parameters, is output.

7. The method for predicting material prepareability as described in claim 6, characterized in that, The determination of the reverse design confidence level used to characterize the predictive reliability of the target material preparation information includes: The step of determining the target confidence level to characterize the fabrication feasibility of the material to be predicted based on the consistency analysis results and the material preparation information is performed to obtain the target confidence level; Determine the quality parameters used to characterize the iterative search quality of the genetic algorithm; Determine the safety parameters used to characterize the feasibility of the preparation information for the target material; The reverse design confidence level is determined based on the target confidence level, the quality parameter, and the safety parameter; the reverse design confidence level is positively correlated with the target confidence level, the quality parameter, and the safety parameter.

8. A device for predicting the prepareability of materials, characterized in that, The device includes: A preparation information acquisition module is used to acquire material preparation information of the material to be predicted; the material preparation information includes at least composition information and process information. The forward prediction module is used to analyze the material preparation information in parallel through multiple material calculation modules to forward predict the microstructure parameters of the material to be predicted and obtain multiple prediction results. The consistency analysis module is used to analyze the differences between different prediction results to obtain consistency analysis results; and to determine the predicted microstructure parameters of the material to be predicted based on the consistency analysis results. The confidence level determination module is used to determine a target confidence level to characterize the fabricatability of the material to be predicted, based on the consistency analysis results and the material preparation information; the target confidence level is positively correlated with the fabricatability. The prediction result output module is used to output the predicted microstructure parameters when the target confidence level is greater than the first threshold.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for predicting the material prepareability according to any one of claims 1 to 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 steps of the method for predicting the material prepareability according to any one of claims 1 to 7.

Images