High-throughput screening method, system, apparatus, and medium for high temperature soft magnetic materials

Abstract

The application discloses a high-throughput screening method, system, device and medium for high-temperature-resistant soft magnetic materials, and comprises the following steps: obtaining intrinsic structure data of the soft magnetic materials, and performing doping treatment on the intrinsic structure data to obtain a batch file package; performing first-principle calculation on the batch file package to obtain a result file package; training and adjusting an initial machine learning model according to the result file package to obtain a target machine learning model, taking the structure characteristics of the soft magnetic materials corresponding to the result file package as input data of the target machine learning model, and taking the high-temperature magnetic stability data and the high-temperature phase stability data of the soft magnetic materials corresponding to the batch file package as output data of the target machine learning model to construct a material structure-property relationship model; and predicting the soft magnetic materials to be screened according to the material structure-property relationship model, and selecting the high-temperature-resistant soft magnetic materials according to the prediction results. The application greatly improves the efficiency and accuracy of material screening.

Description

Technical Field

[0001] This invention relates to the fields of materials science and computational physics, and in particular to high-throughput screening methods, systems, equipment and media for high-temperature resistant soft magnetic materials. Background Technology

[0002] Rare-earth-iron-based soft magnetic materials have attracted widespread attention due to their excellent magnetic properties and industrial applications. The structural properties of these materials significantly influence their physical and chemical properties, with magnetic properties at high temperatures and phase stability being key performance indicators. Traditional experimental screening methods rely on experiments and simple calculations, resulting in low efficiency, high cost, and time-consuming processes, making it difficult to quickly discover new materials with excellent magnetic properties and phase stability. With the development of computational science, first-principles calculations offer an efficient theoretical prediction method. However, first-principles calculations typically involve large computational loads and long processing times, limiting their application in high-throughput screening. Therefore, how to effectively combine first-principles calculations and machine learning for efficient material screening has become an important research topic in materials science. Currently, several domestic patents (CN108319807B, CN109300510A, CN111177915A, CN113722922A) introduce high-throughput screening methods, but these methods only focus on the design and implementation of high-throughput methods and have not formed a self-consistent "theory-experiment-correction theory" approach. Summary of the Invention

[0003] The purpose of this invention is to provide a self-consistent high-throughput material screening method based on "theory-experiment-correction theory". To achieve the above objective, this invention provides a high-throughput screening method, system, device, and medium for high-temperature resistant soft magnetic materials.

[0004] In a first aspect, embodiments of the present invention provide a high-throughput screening method for high-temperature resistant soft magnetic materials, comprising:

[0005] The intrinsic structure data of the soft magnetic material is obtained, and the intrinsic structure data is doped to obtain a batch file package; the intrinsic structure data corresponding to the batch file package includes at least the high-temperature magnetic stability data and the high-temperature phase stability data of the soft magnetic material.

[0006] First-principles calculations are performed on the batch file packages to obtain result file packages; the structural features of the soft magnetic material corresponding to the result file packages include at least the doping elements, doping sites, and defect formation energies.

[0007] The initial machine learning model is trained and adjusted based on the result file package to obtain the target machine learning model. The soft magnetic material structural features corresponding to the result file package are used as the input data of the target machine learning model, and the high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material corresponding to the batch file package are used as the output data of the target machine learning model to construct a material structure-property relationship model.

[0008] The structure-property relationship model is used to predict the soft magnetic materials to be screened, and high-temperature resistant soft magnetic materials are selected based on the prediction results. The prediction results include the predicted high-temperature magnetic stability of the soft magnetic materials to be screened and the predicted high-temperature phase stability of the soft magnetic materials to be screened.

[0009] Preferably, after predicting the soft magnetic materials to be screened based on the material structure-property relationship model and selecting high-temperature resistant soft magnetic materials based on the prediction results, the method further includes:

[0010] Experimental verification was conducted on the soft magnetic materials to be screened, and the experimental results were compared with the predicted results to evaluate the predictive effect of the material structure-property relationship model based on the comparison results.

[0011] The material structure-property relationship model is iteratively optimized based on the experimental results to adjust the material structure-property relationship model.

[0012] Preferably, the step of acquiring intrinsic structure data of the soft magnetic material and performing doping processing on the intrinsic structure data to obtain a batch file package includes:

[0013] Access a materials science database and retrieve intrinsic structure data of soft magnetic materials from the database; the materials science database includes at least AFLOWlib and OQMD.

[0014] The intrinsic structure data is subjected to a first preprocessing, and the preprocessed intrinsic structure data is exported and saved as a CSV file package; the first preprocessing includes at least data cleaning and removal of missing values;

[0015] The CSV file package is subjected to doping processing to obtain a batch file package.

[0016] Preferably, the step of performing doping processing on the CSV file package to obtain a batch file package includes:

[0017] The crystal structure of the CSV file is read, and the preset doping elements are traversed through the placeholders in the crystal structure to obtain multiple doped intrinsic structure data files.

[0018] Preferably, the step of performing first-principles calculations on the batch file packages to obtain the result file packages includes:

[0019] The batch file packages were subjected to first-principles calculations using VASP to obtain the result file packages.

[0020] Preferably, the step of training and adjusting the initial machine learning model based on the result file package to obtain the target machine learning model includes:

[0021] The result file package is subjected to a second preprocessing to obtain the soft magnetic material structural features corresponding to the result file package, and the soft magnetic material structural features are organized into a feature dataset; the second preprocessing includes at least feature selection and normalization.

[0022] The feature dataset is divided into a training set and a test set;

[0023] An initial machine learning model is constructed based on a random forest, and the initial machine learning model is trained using the training set to obtain an intermediate machine learning model;

[0024] The intermediate machine learning model is adjusted based on the test set to obtain the target machine learning model.

[0025] Preferably, adjusting the intermediate machine learning model based on the test set to obtain the target machine learning model includes:

[0026] The intermediate machine learning model is used to predict the test set to obtain the predicted value of the defect formation energy of the test set.

[0027] The defect formation energy prediction value is used to perform first-principles calculations to obtain the defect formation energy calculation value of the test set;

[0028] The variance between the predicted defect formation energy and the calculated defect formation energy is calculated, and the intermediate machine learning model is adjusted based on the variance to obtain the target machine learning model; the adjustment includes at least adjusting hyperparameters, reselecting features, and expanding the training set.

[0029] Secondly, embodiments of the present invention provide a high-throughput screening system for high-temperature resistant soft magnetic materials, comprising:

[0030] The data collection and processing module is used to acquire intrinsic structure data of soft magnetic materials and perform doping processing on the intrinsic structure data to obtain batch file packages; the intrinsic structure data corresponding to the batch file packages includes at least high-temperature magnetic stability data and high-temperature phase stability data of soft magnetic materials.

[0031] The first-principles calculation module is used to perform first-principles calculations on the batch file package to obtain a result file package; the soft magnetic material structural features corresponding to the result file package include at least the doping element, doping site, and defect formation energy;

[0032] The machine learning training module is used to train and adjust the initial machine learning model according to the result file package to obtain the target machine learning model. The soft magnetic material structural features corresponding to the result file package are used as the input data of the target machine learning model, and the high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material corresponding to the batch file package are used as the output data of the target machine learning model to construct a material structure-property relationship model.

[0033] The high-throughput screening module is used to predict the soft magnetic materials to be screened based on the material structure-property relationship model, and select high-temperature resistant soft magnetic materials based on the prediction results; the prediction results include the predicted values ​​of the high-temperature magnetic stability and the high-temperature phase stability of the soft magnetic materials to be screened.

[0034] Thirdly, embodiments of the present invention provide a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the high-throughput filtering method as described above.

[0035] Fourthly, embodiments of the present invention provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the high-throughput screening method as described above.

[0036] Compared with existing technologies, the high-throughput screening method, system, equipment and medium of the present invention for high-temperature resistant soft magnetic materials have the following advantages: a specific theoretical model for efficient screening of high-temperature resistant soft magnetic materials is formed, and an experimental verification step is provided to correct the theoretical model, thereby improving the efficiency and accuracy of material screening, expanding the material design space, and enabling the prediction of the performance of unknown materials and the optimization of the performance of existing materials. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a high-throughput screening method for high-temperature resistant soft magnetic materials according to an embodiment of the present invention.

[0038] Figure 2 This is a schematic diagram of the process for obtaining batch file packages according to an embodiment of the present invention;

[0039] Figure 3 This is a schematic diagram of the process for obtaining the target machine learning model according to an embodiment of the present invention;

[0040] Figure 4 This is a schematic diagram of the process of adjusting the intermediate machine learning model according to an embodiment of the present invention;

[0041] Figure 5 This is an embodiment of the present invention, Y2Fe 15 A schematic diagram of the XRD diffraction results of Cr2 at room temperature;

[0042] Figure 6 This is an embodiment of the present invention, Y2Fe 15 A schematic diagram of the XRD diffraction results of Cr2 at 600K.

[0043] Figure 7 This is a schematic diagram of the structure of a high-throughput screening system for high-temperature resistant soft magnetic materials according to an embodiment of the present invention;

[0044] Figure 8 This is a schematic diagram of the structure of a terminal device according to an embodiment of the present invention. Detailed Implementation

[0045] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0046] like Figure 1 As shown, this embodiment of the invention provides a high-throughput screening method for high-temperature resistant soft magnetic materials, including:

[0047] S1. Obtain the intrinsic structure data of the soft magnetic material and perform doping processing on the intrinsic structure data to obtain a batch file package;

[0048] Specifically, such as Figure 2 As shown, step S1 includes:

[0049] S101. Access the materials science database and retrieve the intrinsic structure data of soft magnetic materials from the materials science database;

[0050] The selected materials science databases include at least AFLOWlib and OQMD. The access interface and method should be chosen based on the REST API or Python library provided by each database. Specifically, sqlite3 and SQLAlchemy should be selected for database access.

[0051] Furthermore, a Python script was written, importing requests (API calls), pandas (data processing), and SQLAlchemy (SQL database). Data was retrieved from the database using API calls or SQL queries to obtain the intrinsic structure data of the soft magnetic material.

[0052] S102. Perform a first preprocessing on the intrinsic structure data, and export and save the preprocessed intrinsic structure data as a CSV file package.

[0053] The retrieved intrinsic structure data is preprocessed using libraries such as pandas. Specifically, the first preprocessing includes at least data cleaning and removal of missing values. The preprocessed intrinsic structure data is then exported to a file and saved as a CSV file.

[0054] S103. Perform doping processing on the CSV file package to obtain a batch file package.

[0055] The crystal structure of the CSV file is read, and the pre-defined doping elements are iterated through the placeholders in the crystal structure to obtain multiple intrinsic structure data files of doped materials, i.e., batch file packages. The intrinsic structure data corresponding to each batch file package includes at least high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material.

[0056] Specifically, the pymatgen library is used to read the crystal structure from the CSV file, and doping is performed based on the lattice occupancy. The doping process is as follows:

[0057] Create a Python script and import the Structure, MPRELAXSet, and Poscar modules. Read the crystal structure from the CSV file and create a Structure object. Select one element (the preset dopant element) and iterate through the occupants on the crystal lattice.

[0058] Furthermore, the doped crystal structure file is imported into the Python environment using the pymatgen library for subsequent calculations and analysis. Additionally, the pymatgen library is used for further crystal structure analysis, calculating lattice parameters, density, and crystal symmetry, performing preliminary doping information processing to evaluate the properties of the doped material.

[0059] S2. Perform first-principles calculations on the batch file packages to obtain the result file packages;

[0060] First-principles calculations were performed on batch file packages using VASP to obtain result file packages. The structural characteristics of the soft magnetic material corresponding to the result file packages include at least the dopant elements, dopant sites, and defect formation energies.

[0061] Specifically, the pymatgen library is used to generate the input files (INCAR, POSCAR, KPOINTS, and POTCAR, etc.) required for VASP calculations based on the doped structure. A Slurm job script is created in the same directory containing the VASP input files. Furthermore, the job files can be submitted and the job status monitored by executing shell commands using Python's subprocess module. The batch file packages are then submitted to the high-performance computing cluster for VASP calculations.

[0062] S3. Train and adjust the initial machine learning model based on the result file package to obtain the target machine learning model. Use the soft magnetic material structural features corresponding to the result file package as the input data of the target machine learning model, and use the high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material corresponding to the batch file package as the output data of the target machine learning model to construct a material structure-property relationship model.

[0063] Specifically, such as Figure 3 As shown, step S3 includes:

[0064] S301. Perform a second preprocessing on the result file package to obtain the soft magnetic material structural features corresponding to the result file package, and organize the soft magnetic material structural features into a feature dataset.

[0065] The second preprocessing includes at least feature selection and normalization. Specifically, the pandas library is used to load the result file package, and necessary data cleaning and missing value handling are performed; functions are written using the pymatgen library to calculate features (doped elements, doped sites, and defect formation energies, etc.) from the result file package; feature selection techniques using embedding methods are used to identify and retain the most informative features, and the features are scaled according to the computational accuracy to make them on the same scale.

[0066] S302. Divide the feature dataset into a training set and a test set;

[0067] Furthermore, the feature dataset is saved to a file in CSV format for later use.

[0068] S303. Construct an initial machine learning model based on random forest, and train the initial machine learning model according to the training set to obtain an intermediate machine learning model;

[0069] S304. Adjust the intermediate machine learning model based on the test set to obtain the target machine learning model.

[0070] Specifically, such as Figure 4 As shown, step S304 includes:

[0071] S304-A: Use an intermediate machine learning model to predict the test set and obtain the predicted value of the defect formation energy of the test set.

[0072] S304-B, First-principles calculations are performed on the predicted defect formation energy values ​​to obtain the calculated defect formation energy values ​​for the test set;

[0073] S304-C: Calculate the variance between the predicted and calculated defect formation energy values, and adjust the intermediate machine learning model based on the variance to obtain the target machine learning model.

[0074] Variance is used to evaluate the difference between the defect formation energy obtained from the machine learning model and the first-principles calculation, and this difference is used to adjust the machine learning model. Specifically, the variance between the predicted and calculated defect formation energy is expressed by the following formula:

[0075]

[0076] Where N represents the number of samples in the test set, P i O represents the predicted defect formation energy of the i-th sample. i This represents the calculated defect formation energy of the i-th sample. The intermediate machine learning model is adjusted based on the analysis of variance results, including at least adjusting hyperparameters, reselecting features, and expanding the training set. Furthermore, different models or algorithms should be considered if necessary, especially when the existing model structure cannot adequately capture the complexity of the data.

[0077] S4. Predict the soft magnetic materials to be screened based on the material structure-property relationship model, and select high-temperature resistant soft magnetic materials based on the prediction results.

[0078] Specifically, the prediction results include predicted high-temperature magnetic stability and high-temperature phase stability of the soft magnetic materials to be screened. Based on the magnitude of the predicted values, the different materials to be screened are ranked, and materials with higher stability are selected.

[0079] It should be noted that in the problem of doping effect, the high-temperature magnetic stability and high-temperature phase stability of the material are predicted as dependent variables, while the characteristics are the information of the doped material (doping element, doping site and defect formation energy) obtained by first-principles calculation.

[0080] Furthermore, to understand which features are most important for predicting material properties, this embodiment modifies different doping elements and doping sites, and performs first-principles calculations based on the prediction results given by the material structure-property relationship model to obtain weight coefficients for the doping elements and doping sites. Based on the calculated weight coefficients, it is possible to identify which features (such as specific doping elements or doping sites) are most important for predicting material properties. It is understandable that features with higher weight coefficients have a greater impact on the prediction results.

[0081] Furthermore, after step S4, the following steps are also included:

[0082] S5. Conduct experimental verification of the soft magnetic materials to be screened, and compare the experimental results with the predicted results to evaluate the predictive effect of the material structure-property relationship model based on the comparison results.

[0083] Specifically, the comparison results include absolute error, relative error, or other relevant error measures. By analyzing the errors, especially the inconsistencies between experimental and predicted results, it is possible to determine which properties are underestimated or overestimated by the material structure-property relationship model.

[0084] S6. Based on the experimental results, iteratively optimize the material structure-property relationship model to adjust the material structure-property relationship model.

[0085] Specifically, iterative optimization of the material structure-property relationship model includes at least adjusting hyperparameters such as model complexity and learning rate; adding more training data to improve the model's generalization ability; and improving feature engineering to extract more relevant features. Furthermore, if necessary, more complex machine learning models or changes to the model architecture should be considered, and the feature selection method should be re-evaluated. Continuous iterative optimization continues until the model's predictions and experimental results achieve the expected consistency.

[0086] To further verify the effectiveness of the high-throughput screening method for high-temperature resistant soft magnetic materials according to an embodiment of the present invention, in a specific embodiment, for a trained material structure-property relationship model, yttrium (Y), iron (Fe), and chromium (Cr) elements are input according to the model's input rules. After training with a large number of rare earth-iron-based soft magnetic materials, the model will output possible stable crystal structures, thermodynamic parameters, and high-temperature magnetic properties based on fixed elements.

[0087] Specifically, stable crystal structures include the crystal structure type as well as the position and amount of doping elements. Material structure-property relationship models analyze the composition and interactions of elements to predict the possible crystal structure types that may form, such as cubic, hexagonal, or other more complex structures. For yttrium (Y), iron (Fe), and chromium (Cr), the model predicts a hexagonal crystal system.

[0088] When considering dopant elements (such as chromium), the model predicts the optimal location and amount of dopant. This includes the specific position of the dopant in the crystal lattice and its impact on the overall crystal structure stability. Specifically, the model predicts doping with two chromium atoms at the 4f and 12k sites of iron.

[0089] Thermodynamic parameters are involved in calculating the stability and reaction tendency of materials under different temperatures and environmental conditions. These parameters, including Gibbs free energy, enthalpy, and entropy, help predict the stability of materials at specific temperatures and pressures. Specifically, the model gives Y₂Fe 15 The Gibbs free energy of Cr2 is 8.2642 eV per atom.

[0090] Furthermore, thermodynamic parameters also include phase transitions. The model predicts the phase transitions of the material at different temperatures, such as transformation from one crystal structure to another, or from magnetic to non-magnetic. Specifically, the model gives the phase transition of Y₂Fe. 15 Cr2 is a stable single phase.

[0091] High-temperature magnetic property prediction assesses the magnetic behavior of materials at high temperatures, which is crucial for the application of soft magnetic materials. This includes magnetization, Curie temperature, permeability, hysteresis loss, magnetic saturation, and magnetic moment. Specifically, the model predicts Y₂Fe 15 The magnetic moment of Cr2 is 26.447 μB.

[0092] Furthermore, regarding Y2Fe 15 XRD diffraction experiments were conducted on Cr2, based on... Figure 5 and Figure 6 The experimental results confirmed that Y2Fe 15 The Cr2 structure is stable at both room temperature and high temperature, and the prediction is correct.

[0093] In another specific embodiment, for a trained material structure-property relationship model, the Curie temperature is input according to the model's input rules. After training with a large number of rare earth-iron-based soft magnetic materials, the model will output possible stable crystal structures and thermodynamic parameters based on fixed properties.

[0094] Specifically, a stable crystal structure includes the crystal structure type as well as the position and amount of doping elements. Based on the Curie temperature, the model assigns SmY3Fe as the rare-earth-iron-based soft magnetic material. 34 、SmYFe 17 CeYFe 17 YEr3Fe 34 Y2Al3Fe 14 Y2Ga3Fe 14 and Y2Ga2Fe 15 There are 18 crystal structures. The Gibbs free energy decreases from 0.034 eV per atom to 0.01 eV per atom.

[0095] Furthermore, VSM analysis was performed on the above 18 crystal structures, and it was found that the Curie temperature was higher than the input value, meaning that the selected crystal structures all met the criteria.

[0096] This invention provides a high-throughput screening method for high-temperature resistant soft magnetic materials. It establishes a specific theoretical model for efficiently screening high-temperature resistant soft magnetic materials and includes an experimental verification step to correct the theoretical model. This improves the efficiency and accuracy of material screening, expands the material design space, and can predict the performance of unknown materials and optimize the performance of existing materials.

[0097] Based on the above high-throughput screening methods, such as Figure 7 As shown, this embodiment of the invention provides a high-throughput screening system for high-temperature resistant soft magnetic materials, comprising:

[0098] Data collection and processing module 1 is used to acquire intrinsic structure data of soft magnetic materials and perform doping processing on the intrinsic structure data to obtain batch file packages; the intrinsic structure data corresponding to the batch file packages includes at least high-temperature magnetic stability data and high-temperature phase stability data of soft magnetic materials.

[0099] First-principles calculation module 2 is used to perform first-principles calculations on batch file packages to obtain result file packages; the structural features of the soft magnetic material corresponding to the result file packages include at least the doping elements, doping sites, and defect formation energies;

[0100] Machine learning training module 3 is used to train and adjust the initial machine learning model based on the result file package to obtain the target machine learning model. The soft magnetic material structural features corresponding to the result file package are used as the input data of the target machine learning model, and the high temperature magnetic stability data and high temperature phase stability data of the soft magnetic material corresponding to the batch file package are used as the output data of the target machine learning model to construct a material structure-property relationship model.

[0101] The high-throughput screening module 4 is used to predict the soft magnetic materials to be screened based on the material structure-property relationship model, and select high-temperature resistant soft magnetic materials based on the prediction results; the prediction results include the predicted values ​​of the high-temperature magnetic stability and the high-temperature phase stability of the soft magnetic materials to be screened.

[0102] It should be noted that each module in the aforementioned high-throughput screening system for high-temperature resistant soft magnetic materials can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module. For specific limitations regarding the high-throughput screening system for high-temperature resistant soft magnetic materials, please refer to the limitations of the high-throughput screening method for high-temperature resistant soft magnetic materials described above; both have the same function and role, and will not be repeated here.

[0103] This invention also provides a terminal device, which includes:

[0104] Processor, memory, and bus;

[0105] The bus is used to connect the processor and the memory;

[0106] The memory is used to store operation instructions;

[0107] The processor is configured to execute operations corresponding to the high-throughput screening method for high-temperature resistant soft magnetic materials described above by invoking the operation instructions.

[0108] In one alternative embodiment, a terminal device is provided, such as Figure 8 As shown, Figure 8 The terminal device 5000 shown includes a processor 5001 and a memory 5003. The processor 5001 and the memory 5003 are connected, for example, via a bus 5002. Optionally, the terminal device 5000 may also include a transceiver 5004. It should be noted that in practical applications, the transceiver 5004 is not limited to one type, and the structure of this terminal device 5000 does not constitute a limitation on the embodiments of the present invention.

[0109] Processor 5001 may be a CPU, a general-purpose processor, a DSP, an ASIC, an FPGA, or other programmable logic device, transistor logic device, hardware component, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. Processor 5001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0110] Bus 5002 may include a path for transmitting information between the aforementioned components. Bus 5002 may be a PCI bus or an EISA bus, etc. Bus 5002 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0111] The memory 5003 may be a ROM or other type of static storage device capable of storing static information and instructions, RAM or other type of dynamic storage device capable of storing information and instructions, or it may be an EEPROM, CD-ROM or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0112] The memory 5003 is used to store application code that executes the present invention, and its execution is controlled by the processor 5001. The processor 5001 is used to execute the application code stored in the memory 5003 to implement the content shown in any of the foregoing method embodiments.

[0113] In this context, "terminal device" refers to the controller.

[0114] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the high-throughput screening method for high-temperature resistant soft magnetic materials described above.

[0115] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the corresponding content in the foregoing method embodiments.

[0116] Furthermore, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0117] In summary, the embodiments of the present invention provide a high-throughput screening method, system, equipment, and medium for high-temperature resistant soft magnetic materials. This forms a specific theoretical model for efficiently screening high-temperature resistant soft magnetic materials and includes an experimental verification step to correct the theoretical model. This improves the efficiency and accuracy of material screening, expands the material design space, and can predict the performance of unknown materials and optimize the performance of existing materials.

[0118] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on its differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0119] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A high-throughput screening method for high-temperature resistant soft magnetic materials, characterized in that, include: The intrinsic structure data of the soft magnetic material is obtained, and the intrinsic structure data is doped to obtain a batch file package; the intrinsic structure data corresponding to the batch file package includes at least the high-temperature magnetic stability data and the high-temperature phase stability data of the soft magnetic material. First-principles calculations are performed on the batch file packages to obtain result file packages; the structural features of the soft magnetic material corresponding to the result file packages include at least the doping elements, doping sites, and defect formation energies. The initial machine learning model is trained and adjusted based on the result file package to obtain the target machine learning model. The soft magnetic material structural features corresponding to the result file package are used as the input data of the target machine learning model, and the high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material corresponding to the batch file package are used as the output data of the target machine learning model to construct a material structure-property relationship model. The structure-property relationship model is used to predict the soft magnetic materials to be screened, and high-temperature resistant soft magnetic materials are selected based on the prediction results. The prediction results include the predicted high-temperature magnetic stability and the predicted high-temperature phase stability of the soft magnetic materials to be screened. Experimental verification was conducted on the soft magnetic materials to be screened, and the experimental results were compared with the predicted results to evaluate the predictive effect of the material structure-property relationship model based on the comparison results. The material structure-property relationship model is iteratively optimized based on the experimental results to adjust the material structure-property relationship model; The step of training and adjusting the initial machine learning model based on the result file package to obtain the target machine learning model includes: The result file package is subjected to a second preprocessing to obtain the soft magnetic material structural features corresponding to the result file package, and the soft magnetic material structural features are organized into a feature dataset; the second preprocessing includes at least feature selection and normalization. The feature dataset is divided into a training set and a test set; An initial machine learning model is constructed based on a random forest, and the initial machine learning model is trained using the training set to obtain an intermediate machine learning model; The intermediate machine learning model is adjusted based on the test set to obtain the target machine learning model, including: The intermediate machine learning model is used to predict the test set to obtain the predicted value of the defect formation energy of the test set. The defect formation energy prediction value is used to perform first-principles calculations to obtain the defect formation energy calculation value of the test set; The variance between the predicted defect formation energy and the calculated defect formation energy is calculated, and the intermediate machine learning model is adjusted based on the variance to obtain the target machine learning model; the adjustment includes at least adjusting hyperparameters, reselecting features, and expanding the training set.

2. The high-throughput screening method according to claim 1, characterized in that, The process of acquiring intrinsic structure data of soft magnetic materials and performing doping processing on the intrinsic structure data to obtain a batch file package includes: Access a materials science database and retrieve intrinsic structure data of soft magnetic materials from the database; the materials science database includes at least AFLOWlib and OQMD. The intrinsic structure data is subjected to a first preprocessing, and the preprocessed intrinsic structure data is exported and saved as a CSV file package; the first preprocessing includes at least data cleaning and removal of missing values; The CSV file package is subjected to doping processing to obtain a batch file package.

3. The high-throughput screening method according to claim 2, characterized in that, The process of doping the CSV file package to obtain a batch of file packages includes: The crystal structure of the CSV file is read, and the preset doping elements are traversed through the placeholders in the crystal structure to obtain multiple doped intrinsic structure data files.

4. The high-throughput screening method according to claim 1, characterized in that, The first-principles calculation of the batch file package to obtain the result file package includes: The batch file packages were subjected to first-principles calculations using VASP to obtain the result file packages.

5. A high-throughput screening system for high-temperature resistant soft magnetic materials, characterized in that, The system is applied to the method as described in any one of claims 1 to 4, comprising: The data collection and processing module is used to acquire intrinsic structure data of soft magnetic materials and perform doping processing on the intrinsic structure data to obtain batch file packages; the intrinsic structure data corresponding to the batch file packages includes at least high-temperature magnetic stability data and high-temperature phase stability data of soft magnetic materials. The first-principles calculation module is used to perform first-principles calculations on the batch file package to obtain a result file package; the soft magnetic material structural features corresponding to the result file package include at least the doping element, doping site, and defect formation energy; The machine learning training module is used to train and adjust the initial machine learning model according to the result file package to obtain the target machine learning model. The soft magnetic material structural features corresponding to the result file package are used as the input data of the target machine learning model, and the high-temperature magnetic stability data and high-temperature phase stability data of the soft magnetic material corresponding to the batch file package are used as the output data of the target machine learning model to construct a material structure-property relationship model. The high-throughput screening module is used to predict the soft magnetic materials to be screened based on the material structure-property relationship model, and select high-temperature resistant soft magnetic materials based on the prediction results; the prediction results include the predicted high-temperature magnetic stability value and the predicted high-temperature phase stability value of the soft magnetic materials to be screened.

6. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the high-throughput screening method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the high-throughput screening method as described in any one of claims 1 to 4.

Images