Thermal conductivity prediction method and device, computer device and storage medium

Abstract

The application relates to a thermal conductivity prediction method and device, computer equipment and a storage medium. The method comprises the following steps: obtaining an initial structure file of a target crystal; a plurality of atomic data are included in the initial structure file; a plurality of perturbation structure files are generated based on the initial structure file and a preset supercell size; atomic information corresponding to each perturbation structure file is obtained based on a plurality of the perturbation structure files and a preset potential function model; and a thermal conductivity prediction result of the target crystal is obtained based on a plurality of the atomic information. The method can improve the thermal conductivity prediction efficiency.

Description

Technical Field

[0001] This application relates to the field of thermal conductivity prediction technology, and in particular to a thermal conductivity prediction method, apparatus, computer equipment, and storage medium. Background Technology

[0002] Thermal conductivity is one of the fundamental physical properties of materials, representing their ability to conduct heat. Higher thermal conductivity indicates a stronger ability to conduct heat, while lower thermal conductivity indicates a weaker ability. Materials with high and low thermal conductivity have important applications in their respective fields.

[0003] In traditional techniques, thermal conductivity is often calculated using first-principles calculations of the energy and forces of atoms in a structure. Based on these energy and forces, the spectrum and dispersion relations of phonons in the structure are determined, thereby obtaining thermal conductivity parameters such as the material's thermal conductivity. However, first-principles calculations are time-consuming and laborious. When predicting the thermal conductivity of a large number of structures with atomic positional shifts, the time required for prediction increases significantly.

[0004] It is evident that current methods for predicting thermal conductivity still suffer from low prediction efficiency. Summary of the Invention

[0005] Therefore, it is necessary to provide a thermal conductivity prediction method, apparatus, computer equipment, and storage medium that can improve the efficiency of thermal conductivity prediction in response to the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for predicting thermal conductivity, the method comprising:

[0007] Obtain the initial structure file of the target crystal; the initial structure file includes multiple atomic data.

[0008] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated;

[0009] Based on multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained;

[0010] Based on the atomic information, the predicted thermal conductivity of the target crystal is obtained.

[0011] In one embodiment, obtaining multiple perturbation structure files based on the initial structure file and the preset supercell size includes:

[0012] The preset supercell size is determined based on the perturbation structure file number threshold and / or supercell threshold;

[0013] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated.

[0014] In one embodiment, before obtaining the atomic information corresponding to each of the perturbation structure files based on multiple perturbation structure files and a preset potential function model, the process includes:

[0015] The positions of each atom in the initial structure file are moved individually and in combination, and a sample structure file is generated based on each type of movement, resulting in multiple sample structure files;

[0016] First-principles calculations are performed on multiple sample structure files to obtain sample atom information corresponding to each sample structure file;

[0017] Based on the sample structure file, the sample atom information, and the deep potential energy molecular dynamics algorithm, the preset potential function model is obtained.

[0018] In one embodiment, the step of moving the positions of different atoms in the initial structure file and generating a sample structure file based on each movement, resulting in multiple sample structure files, includes:

[0019] The positions of 1 to N atoms in the initial structure file are moved in three directions respectively, and a sample structure file is generated based on each movement, resulting in multiple sample structure files, where N is the number of atoms in the initial structure file.

[0020] In one embodiment, obtaining the atomic information corresponding to each of the perturbation structure files based on multiple perturbation structure files and a preset potential function model includes:

[0021] The perturbation structure file is converted into a perturbation structure file in the format of molecular dynamics software;

[0022] The transformed perturbation structure file is input into a preset potential function model to obtain the atomic information corresponding to each perturbation structure file.

[0023] In one embodiment, obtaining the predicted thermal conductivity of the target crystal based on the atomic information includes:

[0024] Based on the information of multiple atoms, the structural force information of the target crystal is calculated;

[0025] Based on the structural force information, preset mesh size and preset mesh density, the phonon properties of the target crystal are calculated, and the predicted thermal conductivity of the target crystal is obtained.

[0026] In one embodiment, the step of calculating the phonon properties of the target crystal based on the structural force information, a preset mesh size, and a preset mesh density, and obtaining the predicted thermal conductivity of the target crystal, includes:

[0027] Based on the structural force information, preset mesh size and preset mesh density, the phonon dispersion relation and phonon density of states of the target crystal are calculated;

[0028] Based on the phonon dispersion relation and phonon density of states, the predicted thermal conductivity of the target crystal is determined.

[0029] Secondly, this application provides a thermal conductivity prediction device, the thermal conductivity prediction device comprising:

[0030] An acquisition module is used to acquire the initial structure file of the target crystal; the initial structure file includes multiple atomic data.

[0031] The generation module is used to generate multiple perturbation structure files based on the initial structure file and the preset supercell size;

[0032] The potential function module is used to obtain the atomic information corresponding to each of the perturbation structure files based on multiple perturbation structure files and a preset potential function model;

[0033] The prediction module is used to obtain the predicted thermal conductivity of the target crystal based on multiple atomic information.

[0034] Thirdly, this application provides a computer device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0035] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0036] The aforementioned thermal conductivity prediction method, apparatus, computer equipment, and storage medium acquire an initial structure file of the target crystal; the initial structure file includes multiple atomic data; based on the initial structure file and a preset supercell size, multiple perturbation structure files are generated; based on the multiple perturbation structure files and a preset potential function model, atomic information corresponding to each perturbation structure file is obtained; based on the multiple atomic information, the thermal conductivity prediction result of the target crystal is obtained. This method can calculate the atomic information of the perturbation structure files using a preset potential function model, avoiding the time-consuming problem of first-principles calculations on multiple perturbation structure files, thereby shortening the overall time of thermal conductivity prediction and improving the efficiency of thermal conductivity prediction. Attached Figure Description

[0037] Figure 1 This is a diagram illustrating the application environment of the thermal conductivity prediction method in one embodiment.

[0038] Figure 2 This is a flowchart illustrating a thermal conductivity prediction method in one embodiment;

[0039] Figure 3 This is a flowchart illustrating the thermal conductivity prediction method in another embodiment;

[0040] Figure 4 This is a structural block diagram of a thermal conductivity prediction device in one embodiment;

[0041] Figure 5 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0043] The thermal conductivity prediction method provided in this application can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. The data storage system can store data that server 104 needs to process, or data that terminal 102 needs to process. The data storage system can be integrated on server 104, or it can be placed in the cloud or on other network servers. Terminal 102 obtains the initial structure file of the target crystal stored in server 104 by communicating with server 104; the initial structure file includes multiple atomic data; based on the initial structure file and a preset supercell size, multiple perturbation structure files are generated; based on the multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained; based on the multiple atomic information, the predicted thermal conductivity of the target crystal is obtained. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0044] In one embodiment, such as Figure 2 As shown, a method for predicting thermal conductivity is provided, and this method is applied to... Figure 1 Taking terminal 102 as an example, the explanation includes the following steps:

[0045] Step S100: Obtain the initial structure file of the target crystal.

[0046] The initial structure file of the target crystal includes multiple atomic data. Atomic data can include atomic position information and related characteristics, such as atom type and lattice constant. The initial structure file can be obtained through laboratory measurements or theoretical simulations; this embodiment does not impose any limitations on this.

[0047] Step S200: Based on the initial structure file and the preset supercell size, generate multiple perturbation structure files.

[0048] A supercell is a structure composed of repeating units of the original unit cell. The size of the supercell has a certain impact on the accuracy and computational complexity of the calculation. For example, the larger the supercell, the greater the required accuracy and computational complexity. The preset supercell size can be set based on prior knowledge or operator predictions, and this embodiment does not impose any limitations on this.

[0049] Based on the initial structure file and a preset supercell size, multiple perturbation structure files are generated. These files can be derived from the initial structure file, with slight variations depending on the choice of supercell size. These slight variations could be minor displacements of atomic positions, lattice distortions, etc. It is understandable that the actual structure of a crystal may not be perfectly aligned with the ideal state in the initial structure file; defects and dislocations may exist. Therefore, by generating perturbation structure files to calculate thermal conductivity, the presence of crystal defects can be considered, allowing for the prediction of corresponding thermal conductivity and thus achieving greater accuracy in thermal conductivity prediction.

[0050] Step S300: Based on multiple perturbation structure files and a preset potential function model, obtain the atomic information corresponding to each perturbation structure file.

[0051] Among them, the potential function model is used to describe the interaction between atoms, and can be an empirical potential function, an embedded atom model, or other models based on density functional theory.

[0052] Based on multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained. This can be achieved by calculating each perturbation structure file using the preset potential function model, thereby inferring the atomic information corresponding to each perturbation structure file. The atomic information may include one or more of the following: atom ID, atom type, atom position, and interatomic interaction forces.

[0053] The preset potential function model can be obtained through pre-training. For example, it can be trained using a specific perturbation structure and corresponding atomic information to obtain a potential function model that can be used to generate atomic information.

[0054] Step S400: Based on multiple atomic information, the predicted thermal conductivity of the target crystal is obtained.

[0055] In this process, the predicted thermal conductivity of the target crystal can be obtained based on information from multiple atoms. This can be achieved by analyzing the behavior of atoms and the propagation characteristics of phonons under different perturbation structures, thereby constructing a statistical model to predict the thermal conductivity. For example, methods such as molecular dynamics simulations (MD) and Monte Carlo simulations (MC) can be used to estimate the mean free path and lifetime of phonons, and then calculate the thermal conductivity.

[0056] This embodiment provides a thermal conductivity prediction method that obtains an initial structure file of the target crystal, which includes multiple atomic data. Based on the initial structure file and a preset supercell size, multiple perturbation structure files are generated. Based on the multiple perturbation structure files and a preset potential function model, atomic information corresponding to each perturbation structure file is obtained. Based on the multiple atomic information, the thermal conductivity prediction result of the target crystal is obtained. This method can calculate the atomic information of the perturbation structure files through the preset potential function model, avoiding the time-consuming problem of first-principles calculations for multiple perturbation structure files, thereby shortening the overall time of thermal conductivity prediction and improving the efficiency of thermal conductivity prediction.

[0057] In one embodiment, based on the initial structure file and the preset supercell size, multiple perturbation structure files are obtained, including:

[0058] The preset supercell size is determined based on the perturbation structure file number threshold and / or supercell threshold;

[0059] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated.

[0060] It is understandable that the size of the supercell is positively correlated with the number of perturbation structure files. The larger the supercell setting, the more perturbation structure files there are, and the greater the computational load required. Therefore, the size of the supercell can be limited by setting a threshold around the number of structure files.

[0061] Since the size of a supercell directly affects computational accuracy and computational cost, a larger supercell can better simulate long-range effects in real crystals, but it also increases the computational burden. Therefore, a maximum value for the preset supercell size can be determined based on a supercell threshold.

[0062] In one specific embodiment, the supercell size corresponding to the threshold number of perturbation file structures can be compared with the supercell size corresponding to the supercell threshold, and the smaller or larger supercell size can be selected as the preset supercell size.

[0063] This embodiment provides a thermal conductivity prediction method that determines a preset supercell size based on a perturbation structure file number threshold and / or a supercell threshold. This allows for the reasonable selection of the preset supercell size, thereby improving the accuracy and efficiency of the prediction results.

[0064] In one embodiment, before obtaining the atomic information corresponding to each perturbation structure file based on multiple perturbation structure files and a preset potential function model, the following steps are included:

[0065] The positions of each atom in the initial structure file are moved individually and in combination, and a sample structure file is generated based on each type of movement, resulting in multiple sample structure files;

[0066] First-principles calculations are performed on multiple sample structure files to obtain sample atom information corresponding to each sample structure file;

[0067] Based on the sample structure file, sample atom information, and deep potential energy molecular dynamics algorithm, a preset potential function model is obtained.

[0068] Individual movement can involve moving a single atom, while combined movement can involve moving multiple atoms. Individual and combined movements of each atom in the initial structure file can be performed by sequentially selecting one or more atoms and moving them individually, with the directions of movement for each atom in each movement being different.

[0069] The sample structure file is the structure file used to train the potential function model. The sample structure file is generated by individual and combined movements, thus obtaining the perturbation structure file based on the initial structure file. In this embodiment, although the sample structure contained in the sample structure file is also a perturbation structure, it is generated by moving the atomic positions, and therefore is not entirely the same as the perturbation structure file generated from the preset supercell size in the above embodiments.

[0070] First-principles calculations are performed on multiple sample structure files to obtain sample atomic information corresponding to each sample structure file. First-principles calculations are a method for calculating material properties based on quantum mechanics, and can use density functional theory or other quantum chemical methods. The obtained sample atomic information can include physical properties such as energy, force, and stress of all atoms in each sample structure file.

[0071] Based on sample structure files, sample atom information, and deep potential energy molecular dynamics algorithms, a preset potential function model is obtained. This can be achieved by using sample structure files as training input features and sample atom information as training output features to train the deep potential energy molecular dynamics algorithm and obtain the preset potential function model.

[0072] This embodiment provides a thermal conductivity prediction method that obtains a sample structure file and its corresponding sample atom information by moving atoms individually or in combination in an initial structure file. Based on the sample structure file, sample atom information, and deep potential energy molecular dynamics algorithm, a preset potential function model is obtained. This method can train a preset potential function model that accurately expresses the correlation between different perturbation structures and atomic information with less first-principles calculation. Therefore, when performing atomic information calculations on the perturbation structure file in the subsequent process, the long first-principles calculation can be avoided, and the prediction results can be obtained more quickly, thereby improving the prediction efficiency.

[0073] In one embodiment, the positions of different atoms in the initial structure file are moved separately, and a sample structure file is generated based on each movement, resulting in multiple sample structure files including:

[0074] The positions of atoms 1 to N in the initial structure file are moved in three directions respectively, and a sample structure file is generated based on each movement, resulting in multiple sample structure files, where N is the number of atoms in the initial structure file.

[0075] In this process, the positions of 1 to N atoms in the initial structure file are moved in three directions, which can be done along the x-axis, y-axis, and z-axis in a Cartesian coordinate system. For example, when only one atom is moved, the movement of each atom in each of the three directions can generate three different sample structure files. By generating one sample structure file for each type of movement, multiple sample structure files can be obtained; when multiple atoms are moved, the movements of multiple atoms in the three different directions can be combined to generate various different sample structure files.

[0076] Furthermore, the selection of the atoms to be moved and the specific direction of movement can be preset. For example, based on prior knowledge or sample statistics, atoms with high offset frequencies and their movement directions can be determined, and atoms with high offset frequencies can be selected for corresponding movement settings. This can reduce the amount of computational data while ensuring the accuracy of training results.

[0077] This embodiment provides a thermal conductivity prediction method that obtains multiple sample structure files by moving the positions of 1 to N atoms in an initial structure file in three directions. This effectively provides sample structure files for training, thereby enabling the training of a preset potential function model and improving prediction efficiency.

[0078] In one embodiment, based on multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained, including:

[0079] Convert the perturbation structure file into a perturbation structure file format suitable for molecular dynamics software;

[0080] The converted perturbation structure file is input into the preset potential function model to obtain the atomic information corresponding to each perturbation structure file.

[0081] For example, if the molecular dynamics software is LAMMPS, then converting the perturbation structure file into a perturbation structure file in the format required by the molecular dynamics software can be done by converting the perturbation structure file into the format required by LAMMPS software.

[0082] The converted perturbation structure file is input into the preset potential function model to obtain the atomic information corresponding to each perturbation structure file. This can be done by writing an input file for LAMMPS, such as the simulation time step, temperature control method, output frequency, etc. After running the software for simulation, the output file is obtained, which is the atomic information corresponding to each perturbation structure.

[0083] This embodiment provides a thermal conductivity prediction method that obtains atomic information corresponding to each perturbation structure file by inputting the converted perturbation structure file into a preset potential function model. This allows for the systematic acquisition of interactions between atoms in different perturbation structure files, thereby improving prediction efficiency.

[0084] In one embodiment, the predicted thermal conductivity of the target crystal is obtained based on multiple atomic information, including:

[0085] Based on information from multiple atoms, the structural force information of the target crystal is calculated.

[0086] Based on structural force information, preset mesh size and preset mesh density, the phonon properties of the target crystal are calculated, and the predicted thermal conductivity of the target crystal is obtained.

[0087] Among them, the structural force information of the target crystal is calculated based on multiple atomic information. This can be done by calculating the force information generated by the interaction between atoms, such as electrostatic force and van der Waals force, based on the atomic information corresponding to the perturbation structure file.

[0088] Phonon properties are the quantized modes of lattice vibrations in a crystal, and the phonon spectrum and dispersion relation are highly correlated with the thermal conductivity of the crystal. The preset mesh size refers to the sampling mesh size set in a three-dimensional spatial grid for calculating phonon properties; the size and density of the mesh determine the accuracy and efficiency of the calculation.

[0089] Based on structural force information, a preset mesh size, and a preset mesh density, the phonon properties of the target crystal are calculated. This can be achieved by calculating the structural force information according to the preset mesh size and density, obtaining phonon dispersion relations, such as the relationship between phonon energy and wave vector. Phonon properties also include phonon lifetime and phonon transport. Phonon lifetime, the average lifetime of phonons under different wave vectors, is an important indicator for evaluating phonon scattering effects. Phonon transport, or phonon transport coefficient, involves factors such as phonon group velocity and mode density.

[0090] Based on the phonon properties of the target crystal, the predicted thermal conductivity of the target crystal can be obtained. This can be achieved by using information such as phonon group velocity, lifetime, and mode density, combined with the formula for phonon thermal conductivity, to calculate the thermal conductivity of the crystal. Considering the characteristic of thermal conductivity changing with temperature, the thermal conductivity at different temperatures can also be calculated, thereby establishing the relationship between thermal conductivity and temperature.

[0091] This embodiment provides a thermal conductivity prediction method that calculates the predicted thermal conductivity of a target crystal using atomic information output from a preset potential function model, thereby improving prediction efficiency.

[0092] In one embodiment, based on structural force information, a preset mesh size, and a preset mesh density, the phonon properties of the target crystal are calculated, and the predicted thermal conductivity of the target crystal is obtained, including:

[0093] Based on structural force information, preset mesh size and preset mesh density, the phonon dispersion relation and phonon density of states of the target crystal are calculated.

[0094] Based on the phonon dispersion relation and phonon density of states, the predicted thermal conductivity of the target crystal is determined.

[0095] Here, the phonon dispersion relation can be the relationship between the phonon energy and the wave vector, and the phonon density of states can be the number of phonon modes per unit energy interval. Based on structural force information, a preset grid size, and a preset grid density, the phonon dispersion relation and phonon density of states of the target crystal can be calculated by transforming the structural force information into reciprocal space through a Fourier transform to obtain the phonon dispersion relation, and by integrating the phonon dispersion relation to obtain the phonon density of states function.

[0096] Based on the phonon dispersion relation and the phonon density of states, the predicted thermal conductivity of the target crystal can be determined by obtaining the phonon frequency from the phonon dispersion relation, estimating the mean free path of the phonon from the group velocity, calculating the thermal conductivity of each phonon mode using the phonon thermal conductivity formula, integrating the thermal conductivity of all phonon modes to obtain the total thermal conductivity, which is then used as the predicted thermal conductivity of the target crystal.

[0097] This embodiment provides a thermal conductivity prediction method that determines the predicted thermal conductivity of a target crystal based on phonon dispersion relations and phonon density of states, thereby improving prediction efficiency.

[0098] To more clearly illustrate the technical solution of this application, a detailed embodiment is also provided.

[0099] In one embodiment, such as Figure 3 As shown, a method for predicting thermal conductivity is provided, taking barium zirconate crystal as an example, including:

[0100] Based on the initial structure file, tiny perturbations are applied to each atom to generate several desired structure files. First-principles calculations are then used to obtain the energy and force corresponding to each structure file, and the results of these calculations are used to obtain a preset potential function model. In this embodiment, the preset potential function model is a deep learning potential function model.

[0101] Applying minute perturbations to each atom involves moving any one atom in the initial structure file, with each atom moving 0.003 nanometers along the x, y, and z directions respectively, generating three different structures. Repeating this step for each atom in the initial structure file yields 3N different structures, where N is the total number of atoms in the initial structure file. Subsequently, moving any two, three, up to N atoms in the initial structure file generates a large number of perturbation-based structure files, i.e., sample structure files.

[0102] In this embodiment, the deep learning potential function model is constructed using DeepMD software based on sample structure files. DeepMD software reads the results of first-principles calculations and constructs the deep learning potential function model. Because this model is based on perturbation structures, compared to general deep learning potential function models, it sacrifices accuracy far from the equilibrium position, while further ensuring accuracy near the equilibrium position. Since the calculation of thermal conductivity is highly sensitive to the energy and forces near the equilibrium position, this model has higher accuracy in predicting the thermal conductivity of materials.

[0103] Using the phonon calculation software Phono3py, the supercell size is set based on the initial structure file to generate the structure files required for calculating thermal conductivity, i.e., the perturbation structure files to be calculated, as well as the phono3py_disp.yaml file. The phono3py_disp.yaml file records the relative displacement information between the structures. In phono3py, the supercell size refers to the size of the supercell in the atomic model used to calculate the lattice thermal conductivity. The supercell size determines the fineness of the atomic arrangement, thus affecting the accuracy and precision of the thermal conductivity calculation. Generally, using a larger supercell can improve the accuracy of the calculation, but it also increases the complexity and time consumption of the calculation. The number of perturbation structure files required for the generated phonon calculation is related to the supercell size and symmetry used. Larger supercells usually result in the generation of more perturbation structures because more atoms are affected and produce different degrees of displacement. In addition, lower crystal symmetry also produces more structures. Therefore, when performing phonon property calculations, the supercell size and crystal symmetry are two key factors affecting the number of generated structures.

[0104] Based on the perturbation structure file, a deep learning potential energy model is used to calculate the forces of each structure in the barium zirconate crystal, i.e., the structural force information, and to calculate the thermal conductivity at different temperatures.

[0105] Specifically, calculating the forces of various structures in a barium zirconate crystal using a deep learning potential energy model can involve preparing a configuration file for the training potential function for each structure file. This means preparing the input and configuration files for the deep learning potential energy model, and generating the corresponding output file for that structure. Preparing the input and configuration files for the deep learning potential energy model can involve preparing the structure files required for use in molecular dynamics software based on the deep learning potential energy model. In this embodiment, Automsk software is used to convert 517 perturbation structure files into LAMMPS files required by the molecular dynamics software. For each LAMMPS file, a corresponding in.lammps runtime parameter file is configured, specifying the name of the perturbation structure file used, the force field file, and other parameters required by molecular dynamics.

[0106] In the application, the deep learning potential function model calculates the interatomic interaction forces based on the structure file required for calculating thermal conductivity and records them in the output file. The output file records the atomic information during the simulation, including the atom's ID, type, position (x, y, z), and force (fx, fy, fz), and is stored in the dump file.

[0107] Calculating the forces in a barium zirconate crystal structure. This can be achieved by converting the output file of a deep learning potential energy model to a different format, and then using the converted file as input to phono3py to calculate the force information of the barium zirconate crystal structure. Converting the energy calculation file output by the deep learning potential energy model involves using a shell script to convert the dump file output by the deep learning potential energy model into XML format for use by phono3py. Phono3py uses the output file of 517 perturbation structures to calculate the second-order and third-order force constants, generating structural force information.

[0108] Thermal conductivity calculations based on structural force information can be performed at different temperatures. Based on the calculated second- and third-order force constants, different grid sizes and densities are used to calculate phonon properties. By adjusting grid parameters, the grid size and distribution used to sample phonon wave vectors during phonon lattice dynamics calculations are controlled, and the thermal conductivity calculation results are saved. Setting grid parameters refers to the grid used in phonon physics. Specifically, for phonon lattice dynamics calculations, a grid needs to be sampled within the first Brillouin zone to calculate the phonon dispersion relation and phonon density of states. This grid is often referred to as the phonon's k-point grid. In phono3py, phonon properties are calculated by setting grid sizes and densities.

[0109] Understandably, traditional methods for calculating phonon interactions typically require significant computational resources and time. Machine learning-based methods, however, can quickly and efficiently predict phonon interaction thermal conductivity by training models, greatly saving computational costs and time. Furthermore, constructing machine learning potential functions based on perturbation structures can significantly improve the accuracy of the calculation results. By flexibly adjusting the cell size to generate perturbation structure cell files, the requirements for both computational cost and accuracy can be met. Using deep learning potential function models, atomic simulation information from perturbation structure cell files can be quickly predicted. These deep learning potential function models, built upon perturbation structures, exhibit higher accuracy near equilibrium positions compared to general deep learning potential function models. Converting traditional perturbation structure files into input files usable by deep learning potential energy models and converting the output files into files readable by phono3py allows for the calculation of material thermal conductivity.

[0110] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0111] Based on the same inventive concept, this application also provides a thermal conductivity prediction device for implementing the thermal conductivity prediction method described above. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more thermal conductivity prediction device embodiments provided below can be found in the limitations of the thermal conductivity prediction method described above, and will not be repeated here.

[0112] In one embodiment, such as Figure 4 As shown, a thermal conductivity prediction device is provided, comprising: an acquisition module 100, a generation module 200, a potential function module 300, and a prediction module 400, wherein:

[0113] The acquisition module 100 is used to acquire the initial structure file of the target crystal; the initial structure file includes multiple atomic data.

[0114] The generation module 200 is used to generate multiple perturbation structure files based on the initial structure file and the preset supercell size;

[0115] Potential function module 300 is used to obtain the atomic information corresponding to each perturbation structure file based on multiple perturbation structure files and a preset potential function model;

[0116] The prediction module 400 is used to obtain the predicted thermal conductivity of the target crystal based on information from multiple atoms.

[0117] In one embodiment, the generation module 200 is further configured to:

[0118] The preset supercell size is determined based on the perturbation structure file number threshold and / or supercell threshold;

[0119] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated.

[0120] In one embodiment, the thermal conductivity prediction system further includes a potential function training module for:

[0121] The positions of each atom in the initial structure file are moved individually and in combination, and a sample structure file is generated based on each type of movement, resulting in multiple sample structure files;

[0122] First-principles calculations are performed on multiple sample structure files to obtain sample atom information corresponding to each sample structure file;

[0123] Based on the sample structure file, sample atom information, and deep potential energy molecular dynamics algorithm, a preset potential function model is obtained.

[0124] In one embodiment, the potential function training module is further used for:

[0125] The positions of atoms 1 to N in the initial structure file are moved in three directions respectively, and a sample structure file is generated based on each movement, resulting in multiple sample structure files, where N is the number of atoms in the initial structure file.

[0126] In one embodiment, the potential function module 300 is further configured to:

[0127] Convert the perturbation structure file into a perturbation structure file format suitable for molecular dynamics software;

[0128] The converted perturbation structure file is input into the preset potential function model to obtain the atomic information corresponding to each perturbation structure file.

[0129] In one embodiment, the prediction module 400 is further configured to:

[0130] Based on information from multiple atoms, the structural force information of the target crystal is calculated.

[0131] Based on structural force information, preset mesh size and preset mesh density, the phonon properties of the target crystal are calculated, and the predicted thermal conductivity of the target crystal is obtained.

[0132] In one embodiment, the prediction module 400 is further configured to:

[0133] Based on structural force information, preset mesh size and preset mesh density, the phonon dispersion relation and phonon density of states of the target crystal are calculated.

[0134] Based on the phonon dispersion relation and phonon density of states, the predicted thermal conductivity of the target crystal is determined.

[0135] Each module in the aforementioned thermal conductivity prediction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0136] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a thermal conductivity prediction method. The display screen can be an LCD screen or an e-ink display screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0137] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0138] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the thermal conductivity prediction method of any of the above embodiments:

[0139] Obtain the initial structure file of the target crystal; the initial structure file includes data for multiple atoms;

[0140] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated;

[0141] Based on multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained;

[0142] Based on information from multiple atoms, the predicted thermal conductivity of the target crystal is obtained.

[0143] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the thermal conductivity prediction method of any of the above embodiments:

[0144] Obtain the initial structure file of the target crystal; the initial structure file includes data for multiple atoms;

[0145] Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated;

[0146] Based on multiple perturbation structure files and a preset potential function model, the atomic information corresponding to each perturbation structure file is obtained;

[0147] Based on information from multiple atoms, the predicted thermal conductivity of the target crystal is obtained.

[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0149] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0150] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0151] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for predicting thermal conductivity, characterized in that, The thermal conductivity prediction method includes: Obtain the initial structure file of the target crystal; the initial structure file includes multiple atomic data. Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated; Based on multiple perturbation structure files and a preset potential function model, atomic information corresponding to each perturbation structure file is obtained. Before obtaining the atomic information corresponding to each perturbation structure file, the process includes: individually and in combination moving the position of each atom in the initial structure file, and generating a sample structure file based on each type of movement, resulting in multiple sample structure files; performing first-principles calculations on the multiple sample structure files to obtain sample atomic information corresponding to each sample structure file; and obtaining the preset potential function model based on the sample structure files, the sample atomic information, and the deep potential energy molecular dynamics algorithm. Based on the atomic information, the predicted thermal conductivity of the target crystal is obtained.

2. The thermal conductivity prediction method according to claim 1, characterized in that, The process of obtaining multiple perturbation structure files based on the initial structure file and the preset supercell size includes: The preset supercell size is determined based on the perturbation structure file number threshold and / or supercell threshold; Based on the initial structure file and the preset supercell size, multiple perturbation structure files are generated.

3. The thermal conductivity prediction method according to claim 1, characterized in that, The step of moving the positions of different atoms in the initial structure file and generating a sample structure file based on each movement, resulting in multiple sample structure files, includes: The positions of 1 to N atoms in the initial structure file are moved in three directions respectively, and a sample structure file is generated based on each movement, resulting in multiple sample structure files, where N is the number of atoms in the initial structure file.

4. The thermal conductivity prediction method according to claim 1, characterized in that, The process of obtaining the atomic information corresponding to each of the perturbation structure files based on multiple perturbation structure files and a preset potential function model includes: The perturbation structure file is converted into a perturbation structure file in the format of molecular dynamics software; The transformed perturbation structure file is input into a preset potential function model to obtain the atomic information corresponding to each perturbation structure file.

5. The thermal conductivity prediction method according to claim 1, characterized in that, The process of obtaining the predicted thermal conductivity of the target crystal based on multiple atomic information includes: Based on the information of multiple atoms, the structural force information of the target crystal is calculated; Based on the structural force information, preset mesh size and preset mesh density, the phonon properties of the target crystal are calculated, and the predicted thermal conductivity of the target crystal is obtained.

6. The thermal conductivity prediction method according to claim 5, characterized in that, The calculation of the phonon properties of the target crystal based on the structural force information, preset mesh size, and preset mesh density, to obtain the predicted thermal conductivity of the target crystal, includes: Based on the structural force information, preset mesh size and preset mesh density, the phonon dispersion relation and phonon density of states of the target crystal are calculated; Based on the phonon dispersion relation and phonon density of states, the predicted thermal conductivity of the target crystal is determined.

7. A thermal conductivity prediction device, characterized in that, The thermal conductivity prediction device includes: An acquisition module is used to acquire the initial structure file of the target crystal; the initial structure file includes multiple atomic data. The generation module is used to generate multiple perturbation structure files based on the initial structure file and the preset supercell size; The potential function module is used to obtain atomic information corresponding to each of the multiple perturbation structure files and a preset potential function model. Before obtaining the atomic information corresponding to each of the multiple perturbation structure files and the preset potential function model, the module includes: individually and in combination moving the position of each atom in the initial structure file, and generating a sample structure file based on each type of movement, resulting in multiple sample structure files; performing first-principles calculations on the multiple sample structure files to obtain sample atomic information corresponding to each sample structure file; and obtaining the preset potential function model based on the sample structure files, the sample atomic information, and the deep potential energy molecular dynamics algorithm. The prediction module is used to obtain the predicted thermal conductivity of the target crystal based on multiple atomic information.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

Images