Ion transport performance evaluation method and device, computer equipment and storage medium

Abstract

The invention relates to an ion transport performance evaluation method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring transport process simulation parameters determined based on a transport interface structure of dissolved ions; based on the transport process simulation parameters and a target machine learning force field obtained by training according to a transport interface structure of the dissolved ions in advance, carrying out molecular dynamics simulation on a dissolution process of the dissolved ions on the transport interface to obtain dissolution stress information of the dissolved ions; and determining an ion transport performance evaluation result of the dissolved ions according to the dissolution stress information. By adopting the method, the calculation cost can be reduced, and the ion transport performance evaluation rate can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of material property analysis, and particularly to an ion transport performance method, device, computer device, storage medium, and computer program product. Background Art

[0002] The ion transport process at the solid-liquid interface has important research value in the fields of materials science and chemical engineering. Understanding the ease of ion transport across the interface is crucial for improving the application performance and properties of materials. For example, in lithium-ion batteries, the ion transport at the electrode-electrolyte interface directly affects important performance parameters such as the charge-discharge rate, capacity, and lifespan of the battery. Therefore, the ion transport properties of various electrodes and electrolytes in the battery have been widely studied.

[0003] When detecting and evaluating ion transport performance, it is usually necessary to use the molecular dynamics simulation method based on first principles to perform kinetic simulations on the movement process of ions in the transport system, so as to evaluate the ease of ion transport across the interface. However, due to the excessive computational cost, the simulation duration and the size of the simulation system are both limited in the molecular dynamics simulation method based on first principles, resulting in problems such as large computational costs and low evaluation rates. Summary of the Invention

[0004] Based on this, it is necessary to provide an ion transport performance evaluation method, device, computer device, computer-readable storage medium, and computer program product that can reduce the computational cost and improve the evaluation rate for the above technical problems.

[0005] In a first aspect, the present application provides an ion transport performance evaluation method, and the method includes:

[0006] Obtain the simulation parameters of the transport process determined based on the transport interface structure of the dissolved ions;

[0007] Based on the simulation parameters of the transport process and the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface to obtain the dissolution force information of the dissolved ions;

[0008] Determine the ion transport performance evaluation result of the dissolved ions according to the dissolution force information.

[0009] In the above embodiment, by introducing a machine learning force field for force prediction analysis in the molecular dynamics simulation process, the computational cost in the ion transport performance evaluation process can be effectively reduced, and the accuracy and rate of determining the ion transport performance evaluation result according to the dissolution force information are improved.

[0010] In one embodiment, based on the transport process simulation parameters and a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, a molecular dynamics simulation is performed on the dissolution process of the dissolved ions at the transport interface to obtain the dissolution force information of the dissolved ions, including:

[0011] Based on the transport process simulation parameters, a molecular dynamics simulation is performed on the dissolution process of the dissolved ions at the transport interface to obtain the transport interface structure during the dissolution process;

[0012] The target machine learning force field pre-trained according to the transport interface structure of the dissolved ions is called through a preset call interface;

[0013] The transport interface structure during the dissolution process is input into the target machine learning force field to obtain the dissolution force information of the dissolved ions.

[0014] In the above embodiment, by using the target machine learning force field pre-trained to analyze the force condition of the dissolved ions during the transport process for multiple frames of transport interface structures during the dissolution process, the dissolution force information of the dissolved ions can be obtained quickly and accurately, effectively reducing the calculation cost in the process of evaluating the ion transport performance, and improving the accuracy and rate of determining the ion transport performance evaluation result based on the dissolution force information subsequently.

[0015] In one embodiment, the obtaining of the transport process simulation parameters determined based on the transport interface structure of the dissolved ions includes:

[0016] Obtain the single-ion dissolution distance during the simulated transport process;

[0017] Based on the single-ion dissolution distance and the transport interface structure of the dissolved ions, determine the simulated dissolution times during the simulated transport process;

[0018] Determine the single-ion dissolution distance, the simulated dissolution times, and the transport interface structure of the dissolved ions as the transport process simulation parameters.

[0019] In the above embodiment, by determining the single-ion dissolution distance, the simulated dissolution times, and the transport interface structure of the dissolved ions as the transport process simulation parameters, it can provide a guiding role for the subsequent simulation process and improve the simulation accuracy of the transport process.

[0020] In one embodiment, the training method of the target machine learning force field includes:

[0021] According to the transport interface structure of the dissolved ions, determine the training simulation parameters for the transport process of the dissolved ions;

[0022] Based on the training simulation parameters, perform molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface to obtain the force field training data set of the initial machine learning force field;

[0023] Perform model learning on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolved ions.

[0024] In the above embodiments, by performing model learning on the initial machine learning force field in the same simulation scenario as in the actual evaluation, a target machine learning force field that can quickly and accurately analyze the force on the dissolved ions during transport in the current transport interface structure can be obtained, providing an analysis and prediction tool for quickly obtaining the accurate dissolution force information of the dissolved ions during actual evaluation, effectively reducing the computational cost in the process of ion transport performance evaluation, and improving the evaluation accuracy and evaluation rate of ion transport performance.

[0025] In one of the embodiments, the training simulation parameters include the single dissolution distance for training, the number of dissolution times for training, and the transport interface structure of the dissolved ions, and the single dissolution distance for training is greater than the single ion dissolution distance during actual simulation;

[0026] The performing molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface based on the training simulation parameters to obtain the force field training data set of the initial machine learning force field includes:

[0027] Update the position of the dissolved ions in the transport interface structure according to the single dissolution distance for training to obtain the training input interface structure;

[0028] Perform molecular dynamics training simulation based on the training input interface structure to obtain multi-frame interface structure data during the training simulation of the training input interface structure, and the structure energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure;

[0029] After the training simulation of the training input interface structure is completed, update the position of the dissolved ions in the training input interface structure based on the single dissolution distance for training to obtain the updated training input interface structure, and return to execute the step of performing molecular dynamics training simulation based on the training input interface structure until the number of updates of the training input interface structure is equal to the number of dissolution times for training;

[0030] Determine the multi-frame interface structure data obtained based on the molecular dynamics training simulation, and the structure energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure as the force field training data set of the initial machine learning force field.

[0031] In the above embodiments, by setting the training single ion dissolution distance greater than the single ion dissolution distance during actual simulation, the computational cost of the training process can be effectively reduced without much impact on the prediction accuracy of the obtained target machine learning force field, thereby effectively reducing the evaluation computational cost of ion transport performance. At the same time, data interaction between the evaluation system and the target machine learning force field can be carried out through a preset call interface to achieve serialization of multiple computational tasks and reduce cumbersome manual operations.

[0032] In one of the embodiments, the model learning of the initial machine learning force field based on the force field training data set to obtain the target machine learning force field of the dissolved ions includes:

[0033] Dividing the force field training data set into a training set and a test set according to a preset division ratio;

[0034] Performing model learning on the initial machine learning force field according to the training set to obtain a machine learning force field to be tested;

[0035] Obtaining the force field parameters of the machine learning force field to be tested;

[0036] When it is determined that the machine learning force field to be tested meets the preliminary verification condition based on the comparison result between the force field parameters and the preset preliminary verification parameter information, testing the machine learning force field to be tested according to the test set;

[0037] When it is determined that the test of the machine learning force field to be tested is successful, determining the machine learning force field to be tested as the target machine learning force field of the dissolved ions.

[0038] In the above embodiments, the machine learning force field to be tested is preliminarily verified through the preliminary verification parameter information. Only when the preliminary verification is successful, will the machine learning force field to be tested be further tested, reducing the test cost of the machine learning force field to be tested. Determining the machine learning force field to be tested with a successful test as the target machine learning force field. Through two tests, the prediction and analysis accuracy of the finally determined target machine learning force field can also be further improved, thereby improving the evaluation accuracy of ion transport performance.

[0039] In one of the embodiments, the construction method of the transport interface structure of the dissolved ions includes:

[0040] Constructing the initial transport interface structure of the dissolved ions according to the construction parameters of the transport interface structure, where the construction parameters include the type of dissolved ions, the surface structure data of the vacuum layer, the solvent molecule structure data, and the solvent molecule density data;

[0041] Perform a single molecular dynamics simulation on the initial transport interface structure to obtain an equilibrium transport interface structure in a pre-equilibrium state;

[0042] Obtain the solvent molecule density data in the middle of the liquid phase in the equilibrium transport interface structure;

[0043] When it is determined according to the solvent molecule density data in the middle of the liquid phase that the equilibrium transport interface structure meets the structural rationality requirements, determine the equilibrium transport interface structure as the transport interface structure of the dissolved ions.

[0044] In the above embodiments, by performing a single molecular dynamics simulation on the constructed initial transport interface structure and evaluating the density rationality of the equilibrium transport interface structure in a pre-equilibrium state, the structural rationality of the finally constructed transport interface structure of the dissolved ions can be effectively improved.

[0045] In one of the embodiments, the method further includes:

[0046] When it is determined according to the solvent molecule density data in the middle of the liquid phase that the equilibrium transport interface structure does not meet the structural rationality requirements, determine the solvent molecule correction data of the equilibrium transport interface structure;

[0047] Based on the solvent molecule correction data, correct the solvent molecule density in the middle of the liquid phase of the equilibrium transport interface structure to obtain the transport interface structure of the dissolved ions.

[0048] In the above embodiments, by correcting the solvent molecule density of the equilibrium transport interface structure, the structural rationality of the finally constructed transport interface structure of the dissolved ions can be effectively improved.

[0049] In one of the embodiments, the determining the evaluation result of the ion transport performance of the dissolved ions according to the dissolution force information includes:

[0050] Generate a free energy change curve of the dissolved ions during the dissolution process according to the dissolution force information;

[0051] Determine the dissolution energy barrier for the dissolved ions to transport in the transport interface structure based on the free energy change curve;

[0052] According to the dissolution energy barrier and the preset corresponding relationship between the energy barrier and the ease of dissolution, determine the evaluation result of the ion transport performance of the dissolved ions.

[0053] In the above embodiments, by presetting the corresponding relationship between the energy barrier and the ease of dissolution, the evaluation system can directly evaluate and determine the ease of dissolution of ions, improving the convenience of ion transport performance evaluation.

[0054] An ion transport performance evaluation device, the device comprising:

[0055] A simulation parameter acquisition module for acquiring simulation parameters of a transport process determined based on the transport interface structure of dissolved ions;

[0056] A simulation module for performing molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the simulation parameters of the transport process and a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, to obtain the dissolution force information of the dissolved ions;

[0057] A performance evaluation module for determining an evaluation result of the ion transport performance of the dissolved ions according to the dissolution force information.

[0058] In a second aspect, the present application further provides a computer device, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the above method when executing the computer program.

[0059] In a third aspect, the present application further provides a computer-readable storage medium, having a computer program stored thereon, and the computer program implementing the steps of the above method when executed by a processor.

[0060] In a fourth aspect, the present application further provides a computer program product, comprising a computer program, and the computer program implementing the steps of the above method when executed by a processor.

[0061] For the above ion transport performance evaluation method, device, computer device, storage medium and computer program product, by acquiring simulation parameters of the transport process determined based on the transport interface structure of dissolved ions, it is possible to guide the dissolution process of dissolved ions at the transport interface. Based on the simulation parameters of the transport process and a target machine learning force field pre-trained according to the transport interface structure of dissolved ions, molecular dynamics simulation is performed on the dissolution process of dissolved ions at the transport interface. Since the target machine learning force field is a machine learning force field pre-trained according to the transport interface structure of dissolved ions, when performing molecular dynamics simulation on dissolved ions, the force condition of dissolved ions during the dissolution process can be quickly and accurately predicted according to the dissolution condition of dissolved ions at the transport interface, to obtain the dissolution force information of dissolved ions. By introducing a machine learning force field for force prediction analysis during the molecular dynamics simulation process, the calculation cost in the ion transport performance evaluation process can be effectively reduced, and the accuracy and rate of determining the ion transport performance evaluation result according to the dissolution force information subsequently are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is an application environment diagram of the ion transport performance evaluation method in an embodiment;

[0063] Figure 2 It is a schematic flowchart of a method for evaluating ion transport performance in an embodiment;

[0064] Figure 3 It is a schematic flowchart of the steps for determining the ion transport performance evaluation result of the dissolving ions according to the dissolution force information in an embodiment;

[0065] Figure 4 It is a schematic flowchart of the construction method of the transport interface structure of the dissolving ions in an embodiment;

[0066] Figure 5 It is a schematic flowchart of the steps for performing molecular dynamics simulation on the dissolution process of the dissolving ions at the transport interface based on the transport process simulation parameters and the target machine learning force field pre-trained according to the transport interface structure of the dissolving ions to obtain the dissolution force information of the dissolving ions;

[0067] Figure 6 It is a schematic flowchart of the steps for obtaining the transport process simulation parameters determined based on the transport interface structure of the dissolving ions in an embodiment;

[0068] Figure 7 It is a schematic flowchart of the training method of the target machine learning force field in an embodiment;

[0069] Figure 8 It is a schematic flowchart of the steps for performing molecular dynamics training simulation on the dissolution process of the dissolving ions at the transport interface based on the training simulation parameters to obtain the force field training data set of the initial machine learning force field in an embodiment;

[0070] Figure 9 It is a schematic flowchart of the steps for performing model learning on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolving ions in an embodiment;

[0071] Figure 10 It is a schematic flowchart of a method for evaluating ion transport performance in another embodiment;

[0072] Figure 11 It is a schematic diagram of the specific model structure of the solid-liquid interface model in an embodiment;

[0073] Figure 12 It is a schematic diagram of the analysis result of the solvent molecular density in the middle of the liquid phase in an embodiment;

[0074] Figure 13 It is a schematic diagram of the test performance of the target machine learning force field in an embodiment;

[0075] Figure 14It is a diagram showing the change of the force in the z - direction during the Li - ion dissolution process in the TiO2 - EC system in an embodiment;

[0076] Figure 15 It is a schematic diagram showing the change of the average force in the z - direction and the change of free energy during the Li dissolution process in the TiO2 - EC system in an embodiment;

[0077] Figure 16 It is a structural block diagram of an ion transport performance evaluation device in an embodiment;

[0078] Figure 17 It is an internal structure diagram of a computer device in an embodiment. Detailed implementation manners

[0079] Hereinafter, embodiments of the technical solutions of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solutions of the present application more clearly, and thus are only examples and should not be used to limit the protection scope of the present application.

[0080] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above - mentioned drawings are intended to cover non - exclusive inclusion.

[0081] Referring to "embodiment" herein means that a specific feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of this application. The phrase appears in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0082] In the description of the embodiments of this application, the term "plurality" means two or more (including two). Similarly, "multiple groups" means two or more groups (including two groups), and "multiple pieces" means two or more pieces (including two pieces).

[0083] In the description of the embodiments of the present application, the orientation or positional relationship indicated by technical terms such as "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the embodiments of the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation on the embodiments of the present application.

[0084] In the description of the embodiments of the present application, unless otherwise clearly specified and limited, technical terms such as "installation", "connection", "connection", "fixation", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can also be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0085] The prediction and evaluation of the ease of ion cross-interface transport are of great significance for studying the application performance of materials and improving material properties. Especially in lithium-ion batteries, the ease of ion transport at the solid-liquid interface, that is, the electrode-electrolyte interface, will directly affect important performance parameters such as the charge and discharge rate, capacity, and life of the battery. Therefore, evaluating the transport performance of ions across the interface can provide important data reference materials for the development of products such as batteries.

[0086] When ions transport across the interface, under the influence of the interaction force, the ease of ion transport is related to the movement of each atom inside the transport structure during the transport process. Therefore, when studying the transport performance of ions across the interface, the first-principles molecular dynamics simulation is usually used to simulate the ion transport process, and the ease of ion transport is determined according to the force condition of the ions during the simulation. However, due to the excessive computational cost, the simulation duration and the size of the simulation system are limited to a certain extent in the molecular dynamics simulation method based on the first principles, and there are problems such as a large computational cost and a low evaluation rate.

[0087] In order to improve the evaluation rate of ion transport performance and reduce the evaluation cost, when evaluating the ion transport performance, a target machine learning force field can be obtained in advance by training the transport interface structure required for ion transport according to the dissolved ions. During the evaluation, only the transport process simulation parameters determined based on the transport interface structure of the dissolved ions need to be obtained, and the target machine learning force field is called to perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the transport process simulation parameters. Then, according to the dissolution situation of the dissolved ions at the transport interface, the force condition of the dissolved ions during the dissolution process can be quickly and accurately predicted, and the dissolution force information of the dissolved ions can be obtained. By introducing the machine learning force field in the molecular dynamics simulation process for force prediction analysis, the calculation cost in the ion transport performance evaluation process can be effectively reduced, and the accuracy and rate of determining the ion transport performance evaluation result based on the dissolution force information are improved.

[0088] The ion transport performance evaluation method provided by the embodiments of the present application can be applied to an application environment such as Figure 1 shown. Among them, the evaluation system 102 communicates with the user terminal 104 through the network. The data storage system can store the data that the evaluation system 102 needs to process. The data storage system can be integrated on the evaluation system 102, or placed in the cloud or other network servers. The evaluation system 102 obtains the transport process simulation parameters determined based on the transport interface structure of the dissolved ions through the user terminal 104, and performs molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the transport process simulation parameters and the target machine learning force field obtained in advance by training the transport interface structure of the dissolved ions, so as to obtain the dissolution force information of the dissolved ions. Among them, the evaluation system 102 can be integrated on the user terminal 104 or the server. The user terminal 104 can be, but is not limited to, various personal computers, laptop computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The portable wearable device can be a smart watch, a smart bracelet, etc. The server can be implemented by an independent server or a server cluster composed of multiple servers.

[0089] In some embodiments, as Figure 2 shown, an ion transport performance evaluation method is provided. Taking the method applied to the evaluation system 102 in Figure 1 as an example, the method includes the following steps:

[0090] S202, obtain the transport process simulation parameters determined based on the transport interface structure of the dissolved ions.

[0091] Among them, the dissolved ions can be transport ions that need to undergo cross-interface transport at the transport interface structure to achieve corresponding chemical properties. The specific types of dissolved ions can be determined according to the actual ion transport performance evaluation scenario. For example, when evaluating the ion transport performance of ions in a lithium battery, the corresponding dissolved ions can be the ions in the lithium battery that need to be transported at the electrode-electrolyte interface to achieve the charge and discharge effects, such as lithium ions (Li + ), manganese ions (Mn 2+ ), etc.

[0092] The transport interface structure is the scenario where ion transport occurs. Generally, there are two interfaces in the transport interface structure, and ions complete cross-interface transport in the transport interface structure. The cross-interface transport of ions generally exists in the solid-liquid interface. Therefore, the transport structure interfaces in this application are all exemplified by solid-liquid structure interfaces, such as the electrode-electrolyte interface in a lithium battery.

[0093] The transport process simulation parameters are guiding parameters used to simulate and guide the dissolution process of dissolved ions at the transport interface. The transport process simulation parameters can include basic parameters before simulation operation, such as the initial position data of dissolved ions, the transport interface structure when the dissolved ions have not started to transport, etc.; flow parameters during the simulation process, such as the number of dissolution times and the dissolution distance during the ion dissolution process; and simulation configuration parameters required for performing molecular dynamics simulation, such as the system configuration data required for performing molecular dynamics simulation using first principles.

[0094] In some alternative embodiments, when ion transport performance evaluation is required, the evaluation system can obtain transport process simulation parameters determined based on the transport interface structure of the dissolved ions.

[0095] In some of these embodiments, the transport process simulation parameters can be determined and generated by the developer according to the constructed transport interface structure and the actual evaluation requirements, and the transport process simulation parameters are sent to the evaluation system based on the user terminal used by the developer, and the evaluation system obtains the transport process simulation parameters based on the user terminal.

[0096] In some of these embodiments, the developer can generate some of the simulation parameters in the transport process simulation parameters according to the actual evaluation requirements. The evaluation system can obtain some of the simulation parameters based on the user terminal, and then determine the transport process simulation parameters according to some of the simulation parameters and the transport interface structure of the dissolved ions.

[0097] S204, based on the transport process simulation parameters and the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface to obtain the dissolution force information of the dissolved ions.

[0098] Among them, the machine learning force field is a model used to characterize the mapping relationship between the transport interface structure and the force field properties during ion transport. The machine learning force field contains the interatomic potential function constructed by machine learning methods. The target machine learning force field is a machine learning force field that is pre-trained based on the transport interface structure of the dissolved ions. Therefore, the target machine learning force field can quickly predict the force conditions of the dissolved ions during the dissolution process based on the movement of the dissolved ions during transport between the transport interface structures, and obtain the dissolution force information of the dissolved ions. The dissolution force information is information data used to characterize the force conditions of the dissolved ions during the dissolution process.

[0099] In some of the embodiments, the target machine learning force field is a machine learning force field trained on the transport interface structure of dissolved ions.

[0100] In other embodiments, the target machine learning force field may be a machine learning force field obtained by training the basic structure of the transport interface structure of dissolved ions, that is, the unexpanded structure. Among them, since the machine learning force field has a transferable characteristic, when training the target machine learning force field, the basic transport interface structure of dissolved ions without expansion can be used for training to reduce the training cost. In the subsequent actual evaluation simulation, the basic transport interface structure can be expanded to obtain the transport interface results of dissolved ions, so that the transport interface structure can be more in line with the actual application scenario. The evaluation simulation of ion transport performance based on the transport interface structure can further improve the evaluation accuracy of ion transport performance.

[0101] The molecular dynamics simulation of the dissolution process of dissolved ions at the transport interface is used to simulate the motion trajectory of each atom in the entire transport interface structure when the dissolved ions are transported at the transport interface, and analyze the interaction mode and interaction energy between atoms to obtain a simulation method of relatively stable structures in various states between molecules. It can be understood that molecular dynamics simulation can be simulated using molecular dynamics software, such as first principle calculation software.

[0102] In some optional embodiments, after obtaining the transport process simulation parameters, the evaluation system can call the target machine learning force field that is pre-trained according to the transport interface structure of the dissolved ions, and perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the transport process simulation parameters. By introducing the machine learning force field in the molecular dynamics simulation process, the force conditions of the dissolved ions in the dissolution process can be quickly and accurately determined, and the dissolution force information of the dissolved ions can be obtained.

[0103] In some embodiments, the target machine learning force field can be pre-trained by developers according to the transport interface structure of the dissolved ions and configured in the evaluation system. When it is necessary to use the target machine learning force field for evaluation, the target machine learning force field corresponding to the dissolved ions and the transport interface structure of the dissolved ions can be called according to the preset force field identifier for use.

[0104] In some embodiments, the target machine learning force field can be a machine learning force field obtained by the evaluation system through training according to the transport interface structure of the dissolved ions during the actual simulation of the dissolution process. When it is necessary to evaluate the ion transport performance, the evaluation system can obtain in real time, according to the transport interface structure of the dissolved ions, the target machine learning force field corresponding to the dissolved ions and the transport interface structure of the dissolved ions. During subsequent actual simulation evaluation, the trained target machine learning force field can be directly called for use.

[0105] S206. Determine the evaluation result of the ion transport performance of the dissolved ions according to the dissolution force information.

[0106] Among them, the ion transport performance evaluation result is result data used to characterize the ease of dissolution of the dissolved ions in the transport interface structure. The type of the ion transport performance evaluation result can be determined according to the actual situation. For example, it can be a numerical result, that is, the dissolution energy barrier of the dissolved ions in the transport interface structure, or it can be a program result, that is, the ease of dissolution of the dissolved ions in the transport interface structure, such as difficult, medium, general, easy, etc.

[0107] In some embodiments, the evaluation system can determine the dissolution energy barrier of the dissolved ions according to the dissolution force information, and determine the evaluation result of the ion transport performance of the dissolved ions with the dissolution energy barrier of the dissolved ions. It can be understood that the higher the dissolution energy barrier, the more difficult it is for the ions to dissolve from the transport interface, and the worse the ion transport performance; the lower the ion energy barrier, the easier it is for the ions to dissolve from the transport interface, and the better the ion transport performance.

[0108] In some embodiments, as Figure 3 shown, determining the evaluation result of the ion transport performance of the dissolved ions according to the dissolution force information includes:

[0109] S302. Generate a free energy change curve of the dissolved ions during the dissolution process according to the dissolution force information.

[0110] After obtaining the dissolution force information of the dissolved ions, the evaluation system can process the dissolution force information according to the average force potential processing method, and generate a free energy change curve of the dissolved ions during the dissolution process based on the average force integral.

[0111] S304. Determine the dissolution energy barrier for the dissolved ions to transport in the transport interface structure based on the free energy change curve.

[0112] After obtaining the free energy change curve, the evaluation system determines the dissolution energy barrier for the dissolved ions to transport in the transport interface structure based on the energy barrier calculation method and according to the free energy change curve.

[0113] S306. According to the dissolution energy barrier and the preset corresponding relationship between the energy barrier and the dissolution difficulty, determine the evaluation result of the ion transport performance of the dissolved ions.

[0114] Among them, developers can pre-determine the range of dissolution energy barriers corresponding to each dissolution difficulty level to obtain the corresponding relationship between the energy barrier and the dissolution difficulty, and pre-configure the corresponding relationship between the energy barrier and the dissolution difficulty in the evaluation system.

[0115] After the evaluation system obtains the dissolution energy barrier for the dissolved ions to transport in the transport interface structure, it calls the preset corresponding relationship between the energy barrier and the dissolution difficulty, searches for the dissolution difficulty level corresponding to the dissolution energy barrier, and determines the dissolution difficulty level corresponding to the dissolution energy barrier as the evaluation result of the ion transport performance of the dissolved ions. For example, when the dissolution energy barrier is 2.0 eV, search for the preset corresponding relationship between the energy barrier and the dissolution difficulty, and determine that 2.0 eV falls within the range of energy barriers corresponding to the difficult dissolution level, then the difficult dissolution can be determined as the evaluation result of the ion transport performance of the dissolved ions. By presetting the corresponding relationship between the energy barrier and the dissolution difficulty, the evaluation system can directly evaluate and determine the dissolution difficulty of the ions, improving the convenience of ion transport performance evaluation.

[0116] In the above ion transport performance evaluation method, obtaining the transport process simulation parameters determined based on the transport interface structure of the dissolved ions can guide the dissolution process of the dissolved ions at the transport interface. Based on the transport process simulation parameters and the target machine learning force field obtained in advance by patrolling according to the transport interface structure of the dissolved ions, perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface. Since the target machine learning force field is a machine learning force field trained in advance according to the transport interface structure of the dissolved ions, when performing molecular dynamics simulation on the dissolved ions, the force situation of the dissolved ions during the dissolution process can be quickly and accurately predicted according to the dissolution situation of the dissolved ions at the transport interface, obtaining the dissolution force information of the dissolved ions. By introducing the machine learning force field for force prediction analysis during the molecular dynamics simulation process, the calculation cost in the ion transport performance evaluation process can be effectively reduced, and the accuracy and rate of determining the ion transport performance evaluation result according to the dissolution force information are improved.

[0117] The transport interface structure of the dissolved ions is the place where the entire ion transport process occurs. Therefore, the accurate construction of the transport interface structure will also affect the subsequent evaluation of ion transport performance.

[0118] In some embodiments, such as Figure 4 shown, the construction method of the transport interface structure of the dissolved ions includes:

[0119] S402. Construct an initial transport interface structure of the dissolved ions according to the construction parameters of the transport interface structure.

[0120] Among them, the construction parameters include the type of dissolved ions, the surface structure data of the vacuum layer, the molecular structure data of the solvent molecules, and the solvent molecule density data.

[0121] The type of dissolved ions is determined by the developer according to the ion type in the actual use scenario. Different types of dissolved ions have different force conditions during dissolution, and the corresponding transport performance is also different.

[0122] The vacuum layer can be regarded as the construction layer for constructing the solvent molecule storage container. The surface structure of the vacuum layer can be determined by the developer according to the actual vacuum layer construction material used. It can be understood that different vacuum layer construction materials correspond to different surface structures of the vacuum layer, and the dissolution direction of the dissolved ions is also different.

[0123] The solvent molecule structure data is the data used to characterize the molecular structure of the solvent molecules. The solvent molecule density data is the data used to characterize the number of solvent molecules contained in a certain volume of storage container constructed by the vacuum layer. The solvent molecule density data can be the number of solvent molecules or the solvent molecule density value. It can be understood that the solvent molecule structure data and the solvent molecule density data are also determined by the developer according to the usage situation of the solvent molecules in the actual use scenario.

[0124] In some alternative embodiments, the evaluation system receives the construction parameters uploaded by the developer based on the user terminal, and constructs an initial transport interface structure of the dissolved ions according to the construction parameters. The initial transport interface structure can be regarded as an interface structure model.

[0125] S404. Perform a single molecular dynamics simulation on the initial transport interface structure to obtain an equilibrium transport interface structure in a pre-equilibrium state.

[0126] In some alternative embodiments, in order to make the transport interface structure more reasonable, the evaluation system performs a single molecular dynamics simulation on the constructed initial transport interface structure, that is, simulates the atomic diffusion movement of each atom in the initial transport interface structure, and obtains an equilibrium transport interface structure in a pre-equilibrium state. Among them, the equilibrium transport interface structure can be regarded as the interface structure in which each ion in the initial transport interface structure moves to an equilibrium state under the action of mutual forces.

[0127] S406. Obtain the solvent molecule density data in the middle of the liquid phase in the equilibrium transport interface structure.

[0128] Among them, when atoms are undergoing diffusion motion, due to the interfacial force being greater than the atomic force between atoms, a larger number of atoms will move to the interface. Therefore, the solvent molecule density in the interfacial part does not conform to the solvent molecule density provided by the developer. The middle part of the liquid phase is the position of the liquid phase far from the interface. Only when the solvent molecule density data in the middle part of the liquid phase is roughly consistent with the theoretical solvent density value of the corresponding interfacial structure can the obtained transport interfacial structure be determined as a reasonable transport interfacial structure that meets the user's requirements.

[0129] After the evaluation system obtains the equilibrium transport interfacial structure in the pre-equilibrium state, it acquires the solvent molecule density data in the middle part of the liquid phase in the equilibrium transport interfacial structure. Similarly, the solvent molecule density data can be the number of solvent molecules or the solvent molecule density value.

[0130] S408. When it is determined that the equilibrium transport interfacial structure meets the requirements of structural rationality based on the solvent molecule density data in the middle part of the liquid phase, the equilibrium transport interfacial structure is determined as the transport interfacial structure of the dissolving ions.

[0131] Among them, the requirements of structural rationality are the requirements for judging whether the equilibrium transport interfacial structure is reasonable, and the requirements of structural rationality can be determined according to the theoretical solvent density value of the corresponding transport interfacial structure.

[0132] The evaluation system can judge whether the equilibrium transport interfacial structure meets the requirements of structural rationality based on the solvent molecule density data in the middle part of the liquid phase of the equilibrium transport interface. When it is determined that the equilibrium transport interfacial structure meets the requirements of structural rationality, the equilibrium transport interfacial structure is determined as the transport interfacial structure of the dissolving ions.

[0133] In the above embodiments, by performing a single molecular dynamics simulation on the constructed initial transport interfacial structure and evaluating the density rationality of the equilibrium transport interfacial structure in the pre-equilibrium state, the structural rationality of the finally constructed transport interfacial structure of the dissolving ions can be effectively improved.

[0134] In some of these embodiments, the evaluation system can compare the solvent molecule density data in the middle part of the liquid phase of the equilibrium transport interface with the theoretical solvent density value of the transport interfacial structure. When the difference between the two falls within a preset difference range, it is determined that the equilibrium transport interfacial structure meets the requirements of structural rationality. Among them, the preset difference range is also determined by the developer according to the theoretical density error range.

[0135] In some of these embodiments, the evaluation system can determine that the equilibrium transport interfacial structure meets the requirements of structural rationality when the solvent molecule density data in the middle part of the liquid phase of the equilibrium transport interface is equal to the theoretical solvent density value of the transport interfacial structure.

[0136] Furthermore, in some embodiments, the ion transport performance evaluation method further includes:

[0137] When it is determined that the equilibrium transport interface structure does not meet the requirements of structural rationality based on the solvent molecule density data in the middle of the liquid phase, the solvent molecule correction data for the equilibrium transport interface structure is determined. Based on the solvent molecule correction data, the solvent molecule density in the middle of the liquid phase of the equilibrium transport interface structure is corrected to obtain the transport interface structure of the dissolved ions.

[0138] In some alternative embodiments, if the equilibrium transport interface structure does not meet the requirements of structural rationality, it indicates that there is a gap between the solvent molecule density in the middle of the liquid phase and the theoretical solvent density value of the transport interface structure at this time, and correction is required. The evaluation system determines the solvent molecule correction data for the equilibrium transport interface structure based on the solvent molecule density in the middle of the liquid phase and the theoretical solvent density value of the transport interface structure. Based on the solvent molecule correction data, the solvent molecule density in the middle of the liquid phase of the equilibrium transport interface structure is corrected to obtain the transport interface structure of the dissolved ions. For example, when the solvent molecule density in the middle of the liquid phase is less than the theoretical solvent density value of the transport interface structure, the number of solvent molecules in the equilibrium transport interface structure is increased. When the solvent molecule density in the middle of the liquid phase is greater than the theoretical solvent density value of the transport interface structure, the number of solvent molecules in the equilibrium transport interface structure is decreased.

[0139] In the above embodiments, by correcting the solvent molecule density of the equilibrium transport interface structure, the structural rationality of the finally constructed transport interface structure of the dissolved ions can be effectively improved.

[0140] Introducing a machine learning force field for force prediction analysis during the molecular dynamics simulation is a key way to reduce the computational cost in the process of evaluating the ion transport performance. In one embodiment, as Figure 5 shown, based on the simulation parameters of the transport process and the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, a molecular dynamics simulation of the dissolution process of the dissolved ions at the transport interface is performed to obtain the dissolution force information of the dissolved ions, including:

[0141] S502, based on the simulation parameters of the transport process, a molecular dynamics simulation of the dissolution process of the dissolved ions at the transport interface is performed to obtain the transport interface structure during the dissolution process.

[0142] Among them, the transport interface structure during the dissolution process refers to the interface structure formed by the movement of each atom in the transport interface structure under the interaction force as the dissolved ions move during the dissolution process. It can be understood that the transport interface structure during the dissolution process can include multiple frames of transport interface structures generated during the dissolution process, and the number of frames of the transport interface structure can be determined by the developer according to specific evaluation accuracy requirements. The multiple frames of transport interface structures can characterize the movement of each atom in the transport interface structure when the dissolved ions are transported and move in the transport interface structure.

[0143] After the evaluation system obtains the simulation parameters of the transport process, it can perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the simulation parameters of the transport process, and obtain multiple frames of transport interface structures during the dissolution process.

[0144] In some embodiments, a first-principles calculation model for performing molecular dynamics simulation is pre-configured in the evaluation system. Based on the simulation parameters of the transport process, the first-principles method can be used to perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface, and multiple frames of transport interface structures during the dissolution process can be obtained. The preset first-principles method may include an ab initio method based on quantum mechanics or a density functional theory method.

[0145] S504, call the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions through a preset call interface.

[0146] Among them, the preset call interface is an interaction interface for communication and interaction between the evaluation system and the target machine learning force field, which is pre-constructed by developers for the evaluation system and the target machine learning force field. Through the preset call interface, the evaluation system can call the pre-trained target machine learning force field to analyze the force field properties of the dissolved ions during the dissolution process.

[0147] In some alternative embodiments, after the evaluation system obtains the transport interface structure of the dissolution process, it can call the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions through the preset call interface, providing an analysis tool for subsequent analysis.

[0148] S506, input the transport interface structure of the dissolution process into the target machine learning force field to obtain the dissolution force information of the dissolved ions.

[0149] The evaluation system can input the obtained transport interface structure of the dissolution process into the target machine learning force field through the preset call interface. The target machine learning force field will analyze the force conditions of the dissolved ions during the transport process according to the received multiple frames of transport interface structures during the dissolution process, and then obtain the dissolution force information of the dissolved ions, and return the dissolution force information to the evaluation system through the preset call interface. The evaluation system obtains the dissolution force information of the dissolved ions returned by the target machine learning force field.

[0150] In the above embodiments, by inputting the transport interface structure of the dissolution process obtained from molecular dynamics simulation into the target machine learning force field for force analysis, compared with the high computational cost generated by using first principles for force analysis, using the pre-trained target machine learning force field to analyze the forces on the dissolution ions during the multi-frame transport interface structure in the dissolution process can quickly and accurately obtain the dissolution force information of the dissolution ions, effectively reducing the computational cost in the process of evaluating ion transport performance and improving the accuracy and rate of determining the ion transport performance evaluation result based on the dissolution force information subsequently.

[0151] As can be seen from the single-ion dissolution distance and the number of simulated dissolutions included in the above transport process simulation parameters, the evaluation process of ion transport performance needs to simulate the movement of the dissolution ions each time and then analyze the forces on the dissolution ions after each movement. Therefore, in some embodiments, based on the transport process simulation parameters and the target machine learning force field pre-trained according to the transport interface structure of the dissolution ions, molecular dynamics simulation is performed on the dissolution process of the dissolution ions at the transport interface to obtain the dissolution force information of the dissolution ions, including:

[0152] The evaluation system inputs the single-ion dissolution distance, the transport interface structure of the dissolution ions, and the number of simulated dissolutions into a preset first-principles calculation model. The first-principles calculation model updates the position of the dissolution ions according to the single-ion dissolution distance and the position and dissolution direction of the dissolution ions in the interface reflected by the transport interface structure of the dissolution ions. Subsequently, it simulates the movement of each atom in the transport interface structure due to the interatomic interaction forces after the position update of the dissolution ions to obtain the transport interface structure of the dissolution process. Then, the first-principles calculation model inputs the transport interface structure of the dissolution process into the target machine learning force field through a preset call interface.

[0153] The target machine learning force field predicts and analyzes the forces on each atom in the structure according to the transport interface structure of the dissolution process to obtain the force information of all atoms in the transport interface structure at the current moving position. The force information of all atoms naturally includes the force information of the dissolution ions at the current moving position. The target machine learning force field returns the force information of all atoms to the first-principles calculation model through a preset call interface.

[0154] The first-principles calculation model updates the position of the dissolved ions according to the single-ion dissolution distance and the transport interface structure of the dissolved ions. Subsequently, based on the force information of all atoms in the transport interface structure at the current moving position, the model simulates the movement of each atom in the updated transport interface structure after the position of the dissolved ions is updated, caused by the interatomic interaction forces, to obtain the updated transport interface structure of the dissolution process. That is, the first-principles calculation model will determine the direction of the next simulation evolution based on the force information predicted by the target machine learning force field.

[0155] The first-principles calculation model inputs the updated transport interface structure of the dissolution process into the target machine learning force field to obtain the force information of all atoms in the transport interface structure at the current moving position. The target machine learning force field returns the force information of all atoms to the first-principles calculation model through a preset call interface.

[0156] The first-principles calculation model returns to execute the step of updating the position of the dissolved ions according to the single-ion dissolution distance and the transport interface structure of the dissolved ions. Subsequently, based on the force information of all atoms in the transport interface structure at the current moving position, the model simulates the movement of each atom in the updated transport interface structure after the position of the dissolved ions is updated, caused by the interatomic interaction forces, to obtain the updated transport interface structure of the dissolution process, until the number of simulations reaches the number of simulated dissolutions, that is, until the dissolved ions complete the cross-interface transport. The target machine learning force field will feedback the dissolution force information of the dissolved ions during the entire dissolution simulation process.

[0157] In the above embodiments, by performing information interaction between the first-principles calculation model and the target machine learning force field, the dissolution force information of the dissolved ions during the ion dissolution process can be obtained quickly and accurately, effectively reducing the calculation cost in the process of evaluating the ion transport performance, and improving the evaluation accuracy and evaluation rate of the ion transport performance.

[0158] During the process of evaluating the ion transport performance, the simulation parameters of the transport process, as guiding parameters, play an important guiding role in the transport. In one embodiment, as Figure 6 shown, the simulation parameters of the transport process determined based on the transport interface structure of the dissolved ions are obtained, including:

[0159] S602, obtaining the single-ion dissolution distance during the simulated transport process.

[0160] Among them, the single-ion dissolution distance refers to the distance parameter that can be moved by each dissolved ion during the transport movement in the transport interface structure during the evaluation of the simulation process. It can be understood that during the simulation of the transport process, the dissolution process of ions is achieved by slowly moving the position coordinates of the dissolved ions in the dissolution direction of the transport interface structure. Therefore, it is necessary to specify the moving distance of the dissolved ions during each movement in the simulation transport process. The single-ion dissolution distance can be determined by the developer according to the simulation experience. The developer can determine the dissolution distance that can provide sufficient accuracy and is within the reasonable calculation cost range as the single-ion dissolution distance.

[0161] In one embodiment, the single-ion dissolution distance during the simulation of the transport process can be 0.2 Å.

[0162] In some alternative embodiments, the developer can upload the single-ion dissolution distance during the simulation of the transport process through the user terminal. The evaluation system obtains the single-ion dissolution distance uploaded by the user terminal.

[0163] S604. Based on the single-ion dissolution distance and the transport interface structure of the dissolved ions, determine the number of simulated dissolutions during the simulation of the transport process.

[0164] Among them, the transport interface structure of the dissolved ions can reflect the interface distance between the two interfaces that the dissolved ions need to cross. The number of simulated dissolutions refers to the number of times the dissolved ions need to move to complete the cross-interface transport from the initial position when each time they can only move the single-ion dissolution distance.

[0165] After the evaluation system determines the single-ion dissolution distance during the simulation of the transport process, it can determine the number of simulated dissolutions during the simulation of the transport process based on the single-ion dissolution distance and the interface distance reflected by the transport interface structure of the dissolved ions. For example, the ratio of the interface distance to the single-ion dissolution distance can be rounded to determine the number of simulated dissolutions.

[0166] S606. Determine the single-ion dissolution distance, the number of simulated dissolutions, and the transport interface structure of the dissolved ions as the simulation parameters of the transport process.

[0167] After the evaluation system determines the single-ion dissolution distance and the number of simulated dissolutions, it can determine the single-ion dissolution distance, the number of simulated dissolutions, and the transport interface structure of the dissolved ions as the simulation parameters for the transport process. Among them, the number of simulated dissolutions and the single-ion dissolution distance can be regarded as the process parameters in the simulation parameters for the transport process, which specify how many simulations are specifically required to complete the dissolution process of the dissolved ions in the entire simulation process. The transport interface structure of the dissolved ions can be regarded as the basic parameter of the simulation parameters for the transport process before the simulation runs, which specifies information such as the initial position of the dissolved ions, the transport interface structure when the dissolved ions have not started to be transported, and the dissolution direction of the dissolved ions.

[0168] In the above embodiments, by determining the single-ion dissolution distance, the number of simulated dissolutions, and the transport interface structure of the dissolved ions as the simulation parameters for the transport process, it can provide a guiding role for the subsequent simulation process and improve the simulation accuracy of the transport process.

[0169] The target machine learning force field is a key tool for reducing the computational cost in the evaluation process of ion transport performance and improving the accuracy and rate of the evaluation results of ion transport performance. The quality of the training of the target machine learning force field will directly affect the performance evaluation effect of the actual simulation process. The following will elaborate on the training process of the target machine learning force field through some embodiments.

[0170] In some embodiments, such as Figure 7 shown, the training method of the target machine learning force field includes:

[0171] S702, determine the training simulation parameters of the dissolved ion transport process according to the transport interface structure of the dissolved ions.

[0172] Among them, it should be noted that the transport interface structure used in the training process can be the same interface structure as the transport interface structure of the dissolved ions used in the actual simulation process, or the basic interface structure corresponding to the transport interface structure of the dissolved ions used in the actual simulation process. It can be understood that since the internal structures of the basic interface structure and the transport interface structure of the dissolved ions are basically the same, only the interface length is different (the interface width that the dissolved ions need to cross is the same), so for the convenience of understanding in the following, it is assumed that the transport interface structure used in the training process is the same as the transport interface structure of the dissolved ions used in the actual simulation process for illustration.

[0173] After the evaluation system obtains the transport interface structure of the dissolved ions, it can determine the training simulation parameters of the dissolved ion transport process according to the transport interface structure of the dissolved ions.

[0174] Similarly, in some of these embodiments, the evaluation system may obtain, based on the user terminal, some training simulation parameters generated by the developer according to the actual training requirements, and then determine the training simulation parameters based on the some training simulation parameters and the transport interface structure of the dissolved ions.

[0175] S704. Based on the training simulation parameters, perform molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface to obtain a force field training data set for the initial machine learning force field.

[0176] Among them, the molecular dynamics training simulation process is a process of performing molecular dynamics simulation and force prediction analysis on the dissolution process of the dissolved ions at the transport interface. That is, it can be considered that after the molecular dynamics training simulation, in addition to obtaining structural data reflecting the movement of each atom at the transport interface during the dissolution process, force information data reflecting the force on each atom during the dissolution process can also be obtained.

[0177] The initial machine learning force field can be considered as an untrained numerical model, in which an initial interatomic potential function is set, and the process of training the initial machine learning force field can be considered as a process of optimizing the parameters in the initial function.

[0178] In some alternative embodiments, after obtaining the training simulation parameters, the evaluation system may, based on the training simulation parameters, perform molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface, and integrate the simulation data obtained from the training simulation into a force field training data set for the initial machine learning force field.

[0179] Similarly, in some of these embodiments, the molecular dynamics training simulation can use molecular dynamics software, such as first-principles calculation software, for training simulation.

[0180] S706. Perform model learning on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolved ions.

[0181] After obtaining the force field training data set, the evaluation system may perform model learning on the initial machine learning force field according to the force field training data set, optimize the parameters in the initial machine learning force field, and finally obtain a target machine learning force field that can be used to analyze the force on the dissolved ions during transport in the current transport interface structure.

[0182] In one of the embodiments, the evaluation system may call a preset call interface preset with the initial machine learning force field, input the force field training data set into the initial machine learning force field for model learning, and obtain the target machine learning force field of the dissolved ions.

[0183] In one embodiment, the evaluation system can partition the force field training data set to obtain a training set and a test set. After initially training the initial machine learning force field based on the training set, the trained initial machine learning force field is then tested based on the test set. In the case of successful testing, the target machine learning force field of the dissolved ions is obtained.

[0184] In the above embodiment, by performing model learning on the initial machine learning force field in the same simulation scenario as during actual evaluation, a target machine learning force field capable of quickly and accurately analyzing the force conditions of dissolved ions during transport in the current transport interface structure can be obtained. This provides an analysis and prediction tool for quickly obtaining accurate dissolution force information of dissolved ions during actual evaluation, effectively reducing the computational cost during the ion transport performance evaluation process, and improving the evaluation accuracy and evaluation rate of the ion transport performance.

[0185] Further, in some embodiments, as Figure 8 shown, the training simulation parameters include the single dissolution distance for training, the number of dissolution times for training, and the transport interface structure of the dissolved ions. The single dissolution distance for training is greater than the single ion dissolution distance during actual simulation.

[0186] Based on the training simulation parameters, a molecular dynamics training simulation of the dissolution process of the dissolved ions at the transport interface is performed to obtain the force field training data set of the initial machine learning force field, including:

[0187] S802, updating the position of the dissolved ions in the transport interface structure according to the single dissolution distance for training to obtain the training input interface structure.

[0188] Among them, the single dissolution distance for training refers to the distance parameter that the dissolved ions can move each time during the transport movement in the transport interface structure during the training process. It can be understood that, similar to the simulation transport process during evaluation, during the training process, the dissolution process of the ions is realized by slowly moving the position coordinates of the dissolved ions in the dissolution direction of the transport interface structure. Therefore, during the training process, it is also necessary to specify the moving distance of the dissolved ions each time during the training simulation transport process. It can be understood that the single dissolution distance for training can be determined by the developer according to experience.

[0189] It should be noted that the single dissolution distance during training will be greater than the single ion dissolution distance during actual simulation. This is because during actual simulation, in order to accurately evaluate the ion transport performance, a smaller single ion dissolution distance will be set, sacrificing a little computational cost to improve the accuracy of ion transport performance evaluation. During the molecular dynamics simulation training process, it is necessary to analyze the force on the dissolved ions during the dissolution process through first principles. This process will consume a large amount of computational cost. Therefore, in order to reduce the computational cost during the training process, the training single dissolution distance can be set to be greater than the single ion dissolution distance during actual simulation, so as to achieve the effect of both generating enough training data to train the initial machine learning force field and obtaining a target machine learning force field that can complete high-precision analysis tasks during actual evaluation simulation, while reducing the training cost.

[0190] In some of these embodiments, the single ion dissolution distance during actual simulation can be set to 0.2 Å, and the setting range of the training single dissolution distance can be 0.6 Å - 1 Å. In this way, the training single dissolution distance will not be too large, resulting in missing data in the middle part of the dissolution process during training simulation and affecting the analysis accuracy of the target machine learning force field, nor will it be too small, resulting in an increase in computational cost.

[0191] The number of training dissolutions refers to the number of times the dissolved ions need to move to complete cross-interface transport from the initial position when only moving the training single dissolution distance each time.

[0192] In some of these embodiments, after the evaluation system obtains the training single dissolution distance and the transport interface structure of the dissolved ions, it can determine the number of training dissolutions during the training process based on the training single dissolution distance and the interface distance reflected by the transport interface structure of the dissolved ions. For example, the ratio of the interface distance to the training single dissolution distance can be rounded to determine the number of training dissolutions.

[0193] In some of these embodiments, the evaluation system can directly obtain the training simulation parameters uploaded by the developer.

[0194] In some alternative embodiments, after the evaluation system obtains the training simulation parameters, it can update the position of the dissolved ions in the transport interface structure based on the training single dissolution distance and the ion dissolution direction reflected by the transport interface structure of the dissolved ions to obtain the training input interface structure, and the training input interface structure can reflect the moving position of the dissolved ions during each movement.

[0195] S804, perform molecular dynamics training simulation based on the training input interface structure to obtain multi-frame interface structure data of the training input interface structure during training simulation, and the structure energy and atomic force information of each atom in the structure corresponding to each frame of the interface structure.

[0196] Among them, the multi-frame interface structure data can characterize the interface structure formed by the movement of atoms in the transport interface structure under the action of mutual forces as the dissolved ions move during the training simulation. It can be understood that the number of frames in the multi-frame interface structure data can be determined by the developer according to specific training accuracy requirements. The multi-frame transport interface structure can characterize the movement of each atom in the transport interface structure when the dissolved ions are transported and move in the transport interface structure.

[0197] The structural energy corresponding to each frame of the interface structure is used to characterize the energy data contained in the current interface structure. The level of the structural energy can reflect the stability of the current frame of the interface structure. The atomic force information of each atom in the structure is used to characterize the force conditions of the mutual forces between each atom under the current structure.

[0198] In some alternative embodiments, the evaluation system can perform an analytical dynamics training simulation based on the training input interface structure, and during the training simulation, intercept multi-frame interface structure data generated by the evolution of the training input interface structure according to the preset number of frames requirement, and at the same time determine the structural energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure.

[0199] In some of these embodiments, the evaluation system can call a preset first-principles calculation model to perform a molecular dynamics training simulation.

[0200] S806, after the training simulation of the training input interface structure is completed, update the position of the dissolved ions in the training input interface structure based on the single training dissolution distance to obtain an updated training input interface structure, and return to execute the step of performing a molecular dynamics training simulation based on the training input interface structure until the number of updates of the training input interface structure is equal to the number of training dissolution times.

[0201] After the training simulation of the training input interface structure is completed, the evaluation system updates the position of the dissolved ions in the training input interface structure based on the single training dissolution distance to obtain an updated training input interface structure, and then returns to execute the step of performing a molecular dynamics training simulation based on the training input interface structure until the number of updates of the training input interface structure is equal to the number of training dissolution times.

[0202] S808, determine the multi-frame interface structure data obtained from the molecular dynamics training simulation, as well as the structural energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure as the force field training data set of the initial machine learning force field.

[0203] When the number of updates to the training input interface structure reaches the number of training dissolution times, the evaluation system will determine the force field training data set of the initial machine learning force field based on the multi-frame interface structure data obtained from the molecular dynamics training simulation, as well as the structural energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure.

[0204] In the above embodiment, by setting the training single dissolution distance greater than the single ion dissolution distance during actual simulation, the computational cost of the training process can be effectively reduced without much impact on the prediction accuracy of the trained target machine learning force field, thereby effectively reducing the evaluation computational cost of ion transport performance. At the same time, through the preset call interface, data interaction between the evaluation system and the target machine learning force field can be carried out to achieve the serialization of multiple computational tasks and reduce cumbersome manual operations.

[0205] Further, in order to improve the prediction and analysis accuracy of the target machine learning force field, in some embodiments, such as Figure 9 shown, model learning is performed on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolved ions, including:

[0206] S902, divide the force field training data set into a training set and a test set according to a preset division ratio.

[0207] Among them, the preset division ratio can be determined by developers according to actual simulation requirements. For example, the preset division ratio can be 8:2.

[0208] After obtaining the force field training data set, the evaluation system can divide the force field training data set into a training set and a test set according to the preset division ratio. The training set is the data set used to train the initial machine learning force field, and the test set is the data set used to perform performance testing on the trained initial machine learning force field to determine whether the training is completed.

[0209] S904, perform model learning on the initial machine learning force field according to the training set to obtain the machine learning force field to be tested.

[0210] The evaluation system inputs the data included in the training set into the initial machine learning force field for model learning to obtain the machine learning force field to be tested.

[0211] In some of these embodiments, the initial machine learning force field performs a predicted analysis of the forces on the structural atoms based on the multi-frame interface structures and structural energies in the training set, obtaining the predicted force information for all atoms in the structure. The predicted force information is compared with the atomic force information of each atom in the structure corresponding to the multi-frame interface structure, and the function parameters in the initial machine learning force field are optimized according to the comparison result, making the predicted force information closer to the atomic force information of each atom in the structure corresponding to the multi-frame interface structure, and thus conducting model learning.

[0212] S906. Obtain the force field parameters of the machine learning force field to be tested.

[0213] Among them, the force field parameters of the machine learning force field to be tested are parameter information used to characterize the predicted analysis performance of the machine learning force field to be tested.

[0214] After the evaluation system trains the initial machine learning force field to obtain the machine learning force field to be tested, it can obtain the force field parameters of the machine learning force field to be tested.

[0215] S908. When it is determined that the machine learning force field to be tested meets the preliminary verification conditions based on the comparison result between the force field parameters and the preset preliminary verification parameter information, test the machine learning force field to be tested according to the test set.

[0216] Among them, the preliminary verification parameter information is verification information used to determine whether the machine learning force field to be tested meets the preliminary prediction analysis accuracy, and is pre-determined and set by the developer according to the theoretical performance parameters of the machine learning force field.

[0217] After the evaluation system obtains the force field parameters of the machine learning force field to be tested, it calls the preset preliminary verification parameter information, compares the force field parameters with the preliminary verification parameter information, determines whether the machine learning force field to be tested meets the preliminary verification conditions according to the comparison result, and when it is determined that the machine learning force field to be tested meets the preliminary verification conditions, tests the machine learning force field to be tested according to the test set.

[0218] In some of these embodiments, when the comparison result between the force field parameters and the preliminary verification parameters falls within the preset verification range, it is determined that the machine learning force field to be tested meets the preliminary verification conditions, where the preset verification range is pre-determined and set by the developer according to the theoretical error range.

[0219] In some of these embodiments, when it is determined that the machine learning force field to be tested does not meet the preliminary verification conditions, continue to train the machine learning force field to be tested.

[0220] S910. When it is determined that the test of the machine learning force field to be tested is successful, determine the machine learning force field to be tested as the target machine learning force field for the dissolving ions.

[0221] In some alternative embodiments, the evaluation system uses a test set to test the machine learning force field to be tested. When the data difference between the test data of the machine learning force field to be tested and the verification data in the test set is less than a preset difference threshold, it is determined that the test of the machine learning force field to be tested is successful, and the evaluation system determines the machine learning force field to be tested as the target machine learning force field for the dissolved ions.

[0222] In other embodiments, when the evaluation system determines that the test of the machine learning force field to be tested is unsuccessful, it continues to train the machine learning force field to be tested.

[0223] In some of these embodiments, the evaluation system inputs the multi-frame interface structure data in the test set into the machine learning force field to be tested for predictive analysis, obtains the corresponding predicted energy and predicted atomic force information. The evaluation system compares the predicted energy with the structural energy in the test set, and compares the predicted atomic force information with the atomic force information corresponding to each frame of the interface structure in the test set. When the comparison differences between the two are both less than the corresponding preset difference thresholds, it is determined that the test of the machine learning force field to be tested is successful.

[0224] In the above embodiments, the machine learning force field to be tested is preliminarily verified through the preliminary verification parameter information. Only when the preliminary verification is successful, will the machine learning force field to be tested be further tested, reducing the test cost of the machine learning force field to be tested. The machine learning force field to be tested with a successful test is determined as the target machine learning force field. Through two tests, the predictive analysis accuracy of the finally determined target machine learning force field can be further improved, thereby improving the evaluation accuracy of the ion transport performance.

[0225] In some embodiments, as Figure 10 shown, a method for evaluating ion transport performance is provided. Taking the example where this method is applied to the dissolution scenario of Li ions at the solid-liquid interface for illustration. The method specifically includes the following steps:

[0226] S1001, the evaluation system obtains the interface model construction parameters.

[0227] The construction parameters specify the ion type, that is, Li ions, the TiO2 (001) surface structure with a 20 Å vacuum layer in the z direction, the solvent molecule is ethylene carbonate (EC), and the molecular structure of EC and the number of EC molecules are given. The number of EC molecules is set to 15.

[0228] S1002, construct a solid-liquid interface model for the dissolution of Li ions according to the model construction parameters. The specific model structure of the solid-liquid interface model is as Figure 11 shown.

[0229] S1003. Perform molecular dynamics simulations on the solid-liquid interface model to obtain a structurally pre-equilibrated solid-liquid interface model.

[0230] Evaluate the system by calling the pre-configured first-principles calculation software VASP and PBE functional, and achieve the structural pre-equilibrium of the solid-liquid interface model through first-principles molecular dynamics simulations.

[0231] S1004. Perform liquid-phase solvent analysis density analysis on the structurally pre-equilibrated solid-liquid interface model to obtain a reasonable solid-liquid interface model.

[0232] The evaluation system obtains the solvent molecule density in the middle of the liquid phase in the structurally pre-equilibrated solid-liquid interface model. If the difference between the solvent molecule density and the experimental theoretical value is not within the density error range, modify the number of solvent molecules in the solid-liquid interface model to obtain a reasonable solid-liquid interface model.

[0233] When the difference between the solvent molecule density and the experimental theoretical value is within the density error range, for example, when obtaining the solvent molecule density in the middle of the liquid phase, the density analysis result is as Figure 12 shown. Figure 12 The curve corresponding to the actual value of the solvent density in is the curve without mean processing, and the curve corresponding to the average value of the solvent density is the curve obtained after mean processing. It can be seen from the figure that the solvent molecule density in the middle of the liquid phase is about 1.36 g / cm 3 , which is 3 0.04 g / cm different from the experimental theoretical value of 1.32 g / cm 3 . Within the density error range caused by one solvent molecule, determine the structurally pre-equilibrated solid-liquid interface model as a reasonable solid-liquid interface model.

[0234] S1005. Collect a training dataset for the solid-liquid interface structure with a reasonable solvent molecule density, and train the initial machine learning force field based on the training dataset to obtain the target machine learning force field.

[0235] The evaluation system obtains training simulation parameters. In the training simulation, the distance of single dissolution is set to 1 Å, and the number of dissolution times is 10. Among them, the interface structure used for the first dissolution is the original solid-liquid interface structure in insoluble S1004. After that, the dissolution structure for each time is based on the final structure obtained from the previous simulation to modify the ion dissolution distance. The interaction between the first-principles molecular dynamics simulation and the machine learning force field is realized by using machine learning software and first-principles calculation software. The training simulation parameters are input into the first-principles calculation software, and each frame of the structure output by the first-principles calculation software and the energy and atomic force information calculated based on the first principles are integrated as a data set. It can be understood that the data format in the data set is determined according to the machine learning software used. Through a preset call interface, the training data set is input into the initial machine learning force field for training to obtain a machine learning force field to be tested.

[0236] When collecting the training data set, some structural data points can also be extracted at intervals during the dissolution process for first-principles calculation to obtain the corresponding energy and atomic forces as a test set to test the prediction performance of the machine learning, and the corresponding prediction error is obtained. When the prediction error meets the preset error range, the machine learning force field to be tested is determined as the target machine learning force field. Figure 13 It is a performance schematic diagram obtained by testing the performance of the target machine learning force field with the first-principles calculation software VASP based on the PBE functional, as Figure 13 shown. The root mean square error of the atomic energy of the target machine learning force field is on the order of 4.86 meV, and the root mean square error of the atomic force is on the order of 147 meV.

[0237] S1006, obtain the simulation parameters of the transport process for evaluating the ion transport performance.

[0238] In the simulation parameters of the transport process, the distance of single dissolution is set to 0.2 Å, and the number of dissolution times is 40. Among them, the interface structure used for the first dissolution includes but is not limited to the original solid-liquid interface structure obtained in S1004, and can also be the supercell structure obtained based on S1004. After that, the dissolution structure for each time is based on the final structure obtained from the previous simulation to modify the ion dissolution distance.

[0239] S1007, according to the simulation parameters of the transport process, introduce the target machine learning force field to perform molecular dynamics simulation of the transport process to obtain the force information of Li ion dissolution.

[0240] The evaluation system performs molecular dynamics simulation of the Li ion dissolution process according to the simulation parameters of the transport process, using the first-principles calculation software combined with the target machine learning force field to obtain the force information of Li ions during the dissolution process.

[0241] S1008. Determine the free energy change curve of Li ions based on the dissolution force information.

[0242] Based on the dissolution force information of the dissolved ions, the evaluation system calculates the average force change of Li ions in the z direction during the dissolution process. Based on the integral of the final average force over the dissolution distance, the free energy change curve is obtained. It can be understood that the force change during the ion dissolution process and the free energy change curve are obtained by the Potential of Mean Force (PMF) method. Figure 14 It is a diagram showing the force change of Li ions in the z direction during the dissolution process in the TiO2-EC system. From top to bottom, it shows the force change of ions during the increase of the ion dissolution distance. Figure 15 It is a schematic diagram of the average force change and free energy change of Li ions in the z direction during the Li dissolution process in the TiO2-EC system.

[0243] S1009. Obtain the dissolution energy barrier of Li ions based on the free energy change curve and determine the evaluation result of the transport performance of Li ions.

[0244] Based on the free energy change curve of Li ions, the evaluation system calculates that the energy barrier for Li ions to dissolve from the solid phase to the liquid phase is about 2.0 eV. According to the preset correspondence between the energy barrier and the transport degree, it is determined that Li ions are not easily dissolved at the TiO2-EC solid-liquid interface.

[0245] For the ion transport performance evaluation method of the above embodiments, since a machine learning force field is used, this method can accelerate the simulation of the solid-liquid interface of materials and reduce the time and computational cost required for computational simulation. At the same time, the transferable characteristics of the machine learning force field can be utilized to apply the machine learning force field with high prediction accuracy trained based on the first-principles calculation data of a small system to the corresponding extended system for computational simulation. Furthermore, by predicting the energy barrier of ion dissolution, the ease of ion dissolution at the interface is evaluated, providing an important basis for the performance evaluation and material design of materials.

[0246] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed 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 executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0247] Based on the same inventive concept, an embodiment of the present application further provides an ion transport performance evaluation device for implementing the ion transport performance evaluation method involved above. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the ion transport performance evaluation device provided below can refer to the limitations on the ion transport performance evaluation method in the above text, and will not be repeated here.

[0248] In some embodiments, such as Figure 16 shown, an ion transport performance evaluation device 1600 is provided, including: a simulation parameter acquisition module 1601, a simulation module 1602, and a performance evaluation module 1603, where:

[0249] The simulation parameter acquisition module 1601 is configured to acquire simulation parameters for the transport process determined based on the transport interface structure of the dissolved ions.

[0250] The simulation module 1602 is configured to perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the simulation parameters of the transport process and a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, to obtain the dissolution force information of the dissolved ions.

[0251] The performance evaluation module 1603 is configured to determine the ion transport performance evaluation result of the dissolved ions according to the dissolution force information.

[0252] The above ion transport performance evaluation device acquires simulation parameters for the transport process determined based on the transport interface structure of the dissolved ions, which can guide the dissolution process of the dissolved ions at the transport interface. Based on the simulation parameters of the transport process and a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, molecular dynamics simulation is performed on the dissolution process of the dissolved ions at the transport interface. Since the target machine learning force field is a machine learning force field pre-trained according to the transport interface structure of the dissolved ions, when performing molecular dynamics simulation on the dissolved ions, the force situation of the dissolved ions during the dissolution process can be quickly and accurately predicted according to the dissolution situation of the dissolved ions at the transport interface, to obtain the dissolution force information of the dissolved ions. By introducing a machine learning force field for force prediction analysis during the molecular dynamics simulation process, the calculation cost in the ion transport performance evaluation process can be effectively reduced, and the accuracy and rate of determining the ion transport performance evaluation result according to the dissolution force information subsequently are improved.

[0253] In some embodiments, the simulation module is further configured to: perform a molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the transport process simulation parameters to obtain the transport interface structure of the dissolution process; call, through a preset call interface, a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions; and input the transport interface structure of the dissolution process into the target machine learning force field to obtain the dissolution force information of the dissolved ions.

[0254] In some embodiments, the simulation parameter acquisition module is further configured to: obtain the single-ion dissolution distance during the simulated transport process; determine the simulated dissolution times during the simulated transport process based on the single-ion dissolution distance and the transport interface structure of the dissolved ions; and determine the single-ion dissolution distance, the simulated dissolution times, and the transport interface structure of the dissolved ions as the transport process simulation parameters.

[0255] In some embodiments, the ion transport performance evaluation device further includes: a force field training module, configured to determine the training simulation parameters of the dissolution process of the dissolved ions according to the transport interface structure of the dissolved ions; perform a molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface based on the training simulation parameters to obtain a force field training data set of the initial machine learning force field; and perform model learning on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolved ions.

[0256] In some embodiments, the training simulation parameters include the training single-ion dissolution distance, the training dissolution times, and the transport interface structure of the dissolved ions, and the training single-ion dissolution distance is greater than the single-ion dissolution distance during the actual simulation. The force field training module is further configured to: update the position of the dissolved ions in the transport interface structure according to the training single-ion dissolution distance to obtain a training input interface structure; perform a molecular dynamics training simulation based on the training input interface structure to obtain multi-frame interface structure data of the training input interface structure during the training simulation, and the structure energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure; after the training simulation of the training input interface structure is completed, update the position of the dissolved ions in the training input interface structure according to the training single-ion dissolution distance to obtain an updated training input interface structure, and return to execute the step of performing a molecular dynamics training simulation based on the training input interface structure until the update times of the training input interface structure are equal to the training dissolution times; and determine the multi-frame interface structure data obtained based on the molecular dynamics training simulation, and the structure energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure as the force field training data set of the initial machine learning force field.

[0257] In some embodiments, the force field training module is further configured to: divide the force field training data set into a training set and a test set according to a preset division ratio; perform model learning on the initial machine learning force field according to the training set to obtain a machine learning force field to be tested; obtain the force field parameters of the machine learning force field to be tested; when it is determined that the machine learning force field to be tested meets the preliminary verification conditions based on the comparison result between the force field parameters and the preset preliminary verification parameter information, test the machine learning force field to be tested according to the test set; when it is determined that the test of the machine learning force field to be tested is successful, determine the machine learning force field to be tested as the target machine learning force field of the dissolved ions.

[0258] In some embodiments, the ion transport performance evaluation device further includes: an interface structure construction module, configured to construct an initial transport interface structure of the dissolved ions according to the construction parameters of the transport interface structure, where the construction parameters include the type of dissolved ions, the surface structure data of the vacuum layer, the solvent molecule structure data, and the solvent molecule density data; perform a single molecular dynamics simulation on the initial transport interface structure to obtain an equilibrium transport interface structure in a pre-equilibrium state; obtain the solvent molecule density data in the middle of the liquid phase in the equilibrium transport interface structure; when it is determined that the equilibrium transport interface structure meets the structural rationality requirements according to the solvent molecule density data in the middle of the liquid phase, determine the equilibrium transport interface structure as the transport interface structure of the dissolved ions.

[0259] In some embodiments, the ion transport performance evaluation device further includes: an interface structure correction module, configured to determine the solvent molecule correction data of the equilibrium transport interface structure when it is determined that the equilibrium transport interface structure does not meet the structural rationality requirements according to the solvent molecule density data in the middle of the liquid phase; based on the solvent molecule correction data, correct the solvent molecule density in the middle of the liquid phase of the equilibrium transport interface structure to obtain the transport interface structure of the dissolved ions.

[0260] In some embodiments, the performance evaluation module is further configured to: generate a free energy change curve of the dissolved ions during the dissolution process according to the dissolution force information; determine the dissolution energy barrier for the dissolved ions to transport in the transport interface structure based on the free energy change curve; determine the ion transport performance evaluation result of the dissolved ions according to the dissolution energy barrier and the preset correspondence between the energy barrier and the dissolution difficulty.

[0261] Each module in the above ion transport performance evaluation device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above respective modules.

[0262] In some embodiments, a computer device is provided. The computer device may be a server integrated with an evaluation system, and its internal structure diagram may be as shown in Figure 17 . The computer device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data such as transport process simulation parameters, dissolution force information, and ion transport performance evaluation results. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements an ion transport performance evaluation method.

[0263] Those skilled in the art can understand that Figure 17 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0264] In some embodiments, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, it implements the specific implementation steps of the above ion transport performance evaluation method.

[0265] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, it implements the specific implementation steps of the above ion transport performance evaluation method.

[0266] In some embodiments, a computer program product is provided, including a computer program. When the computer program is executed by the processor, it implements the specific implementation steps of the above ion transport performance evaluation method.

[0267] 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 for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or fully authorized by all parties.

[0268] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random access memories (ReRAM), magnetoresistive random access memories (MRAM), ferroelectric random access memories (FRAM), phase change memories (PCM), graphene memories, etc. Volatile memories can include random access memory (RAM) or external cache memories, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0269] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, 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, it should be considered as the scope recorded in this specification.

[0270] The above-described embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for evaluating ion transport performance, characterized in that, The method includes: Obtaining simulation parameters of the transport process determined based on the transport interface structure of the dissolved ions; Based on the simulation parameters of the transport process and the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, performing molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface to obtain the dissolution force information of the dissolved ions; Determining the evaluation result of the ion transport performance of the dissolved ions according to the dissolution force information.

2. The method according to claim 1, wherein The performing molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the simulation parameters of the transport process and the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions to obtain the dissolution force information of the dissolved ions includes: Based on the simulation parameters of the transport process, performing molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface to obtain the transport interface structure during the dissolution process; Calling the target machine learning force field pre-trained according to the transport interface structure of the dissolved ions through a preset call interface; Inputting the transport interface structure during the dissolution process into the target machine learning force field to obtain the dissolution force information of the dissolved ions.

3. The method according to claim 1 or 2, characterized in that, The obtaining simulation parameters of the transport process determined based on the transport interface structure of the dissolved ions includes: Obtaining the single-ion dissolution distance during the simulated transport process; Based on the single-ion dissolution distance and the transport interface structure of the dissolved ions, determining the number of simulated dissolution times during the simulated transport process; Determining the single-ion dissolution distance, the number of simulated dissolution times, and the transport interface structure of the dissolved ions as the simulation parameters of the transport process.

4. The method according to any one of claims 1-3, characterized in that The training method of the target machine learning force field includes: According to the transport interface structure of the dissolved ions, determining the training simulation parameters of the transport process of the dissolved ions; Based on the training simulation parameters, performing molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface to obtain the force field training data set of the initial machine learning force field; Performing model learning on the initial machine learning force field according to the force field training data set to obtain the target machine learning force field of the dissolved ions.

5. The method according to claim 4, characterized in that, The training simulation parameters include the training single-ion dissolution distance, the number of training dissolution times, and the transport interface structure of the dissolved ions, and the training single-ion dissolution distance is greater than the single-ion dissolution distance during actual simulation; The performing molecular dynamics training simulation on the dissolution process of the dissolved ions at the transport interface based on the training simulation parameters to obtain the force field training data set of the initial machine learning force field includes: Updating the position of the dissolved ions in the transport interface structure according to the training single-ion dissolution distance to obtain the training input interface structure; Based on the training input interface structure, performing molecular dynamics training simulation to obtain the multi-frame interface structure data of the training input interface structure during the training simulation, and the structure energy corresponding to each frame of the interface structure and the atomic force information of each atom in the structure; After the training input interface structure training simulation is completed, based on the single - dissolution distance in the training, update the position of the dissolved ions in the training input interface structure to obtain an updated training input interface structure, and return to execute the step of performing molecular dynamics training simulation based on the training input interface structure until the number of updates of the training input interface structure is equal to the number of training dissolution times; Determine the multi - frame interface structure data obtained from the molecular dynamics training simulation, as well as the structural energy and atomic force information of each atom in the structure corresponding to each frame of the interface structure, as the force field training data set of the initial machine - learning force field.

6. The method according to any one of claims 4 or 5, characterized in that The model learning of the initial machine - learning force field according to the force field training data set to obtain the target machine - learning force field of the dissolved ions includes: Divide the force field training data set into a training set and a test set according to a preset division ratio; Perform model learning on the initial machine - learning force field according to the training set to obtain a machine - learning force field to be tested; Obtain the force field parameters of the machine - learning force field to be tested; When it is determined that the machine - learning force field to be tested meets the preliminary verification conditions based on the comparison result between the force field parameters and the preset preliminary verification parameter information, test the machine - learning force field to be tested according to the test set; When it is determined that the test of the machine - learning force field to be tested is successful, determine the machine - learning force field to be tested as the target machine - learning force field of the dissolved ions.

7. The method according to any one of claims 1-6, characterized in that The construction method of the transport interface structure of the dissolved ions includes: Construct the initial transport interface structure of the dissolved ions according to the construction parameters of the transport interface structure, where the construction parameters include the type of dissolved ions, the surface structure data of the vacuum layer, the solvent molecule structure data, and the solvent molecule density data; Perform a single - molecule dynamics simulation on the initial transport interface structure to obtain an equilibrium transport interface structure in a pre - equilibrium state; Obtain the solvent molecule density data in the middle of the liquid phase in the equilibrium transport interface structure; When it is determined that the equilibrium transport interface structure meets the structural rationality requirements according to the solvent molecule density data in the middle of the liquid phase, determine the equilibrium transport interface structure as the transport interface structure of the dissolved ions.

8. The method according to claim 7, wherein The method further includes: When it is determined that the equilibrium transport interface structure does not meet the structural rationality requirements according to the solvent molecule density data in the middle of the liquid phase, determine the solvent molecule correction data of the equilibrium transport interface structure; Based on the solvent molecule correction data, correct the solvent molecule density in the middle of the liquid phase of the equilibrium transport interface structure to obtain the transport interface structure of the dissolved ions.

9. The method according to any one of claims 1-8, characterized in that, The determination of the ion transport performance evaluation result of the dissolved ions according to the dissolution force information includes: Generate a free - energy change curve of the dissolved ions during the dissolution process according to the dissolution force information; Determine the dissolution energy barrier for the dissolved ions to transport in the transport interface structure based on the free - energy change curve; Determine the evaluation result of the ionic transport performance of the dissolved ions according to the dissolution energy barrier and the preset corresponding relationship between the energy barrier and the ease of dissolution.

10. An ion transport performance evaluation device, characterized in that, The device includes: A simulation parameter acquisition module, configured to acquire simulation parameters of the transport process determined based on the transport interface structure of the dissolved ions; A simulation module, configured to perform molecular dynamics simulation on the dissolution process of the dissolved ions at the transport interface based on the simulation parameters of the transport process and a target machine learning force field pre-trained according to the transport interface structure of the dissolved ions, to obtain the dissolution force information of the dissolved ions; A performance evaluation module, configured to determine the evaluation result of the ionic transport performance of the dissolved ions according to the dissolution force information.

11. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images