CN116092593A

CN116092593A - Electrolyte composition prediction method, device and computer equipment

Info

Publication number: CN116092593A
Application number: CN202310368977.8A
Authority: CN
Inventors: 黄杜斌; 常希望; 李爱军; 杨扬; 徐伟恒
Original assignee: Beijing Jinyu New Material Technology Co ltd
Current assignee: Beijing Jinyu New Material Technology Co ltd
Priority date: 2023-04-10
Filing date: 2023-04-10
Publication date: 2023-05-09
Anticipated expiration: 2043-04-10
Also published as: CN116092593B

Abstract

The application provides an electrolyte component prediction method, an electrolyte component prediction device and computer equipment, and belongs to the technical field of computers. The method comprises the following steps: determining a front track energy level difference corresponding to the plurality of first sample material sets; according to the component proportion and concentration of each first sample material group and the front line track energy level difference corresponding to each first sample material group, performing iterative training and correction on the initial track energy level difference prediction model to obtain an intermediate track energy level difference prediction model; performing iterative training and correction on the intermediate orbit energy level difference prediction model according to each second sample material group to obtain a target orbit energy level difference prediction model; and inputting the plurality of material groups to be predicted into a target orbit energy level difference prediction model, and determining at least one electrolyte component according to the final predicted front orbit energy level difference corresponding to each material group to be predicted. The effect of improving efficiency and accuracy can be achieved when the optimal components of the electrolyte with strong oxidizing and reducing properties need to be determined.

Description

Electrolyte composition prediction method, device and computer equipment

Technical Field

The application relates to the technical field of computers, in particular to a method and a device for predicting electrolyte components and computer equipment.

Background

The lithium ion battery has the characteristics of high energy density, high safety, long service life and the like, and is widely applied to mobile phones, tablet personal computers, electric automobiles and electric energy storage equipment. However, the electrolyte is an important component of the lithium ion battery, so that it is required to configure the lithium ion battery with the electrolyte having higher performance to ensure the performance of the lithium ion battery.

In the related art, generally, a related technician is required to prepare a plurality of electrolyte samples with different concentrations and different components by using a solvent and at least one lithium salt, then put each electrolyte sample into a battery model to perform a series of redox experiments to obtain detection results for indicating the oxidizing property and the reducing property, fine-tune the components of each electrolyte sample according to the detection results, and repeat the above steps for a plurality of times to detect the oxidizing property and the reducing property of each fine-tuned electrolyte sample. Thus, the optimal components of the electrolyte with strong oxidability and reducibility can be determined.

However, the solution flow of the related art is complicated, and it takes a lot of time to manually prepare the electrolyte samples, and many times of detection and fine adjustment are performed on each electrolyte sample, so that errors are unavoidable. Therefore, the related art scheme has problems of low efficiency and poor accuracy when it is required to determine the optimal components of the electrolyte having strong oxidizing and reducing properties.

Disclosure of Invention

The purpose of the application is to provide a method, a device and computer equipment for predicting electrolyte components, which can achieve the effect of improving efficiency and accuracy when the optimal components of the electrolyte with stronger oxidability and reducibility need to be determined.

Embodiments of the present application are implemented as follows:

in a first aspect of embodiments of the present application, there is provided a method for predicting an electrolyte composition, the method including:

determining front-line orbital energy level differences corresponding to a plurality of first sample material groups based on a molecular dynamics algorithm, the materials of each of the first sample material groups including at least one lithium salt, and the composition ratio and concentration of each of the materials in each of the first sample material groups being different;

performing iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front line track energy level difference corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model;

generating a plurality of random material groups based on a random algorithm, determining a plurality of second sample material groups according to the plurality of random material groups and the first sample material groups, and performing iterative training and correction on the intermediate orbit energy level difference prediction model according to the second sample material groups to obtain a target orbit energy level difference prediction model, wherein the target orbit energy level difference prediction model is used for predicting the orbit energy level difference of electrolyte manufactured based on the material groups;

And inputting a plurality of material groups to be predicted into the target track energy level difference prediction model, and determining at least one electrolyte component from each material group to be predicted according to the final prediction front track energy level difference corresponding to each material group to be predicted, which is output by the target track energy level difference prediction model.

Optionally, the determining, based on the molecular dynamics algorithm, a front-line orbital energy level difference corresponding to the plurality of first sample material groups includes:

performing structural adjustment on the conformation of each ion and each molecule included in the first sample material group, and determining structural information of each ion and each molecule under the condition of lowest energy;

acquiring force field information corresponding to each atom included in the first sample material group;

establishing a sample electrolyte virtual model based on the structural information, the force field information and the component proportion and concentration of the first sample material group;

controlling the operation of the sample electrolyte virtual model based on a molecular dynamics algorithm, and obtaining a front line orbit energy level difference of the sample electrolyte virtual model according to an operation result of the sample electrolyte virtual model;

and taking the front line track energy level difference of each sample electrolyte virtual model as the front line track energy level difference of each corresponding first sample material group.

Optionally, the controlling the operation of the sample electrolyte virtual model based on the molecular dynamics algorithm, and obtaining the front line track energy level difference of the sample electrolyte virtual model according to the operation result of the sample electrolyte virtual model includes:

controlling the operation of the sample electrolyte virtual model based on a preset energy minimization algorithm so that the sample electrolyte virtual model is in an energy minimization state;

controlling the sample electrolyte virtual model to perform at least one simulation treatment and at least one annealing simulation treatment based on a dynamics simulation treatment algorithm and an annealing simulation treatment algorithm to obtain a treatment result of the sample electrolyte virtual model;

determining a front line rail energy level difference of the sample electrolyte virtual model based on the processing result;

wherein the dynamics simulation processing algorithm comprises at least one of the following: the annealing simulation processing algorithm is an annealing algorithm based on the regular ensemble.

Optionally, the determining the front track energy level difference based on the processing result includes:

Determining a first solvent sheath radius of the lithium ion and coordination numbers of other particles except the lithium ion based on radial distribution function values of the other particles in the processing result;

determining a target solvation structure corresponding to the sample electrolyte virtual model according to the first solvent sheath radius and the coordination number;

and performing density functional theoretical calculation on the target solvated structure to obtain the front line orbit energy level difference of the sample electrolyte virtual model.

Optionally, the performing iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front track energy level difference corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model, including:

respectively establishing an initial training set and an initial testing set by taking the component proportion of each first sample material group as a first characteristic, the concentration of each first sample material group as a second characteristic and the front line track energy level difference corresponding to each first sample material group as a first label, wherein the first sample material group contained in the initial training set is different from the first sample material group contained in the initial testing set;

Inputting the initial orbit energy level difference prediction model based on a first sample material group included in the initial training set, and training the initial orbit energy level difference prediction model;

inputting the first sample material group included in the initial test set into a trained initial track energy level difference prediction model to obtain an initial predicted track energy level difference;

determining whether the initial track energy level difference prediction model meets a first preset condition according to the initial predicted track energy level differences and the front track energy level differences corresponding to the first sample material groups, if not, respectively adjusting the initial training set and the initial track energy level difference prediction model, and retraining the adjusted initial track energy level difference prediction model based on the adjusted initial training set;

and if so, taking the initial orbit energy level difference prediction model meeting the first preset condition as the intermediate orbit energy level difference prediction model.

Optionally, the determining a plurality of second sample material sets according to the plurality of random material sets and each of the first sample material sets includes:

inputting each random material group into the intermediate orbit energy level difference prediction model to obtain a predicted orbit energy level difference corresponding to each random material group;

Determining a preset number of to-be-selected material groups with the largest predicted orbit energy level difference from the random material groups, and establishing a to-be-selected electrolyte virtual model of each to-be-selected material group;

controlling the operation of each virtual model of the electrolyte to be selected based on a molecular dynamics algorithm to determine the front line track energy level difference of the virtual model of the electrolyte to be selected, and taking the front line track energy level difference of each virtual model of the electrolyte to be selected as the front line track energy level difference of the corresponding material group to be selected;

and taking a plurality of material groups with the largest front track energy level difference in the material groups to be selected and the first sample material groups as the second sample material groups.

Optionally, the inputting the plurality of material groups to be predicted into the target track energy level difference prediction model, determining at least one electrolyte component from each material group to be predicted according to the final prediction front track energy level difference corresponding to each material group to be predicted output by the target track energy level difference prediction model, including:

inputting a plurality of material groups to be predicted into the target track energy level difference prediction model, and determining a final prediction front track energy level difference corresponding to each material group to be predicted;

Determining convergence conditions of the material groups to be predicted according to the component proportion and concentration of the material groups to be predicted and the final prediction front line track energy level difference;

and if the convergence condition of each material group to be predicted meets a second preset condition, screening at least one material group to be predicted with the largest final prediction front line track energy level difference from each material group to be predicted as the electrolyte component.

In a second aspect of embodiments of the present application, there is provided an electrolyte composition prediction apparatus, the apparatus including:

a determining module for determining a front line orbital energy level difference corresponding to a plurality of first sample material groups based on a molecular dynamics algorithm, the materials of each of the first sample material groups including at least one lithium salt, and the composition ratio and concentration of each of the materials in each of the first sample material groups being different;

the first iterative training module is used for carrying out iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front track energy level difference corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model;

the second iterative training module is used for generating a plurality of random material groups based on a random algorithm, determining a plurality of second sample material groups according to the plurality of random material groups and each first sample material group, and carrying out iterative training and correction on the intermediate orbit energy level difference prediction model according to each second sample material group to obtain a target orbit energy level difference prediction model, wherein the target orbit energy level difference prediction model is used for predicting the orbit energy level difference of electrolyte manufactured based on each material group;

And the prediction module is used for inputting a plurality of material groups to be predicted into the target track energy level difference prediction model, and determining at least one electrolyte component from each material group to be predicted according to the final prediction front line track energy level difference corresponding to each material group to be predicted, which is output by the target track energy level difference prediction model.

In a third aspect of embodiments of the present application, there is provided a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the computer program implementing the electrolyte composition prediction method of the first aspect described above when executed by the processor.

In a fourth aspect of the embodiments of the present application, there is provided a computer-readable storage medium storing a computer program which, when executed by a processor, implements the electrolyte composition prediction method described in the first aspect.

The beneficial effects of the embodiment of the application include:

according to the method for predicting the electrolyte components, the front line track energy level difference can indicate the reducibility and/or the oxidability of the electrolyte supported by each first sample material group, the front line track energy level differences corresponding to a plurality of first sample material groups are determined based on a molecular dynamics algorithm, a neural network model capable of accurately predicting the front line track energy level differences of different material groups can be trained conveniently and trained subsequently, and the reducibility and/or the oxidability of the different material groups are determined through the front line track energy level differences predicted by the neural network model.

And carrying out iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front line track energy level difference corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model. Therefore, the accuracy of the initial track energy level difference prediction model for predicting the previous track energy level differences of different material groups can be improved, and the reliability and the accuracy of the obtained intermediate track energy level difference prediction model for predicting the previous track energy level differences can be ensured.

And generating a plurality of random material groups based on a random algorithm, determining a plurality of second sample material groups according to the plurality of random material groups and each first sample material group, and carrying out iterative training and correction on the intermediate orbit energy level difference prediction model according to each second sample material group to obtain a target orbit energy level difference prediction model.

Since the second sample material sets are selected from the first sample material sets and the random material sets, the front track energy level difference is large, and the random material sets are randomly generated material sets which are the same as the first sample material sets but have different composition ratios and concentrations. Thus, the diversity and the randomness of samples for training the intermediate orbit energy level difference prediction model can be improved, and the accuracy of predicting the material groups comprising different materials by the intermediate orbit energy level difference prediction model and/or the target orbit energy level difference prediction model can be further improved. Thus, the robustness of the target track energy level difference prediction model obtained through training is improved.

And inputting the plurality of material groups to be predicted into the target track energy level difference prediction model, and determining at least one electrolyte component from each material group to be predicted according to the final prediction front track energy level difference corresponding to each material group to be predicted, which is output by the target track energy level difference prediction model. In this way, at least one group of electrolyte components with smaller final predicted front line orbit energy level difference can be obtained, and each material indicated by each electrolyte component, the component proportion and concentration of each material and the total particle number of the electrolyte can be accurately determined, so that the components of the electrolyte with higher oxidability and reducibility can be accurately determined, and the detection, experiment and fine adjustment of the oxidability of each material group can be carried out for many times without spending a large amount of resources, time and manpower.

Thus, the effect of improving efficiency and accuracy can be achieved when the optimal components of the electrolyte having strong oxidizing and reducing properties need to be determined. And, it is also possible to ensure the robustness of determining the optimal composition of the electrolyte having strong oxidizing and reducing properties.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the embodiments will be briefly described below, it being understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered limiting the scope, and that other related drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a first method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 2 is a flow chart of a second method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 3 is a flow chart of a third method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 4 is a flow chart of a fourth method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 5 is a schematic diagram of a radial distribution function according to an embodiment of the present application;

FIG. 6 is a flowchart of a fifth method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 7 is a flowchart of a sixth method for predicting electrolyte composition according to an embodiment of the present application;

FIG. 8 is a flowchart of a seventh method for predicting electrolyte composition according to an embodiment of the present application;

fig. 9 is a schematic structural diagram of an electrolyte component predicting device according to an embodiment of the present application;

fig. 10 is a schematic structural diagram of a computer device according to an embodiment of the present application.

Detailed Description

For the purposes of making the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments. The components of the embodiments of the present application, which are generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations.

Thus, the following detailed description of the embodiments of the present application, as provided in the accompanying drawings, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.

In the description of the present application, it should be noted that the terms "first," "second," "third," and the like are used merely to distinguish between descriptions and are not to be construed as indicating or implying relative importance.

In the related art, generally, a related technician is required to prepare a plurality of electrolyte samples having different concentrations and different components by using a solvent and at least one lithium salt, then put each electrolyte sample into a battery model to perform a series of redox experiments to obtain detection results for indicating the oxidizing property and the reducing property, fine-tune the components of each electrolyte sample according to the detection results, and repeat the above steps a plurality of times to detect the oxidizing property and the reducing property of each fine-tuned electrolyte sample. Thus, the optimal components of the electrolyte with strong oxidability and reducibility can be determined.

Therefore, the embodiment of the application provides an electrolyte component prediction method, which comprises the steps of determining front track energy level differences corresponding to a plurality of first sample material groups based on a molecular dynamics algorithm, performing iterative training and correction on an initial track energy level difference prediction model according to component proportion and concentration of each first sample material group and the front track energy level differences corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model, generating a plurality of random material groups based on a random algorithm, determining a plurality of second sample material groups according to the plurality of random material groups and each first sample material group, performing iterative training and correction on the intermediate track energy level difference prediction model according to each second sample material group to obtain a target track energy level difference prediction model, inputting a plurality of material groups to be predicted into the target track energy level difference prediction model, and determining at least one electrolyte component from each material group to be predicted according to the final predicted front track energy level differences corresponding to each material group output by the target track energy level difference prediction model. The effect of improving efficiency and accuracy can be achieved when the optimal components of the electrolyte with strong oxidizing and reducing properties need to be determined.

The embodiment of the application will be described by taking an electrolyte composition prediction method applied to computer equipment as an example. It is not intended that the embodiments of the present application be limited to computer equipment for electrolyte composition prediction.

The method for predicting the electrolyte composition provided in the examples of the present application will be explained in detail.

Fig. 1 is a flowchart of an electrolyte composition prediction method provided in the present application, where the method may be applied to the above-mentioned computer device, and the computer device may be the aforementioned terminal device or server. Referring to fig. 1, an embodiment of the present application provides a method for predicting an electrolyte component, including:

step 1001: front-line orbital energy level differences corresponding to the plurality of first sample material sets are determined based on a molecular dynamics algorithm.

Alternatively, the molecular dynamics algorithm is a molecular simulation algorithm that can rely on newtonian mechanics to simulate the movement of a molecular system to calculate the thermodynamic or other physical quantities of the molecular system in different states of the molecular system.

Optionally, the material of each first sample material group includes at least one lithium salt, and the material of each first sample material group may further include an ether solvent. And the composition ratio and concentration of each material in each first sample material group are different.

The lithium salt may be a salt containing a lithium element, for example, a salt such as lithium bis (trifluoromethanesulfonyl) imide (LiTFSI), lithium bis (fluorosulfonyl) imide (LiFSI), lithium hexafluorophosphate (LiPF 6), and the like, and the examples of the present application are not limited thereto.

The ether solvent may be dimethyl ether (DME), propylene Oxide (PO), diethylene glycol diethyl ether (DEE), tetrahydrofuran, 1, 3-dioxolane, dioxolane (DOL), or the like, which is not limited in the examples herein.

Alternatively, the front orbital energy level difference may refer to an energy difference between a highest occupied molecular orbital (Highest Occupied Molecular Orbital, abbreviated as HOMO) and a lowest unoccupied molecular orbital (Lowest Unoccupied Molecular Orbital, abbreviated as LUMO), and may also be referred to as a band gap, i.e., HOMO-LUMO energy level.

In short, the stability of an electrolyte is mainly determined by the ability of a molecule or particle to transfer electrons, and the energy possessed by a molecule or particle when occupying the highest molecular orbital is called the HOMO value, and the energy possessed by a molecule or particle when occupying the lowest molecular orbital is called the LUMO value.

The electrolyte may be a liquid electrolyte, generally the electrolyte may be calculated by simulation according to molecular dynamics, and the flow of simulating the electrolyte may be performed by a stage of heating and cooling, which is not limited in the embodiment of the present application.

Optionally, the front rail energy level difference may also be used to indicate the reducibility and/or the oxidability of the electrolyte or electrolytes supported according to each first sample material set.

Generally, the higher the energy of the HOMO, the more likely the particle or molecule loses electrons, and the more reducing the particle or molecule. The lower the energy of the LUMO, the more readily available electrons the particle or molecule is represented and the more oxidizing the particle or molecule is. Therefore, in general, the larger the front-line orbital energy level difference, the more oxidizing and reducing the particles or molecules are represented.

Alternatively, the front track energy level difference may be obtained by actually measuring each first sample material group, or may be obtained by performing simulation calculation on each first sample material group, which is not limited in the embodiment of the present application.

Illustratively, a total of 3 materials of two lithium salts and one solvent may be included in each first sample material group. The two lithium salts may be LiFSI and LiTFSI, respectively, and the solvent may be DME. Where Li refers to a cation, FSI and TFSI may each refer to an anion.

For example, the amount fraction of the substance may be defined as a unit of measurement, and in any mixture, the amount fraction of the substance of component a=the amount of the substance of component a ≡total amount of the substance. The first sample material sets were designed by adding different amounts of LiFSI and LiTFSI to the DME solvent, and the concentration of the first sample material sets was controlled by adjusting the amount of DME solvent, or the total amount of LiFSI and LiTFSI.

Specifically, the composition ratio of each material in each first sample material group, and the concentration and total particle count of each first sample material group may be set as in table 1 below.

TABLE 1

It can be seen that the sum of the mass fractions of LiFSI and LiTFSI in table 1 is 1, and in general, the total particle number in each first sample material group may be greater than 500 and less than 1500, and specifically, the total particle number may be controlled to be about 1000, which is not limited in the embodiment of the present application. And the concentration of each first sample material set can be controlled by adjusting the amount of solvent DME.

The concentration may refer to a concentration, a mole fraction, a mass fraction, or a mass mole fraction of an amount of a substance, which is not limited in the examples herein.

In addition, the interval of the mass fractions of LiFSI and LiTFSI in each first sample material group may be set according to the range of the related art pre-test, or may be randomly set, which is not limited in the embodiment of the present application.

It is noted that, since the front rail energy level difference may indicate the reducibility and/or the oxidability of the electrolyte or the electrolyte supported according to each first sample material group, determining the front rail energy level difference corresponding to each first sample material group facilitates performing a subsequent operation, and in particular, training the model according to the front rail energy level difference corresponding to each first sample material group.

Step 1002: and carrying out iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front line track energy level difference corresponding to each first sample material group to obtain an intermediate track energy level difference prediction model.

Alternatively, with continued reference to Table 1, the compositional ratio of the first sample material set refers to the fractional amount of the substance of each lithium salt in the first sample material set.

For example, the composition ratio of the first sample material group of group 1 in Table 1 means that the mass fraction of LiFSI is 0.992308 and the mass fraction of LiTFSI is 0.007692. The concentration of the first sample material set of group 1 was 13.

Alternatively, the initial orbit energy level difference prediction model and the intermediate orbit energy level difference prediction model may be neural network models.

Moreover, the initial orbit energy level difference prediction model and the intermediate orbit energy level difference prediction model may be models that use gaussian process regression as a machine learning regression function, which is not limited in the embodiment of the present application.

The initial trajectory energy level difference prediction model may be used to predict the front trajectory energy level differences corresponding to different material groups.

Optionally, the intermediate orbit energy level difference prediction model is an initial orbit energy level difference prediction model which meets certain performance conditions after training. Generally, the performance condition can be set or set by a related technician according to actual needs, and can be used for screening out an initial track energy level difference prediction model with smaller error of predicting the front track energy level difference.

It should be noted that, after the initial track energy level difference prediction model is iteratively trained and corrected according to the component proportion and the concentration of each first sample material group and the front track energy level difference corresponding to each first sample material group, the accuracy of predicting the front track energy level difference of different material groups by the initial track energy level difference prediction model can be improved, so that the reliability and the accuracy of predicting the front track energy level difference by the obtained intermediate track energy level difference prediction model can be ensured.

In addition, the performance of the intermediate orbit energy level difference prediction model for predicting the previous orbit energy level differences of different material groups is generally better than that of the initial orbit energy level difference prediction model, that is, the previous orbit energy level difference of any material group output by the intermediate orbit energy level difference prediction model is closer to the actual previous orbit energy level difference of any material group.

Step 1003: and generating a plurality of random material groups based on a random algorithm, determining a plurality of second sample material groups according to the plurality of random material groups and each first sample material group, and carrying out iterative training and correction on the intermediate orbit energy level difference prediction model according to each second sample material group to obtain a target orbit energy level difference prediction model.

Alternatively, the random algorithm may be an algorithm that randomly generates a plurality of material sets of different proportions, concentrations, based on the input material or composition, which embodiments of the present application are not limited to.

Alternatively, the materials in the random material sets may be consistent with the materials in the first sample material sets, but the composition ratio of the materials in the random material sets may be different from the composition ratio of the materials in the first sample material sets, and the concentration and total particle count of the random material sets may be different from the concentration and total particle count of the first sample material sets.

The materials in each random material group can be different from the materials in each first sample material group, so that the randomness of obtaining each second sample material group can be improved, and the robustness of the target track energy level difference prediction model is further improved.

The proportions of the components of the random material groups may be different from each other.

Illustratively, if two lithium salts of LiFSI and LiTFSI are included in each first sample material set and DME is used as the solvent, then each random material set should also include two lithium salts of LiFSI and LiTFSI and DME is used as the solvent. In this way, the uniformity of the determined second sample material sets can be ensured.

Alternatively, each of the second sample material sets may be a plurality of material sets having a large difference in front line track energy level selected from each of the random material sets and each of the first sample material sets. That is, each second sample material set may include at least one random material set and at least one first sample material set, or may include only at least one first sample material set or at least one first sample material set random material set, which is not limited in the embodiments of the present application.

The number of second sample material sets may be any number, and the embodiments of the present application are not limited in this regard.

Optionally, the target orbital level difference prediction model is used to predict an orbital level difference of an electrolyte fabricated based on each material group.

It is noted that since the second sample material set is a plurality of material sets having a large difference in front line track energy level selected from the first sample material sets and the random material sets, the random material sets are randomly generated material sets having the same material as the first sample material sets but different composition ratios and concentrations. Thus, the diversity and the randomness of samples for training the intermediate orbit energy level difference prediction model can be improved, and the accuracy of predicting the material groups comprising different materials by the intermediate orbit energy level difference prediction model and/or the target orbit energy level difference prediction model can be further improved.

Thus, the robustness of the target track energy level difference prediction model obtained through training is improved.

Step 1004: and inputting the plurality of material groups to be predicted into the target track energy level difference prediction model, and determining at least one electrolyte component from each material group to be predicted according to the final prediction front track energy level difference corresponding to each material group to be predicted, which is output by the target track energy level difference prediction model.

Alternatively, each set of materials to be predicted may refer to any set of materials for which a front-line orbital energy level difference prediction is desired.

In general, the materials in each of the sets of materials to be predicted may be the same as or similar to the materials in each of the first sample sets of materials and/or the materials in each of the second sample sets of materials, e.g., liFSI, liTFSI, DME, etc. In this way, the accuracy and reliability of the predicted final predicted track level difference can be ensured.

In addition, the materials in each material group to be predicted may also be different from the materials in each first sample material group and/or the materials in each second sample material group, for example, each material group to be predicted may further include two lithium salts, namely LiTFSI and LiPF6, and PO is used as a solvent. In this way, the robustness and practicality of the target track level difference prediction model can be ensured.

For example, if the material in each of the sets of materials to be predicted may be the same as or similar to the material in each of the first sets of sample materials and/or the material in each of the second sets of sample materials, each of the sets of materials to be predicted may be a step between decreasing the constituent proportions of each of the materials in each of the second sets of sample materials. In this way, groups of materials to be predicted having a smaller difference in composition ratio of the materials between the groups, a larger number of groups of materials having composition ratios different from each other, can be obtained based on the groups of second sample materials.

Alternatively, the final predicted front track energy level difference may refer to the front track energy level differences of each of the groups of materials to be predicted through the target track energy level difference prediction model.

Alternatively, each electrolyte composition may refer to at least one material group to be predicted whose corresponding final predicted front rail energy level difference determined from each material group to be predicted is large according to each final predicted front rail energy level difference.

Alternatively, each electrolyte component may be used to indicate each material, the ratio and concentration of each material component, and/or the total particle count of the electrolyte or electrolyte that is required when preparing an electrolyte or electrolyte that is more oxidizing and reducing. The embodiments of the present application are not limited in this regard.

It should be noted that, in this way, at least one set of electrolyte components with smaller final predicted front line energy level differences can be obtained, and each material indicated by each electrolyte component, the component proportion and concentration of each material, and the total particle number of the electrolyte can be accurately determined, so that the components of the electrolyte with higher oxidability and reducibility can be accurately determined, and the detection, experiment and fine adjustment of the oxidability of various material groups can be performed many times without spending a great deal of resources, time and manpower.

Thus, the effect of improving efficiency and accuracy can be achieved when the optimal components of the electrolyte having strong oxidizing and reducing properties need to be determined.

In the embodiment of the application, the front track energy level difference can indicate the reducibility and/or the oxidability of the electrolyte supported by each first sample material group, the front track energy level differences corresponding to a plurality of first sample material groups are determined based on a molecular dynamics algorithm, a neural network model capable of accurately predicting the front track energy level differences of different material groups can be trained conveniently and then the reducibility and/or the oxidability of the different material groups are determined through the front track energy level differences predicted by the neural network model.

In one possible implementation, referring to fig. 2, determining the front line orbital energy level differences corresponding to the plurality of first sample material groups based on a molecular dynamics algorithm includes:

Step 1005: the conformations of the ions and molecules included in the first sample material set are adjusted and structural information of the ions and molecules is determined at the lowest energy.

Optionally, each ion comprises each anion and each cation comprised in the first set of sample materials. For example, if the first sample material set includes LiFSI, liTFSI two lithium salts and DME solvent, then each ion in the first sample material set includes lithium ion (Li+), bis-fluoro-sulfonimide anion (FSI-), bis-trifluoro-methanesulfonimide anion (TFSI-), and each molecule may include DME molecules.

Alternatively, the conformation is a spatial geometric arrangement of atoms or groups due to rotation of chemical bonds.

The structural information may include information on conformation, bond length, bond angle, bond energy, and the like.

By way of example, the conformation of the cations, anions and molecules can be optimized by means of the DMol3 module by inputting the Materials, the composition ratios of the Materials and the concentrations of the Materials in the first sample material groups into the Materials Studio software, respectively, until the conformation of the cations, anions and molecules is determined to be the least energetic conformation in the case where the total energy of the cations, anions and molecules is determined, and the structural information of the ions and molecules in this case is determined.

Step 1006: and acquiring force field information corresponding to each atom included in the first sample material group.

Alternatively, each atom in the first sample material group may include each lithium salt and all atoms included in the solvent.

For example, the first sample material group includes LiFSI, liTFSI two lithium salts and solvent DME, and then the first sample material group includes fluorine (F), nitrogen (N), oxygen (O), sulfur (S), lithium (Li), carbon (C) and hydrogen (H).

Alternatively, the force field information may refer to force fields to which various atoms are subjected in different ions or different molecules.

For example, the methyl group has the chemical formula of-CH 3, the ethyl group has the chemical formula of-C2H 5, and both the methyl group and the ethyl group can be seen to contain carbon atoms, but the force field suffered by the carbon atoms in the methyl group is different from the force field received by the carbon atoms in the ethyl group.

In short, the force field information is information for indicating force fields to which each atom is subjected in different ions, and specifically may also indicate parameters to which each atom is subjected to various force fields, such as subjected van der waals forces, and the like.

For example, each material in the first sample material set may be input to corresponding software, and a Compass II force field may be used to obtain a potential file of a corresponding ion or molecule, and then determine a force field of each atom according to the potential file, which is not limited in this embodiment of the present application.

Step 1007: and establishing a sample electrolyte virtual model based on the structural information, the force field information and the component proportion and concentration of the first sample material group.

Alternatively, the sample electrolyte virtual model may refer to a simulation model of an electrolyte structure built from the structural information, the force field information, the constituent proportions and concentrations of the first sample material set.

Specifically, the sample electrolyte virtual model may be an electrolyte simulation model with a boundary length of 0.5nm and a cubic shape.

For example, the structural information, the force field information, the component proportion and concentration of the first sample material group may be input into an amorphorus Cell module, the amorphorus Cell module divides lithium salt in the first sample material group into lithium ions and corresponding anions, and determines the total number of particles contained in the sample electrolyte virtual model according to the component proportion and concentration of the first sample material group, so as to construct a cubic virtual model with a boundary length of 0.5 nm.

Therefore, the real electrolyte or the electrolyte is not required to be prepared manually according to the first sample material group for experiments, so that the problem that errors occur in the process of preparing the electrolyte or carrying out experiments manually can be avoided, and the efficiency and the accuracy can be improved.

Step 1008: and controlling the operation of the sample electrolyte virtual model based on a molecular dynamics algorithm, and obtaining the front line orbit energy level difference of the sample electrolyte virtual model according to the operation result of the sample electrolyte virtual model.

Alternatively, the operation result may be used to indicate the arrangement or spatial position of each molecule and each ion in the sample electrolyte virtual model after the simulation operation based on molecular dynamics.

In addition, the operation result may be a result obtained in a state where the sample electrolyte virtual model is at the lowest energy.

It is noted that after the sample electrolyte virtual model is created in step 1007, there may be internal stress in the sample electrolyte virtual model that is just created, at which time the energy of the sample electrolyte virtual model is not the lowest state. Therefore, relaxation needs to be performed on the sample electrolyte virtual model, so that the sample electrolyte virtual model enters an equilibrium state and is closer to a real situation, and the accuracy of the obtained front line track energy level difference of the sample electrolyte virtual model is further ensured.

The process of controlling the operation of the sample electrolyte virtual model based on the molecular dynamics algorithm is equivalent to relaxation treatment of the sample electrolyte virtual model.

Step 1009: and taking the front line track energy level difference of each sample electrolyte virtual model as the front line track energy level difference of each corresponding first sample material group.

It is worth to say that, in this way, the experiment of repeatedly preparing the real electrolyte can be performed without wasting a great deal of manpower, material resources and time, but the simulation is performed based on the molecular dynamics and the sample electrolyte virtual model, and then the front line track energy level difference of each sample electrolyte virtual model under the condition of lowest energy is accurately obtained. Thus, the front line track energy level difference of each first sample material group can be accurately determined through simulation of the sample electrolyte virtual model.

The operation of the sample electrolyte virtual model can be controlled based on Lammps software, and one possible implementation manner is provided for this purpose, referring to fig. 3, the operation of the sample electrolyte virtual model is controlled based on a molecular dynamics algorithm, and a front line track energy level difference of the sample electrolyte virtual model is obtained according to an operation result of the sample electrolyte virtual model, which includes:

step 1010: and controlling the operation of the sample electrolyte virtual model based on a preset energy minimization algorithm so that the sample electrolyte virtual model is in an energy minimization state.

Optionally, the preset energy minimization algorithm may include at least one of: the steepest descent method (SD), conjugate gradient method (CG), newton-larsen method (NR), and any other algorithm that can bring the sample electrolyte virtual model into an energy minimized state.

Generally, the operation of the sample electrolyte virtual model can be controlled by combining three algorithms, namely a steepest descent method, a conjugate gradient method and a Newton-Larson method, which is not limited in the embodiment of the application.

In this way, the influence of the internal stress in the sample electrolyte virtual model can be eliminated.

In addition, when the sample electrolyte virtual model is controlled to run based on a preset energy minimization algorithm, the sample electrolyte virtual model may be controlled to run once, or the sample electrolyte virtual model may be controlled to run twice or more times, so as to ensure that the influence of internal stress can be completely eliminated.

Step 1011: and controlling the sample electrolyte virtual model to perform at least one simulation treatment and at least one annealing simulation treatment based on the dynamics simulation treatment algorithm and the annealing simulation treatment algorithm to obtain a treatment result of the sample electrolyte virtual model.

Optionally, the dynamics simulation processing algorithm includes at least one of: a canonical ensemble (NVT) based dynamics algorithm, an isothermal-isobaric ensemble (NPT) based dynamics algorithm, a micro-canonical ensemble (NVE) based dynamics algorithm.

The annealing simulation processing algorithm is an annealing algorithm based on the regular ensemble.

Optionally, if multiple simulation processes and multiple annealing simulation processes are required, the sample electrolyte virtual model may be controlled to alternately perform the annealing simulation process and the simulation process.

The processing results may be obtained by performing five simulation processes and two annealing simulation processes, for example. Specifically, two simulation processes may be performed first, in the first simulation process, a kinetic algorithm based on NVT may be used for simulation, the temperature is set to 300 kelvin (K), the simulation duration is set to 0.1 nanosecond (ns), and then, in the second simulation process, a kinetic algorithm based on NPT is used for simulation, and the temperature is set to 500K, and the simulation duration is set to 0.1ns. Then, carrying out primary annealing simulation treatment, adopting an NVT-based annealing algorithm to carry out simulation, setting the high temperature in the annealing process to be 500K, setting the heating and cooling cycle times to be 2-3 times, and setting the annealing simulation duration to be 0.2ns.

And then performing third simulation treatment, and setting the temperature to be 500K and the simulation duration to be 0.1ns by adopting a dynamics algorithm based on NPT. Then, carrying out a second annealing simulation treatment, adopting an NVT-based annealing algorithm to simulate, setting the high temperature in the annealing process to be 500K, setting the heating and cooling cycle times to be 2-3 times, and setting the annealing simulation time length to be 0.2ns.

Then, two simulation processes are performed, simulation is performed by adopting a dynamics algorithm based on NVT in the fourth simulation process, the temperature is set to 300K, and the simulation duration is set to 0.2ns. After the fourth simulation is completed, performing a fifth simulation, performing simulation by adopting a NET-based dynamics algorithm, and setting the temperature to 300K and the simulation time length to 0.1ns.

The above is merely one possible molecular dynamics simulation procedure provided in the embodiments of the present application, and is not meant to represent the only way to simulate molecular dynamics in a manner that only five simulation processes and two annealing simulation processes can be performed. For example, any other possible number of simulation processes and annealing simulation processes may be performed, and different simulation ensembles, simulation temperatures, and/or simulation durations may also be set, so that each ion and each molecule in the sample electrolyte virtual model fully move in the state of lowest energy, which is not limited in the embodiments of the present application.

Therefore, the ions and molecules in the sample electrolyte virtual model can fully move in the state of lowest energy, so that the sample electrolyte virtual model can be kept in a natural state, and further the processing result can be ensured to more truly simulate the distribution condition of the ions and molecules in the electrolyte prepared based on the first sample material group.

Step 1012: and determining a front line orbit energy level difference of the sample electrolyte virtual model based on the processing result.

Thus, the front line orbit energy level difference of the sample electrolyte virtual model can be accurately determined through the processing result obtained through simulation processing.

In one possible implementation, referring to fig. 4, determining the front track energy level difference based on the processing result includes:

step 1013: the first solvent sheath radius of the lithium ion and coordination number of the other particles and the lithium ion are determined based on the radial distribution function value of the other particles except the lithium ion in the processing result.

Alternatively, the other particles may refer to all anions and all molecules except lithium ions, such as FSI-, TFSI-, and DME molecules.

Alternatively, the radial distribution function value (Radical distribution function, abbreviated as RDF) may refer to a given space where one object is centered on the probability of finding other objects around. The RDF value of particle a can be used to characterize the ratio of the number density of particle a relative to the number density of all particles in the whole space over a range of distances from the lithium ion.

Additionally, the solvent sheath may be used to characterize the process of solvent particle binding to solute particle, while the first solvent sheath radius of the lithium ion may be used to characterize the extent of the last layer of lithium ion binding solvent.

Alternatively, the coordination number of any other particle with the lithium ion may refer to the number of coordination of the any other example around the lithium ion.

Illustratively, referring to fig. 5, as can be seen from (a) in fig. 5, the first solvent sheath radius of the lithium ion is located near 0.27nm, and the order of other particles from near to far distance of the lithium ion is: solvent molecules (DME), first anions (TFSI-), second anions (FSI-).

As can be seen from fig. 5 (b), at the first solvent sheath radius of the lithium ion, the ratio of the coordination number of other particles to the lithium ion is the solvent molecule (DME): first anion (TFSI-): second anion (FSI-) =3.3:0.33:0.001.

Thus, the structure of the first solvent sheath of the lithium ions can be analyzed and determined.

Step 1014: and determining a target solvation structure corresponding to the virtual model of the sample electrolyte according to the radius of the first solvent sheath and the coordination number.

Alternatively, the target solvated structure may be an electrically neutral-locally solvated structure in the virtual model of the sample electrolyte.

The target solvation structure may be used to characterize a HOMO-LUMO energy level of the sample electrolyte virtual model, and a front-line orbital energy level difference of the sample electrolyte virtual model may be determined from the target solvation structure.

It is noted that 3 or more different solvent clusters can be screened out in a corresponding manner, and then corresponding average values are obtained according to the parameters of the 3 or more different solvent clusters screened out, so as to represent the target solvated structure corresponding to the sample electrolyte virtual model by using the final average values.

Step 1015: and carrying out density functional theoretical calculation on the target solvated structure to obtain the front line orbit energy level difference of the sample electrolyte virtual model.

Alternatively, the basic idea of density functional theory (density functional theory, abbreviated as DFT) is to describe the ground state physical properties of solids, molecules and atoms by means of electron density functions, in particular the ground state properties of a system can be represented by means of the particle number density.

Illustratively, after performing a density functional theoretical calculation on the target solvated structure corresponding to the 9 first sample material groups based on table 1 above, 9 sets of front rail energy levels may be obtained as shown in table 2 below.

TABLE 2

The front orbital level difference of the sample electrolyte virtual model can be obtained by subtracting the average LUMO from the average HOMO of each group in table 2.

Thus, the front line track energy level difference of the sample electrolyte virtual model corresponding to each first sample material group can be accurately calculated, and the reliability of training the initial track energy level difference prediction model and the intermediate track energy level difference prediction model is further improved.

In a possible implementation manner, referring to fig. 6, according to the component proportion and concentration of each first sample material group and the front line track energy level difference corresponding to each first sample material group, performing iterative training and correction on the initial track energy level difference prediction model to obtain an intermediate track energy level difference prediction model, including:

step 1016: the initial training set and the initial testing set are respectively established by taking the component proportion of each first sample material group as a first characteristic, the concentration of each first sample material group as a second characteristic and the front line track energy level difference corresponding to each first sample material group as a first label.

Optionally, the first set of sample materials included in the initial training set is different from the first set of sample materials included in the initial test set.

Typically, the number of initial training sets is greater than the number of initial test sets, so that the training accuracy of the initial trajectory level difference prediction model may be improved.

Illustratively, with continued reference to Table 1 above, each initial training set may be the composition ratio and concentration of each material in any 6 of the first sample material sets of Table 1, as well as the corresponding front line rail energy level differences for those 6 first sample material sets. Each initial test set may be the composition ratio and concentration of each material in the 3 first sample material sets other than the arbitrary 6 sets, and the corresponding front line track energy level differences for the 3 first sample material sets.

In this way, a plurality of training sets and test sets may be established based on the constituent proportions, concentrations, and front line rail energy level differences for each first sample material set to train the initial rail energy level difference prediction model.

Step 1017: the initial trajectory energy level difference prediction model is trained based on the first set of sample materials included in the initial training set input to the initial trajectory energy level difference prediction model.

Alternatively, the initial orbit energy level difference prediction model may be trained using a gaussian process regression function, which is not limited by the embodiments of the present application.

Step 1018: and inputting the first sample material group included in the initial test set into a trained initial track energy level difference prediction model to obtain an initial predicted track energy level difference.

Optionally, the initial predicted track energy level difference refers to the initial track energy level difference prediction model trained by the initial training set, and the predicted previous track energy level difference of the first sample material group in the initial test set is obtained.

Step 1019: and determining whether the initial track energy level difference prediction model meets a first preset condition according to the initial predicted track energy level differences and the front track energy level differences corresponding to the first sample material groups, if not, respectively adjusting the initial training set and the initial track energy level difference prediction model, and retraining the adjusted initial track energy level difference prediction model based on the adjusted initial training set.

Alternatively, the first preset condition may be preset for detecting the prediction accuracy of the initial track level difference prediction model. The first preset condition may specifically be that the covariance or average value of the difference between the energy level differences of each initial predicted track and the front track energy level differences corresponding to each first sample material group is smaller than a certain threshold, or that the iteration number reaches a preset number, which is not limited in the embodiment of the present application.

If the initial orbit energy level difference prediction model does not meet the first preset condition, the prediction accuracy of the initial orbit energy level difference prediction model is lower, and the training is needed to be carried out again.

Additionally, adjusting the initial training set may refer to exchanging a portion of the first set of sample materials in each initial training set with the first set of sample materials in each initial testing set.

The adjusting of the initial trajectory energy level difference prediction model may be adjusting a parameter or a kernel function of the initial trajectory energy level difference prediction model.

Step 1020: if yes, the initial orbit energy level difference prediction model meeting the first preset condition is used as the intermediate orbit energy level difference prediction model.

It should be noted that, if the initial track energy level difference prediction model meets the first preset condition, it may be determined that the prediction accuracy of the initial track energy level difference prediction model is higher. Thus, it is possible to ensure a high prediction accuracy of the intermediate orbit energy level difference prediction model.

In one possible implementation, referring to fig. 7, determining a plurality of second sample material sets from the plurality of random material sets and each first sample material set includes:

step 1021: and inputting each random material group into the intermediate orbit energy level difference prediction model to obtain the predicted orbit energy level difference corresponding to each random material group.

Alternatively, the predicted orbital level difference corresponding to each random material group may refer to a previous-stage orbital level difference predicted by the intermediate orbital level difference prediction model according to the composition ratio and concentration of each material in each random material group.

Step 1022: and determining a preset number of candidate material groups with the largest predicted orbit energy level difference from the random material groups, and establishing a candidate electrolyte virtual model of each candidate material group.

Alternatively, the preset number may be any positive integer, and in general, the preset number may be the same as the number of the respective first sample material groups.

Optionally, the virtual model of the electrolyte to be selected may refer to a simulation model of an electrolyte structure established according to structural information, force field information, component proportion and concentration of the material group to be selected, which corresponds to the material group to be selected.

The method for establishing the virtual model of the electrolyte to be selected for each material group to be selected is the same as the method for establishing the virtual model of the electrolyte to be selected for each sample, and specifically, the contents of steps 1005-1007 can be referred to, and details are not described here.

It is worth to say that, because each material group to be selected is the maximum of the energy level difference of the prediction orbit in each random material group, the accuracy of each second sample material group obtained later can be further improved, and the accuracy of the intermediate orbit energy level difference prediction model is further improved.

Step 1023: and controlling the operation of each virtual model of the electrolyte to be selected based on a molecular dynamics algorithm so as to determine the front line track energy level difference of the virtual model of the electrolyte to be selected, and taking the front line track energy level difference of each virtual model of the electrolyte to be selected as the front line track energy level difference of the corresponding material group to be selected.

For example, the step of controlling the operation of the virtual model of the electrolyte to be selected based on the molecular dynamics algorithm to obtain the front rail energy level difference of the virtual model of the electrolyte to be selected may be specifically referred to in step 1008 and the other corresponding steps, which are not described herein.

Step 1024: and taking the material groups with the largest front line track energy level difference in the material groups to be selected and the first sample material groups as second sample material groups.

In a possible implementation manner, referring to fig. 8, a plurality of material groups to be predicted are input into the target track energy level difference prediction model, and at least one electrolyte component is determined from each material group to be predicted according to a final prediction front track energy level difference corresponding to each material group to be predicted output by the target track energy level difference prediction model, including:

Step 1025: and inputting a plurality of material groups to be predicted into the target track energy level difference prediction model, and determining the final prediction front track energy level difference corresponding to each material group to be predicted.

After multiple iterative training, the prediction accuracy of the target orbit energy level difference prediction model is higher, so that the obtained final prediction front orbit energy level difference corresponding to each material group to be predicted is very close to or even the same as the actual front orbit energy level difference of the electrolyte prepared based on each material group to be predicted, and the accuracy of determining the optimal component of the electrolyte with stronger oxidizing property and reducibility according to the final prediction front orbit energy level difference corresponding to each material group to be predicted can be improved.

Step 1026: and determining convergence conditions of the material groups to be predicted according to the component proportion and concentration of the material groups to be predicted and the energy level difference of the final prediction front line orbit.

Alternatively, the convergence may be used to characterize the similarity of several groups of materials to be predicted for which the final predicted front rail energy level difference is the greatest among the groups of materials to be predicted, and to indicate whether each group of materials to be predicted is concentrated on the same group of materials to be predicted or on close groups of materials to be predicted.

For example, if the component proportions, concentrations, or the like of several groups of materials to be predicted having the largest difference in energy level of the final predicted front trajectory are similar or identical, it is indicated that the convergence of these groups of materials to be predicted is superior. Otherwise, it indicates that the convergence of the groups of materials to be predicted is poor.

Alternatively, if there are multiple groups of materials to be predicted with similar or identical composition ratios and concentrations, the difference between the final prediction front line track energy level differences is small, it may also indicate that the convergence of the groups of materials to be predicted is better.

Step 1027: and if the convergence condition of each material group to be predicted meets a second preset condition, screening at least one material group to be predicted with the largest final prediction front line track energy level difference from each material group to be predicted as the electrolyte component.

Alternatively, the second preset condition may be set according to actual needs, which is the case in the embodiments of the present application. For example, the second preset condition may be that the error between the component ratios of the N groups of materials to be predicted with the largest difference in energy level of the final predicted front line track is less than 0.01, and the error between the concentrations is less than 1, which is not limited in the embodiment of the present application.

In addition, after any one of the groups of materials to be predicted is taken as the electrolyte component, each material in the any one of the groups of materials to be predicted may be taken as a material in the electrolyte component, and the component ratio and concentration of each material in the any one of the groups of materials to be predicted may be taken as the component ratio and concentration of each material in the electrolyte component.

Notably, the greater the difference in front orbital energy levels, the more oxidizing and reducing the particles or molecules are represented. Therefore, by screening at least one material group to be predicted with the largest energy level difference of the final predicted front line track from the material groups to be predicted as the electrolyte component, the determined electrolyte component can be ensured to have stronger oxidizing property and reducing property.

Thus, each material of each electrolyte component, the component proportion and concentration of each material in each electrolyte component can be accurately determined, and the optimal components of the electrolyte with strong oxidability and reducibility can be further determined.

In a possible implementation manner, performing iterative training and correction on the intermediate track energy level difference prediction model according to each second sample material group to obtain a target track energy level difference prediction model, including:

and respectively establishing a target training set and a target test set by taking the component proportion of each second sample material group as a third characteristic, the concentration of each second sample material group as a fourth characteristic and the front line track energy level difference corresponding to each second sample material group as a second label.

Optionally, the second set of sample materials included in each target training set is different from the second set of sample materials included in each target test set.

The intermediate orbit energy level difference prediction model is trained based on the second set of sample materials included in the target training set input to the intermediate orbit energy level difference prediction model.

And inputting a second sample material group included in the target test set into the intermediate orbit energy level difference prediction model to obtain a target predicted orbit energy level difference.

Optionally, the target predicted track energy level difference refers to the intermediate track energy level difference prediction model trained by the target training set, and predicts the obtained previous track energy level difference of the second sample material group in the target test set.

And determining whether the intermediate orbit energy level difference prediction model meets a third preset condition according to the target predicted orbit energy level differences and the orbit energy level differences corresponding to the second sample material groups, if not, respectively adjusting the target training set and the intermediate orbit energy level difference prediction model, and retraining the adjusted intermediate orbit energy level difference prediction model based on the adjusted target training set.

Alternatively, the third preset condition may be that the covariance or average value of the difference between the target predicted track energy level differences and the front track energy level differences corresponding to the second sample material sets is smaller than a certain threshold, or that the iteration number reaches a preset number, which is not limited in the embodiment of the present application.

If yes, the intermediate orbit energy level difference prediction model meeting the third preset condition is taken as the target orbit energy level difference prediction model.

Since the method of performing iterative training and correction on the intermediate track level difference prediction model according to each second sample material set to obtain the target track level difference prediction model is similar to the method of performing iterative training and correction on the initial track level difference prediction model to obtain the intermediate track level difference prediction model, the above-mentioned steps 1016-1020 can be performed, and details thereof will not be repeated here.

In this way, it is possible to ensure that the prediction accuracy of the obtained target track level difference prediction model is high.

The following describes a device, equipment, a computer readable storage medium, etc. for executing the method for predicting electrolyte components provided in the present application, and specific implementation processes and technical effects thereof are referred to above, which are not described in detail below.

Fig. 9 is a schematic structural diagram of an electrolyte composition prediction apparatus according to an embodiment of the present application, referring to fig. 9, the apparatus includes:

a determination module 201 is configured to determine a front line orbital energy level difference corresponding to the plurality of first sample material groups based on a molecular dynamics algorithm.

Optionally, the materials of each first sample material group include at least one lithium salt, and the composition ratio and concentration of each material in each first sample material group are different.

The first iterative training module 202 is configured to perform iterative training and correction on the initial track energy level difference prediction model according to the component proportion and concentration of each first sample material group and the front track energy level difference corresponding to each first sample material group, so as to obtain an intermediate track energy level difference prediction model.

The second iterative training module 203 is configured to generate a plurality of random material sets based on a random algorithm, determine a plurality of second sample material sets according to the plurality of random material sets and each first sample material set, and perform iterative training and correction on the intermediate track energy level difference prediction model according to each second sample material set to obtain a target track energy level difference prediction model.

The prediction module 204 is configured to input a plurality of material groups to be predicted into the target track energy level difference prediction model, and determine at least one electrolyte component from each material group to be predicted according to a final prediction front line track energy level difference corresponding to each material group to be predicted output by the target track energy level difference prediction model.

The foregoing apparatus is used for executing the method provided in the foregoing embodiment, and its implementation principle and technical effects are similar, and are not described herein again.

The above modules may be one or more integrated circuits configured to implement the above methods, for example: one or more application specific integrated circuits (Application Specific Integrated Circuit, abbreviated as ASICs), or one or more microprocessors, or one or more field programmable gate arrays (Field Programmable Gate Array, abbreviated as FPGAs), etc. For another example, when a module above is implemented in the form of a processing element scheduler code, the processing element may be a general-purpose processor, such as a central processing unit (Central Processing Unit, CPU) or other processor that may invoke the program code. For another example, the modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

Fig. 10 is a schematic structural diagram of a computer device according to an embodiment of the present application. Referring to fig. 10, a computer apparatus includes: memory 301, processor 302, memory 301 stores a computer program executable on processor 302, and processor 302 implements the steps of any of the various method embodiments described above when executing the computer program.

The present application also provides a computer readable storage medium storing a computer program which, when executed by a processor, implements the steps of the respective method embodiments described above.

Optionally, the present application also provides a program product, such as a computer readable storage medium, comprising a program for performing any of the electrolyte composition prediction method embodiments described above when being executed by a processor.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of elements is merely a logical functional division, and there may be additional divisions of actual implementation, e.g., multiple elements or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.

The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in hardware plus software functional units.

The integrated units implemented in the form of software functional units described above may be stored in a computer readable storage medium. The software functional unit is stored in a storage medium, and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (english: processor) to perform part of the steps of the methods of the embodiments of the invention. And the aforementioned storage medium includes: u disk, mobile hard disk, read-Only Memory (ROM), random access Memory (Random Access Memory, RAM), magnetic disk or optical disk, etc.

The foregoing is merely a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think about changes or substitutions within the technical scope of the present application, and the changes or substitutions are covered by the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method of predicting electrolyte composition, the method comprising:

2. The electrolyte composition prediction method of claim 1, wherein the determining the front line orbital energy level differences corresponding to the plurality of first sample material groups based on the molecular dynamics algorithm comprises:

3. The method for predicting electrolyte composition according to claim 2, wherein the controlling the operation of the sample electrolyte virtual model based on the molecular dynamics algorithm, and obtaining the front line rail energy level difference of the sample electrolyte virtual model based on the operation result of the sample electrolyte virtual model, comprises:

4. The electrolyte composition prediction method according to claim 3, wherein the determining a front rail energy level difference based on the processing result comprises:

5. The method of predicting electrolyte composition according to claim 1, wherein the performing iterative training and correction on the initial trajectory level difference prediction model according to the composition ratio and concentration of each of the first sample material groups and the front trajectory level difference corresponding to each of the first sample material groups to obtain the intermediate trajectory level difference prediction model comprises:

6. The electrolyte composition prediction method of claim 1, wherein said determining a plurality of second sample material sets from said plurality of random material sets and each of said first sample material sets comprises:

7. The electrolyte composition prediction method according to any one of claims 1 to 6, wherein the inputting the plurality of material groups to be predicted into the target orbit energy level difference prediction model, determining at least one electrolyte composition from each of the material groups to be predicted based on a final prediction front orbit energy level difference corresponding to each of the material groups to be predicted output from the target orbit energy level difference prediction model, comprises:

8. An electrolyte composition prediction apparatus, comprising:

9. A computer device, comprising: memory, a processor, in which a computer program is stored which is executable on the processor, when executing the computer program, realizing the steps of the method of any of the preceding claims 1 to 7.

10. A computer readable storage medium, characterized in that it has stored thereon a computer program which, when executed by a processor, implements the steps of the method according to any of claims 1 to 7.