Electrolyte formula generation method, storage medium and processor

Abstract

The embodiment of the invention provides an electrolyte formula generation method, a storage medium and a processor, and belongs to the technical field of secondary batteries. The electrolyte formula generation method comprises the following steps: determining a target physical property parameter index of an electrolyte and at least one battery performance index according to an application requirement of a battery; according to the electrolyte component prediction model, screening out a formula conforming to a target physical property parameter index from the electrolyte component feasible region, and taking the formula as a group of candidate electrolyte formulas; estimating a performance index estimation value of each of the group of candidate electrolyte formulas according to a performance index estimation model, and solving a weighted sum of the performance index estimation values in the at least one battery performance index; and taking the corresponding electrolyte formula of which the weighted sum meets a preset standard as a recommended electrolyte formula. According to the electrolyte formula generation method disclosed by the invention, the optimal battery electrolyte formula can be recommended by comprehensively considering cost, physical property factors and battery performance, and high-efficiency and low-cost design of the electrolyte is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of secondary batteries, and in particular to an electrolyte formula generation method, a storage medium and a processor. Background Art

[0002] Secondary batteries represented by lithium-ion batteries are used in many fields such as mobile phones, electric vehicles, electric bicycles, electric tools, and energy storage markets, and the requirements for their energy density, cycle performance, safety performance and other indicators are becoming increasingly higher. The electrolyte in the battery carries important functions such as ion transmission, electronic insulation, and the formation of solid electrolyte membranes at the interface of positive and negative electrodes. The electrolyte of secondary batteries is generally prepared in a certain proportion under certain conditions from raw materials such as solvents, electrolyte lithium salts, and necessary additives.

[0003] Currently, the methods for finding the optimal formula of electrolyte are mostly based on manual experience and continuous optimization based on the existing formula. Due to the complex composition of the electrolyte, it often takes a lot of work and a long time period to design an electrolyte that meets actual needs.

[0004] In recent years, some methods have emerged that use mathematical statistics and machine learning algorithms to recommend the optimal battery electrolyte formula, but the effect of optimizing the electrolyte components is very limited, and it is usually impossible to take into account multiple battery performances. For example, a method for optimizing electrolyte components based on a mass triangle model (CN102931436A) is only applicable to electrolyte formulas containing three components, and only considers the optimization of electrolyte ion conductivity. The conductivity mainly affects the rate performance and low temperature performance of the battery. This method has great limitations on the effect of optimizing the electrolyte components and cannot meet the requirements of various complex performance indicators of the battery. Summary of the invention

[0005] The purpose of the embodiment of the present invention is to provide a method that can comprehensively consider physical factors and multiple battery performances to recommend an optimal battery electrolyte formulation, thereby achieving a high-efficiency, low-cost design of the electrolyte.

[0006] In order to achieve the above object, an embodiment of the present invention provides a method for generating an electrolyte formula, comprising:

[0007] Determine target physical property parameter index of the electrolyte and at least one battery performance index according to the application requirements of the battery;

[0008] According to the electrolyte composition prediction model, the formulas that meet the target physical property parameter indicators are screened out from the feasible domain of electrolyte components and used as a group of candidate electrolyte formulas;

[0009] The performance index estimation model is used to estimate the performance index estimation value of each of a group of candidate electrolyte formulas, and the weighted sum of the estimated values ​​of each performance index in at least one battery performance index is calculated. The corresponding electrolyte formula whose weighted sum meets the preset standard is used as the recommended electrolyte formula.

[0010] Preferably, the electrolyte composition prediction model is obtained by the following steps:

[0011] Step 1: Determine the first group of electrolyte formulas according to the target physical property parameter indicators;

[0012] Step 2: preparing an electrolyte solution according to each of the first set of electrolyte formulas, and testing the physical property parameters of each electrolyte; and

[0013] Step 3: Based on the correspondence between each electrolyte formula and the corresponding measured physical property parameters, an electrolyte composition prediction model is established through a data fitting method.

[0014] Furthermore, the electrolyte formula generation method also verifies the electrolyte composition prediction model through the following steps:

[0015] Step 4: according to the method for generating the first set of electrolyte formulas, obtain the second set of electrolyte formulas, prepare corresponding electrolyte solutions, and test the physical property parameters of each electrolyte to verify the accuracy of the electrolyte composition prediction model; and

[0016] Step 5: If the test results indicate that the electrolyte composition prediction model exceeds the preset error range, a set of supplementary electrolyte formulas is determined according to the generation method of the first set of electrolyte formulas, and steps 2-4 are re-executed based on the supplementary electrolyte formulas until the electrolyte composition prediction model meets the preset error range.

[0017] Optionally, the data fitting method is selected from the following ones with the best fitting effect: least squares method, gradient descent method, iterative least squares, and Lemberg-Maoist algorithm.

[0018] Optionally, the method for determining the feasible domain of the electrolyte components includes: determining the material composition of the electrolyte and the upper and lower limits of the mass fraction of each material in the material composition according to the target physical property parameter indicators, and the sum of the mass fractions of each material in the material composition is 1.

[0019] Furthermore, the method for determining the first group of electrolyte formulas includes: generating the first group of electrolyte formulas from the feasible domain of electrolyte components by enumeration method or at least one experimental design method.

[0020] Optionally, the experimental design method includes at least one of the following: classical screening design, response surface design, Taguchi design, orthogonal supersaturated design, filling design, definitive screening design, and mixture design.

[0021] Preferably, the training of the performance indicator valuation model includes:

[0022] The electrolyte formula in the training data set is constructed into an electrolyte formula matrix x according to the order of material composition;

[0023] Construct a performance indicator value output vector B for each performance indicator in the battery performance indicator α (x), where B α The i-th component B of (x) α (x i ) is the i-th electrolyte formula x i The battery performance index value under the performance index α;

[0024] Based on x and B α (x) Construct the logarithmic marginal likelihood function logP(B) corresponding to one of the battery performance indicators α α (x)), and initialize the hyperparameter set ξ of the performance indicator estimation model;

[0025] Perform gradient descent on ξ and maximize the log-marginal likelihood function logP(B α (x)), in order to iteratively update the hyperparameter ξ value, and obtain the final logarithmic marginal likelihood function logP(B α (x)) as the prior model of the performance indicator estimation model corresponding to the performance indicator α,

[0026] The training data set includes: a set of candidate electrolyte formulas and verified electrolyte formulas, and the corresponding battery performance index values ​​of each electrolyte formula.

[0027] Furthermore, the log marginal likelihood function is calculated by the following formula:

[0028]

[0029] Where I is the unit matrix; σ 2 For B α (x) is the variance of the noise; n is the number of samples of electrolyte formula in the training data set; P(B α (x)) is the measured cycle performance test index value B under the hyperparameter set ξ α (x) probability of occurrence;

[0030] μ(x)=[μ(x1),μ(x2),...,μ(x n )], where μ(x i )=ax i +b, a and b are the hyperparameters in the hyperparameter set ξ;

[0031] k(x i ,x j ) is the i-th electrolyte formula x i With the jth electrolyte formula x j The covariance function, σ f and l are the hyperparameters in the hyperparameter set ξ, and σ f is the standard deviation of the Gaussian kernel function, and l is the characteristic length scale of the Gaussian kernel function.

[0032] Preferably, the step of obtaining the estimated value of the performance indicator includes:

[0033] Obtain an electrolyte formula x* from a set of candidate electrolyte formulas, and calculate a posterior distribution formula of the electrolyte formula;

[0034] Determine the conditional probability distribution formula of the battery performance index value at the performance index α based on the posterior distribution formula; and

[0035] According to the conditional probability distribution formula, estimate the mean value of the battery performance index value of x* at performance index α and variance And the mean As the performance index estimate of the electrolyte formula x* at the performance index α.

[0036] Preferably, the weighted sum also includes a weighted value of the production cost of the electrolyte.

[0037] Optionally, the corresponding electrolyte formula whose weighted sum meets the preset standard is used as the recommended electrolyte formula, including: sorting the weighted sum from large to small, and selecting the corresponding electrolyte formula from the front to the back in the obtained sequence as the recommended electrolyte formula.

[0038] Optionally, at least one battery performance indicator includes at least one of the following: a cycle performance indicator, a DCR indicator, a rate performance indicator, a storage performance indicator, and a safety performance indicator.

[0039] Optionally, the corresponding physical property parameters of the target physical property parameter index include at least one of dielectric constant, viscosity, melting point, boiling point, and ionic conductivity.

[0040] Preferably, the material composition of the electrolyte formula includes at least a solvent, an electrolyte salt, and an additive, and the additive includes at least one of a film-forming additive, a flame retardant additive, an anti-overcharge additive, and an overcharge protection additive;

[0041] The electrolyte salt includes at least one of hexafluorophosphate, bisfluorosulfonyl imide salt, nitrate, nitrite, fluoride salt, chloride salt, bromide salt, iodide salt, difluorophosphate, difluorooxalatoborate, dioxalatoborate, tetrafluorooxalatoborate, bisfluorosulfonyl imide salt, and bistrifluoromethanesulfonyl imide salt; and

[0042] The cation in the electrolyte salt includes at least one of the following: Li + 、Na + 、K+、Mg 2+ , Ca 2+ 、Zn 2+ 、Al 3+ .

[0043] On the other hand, the present invention provides a machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute the electrolyte formula generating method of the present application.

[0044] On the other hand, the present invention provides a processor for running a program, wherein the program, when run, is used to execute the electrolyte formula generating method of the present application.

[0045] Through the above technical scheme, a method for generating electrolyte formula based on hierarchical statistical learning is realized. First, a group of candidate electrolyte formulas are preliminarily screened out through the target physical property parameters of the electrolyte and the corresponding electrolyte composition prediction model, and then the estimated values ​​of each performance indicator of each formula in a group of candidate electrolyte formulas in at least one battery performance indicator are estimated through the performance indicator valuation model. Finally, the optimal electrolyte formula in a group of candidate electrolyte formulas is recommended by the weighted sum of the estimated values ​​of each formula under each performance indicator. The weighted sum can further include the cost weighted value of the electrolyte, so that the optimal battery electrolyte formula can be recommended by comprehensively considering the cost, physical property factors, and battery performance, thereby realizing high-efficiency and low-cost design of the electrolyte. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0047] Figure 1 is a flow chart of an embodiment of a method for generating an electrolyte formula of the present application;

[0048] Figure 2 is a flow chart of the initial screening stage in another embodiment of the method for generating an electrolyte formula of the present application; and

[0049] Figure 3 yes Figure 2In the embodiment, a flow chart of screening out a recommended formulation from a group of candidate electrolyte formulations obtained in the initial screening stage. DETAILED DESCRIPTION

[0050] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0051] The embodiment of the present invention provides a method for generating an electrolyte formula. The electrolyte generated by the method can be an electrolyte for a lithium-ion battery or other power batteries. The implementation process is as follows: Figure 1 As shown, including:

[0052] Step 1: Determine the target physical property parameter index of the electrolyte and at least one battery performance index according to the application requirements of the battery;

[0053] Step 2: According to the electrolyte composition prediction model, the formulas that meet the target physical property parameter indicators are screened out from the feasible domain of electrolyte components and used as a group of candidate electrolyte formulas;

[0054] Step 3: Estimate the performance index estimated value of each of a group of candidate electrolyte formulas according to the performance index valuation model, calculate the weighted sum of the estimated values ​​of each performance index in at least one battery performance index, and use the corresponding electrolyte formula whose weighted sum meets the preset standard as the recommended electrolyte formula.

[0055] It should be noted that steps 1 and 2 are the initial screening stages of the electrolyte formula, which can obtain a group of candidate formulas that meet the target physical property parameter indicators. Step 3 is to further select the optimal formula that meets at least one battery performance indicator from the group of candidate formulas obtained from the initial screening through a performance indicator valuation model.

[0056] In step one, the target physical property parameter index of the electrolyte and the battery performance index are set according to the application requirements of the battery to be designed, wherein the target physical property parameter index is used for the initial screening stage of the electrolyte formula, and the battery performance index is used for further screening based on the initial screening.

[0057] The initial screening process may be, for example, to initially screen out a group of candidate electrolyte formulas that meet the target physical property parameter index from a group of electrolyte formulas including a large number of randomly generated electrolyte formulas through the electrolyte composition prediction model in step 2. The group of candidate electrolyte formulas thus obtained does not rely on the technician's existing experience or the existing formula basis, so usually, a group of candidate electrolyte formulas outside the expected range of the technician can be obtained in the initial screening stage.

[0058] In step 2, the method for determining the feasible domain of electrolyte components includes: determining the material composition of the electrolyte and the upper and lower limits of the mass fraction of each material in the material composition according to the target physical property parameter indicators, and the sum of the mass fractions of each material in the material composition is 1.

[0059] The further screening process based on the initial screening corresponds to the above step three, which estimates the performance index of each of a group of candidate electrolyte formulas through the performance index valuation model, and obtains the weighted sum of the estimated values ​​of each performance index, so as to obtain a comprehensive evaluation of each performance index of each candidate formula in at least one battery performance index in a group of candidate electrolyte formulas. Therefore, the best formula that meets each performance index in at least one battery performance index can be further screened out from the group of candidate electrolyte formulas obtained in the initial screening stage for recommendation.

[0060] It should be noted that the above-mentioned optimal formula can be multiple formulas with high comprehensive evaluation rankings. For example, if the estimated values ​​of the performance indicators are not much different, or the electrolytes are configured according to the above-mentioned multiple optimal formulas, and the measured parameter indicators are not much different or each has its own advantages and disadvantages, then multiple recommended formulas with better performance can be selected from a group of candidate electrolyte formulas.

[0061] Furthermore, based on the above multiple recommended formulas, other factors such as production cost, environmental pollution, etc. can be considered to further select the final recommended formula.

[0062] In some embodiments, at least one battery performance indicator includes at least one of the following: a cycle performance indicator, a DCR indicator, a rate performance indicator, a storage performance indicator, and a safety performance indicator.

[0063] In some embodiments, the corresponding physical property parameter of the target physical property parameter indicator includes at least one of dielectric constant, viscosity, melting point, boiling point, and ionic conductivity.

[0064] In some embodiments, the material composition of the electrolyte formula includes at least a solvent, an electrolyte salt, and an additive, and the additive includes at least one of a film-forming additive, a flame retardant additive, an anti-overcharge additive, and an overcharge protection additive;

[0065] The electrolyte salt includes at least one of hexafluorophosphate, bisfluorosulfonyl imide salt, nitrate, nitrite, fluoride salt, chloride salt, bromide salt, iodide salt, difluorophosphate, difluorooxalatoborate, dioxalatoborate, tetrafluorooxalatoborate, bisfluorosulfonyl imide salt, and bistrifluoromethanesulfonyl imide salt;

[0066] The cation in the electrolyte salt includes at least one of the following: Li + 、Na + 、K+、Mg2+ , Ca 2+ 、Zn 2+ 、Al 3+ .

[0067] Compared with the prior art, the method for generating electrolyte formula in this embodiment has the following technical advantages:

[0068] (1) Through the electrolyte composition prediction model, a group of candidate electrolyte formulas that meet the target physical property parameter indicators are obtained in the initial screening stage. On the one hand, this group of candidate electrolyte formulas does not rely on the technicians' existing experience or the existing formula basis, and can obtain electrolyte candidate formulas beyond the technicians' expected range.

[0069] On the other hand, the electrolyte composition prediction model can be used to preliminarily screen a large number of randomly generated electrolyte formulas, thereby avoiding aimless selection from a large number of electrolyte formulas, thereby avoiding the manpower and material costs of configuring a large number of electrolytes and testing electrolyte parameters, and preliminarily screening from a large number of random formulas to avoid missing possible excellent formulas;

[0070] (2) In the process of re-screening through the performance index valuation model, it is possible to find a formula that meets both the physical parameter index and the battery performance index among the candidate formulas obtained in the initial screening, thus avoiding the manpower and material costs of configuring a large amount of electrolyte and testing electrolyte parameters. At the same time, it can also help technicians further discover the corresponding relationship between electrolyte composition and battery performance index.

[0071] Based on this embodiment, the implementation effect of the initial screening stage can be optimized through the following preferred implementation methods, and the process is as follows: Figure 2 shown.

[0072] In some embodiments, a reliable electrolyte composition prediction model is obtained by the following steps:

[0073] Step 1: Determine the first group of electrolyte formulas according to the target physical property parameter indicators;

[0074] Step 2: preparing an electrolyte solution according to each of the first set of electrolyte formulas, and testing the physical property parameters of each electrolyte; and

[0075] Step: 3: Based on the correspondence between each electrolyte formula and the corresponding measured physical property parameters, an electrolyte composition prediction model is established through a data fitting method.

[0076] Among them, the data fitting method selects the one with the best fitting effect from the following: least squares method, gradient descent method, iterative least squares method, and Lemberg-Maoist algorithm.

[0077] In some embodiments, the correctness of the obtained electrolyte composition prediction model can be further verified and optimized by the following steps:

[0078] Step 4: according to the method for generating the first set of electrolyte formulas, obtain the second set of electrolyte formulas, prepare corresponding electrolyte solutions, and test the physical property parameters of each electrolyte to verify the accuracy of the electrolyte composition prediction model; and

[0079] Step 5: If the test results indicate that the electrolyte composition prediction model exceeds the preset error range, a set of supplementary electrolyte formulas is determined according to the generation method of the first set of electrolyte formulas, and steps 2-4 are re-executed based on the supplementary electrolyte formulas until the electrolyte composition prediction model meets the preset error range.

[0080] The first group of electrolyte formulas can be determined by the following steps:

[0081] A first set of electrolyte formulations is generated from the feasible domain of electrolyte components by enumeration or at least one experimental design method, wherein the experimental design method can be at least one of the following: classical screening design, response surface design, Taguchi design, orthogonal supersaturated design, filling design, deterministic screening design, and mixture design.

[0082] To further illustrate the process of the initial screening stage in the electrolyte formula generation method of the present application, the following describes the actual process of obtaining a set of candidate electrolyte formulas in an implemented case.

[0083] 1. The battery application is a low-temperature fast-charging scenario. The selected solvents are ethylene carbonate (EC), propylene carbonate (PC), ethyl methyl carbonate (EMC), dimethyl carbonate (DMC), and diethyl carbonate (DEC). The most commonly used lithium hexafluorophosphate (LiPF6) is selected as the lithium salt. The concentration of the lithium salt is 1 mol·L -1 ;

[0084] 2. According to the battery application scenario, the mass fractions of EC, PC, EMC, DEC, and DMC are set to 0.1-0.3, 0-0.1, 0-0.7, 0-0.3, and 0-0.3, respectively. The mass fractions of each solvent satisfy the following relationship:

[0085] EC+PC+DMC+DEC+EMC=1 (1)

[0086] 3. The experimental design method is mixed material design, and the sum of the specified variable ratios is guaranteed to be equal to 1. A set of electrolyte formulas shown in Table 1 below is obtained (for the solvent part, the lithium salt has been selected to have a concentration of 1 mol·L in the previous step). -1 of lithium hexafluorophosphate).

[0087] Table 1 Mixing design solvent composition

[0088]

[0089]

[0090] 4. After preparing the electrolyte in the order in Table 1, measure the conductivity at -10°C;

[0091] 5. Select the method with the smallest fitting error among the least squares method, gradient descent method, iterative least squares method, Lev-Mao algorithm and other fitting methods to establish an electrolyte composition prediction model for the electrolyte formula composition corresponding to the physical property parameters. In this embodiment, the least squares method has a smaller root mean square error and is effective. The formula is as follows:

[0092]

[0093] Wherein, m represents the number of samples, and n represents the parameter dimension. For example, in this embodiment, the parameter dimension includes the main effect of each solvent component and the interaction between two solvents. The above formula is quantized to obtain:

[0094] Xβ=y (3)

[0095]

[0096] To find the best estimate of β The least squares method can be used and the problem is transformed into the following:

[0097] S(β)=||Xβ-y|| (5)

[0098] By differentiating S(β) to find the maximum value, we can get:

[0099] X T Xβ=X T y (6)

[0100] If X T If X is non-singular, then β has a unique solution:

[0101]

[0102] 6. Randomly select a number of predicted electrolyte formulas to prepare electrolyte solutions and conduct mathematical model accuracy tests. The experimental errors are all within ±0.11ms / cm, and the predicted sample conductivity order is consistent with the experimental results, which ensures the validity of the simulation results. The results are shown in Table 2 below:

[0103] Table 2 Predicted values ​​and measured values ​​of simulation results

[0104]

[0105] 7. According to the battery application requirements, the performance indicators are set as follows: the conductivity requirement at -10°C is >6ms / cm;

[0106] 8. Randomly and uniformly generate 10,000 sets of electrolyte formulas, predict the conductivity at -10°C using the electrolyte composition prediction model, and screen out the electrolyte formulas that meet the requirements.

[0107] exist Figure 1 Based on the embodiment shown, the implementation effect of the re-screening stage can be optimized through the following preferred implementation method, and the process is as follows Figure 3 shown.

[0108] In some embodiments, the training method of the performance indicator estimation model is a Gaussian process regression method, and the process includes:

[0109] The electrolyte formula in the training data set is constructed into an electrolyte formula matrix x according to the order of material composition;

[0110] Construct a performance indicator value output vector B for each performance indicator in the battery performance indicator α (x), where B α The i-th component B of (x) α (x i ) is the i-th electrolyte formula x i The battery performance index value under the performance index α;

[0111] Based on x and B α (x) Construct the logarithmic marginal likelihood function logP(B) corresponding to one of the battery performance indicators α α (x)), and initialize the hyperparameter set ξ of the performance indicator estimation model;

[0112] Perform gradient descent on ξ and maximize the log-marginal likelihood function logP(B α (x)), in order to iteratively update the hyperparameter ξ value, and obtain the final logarithmic marginal likelihood function logP(B α (x)) as the prior model of the performance indicator estimation model corresponding to the performance indicator α,

[0113] The training data set includes: a set of candidate electrolyte formulas and verified electrolyte formulas, and the corresponding battery performance index values ​​of each electrolyte formula. It should be noted that the training data set should select electrolyte formulas that have been experimentally verified to meet the application requirements of the battery under design, and should at least meet the corresponding target physical property parameter indicators or performance indicators.

[0114] Furthermore, the log marginal likelihood function is calculated by the following formula:

[0115]

[0116] Where I is the unit matrix; σ 2 For B α (x) is the variance of the noise; n is the number of samples of electrolyte formula in the training data set; P(B α (x)) is the measured cycle performance test index value B under the hyperparameter set ξ α (x) probability of occurrence;

[0117] μ(x)=[μ(x1),μ(x2),...,μ(x n )], where μ(x i )=ax i +b, a and b are the hyperparameters in the hyperparameter set ξ;

[0118] k(x i ,x j ) is the i-th electrolyte formula x i With the jth electrolyte formula x j The covariance function, σ f and l are the hyperparameters in the hyperparameter set ξ, and σ f is the standard deviation of the Gaussian kernel function, and l is the characteristic length scale of the Gaussian kernel function.

[0119] In some implementations, the step of obtaining the estimated value of the performance indicator includes:

[0120] Obtain an electrolyte formula x* from a set of candidate electrolyte formulas, and calculate a posterior distribution formula of the electrolyte formula;

[0121] Determine the conditional probability distribution formula of the battery performance index value at the performance index α based on the posterior distribution formula; and

[0122] According to the conditional probability distribution formula, estimate the mean value of the battery performance index value of x* at performance index α and variance And the mean As the performance index estimate of the electrolyte formula x* at the performance index α.

[0123] In some embodiments, the weighted sum also includes a weighted value of the production cost of the electrolyte, the weighted sums of the candidate formulas obtained are sorted from large to small, and the corresponding electrolyte formula is selected from the front to back in the obtained sequence as the recommended electrolyte formula.

[0124] To further illustrate the process of the re-screening stage in the electrolyte formula generation method of the present application, the following describes the actual process of obtaining a recommended formula from a group of candidate electrolyte formulas in an implemented case.

[0125] Step 1, data preprocessing: processing the electrolyte formula into a matrix form in the order of solvent, electrolyte salt, and additive;

[0126] Step 2: Input the electrolyte formula matrix and the battery performance index value D = {x, B(x)}. The battery performance index value includes any item in the cycle performance test, DCR test, rate performance test, storage performance test, safety test, etc. Since the prediction method of each battery performance index value and confidence interval is the same, the subsequent steps take the cycle performance test index value and its confidence interval prediction as an example. The subsequent steps will use the new data set D = {x, B α (x)}, where x represents the input electrolyte formula vector matrix, B α (x) represents the output vector of the cycle performance test index value in the Gaussian model;

[0127] Step 3: Train and obtain the cycle performance test index value model based on Gaussian process regression (GPR). The calculation process is as follows:

[0128] 3.1) The cycle performance test index value includes the determined cycle performance test index value and random noise, and the calculation formula is as follows:

[0129] B α (x i )=f α (x i )+ε i (8)

[0130] Where, i∈[1,n], i∈Z, x i For any input electrolyte formula, ε i is Gaussian noise and is independent of each other, with mean 0 and variance σ 2 Gaussian distribution, that is, ε i ~N(0,σ 2 ). α (x i ) indicates that x i The determined cycle performance test index value is given by a mean function μ(x i ) and a covariance function k(x i ,x j ) Determine, determine the function f α (x i ) is a Gaussian process, expressed as follows:

[0131] fα (x i )~N[(μ(x i ),k(x i ,x j )] (9)

[0132] 3.2) In practical applications, in order to define a Gaussian process, it is necessary to specify the mean function and covariance function terms. Here, the Gaussian kernel function is used as the covariance kernel function. In this study, the mean function and covariance function expressions used are as follows:

[0133] μ(x i )=ax i +b (10)

[0134]

[0135] Among them, a and b represent the hyperparameters of the Gaussian model, σ f represents the function standard deviation of the Gaussian kernel function, and l represents the characteristic length scale of the kernel function;

[0136] 3.3) According to the properties of joint Gaussian distribution, for the sample set {x,B α (x)}, B α (x) also obeys the Gaussian distribution, as follows:

[0137] B α (x)~N[(μ(x),k(x,x)+σ 2 I] (12)

[0138] Where I is the identity matrix, σ 2 is the variance term caused by noise, μ(x) is the mean function, k(x,x) is the covariance term, and the expression of the mean function μ(x) is as follows:

[0139] μ(x)=[μ(x1),μ(x2),...,μ(x n )] (13)

[0140] Calculate the covariance matrix k(x,x) of the input electrolyte formula sample set using the following formula:

[0141]

[0142] 3.4) Let the hyperparameter set Initialize the hyperparameters and calculate the log-marginal likelihood function using the following formula:

[0143]

[0144] Where n is the sample size, P(B α(x)) represents the measured cycle performance test index value B under the hyperparameter set ξ α The probability of (x) occurring is calculated by maximizing the log-marginal likelihood function by performing gradient descent on ξ, thereby iteratively updating the value of the hyperparameter ξ.

[0145] Step 4: Input the target input, i.e. the electrolyte formula x obtained in the preliminary screening of physical parameters * , from the description of step 3.1, we can know that B α (x * ) represents the cycle performance test index value of the electrolyte to be designed,

[0146] B α (x * )=f α (x * )+ε (16)

[0147] where f α (x * ) obeys a Gaussian distribution, and its mean function μ(x * )=ax * +b, where a and b have been calculated in step 3.4, k(x * ,x * ) is calculated using the Gaussian kernel function.

[0148] According to the properties of Gaussian distribution, [B α (x),f α (x * )] is still a Gaussian distribution, and the corresponding joint Gaussian distribution prior expression is as follows:

[0149]

[0150] Among them, k(x,x * ) is the training set input electrolyte formula x and the target input electrolyte formula x * The covariance matrix of k(x * ,x) is k(x,x * ) is the transposed matrix; with k(x,x * ) as an example, the calculation formula is as follows:

[0151]

[0152] Among them, k(x * ,x i ) are calculated by formula (11) from which the hyper parameters have been derived;

[0153] Through Bayesian inference calculation, we know that f α (x * ) is:

[0154] P(f α (x * )B α (x))~N[(μ * ,(σ * ) 2 )] (19)

[0155] in:

[0156] μ * = k(x * ,x)[k(x,x)+σ 2 I] -1 [B α (x)-μ(x)]+μ(x * ) (20)

[0157] (σ * ) 2 = k(x * ,x * )-k(x * ,x)[k(x,x)+σ 2 I] -1 k(x,x * ) (twenty one)

[0158] In the formula, μ * is the mean of the predicted values, (σ*) 2 is the variance of the predicted value.

[0159] The electrolyte formula x given the target input * , the predicted average value μ of the cycle performance test index value of the target input electrolyte formula can be estimated based on the constructed GPR-based cycle performance test index value model * and variance (σ * ) 2 +σ 2 Similarly, according to steps 1 to 4, other battery performance index values ​​can also be estimated.

[0160] Step 5, calculating the cost of each electrolyte formula;

[0161] Step 6: According to the requirements, the results of the battery performance indicators and the weighted sum of the costs are calculated, the electrolyte formulas are sorted, and a recommended electrolyte formula is given.

[0162] An embodiment of the present invention provides a storage medium on which a program is stored. When the program is executed by a processor, the method for generating an electrolyte formula of the present invention is implemented.

[0163] An embodiment of the present invention provides a processor, which is used to run a program, wherein the electrolyte formula generation method of the present invention is executed when the program is run.

[0164] The embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor, and the method for generating an electrolyte formula of the present invention is implemented when the processor executes the program. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0165] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program initialized with the steps of the electrolyte formula generation method of the present invention.

[0166] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0167] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0168] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0169] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0170] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0171] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0172] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0173] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0174] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for generating an electrolyte formula, comprising: Determine target physical property parameter index of the electrolyte and at least one battery performance index according to the application requirements of the battery; According to the electrolyte composition prediction model, the formulas that meet the target physical property parameter indicators are screened out from the feasible domain of electrolyte components and used as a group of candidate electrolyte formulas; The performance indicator estimation model is used to estimate the performance indicator estimation value of each of the set of candidate electrolyte formulas, and the weighted sum of the estimated values ​​of the performance indicators in the at least one battery performance indicator is calculated, and the corresponding electrolyte formula whose weighted sum meets the preset standard is used as the recommended electrolyte formula.

2. The method for generating an electrolyte formula according to claim 1, characterized in that: The electrolyte composition prediction model is obtained by the following steps: Step 1: Determine the first group of electrolyte formulas according to the target physical property parameter indicators; Step 2: preparing an electrolyte solution according to each of the first group of electrolyte formulas, and testing the physical property parameters of each electrolyte; and Step: 3: Based on the correspondence between each electrolyte formula and the corresponding measured physical property parameters, the electrolyte composition prediction model is established through a data fitting method.

3. The method for generating an electrolyte formula according to claim 2, characterized in that: The electrolyte composition prediction model is further verified by the following steps: Step 4: according to the method for generating the first group of electrolyte formulas, obtain the second group of electrolyte formulas, prepare corresponding electrolyte solutions, and test the physical property parameters of each electrolyte to verify the accuracy of the electrolyte composition prediction model; and Step 5: If the test results indicate that the electrolyte composition prediction model exceeds a preset error range, a set of supplementary electrolyte formulas is determined according to the generation method of the first set of electrolyte formulas, and steps 2-4 are re-executed based on the supplementary electrolyte formulas until the electrolyte composition prediction model meets the preset error range.

4. The method for generating an electrolyte formula according to claim 2, characterized in that: The data fitting method selects one with the best fitting effect from the following: least squares method, gradient descent method, iterative least squares method, and Lemberg-Maoist algorithm.

5. The method for generating an electrolyte formula according to claim 2, characterized in that: The method for determining the feasible domain of the electrolyte composition comprises: determining the material composition of the electrolyte, and the upper limit and the lower limit of the mass fraction of each material in the material composition according to the target physical property parameter index, and the sum of the mass fractions of each material in the material composition is 1; and The method for determining the first group of electrolyte formulas includes: The first group of electrolyte formulations is generated from the feasible domain of electrolyte components by enumeration or at least one experimental design method.

6. The method for generating an electrolyte formula according to claim 5, characterized in that: The experimental design method includes at least one of the following: classical screening design, response surface design, Taguchi design, orthogonal supersaturated design, filling design, definitive screening design, and mixture design.

7. The method for generating an electrolyte formula according to claim 1, characterized in that: The training of the performance indicator valuation model includes: The electrolyte formulas in the training data set are constructed into an electrolyte formula matrix x according to the order of material composition; Construct a performance indicator value output vector B for each performance indicator in the battery performance indicator α (x), where B α The i-th component B of (x) α (x i ) is the i-th electrolyte formula x i The battery performance index value under the performance index α; Based on the x and the B α (x) constructing a logarithmic marginal likelihood function logP(B) corresponding to one of the performance indicators α of the battery performance indicators α (x)), and initialize the hyperparameter set ξ of the performance indicator estimation model; Gradient descent is performed on the ξ to maximize the logarithmic marginal likelihood function logP(B α (x)), in order to iteratively update the hyperparameter ξ value, and obtain the final logarithmic marginal likelihood function logP(B α (x)) as a priori model of the performance indicator estimation model corresponding to the performance indicator α, The training data set includes: a group of candidate electrolyte formulas and verified electrolyte formulas, and battery performance index values ​​corresponding to each electrolyte formula.

8. The method for generating an electrolyte formula according to claim 7, characterized in that: The log marginal likelihood function is calculated by the following formula: Where I is the unit matrix; σ 2 For B α (x) is the variance of the noise; n is the number of samples of the electrolyte formula in the training data set; P(B α (x)) is the measured cycle performance test index value B under the hyperparameter set ξ α (x) probability of occurrence; μ(x)=[μ(x1),μ(x2),...,μ(x n )], where μ(x i )=ax i + b, where a and b are hyperparameters in the hyperparameter set ξ; k(x i ,x j ) is the i-th electrolyte formula x i With the jth electrolyte formula x j The covariance function, σ f and l are the hyperparameters in the hyperparameter set ξ, and σ f is the standard deviation of the Gaussian kernel function, and l is the characteristic length scale of the Gaussian kernel function.

9. The method for generating an electrolyte formula according to claim 1, characterized in that: The step of obtaining the estimated value of the performance indicator includes: Obtaining an electrolyte formula x* from the set of candidate electrolyte formulas, and calculating a posterior distribution formula of the electrolyte formula; Based on the posterior distribution formula, determine the conditional probability distribution formula of the battery performance indicator value at the performance indicator α; and According to the conditional probability distribution formula, estimate the mean value of the battery performance index value of x* at the performance index α and variance And the mean is the performance indicator estimation value of the electrolyte formula x* at the performance indicator α.

10. The method for generating an electrolyte formula according to claim 1, characterized in that: The weighted sum also includes a weighted value of the production cost of the electrolyte.

11. The method for generating an electrolyte formula according to claim 1, characterized in that: The step of taking the corresponding electrolyte formula whose weighted sum meets the preset standard as the recommended electrolyte formula comprises: The weighted sums are sorted from large to small, and the corresponding electrolyte formula is selected from the obtained sequence from front to back as the recommended electrolyte formula.

12. The method for generating an electrolyte formula according to claim 1, characterized in that: The at least one battery performance indicator includes at least one of the following: a cycle performance indicator, a DCR indicator, a rate performance indicator, a storage performance indicator, and a safety performance indicator.

13. The method for generating an electrolyte formula according to claim 1, characterized in that: The corresponding physical property parameters of the target physical property parameter index include at least one of dielectric constant, viscosity, melting point, boiling point, and ionic conductivity.

14. The method for generating an electrolyte formula according to claim 1, characterized in that: The material composition of the electrolyte formula at least includes a solvent, an electrolyte salt, and an additive, and the additive includes at least one of a film-forming additive, a flame retardant additive, an anti-overcharge additive, and an overcharge protection additive; The electrolyte salt includes at least one of hexafluorophosphate, bisfluorosulfonyl imide salt, nitrate, nitrite, fluoride salt, chloride salt, bromide salt, iodide salt, difluorophosphate, difluorooxalatoborate, dioxalatoborate, tetrafluorooxalatoborate, bisfluorosulfonyl imide salt, and bistrifluoromethanesulfonyl imide salt; and The cations in the electrolyte salt include at least one of the following: Li + 、Na + 、K+、Mg 2+ , Ca 2+ 、Zn 2+ 、Al 3+ .

15. A machine-readable storage medium having instructions stored thereon, the instructions being used to enable a machine to execute: the method for generating an electrolyte formula as described in any one of claims 1 to 14.

16. A processor, characterized in that: Used to run a program, wherein the program, when run, is used to execute: the electrolyte formula generation method as described in any one of claims 1-14.

Images