Material element importance sorting method and system

Abstract

The invention discloses a material element importance sorting method and system, and relates to the cross technical field of material science and machine learning, and the method comprises the steps: obtaining a material current density and energy density combined data set under different element content proportions; expanding the combined data set through the prediction model; screening the expanded data set to obtain an optimal data set; performing multi-collinearity analysis on different elements; and performing correlation coefficient evaluation on the optimal data set based on a multi-collinearity analysis result to obtain a material element importance sorting result. By comprehensively evaluating the energy density and the charge-discharge rate performance, a researcher can more accurately select the positive electrode material, so that the overall performance of the liquid metal battery is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the interdisciplinary field of materials science and machine learning, and more specifically to a material element importance ranking method and system. Background Art

[0002] Liquid Metal Batteries (LMBs) have become a hot topic in energy storage technology research due to their high energy density and fast charge and discharge capabilities. With the rapid development of renewable energy, the demand for efficient and long-lasting battery technology is increasing. The selection of positive electrode materials for liquid metal batteries is crucial because their performance directly affects the overall performance of the battery. Although many studies have focused on the energy density of batteries, performance evaluation of charge and discharge rates has been relatively rare. This has led to a failure to fully consider the multi-dimensional effects of various elements on battery performance when designing positive electrode materials.

[0003] In recent years, with the help of machine learning and data mining technology, researchers have begun to explore the performance evaluation of new battery materials. By integrating existing literature and experimental data, researchers can make a more comprehensive evaluation of the performance of different chemical elements. However, there is currently a lack of a systematic method to simultaneously consider energy density and charge and discharge rate, which limits the optimization of material design.

[0004] Therefore, how to propose a method and system for ranking the importance of material elements, systematically evaluate the importance of different chemical elements in liquid metal battery positive electrode materials, not only focus on the improvement of energy density, but also comprehensively consider the battery's charge and discharge rate performance, so as to provide a more comprehensive basis for material selection is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0005] In view of this, the present invention provides a material element importance ranking method and system, which systematically evaluates the importance of different chemical elements in liquid metal battery positive electrode materials, not only focusing on the improvement of energy density, but also comprehensively considering the battery's charge and discharge rate performance, thereby providing a more comprehensive basis for material selection. In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] A material element importance ranking method, comprising:

[0007] Obtain a combined data set of material current density and energy density at different element content ratios;

[0008] augmenting the combined dataset with predictive models;

[0009] Screen the expanded data set to obtain the optimal data set;

[0010] Multicollinearity analysis was performed on different elements;

[0011] Based on the results of multicollinearity analysis, the correlation coefficient of the optimal data set was evaluated to obtain the importance ranking results of material elements.

[0012] Optionally, the expanding the combined data set includes: combining the n elements by setting different proportion selection ranges, step sizes and current density ranges.

[0013] Optionally, screening the expanded data set to obtain the optimal data set includes:

[0014] For each piece of data in the expanded data set, the partial derivative is calculated using the numerical method to determine whether the partial derivatives on both sides are in the same direction and less than 0.

[0015] Calculate the partial derivatives of all data points under different current density values, and obtain the distribution of partial derivatives on the left and right sides of all data points through the prediction model;

[0016] The prediction effect of the prediction model and the data collection scope are determined based on the partial derivative distribution results.

[0017] Optionally, screening the expanded data set to obtain the optimal data set further includes:

[0018] Refine the data acquisition range to find the optimal current density value;

[0019] The functional relationship between n elements in the original data and the current density and energy density is obtained by using the functional relationship model, and the partial derivative value is obtained according to the functional relationship;

[0020] By calculating the average partial derivative of the prediction model at different current density values, the current density value when the average partial derivative is closest to the partial derivative value obtained by the functional relationship model is the optimal current density value.

[0021] Optionally, the performing multicollinearity analysis on different elements includes:

[0022] Considering the interactions between different elements, multicollinearity analysis was performed on the elements:

[0023]

[0024] Among them, X i is the i-th predictor variable, R i 2 Yes X i The VIF value is the coefficient of determination regressed on all other predictor variables, which indicates the extent to which the variance of the regression coefficient is inflated due to multicollinearity with other predictor variables.

[0025] Optionally, the evaluating the correlation coefficient of the optimal data set based on the multicollinearity analysis result includes: calculating the correlation coefficient of the data set obtained by the prediction model when selecting the optimal current density value.

[0026] Optionally, the correlation coefficient includes a Pearson correlation coefficient and a Spearman correlation coefficient.

[0027] Optionally, a material element importance ranking system, including:

[0028] Acquisition module: used to obtain the combined data set of material current density and energy density at different element content ratios;

[0029] Augmentation module: used to augment the combined dataset through the prediction model;

[0030] Screening module: used to screen the expanded data set to obtain the optimal data set;

[0031] Analysis module: used to perform multicollinearity analysis on different elements;

[0032] Sorting module: used to evaluate the correlation coefficient of the optimal data set based on the multicollinearity analysis results and obtain the ranking results of the importance of material elements.

[0033] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a material element importance ranking method and system, which has the following beneficial effects:

[0034] The present invention proposes a material element importance ranking method, including: obtaining a material current density and energy density combined data set under different element content ratios; expanding the combined data set through a prediction model; screening the expanded data set to obtain an optimal data set; performing multicollinearity analysis on different elements; and evaluating the correlation coefficient of the optimal data set based on the multicollinearity analysis results to obtain a material element importance ranking result. The material element importance ranking method proposed in the present invention will produce significant positive effects and promote the research and application of liquid metal batteries. By comprehensively evaluating the energy density and charge and discharge rate performance, researchers can more accurately select positive electrode materials, thereby significantly improving the overall performance of liquid metal batteries. This method points the way for the development of new battery materials, especially emphasizing the importance of tellurium (Te) and selenium (Se) as key elements, while recommending trace amounts of lead (Pb) and tin (Sn) to improve the charge and discharge rate. This systematic evaluation will promote the development of liquid metal batteries towards higher energy density and faster charge and discharge rates to meet the growing market demand. As battery performance improves, the commercial application of related technologies will also accelerate, especially in the fields of electric vehicles and renewable energy storage. This will not only help reduce production costs, but also enhance the market competitiveness of products and further promote the advancement and development of sustainable energy technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0036] Figure 1 A schematic flow chart of a material element importance ranking method provided by the present invention.

[0037] Figure 2 A schematic diagram of the partial derivative of energy density (E) with respect to current density (J) provided by the present invention.

[0038] Figure 3 This is a distribution diagram of the partial derivative values ​​on the left and right sides of all the data provided by the present invention under different models.

[0039] Figure 4 This is a graph of average partial derivative numerical results of MLP and GPR provided by the present invention.

[0040] Figure 5 This is a graph of multicollinearity results for eight elements provided by the present invention. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The embodiment of the present invention discloses a method for ranking the importance of material elements, such as Figure 1 As shown, including:

[0043] Obtain a combined data set of material current density and energy density at different element content ratios;

[0044] augmenting the combined dataset with predictive models;

[0045] Screen the expanded data set to obtain the optimal data set;

[0046] Multicollinearity analysis was performed on different elements;

[0047] Based on the results of multicollinearity analysis, the correlation coefficient of the optimal data set was evaluated to obtain the importance ranking results of material elements.

[0048] In a specific implementation, the expanded combined data set includes:

[0049] By combining n elements with different proportions, the proportion selection range is 0 to 1.0, the step size is 0.1, and the different current density range is 0 to 1000, the step size is 100.

[0050] In a specific implementation manner, screening the expanded data set to obtain the optimal data set includes:

[0051] For each piece of data in the expanded data set, the partial derivative is calculated using the numerical method to determine whether the partial derivatives on both sides are in the same direction and less than 0.

[0052] Calculate the partial derivatives of all data points under different current density values, and obtain the distribution of partial derivatives on the left and right sides of all data points predicted by the Gaussian process regression model and the multi-layer perceptron model;

[0053] Determine the prediction effect of the model and the reasonable data range based on the distribution results.

[0054] In a specific implementation manner, screening the expanded data set to obtain the optimal data set includes:

[0055] Refine the reasonable data range and find the optimal current density value;

[0056] The linear relationship between n elements in the original data and current density and energy density was obtained by using a linear regression model:

[0057] F(Elements, J) = 1.785Se + 1.007Sb + 0.082Cu - 0.753Bi - 2.795Pb - 2.094Sn + 2.791Te - 0.023Zn - 0.069J + 2.579, where J represents the current density and its coefficient is used as the benchmark;

[0058] The coefficient corresponding to the current density J is -0.069, which indicates the partial derivative value obtained by the linear regression model. By calculating the average partial derivatives of the Gaussian process regression model and the multilayer perceptron model at different current density values, the current density value when the average partial derivative is closest to the partial derivative value obtained by the linear regression model is the optimal current density value, which is used for subsequent calculations.

[0059] In a specific implementation, it also includes: calculating the average partial derivative value of all data obtained by the Gaussian process regression model and the multi-layer perceptron model at different current density values When the current density is 10 -3 When , the average partial derivative obtained is closest to the partial derivative value obtained by the linear regression model.

[0060] Furthermore, the multicollinearity analysis of different elements includes:

[0061] Considering the interactions between different elements, multicollinearity analysis was performed on the elements:

[0062]

[0063] Among them, X i is the i-th predictor variable, R i 2 Yes X i The VIF value is the coefficient of determination regressed on all other predictor variables, which indicates the extent to which the variance of the regression coefficient is inflated due to multicollinearity with other predictor variables.

[0064] In a specific embodiment, the correlation coefficient evaluation of the optimal data set based on the multicollinearity analysis results includes: selecting a current density value of 10 -3 The correlation coefficient is calculated for the data set obtained by the multilayer perceptron model.

[0065] Furthermore, the correlation coefficient includes the Pearson correlation coefficient and the Spearman correlation coefficient.

[0066] In a specific embodiment, a material element importance ranking system, such as Figure 2As shown, including:

[0067] Acquisition module: used to obtain the combined data set of material current density and energy density at different element content ratios;

[0068] Augmentation module: used to augment the combined dataset through the prediction model;

[0069] Screening module: used to screen the expanded data set to obtain the optimal data set;

[0070] Analysis module: used to perform multicollinearity analysis on different elements;

[0071] Sorting module: used to evaluate the correlation coefficient of the optimal data set based on the multicollinearity analysis results and obtain the ranking results of the importance of material elements.

[0072] In a specific embodiment, a material element importance ranking method systematically evaluates the importance of different chemical elements in liquid metal battery positive electrode materials. This method not only focuses on the improvement of energy density, but also comprehensively considers the battery's charge and discharge rate performance, thereby providing a more comprehensive basis for material selection, including:

[0073] Develop a comprehensive evaluation process: Establish a new workflow so that researchers can simultaneously evaluate the energy density and rate performance of materials to provide scientific guidance for the design of cathode materials.

[0074] Recommend key elements based on rate performance: Based on energy density and charge / discharge rate, tellurium (Te) and selenium (Se) are recommended as potential cathode materials. If the main goal of the researcher is to increase the charge / discharge rate, it is recommended to add trace amounts of lead (Pb) and tin (Sn) to the material.

[0075] The specific steps are as follows:

[0076] Step 1: Collect data from literature and data obtained through experiments, as shown in Table 1, a total of 63 data, including the content ratio of 8 different elements, current density and energy density of the material.

[0077] Table 1

[0078]

[0079]

[0080] Step 2: Considering the limited amount of data, two models, Gaussian process regression (GPR) and multi-layer perceptron (MLP), were used to expand the data set. By combining 8 elements with different proportions (the proportion selection range is 0 to 1.0, the step size is 0.1) and different current densities (the range is 0 to 1000, the step size is 100), 213928 On this basis, the two models of GPR and MLP are used to predict these data respectively. Then, these data sets need to be judged to determine whether these data meet the requirements. The judgment process is as follows:

[0081] (1) For each data point, use the numerical method to find its partial derivative (i.e., rate performance) to determine whether the data satisfies the requirement that the partial derivatives on both sides are in the same direction and less than 0 (the partial derivative represents the rate, i.e., the energy density should decrease with the increase of current density, so its partial derivative should be less than 0. If it is greater than 0, it does not conform to the theory and should be discarded). The schematic diagram for finding the partial derivative is as follows: Figure 2 As shown, it is a schematic diagram of the partial derivative of energy density (E) relative to current density (J). The green dotted line and the orange dotted line are the slopes of the tangent lines corresponding to the curve; (For example, 10 -2 , 10 -3 etc.) represent the current density values ​​used to calculate the partial derivatives.

[0082] (2) Using the above method, we can find the difference The partial derivative values ​​of all data points under the value are obtained to obtain the distribution of partial derivative values ​​on both sides of all data points predicted by MLP and GPR, such as Figure 3 As shown in , the distribution of partial derivatives on both sides of all data under different models. and (yellow area), it is consistent with expectations; other cases are contrary to theory and will not be used in subsequent correlation calculations. Figure 3 It can be clearly found that the data obtained by MLP is more reasonable than that obtained by GPR, so the data obtained by MLP will be used for evaluation in the future. Value(10 -2 ~10 -6 ), it can be found in MLP that -2 ~10 -4 The data are all reasonable when these three values ​​are met.

[0083] Although In 10 -2 ~10 -4 The data sets obtained when the experiment is performed are consistent with the theory, but only one of the data sets is used in the experiment. The linear relationship between the eight elements of the original data and the current density and energy density was obtained by using the linear regression (LR) model, as shown in the following formula. J represents the current density, and its coefficient will be used as a benchmark. Then the coefficient of J will be used as a benchmark to evaluate the best

[0084] F(Elements,J)=1.785Se+1.007Sb+0.082Cu-0.753Bi-2.795Pb-2.094Sn+2.791Te-0.023Zn-0.069J+2.579;

[0085] From the above formula, it can be found that the coefficient corresponding to the current density J is -0.069, which represents the partial derivative value obtained by the LR model. By calculating other models (GPR, MLP) in different The average partial derivative under the value of When the partial derivative value is the closest to the partial derivative value obtained by LR, then This is the best result, and subsequent experimental data will be conducted based on this. Figure 4 It can be found that when When , the average partial derivatives obtained by both MLP and GPR are closest to LR (benchmark), so we choose The data set obtained by MLP is used for the subsequent correlation analysis.

[0086] Step 3: In order to find out the importance ranking of element proportions for energy density and its partial derivatives, two different correlation coefficients were used for evaluation: Pearson correlation coefficient (P) and Spearman correlation coefficient (S). Considering that there may be interactions between different elements, which may affect the final importance ranking of elements for energy density or its rate performance, it is also necessary to perform multicollinearity (VIF) on these elements:

[0087] Where X i is the i-th predictor variable, R i 2 Yes X i The coefficient of determination for the regression of all other predictor variables. The VIF value indicates the degree to which the variance of the regression coefficients is inflated due to multicollinearity with other predictor variables. A VIF value of 1 indicates no multicollinearity, values ​​between 1 and 5 indicate moderate multicollinearity, and values ​​greater than 5 indicate high multicollinearity that may be problematic. In particular, a VIF greater than 10 is generally considered a sign of severe multicollinearity, which may require corrective action such as removing or combining variables.

[0088] The VIF values ​​of the eight elements are as follows: Figure 5 As shown, it can be found that the variance inflation factors of all elements are between 1 and 3, and there is moderate multicollinearity between the elements, but it is within an acceptable range; however, the variance inflation factors of Cu and Se are larger than those of other elements, which means that they have greater multicollinearity with other elements, which may make the calculation results of the subsequent regression model for these two elements less reliable.

[0089] Step 4: Use the MLP model with a partial derivative current density of 10 -3 The data set obtained when calculating two kinds of correlation coefficients, the specific values ​​and their corresponding ranking results are shown in Table 2, which are the energy density (E) and its partial derivatives of different elements. The corresponding Pearson coefficient (P) and Spearman coefficient (S) and the importance ranking of their elements. The elements in bold font indicate that they have a high degree of multicollinearity.

[0090] Table 2

[0091]

[0092] Finally, the present invention obtained a comprehensive ranking result of eight elements, namely, Te, Se, Sb, Cu, Zn, Bi, Sn and Pb, and proposed that tellurium (Te) and selenium (Se) are elements that are expected to improve the energy density of battery materials.

[0093] The material element importance ranking method proposed in the present invention will produce significant positive effects and promote the research and application of liquid metal batteries. By comprehensively evaluating the energy density and charge and discharge rate performance, researchers can more accurately select positive electrode materials, thereby significantly improving the overall performance of liquid metal batteries. This method points the way for the development of new battery materials, especially emphasizing the importance of tellurium (Te) and selenium (Se) as key elements, while recommending trace amounts of lead (Pb) and tin (Sn) to improve the charge and discharge rate. This systematic evaluation will promote the development of liquid metal batteries towards higher energy density and faster charge and discharge rates to meet the growing market demand. With the improvement of battery performance, the commercial application of related technologies will also be accelerated, especially in the fields of electric vehicles and renewable energy storage, which will not only help reduce production costs, but also enhance the market competitiveness of products, and further promote the progress and development of sustainable energy technologies.

[0094] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0095] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A material element importance ranking method, characterized in that: include: Obtain a combined data set of material current density and energy density at different element content ratios; augmenting the combined dataset with predictive models; Screen the expanded data set to obtain the optimal data set; Multicollinearity analysis was performed on different elements; Based on the results of multicollinearity analysis, the correlation coefficient of the optimal data set was evaluated to obtain the importance ranking results of material elements.

2. A material element importance ranking method according to claim 1, characterized in that: The expanding the combined data set includes: combining n elements by setting different proportion selection ranges, step sizes and current density ranges.

3. A material element importance ranking method according to claim 1, characterized in that: The screening of the expanded data set to obtain the optimal data set includes: For each piece of data in the expanded data set, the partial derivative is calculated using the numerical method to determine whether the partial derivatives on both sides are in the same direction and less than 0. Calculate the partial derivatives of all data points under different current density values, and obtain the distribution of partial derivatives on the left and right sides of all data points through the prediction model; The prediction effect of the prediction model and the data collection scope are determined based on the partial derivative distribution results.

4. A material element importance ranking method according to claim 1, characterized in that: The screening of the expanded data set to obtain the optimal data set further includes: Refine the data acquisition range to find the optimal current density value; The functional relationship between n elements in the original data and the current density and energy density is obtained by using the functional relationship model, and the partial derivative value is obtained according to the functional relationship; By calculating the average partial derivative of the prediction model at different current density values, the current density value when the average partial derivative is closest to the partial derivative value obtained by the functional relationship model is the optimal current density value.

5. A material element importance ranking method according to claim 1, characterized in that: The multicollinearity analysis of different elements includes: Considering the interactions between different elements, multicollinearity analysis was performed on the elements: Among them, X i is the i-th predictor variable, R i 2 Yes X i The VIF value is the coefficient of determination regressed on all other predictor variables, which indicates the extent to which the variance of the regression coefficient is inflated due to multicollinearity with other predictor variables.

6. A material element importance ranking method according to claim 1, characterized in that: The evaluating the correlation coefficient of the optimal data set based on the multicollinearity analysis result includes: calculating the correlation coefficient of the data set obtained by the prediction model when selecting the optimal current density value.

7. A material element importance ranking method according to claim 6, characterized in that: The correlation coefficients include Pearson's correlation coefficient and Spearman's correlation coefficient.

8. A material element importance ranking system, characterized in that: include: Acquisition module: used to obtain the combined data set of material current density and energy density at different element content ratios; Augmentation module: used to augment the combined dataset through the prediction model; Screening module: used to screen the expanded data set to obtain the optimal data set; Analysis module: used to perform multicollinearity analysis on different elements; Sorting module: used to evaluate the correlation coefficient of the optimal data set based on the multicollinearity analysis results and obtain the ranking results of the importance of material elements.

Images