Sodium-ion battery positive electrode material screening method and equipment based on feature engineering
Through feature engineering and Bayesian neural network models, high-performance sodium-ion battery positive electrode materials were screened out, solving the problems of low screening efficiency and inaccurate prediction in existing technologies, and achieving efficient material screening and energy density prediction.
Patent Information
- Application Number
- CN202510798808.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-16
- Publication Date
- 2025-09-23
AI Technical Summary
Existing technologies make it difficult to quickly screen high-performance sodium-ion battery positive electrode materials, and machine learning has shortcomings in feature selection and model uncertainty quantification, resulting in low accuracy and reliability of prediction results.
A feature engineering-based method was used to obtain parameter data of sodium-ion battery positive electrode materials, and significant characteristic parameters were screened out using feature calculation, correlation analysis, and independence evaluation. A Bayesian neural network model was constructed for energy density prediction, and the Monte Carlo discarding method was used to quantify uncertainty.
The accuracy and interpretability of energy density predictions for sodium-ion battery cathode materials have been improved, the characterization capability of predicted values has been enhanced, R&D costs have been reduced, and screening efficiency has been improved.
Smart Images
Figure CN120690353A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of sodium ion batteries, and in particular to a method and device for screening positive electrode materials for sodium ion batteries based on feature engineering. Background Art
[0002] Sodium-ion batteries have attracted much attention due to their abundant resources, low cost and excellent low-temperature performance. They are considered to be a powerful alternative to lithium-ion batteries due to their abundant resources and low cost. However, the energy density of their positive electrode materials is generally low, which limits their widespread application. In some cases, the research on sodium-ion batteries has mainly focused on polyanion compounds. These materials have a three-dimensional framework structure and can achieve rapid sodium ion insertion / extraction, but have problems such as insufficient energy density and dependence on toxic vanadium elements. Therefore, finding suitable sodium-ion battery positive electrode materials is an important issue to be solved in the development of sodium-ion battery technology. The research and development of sodium-ion battery positive electrode materials relies on complex experimental and characterization methods, which is not only time-consuming and labor-intensive, but also difficult to quickly screen out high-performance materials.
[0003] In some cases, during the screening process of sodium-ion battery positive electrode materials, material properties are often predicted through automatic machine learning, which can only optimize a single performance and cannot take into account the horizontal connections and influences between various target performances. In addition, feature selection depends on preset parameters and does not fully consider the correlation and independence between features, resulting in limited model prediction accuracy. The accuracy of the prediction results is also difficult to quantify and evaluate, resulting in low prediction reliability. Summary of the Invention
[0004] The purpose of this application is to provide a method and equipment for screening sodium-ion battery positive electrode materials based on feature engineering, which can improve the accuracy, interpretability and reliability of the energy density prediction of sodium-ion battery positive electrode materials and enhance the characterization ability of the energy density prediction value.
[0005] To achieve the above objectives, this application provides the following solutions:
[0006] In a first aspect, the present application provides a method for screening positive electrode materials for sodium ion batteries based on feature engineering, the method comprising: obtaining parameter data of positive electrode materials for sodium ion batteries; the parameter data comprising: chemical composition, crystal structure and electrochemical properties; using feature engineering to screen out characteristic parameters affecting the energy density of positive electrode materials for sodium ion batteries in the parameter data to obtain a characteristic parameter set; the feature engineering comprising: feature calculation, correlation analysis, independence assessment and feature selection; inputting the characteristic parameter set into an energy density prediction model to predict the energy density of each positive electrode material to obtain an energy density prediction value of each positive electrode material; the energy density prediction model is constructed based on a Bayesian neural network; based on the energy density prediction value of each positive electrode material, a target positive electrode material is screened; the target positive electrode material is a positive electrode material whose energy density prediction value exceeds an energy density threshold.
[0007] In a second aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-mentioned method for screening sodium-ion battery positive electrode materials based on feature engineering.
[0008] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0009] The present application obtains parameter data of sodium-ion battery positive electrode materials; uses feature engineering to screen out characteristic parameters that affect the energy density of sodium-ion battery positive electrode materials in the parameter data to obtain a characteristic parameter set; uses feature engineering to systematically analyze the chemical composition, crystal structure and physicochemical properties of the materials in the parameter data; screens out characteristic parameters that significantly affect energy density, thereby providing a high-quality data basis for an energy density prediction model; then inputs the high-quality characteristic parameter set into the energy density prediction model to predict the energy density of each positive electrode material, thereby obtaining an energy density prediction value of each positive electrode material; based on the energy density prediction value of each positive electrode material, the target positive electrode material is screened, thereby improving the accuracy, interpretability and reliability of the energy density prediction of the sodium-ion battery positive electrode material and enhancing the characterization capability of the energy density prediction value. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0011] Figure 1A schematic flow chart of a method for screening positive electrode materials for sodium ion batteries based on feature engineering is provided in an embodiment of the present application.
[0012] Figure 2 A flowchart of the feature forward selection method provided in an embodiment of the present application.
[0013] Figure 3 Schematic diagram of the energy density prediction process of the energy density prediction model provided in the embodiment of the present application.
[0014] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0015] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0016] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0017] Machine learning is widely used in the design and optimization of high-performance materials for sodium-ion battery cathode materials. By building data-driven models, machine learning can rapidly identify materials with specific properties, significantly improving R&D efficiency. For example, machine learning has proven to be an effective tool in areas such as alloy design, catalyst screening, and battery electrolyte optimization. However, machine learning still has shortcomings in feature selection and quantifying model uncertainty.
[0018] The Importance of Feature Engineering in Materials Prediction: Feature engineering is a key technique in machine learning. By selecting and optimizing the characteristic parameters of the input model, it can significantly improve the model's prediction accuracy. In materials science, feature engineering involves systematic analysis of a material's chemical composition, crystal structure, and physicochemical properties. Through methods such as correlation analysis, random sampling, and independence assessment, characteristic parameters that significantly influence target properties (such as energy density) can be screened, providing a high-quality data foundation for model training.
[0019] The following is an explanation of the terms used in this application:
[0020] Bootstrap random sampling, also known as the bootstrap method, is a statistical resampling technique widely used in data analysis and machine learning. It generates multiple new sample sets of the same size as the original dataset by randomly selecting samples with replacement from the original dataset. This is used to estimate the distribution of a statistic or evaluate model performance.
[0021] Feature Engineering: Feature engineering is an important step in machine learning, which aims to improve the predictive performance of the model by selecting, transforming and optimizing the feature parameters of the input model.
[0022] Bayesian Neural Networks (BNNs): Bayesian Neural Networks are a type of neural network model that incorporates Bayesian statistical methods and can quantify the uncertainty of model parameters. Unlike traditional deterministic neural networks, BNNs treat model parameters as random variables and represent parameter uncertainty through probability distributions.
[0023] High-entropy materials: These materials are composed of multiple main elements in near-equiatomic ratios and typically possess complex microstructures and excellent physical and chemical properties. In sodium-ion battery cathode materials, high-entropy materials, by introducing multiple metal elements (such as Mn, V, Ti, and Zr), can increase the material's entropy, thereby improving its energy density and cycle stability.
[0024] Example 1, as Figure 1 As shown, this embodiment provides a method for screening positive electrode materials for sodium ion batteries based on feature engineering. The method for screening positive electrode materials for sodium ion batteries based on feature engineering includes the following steps.
[0025] S1. Obtain parameter data of the sodium ion battery positive electrode material; the parameter data includes: chemical composition, crystal structure and electrochemical performance.
[0026] Alternatively, parameter data of Na-ion battery cathode materials are collected from published literature.
[0027] S2. Use feature engineering to screen out characteristic parameters that affect the energy density of the sodium-ion battery positive electrode material in the parameter data to obtain a characteristic parameter set; the feature engineering includes: feature calculation, correlation analysis, independence assessment and feature selection.
[0028] Step S2 specifically includes the following steps.
[0029] S21. Calculate characteristic parameters of the parameter data to obtain an initial characteristic parameter set, which specifically includes the following steps.
[0030] S211. Clean the parameter data using an abnormal data removal method to obtain cleaned parameter data; the abnormal data is an abnormal value or duplicate data.
[0031] In practical applications, abnormal data removal can ensure the accuracy and consistency of data.
[0032] S212. Perform characteristic calculation on the parameter data after cleaning and extract characteristic parameters, which include: relative molecular mass, average ion radius, electronegativity and entropy characteristic parameters.
[0033] S213. Normalize the feature parameters to obtain an initial feature parameter set.
[0034] In practical applications, normalized feature parameters can ensure the comparability between different features, and the obtained preprocessed high-quality data set (initial feature parameter set) can provide a basis for subsequent feature engineering and model training.
[0035] S22. Use the correlation analysis method to perform correlation analysis on each characteristic parameter in the initial characteristic parameter set and the energy density, screen out the target feature, and obtain the correlation coefficient mean and correlation coefficient standard deviation of the target feature; the target feature is the feature whose correlation coefficient exceeds the threshold.
[0036] Optionally, the sample sampling method used in the correlation analysis process is the Bootstrap random sampling method.
[0037] In practical application, the Bootstrap random sampling method selects 50 samples each time and repeats the sampling 300 times to obtain the mean and standard deviation of the correlation coefficient between each characteristic parameter and energy density.
[0038] S23. Use the Gini coefficient to evaluate the independence of the target feature and obtain the independence coefficient of the target feature.
[0039] In practical applications, redundant features can be removed through independence evaluation.
[0040] S24. Use the correlation coefficient mean, correlation coefficient standard deviation and independence coefficient of the target features to calculate the importance of each target feature, and sort the target features from high to low according to importance to obtain the sorting result.
[0041] Furthermore, the calculation formula for the importance of the target feature is as follows.
[0042] I=ρ·(1-σ)·α·(1-σ)·(1+G).
[0043] Among them, i is the importance of the target feature; ρ is the mean of the correlation coefficient; σ is the standard deviation of the correlation coefficient; α is the independence coefficient; G is the Gini coefficient.
[0044] S25. Figure 2 As shown in FIG, according to the sorting results, the feature selection method is used to perform error analysis on the target features in turn, and the target features are screened according to the error changes to obtain a feature parameter set.
[0045] Optionally, the feature selection method is a forward selection method.
[0046] S3. Figure 3 As shown, the characteristic parameter set is input into the energy density prediction model to predict the energy density of each cathode material, and the energy density prediction value of each cathode material is obtained; the energy density prediction model is constructed based on a Bayesian neural network.
[0047] Furthermore, the energy density prediction model includes an input layer, a hidden layer and an output layer connected in sequence; the input layer is used to receive a set of feature parameters; the hidden layer is used to extract features of the feature parameters using an attention mechanism; the output layer is used to predict the energy density prediction value of each positive electrode material, and quantify the uncertainty of the energy density prediction value of each positive electrode material based on the Monte Carlo rejection method.
[0048] In practical applications, the Monte Carlo dropout method was introduced to quantify the uncertainty of model predictions, providing a more reliable basis for material screening. This method can effectively assess the reliability of model predictions, especially in the case of small sample data, significantly improving the practicality and credibility of the model.
[0049] Optionally, the energy density prediction model has high accuracy and reliability.
[0050] Optionally, the energy density prediction model training and optimization process is as follows.
[0051] 1) Dataset division: Divide the historical feature parameter set into a training set and a test set.
[0052] 2) Grid Search Cross-Validation: Define the search range of hyperparameters, evaluate model performance through cross-validation, and select the best hyperparameter combination.
[0053] 3) Model training: Use the training set to train the model and optimize the model parameters.
[0054] 4) Model validation: Use the test set to verify the model performance and evaluate the model's prediction accuracy and generalization ability.
[0055] 5) Model optimization: Adjust model parameters based on the verification results to further improve the performance of the model.
[0056] 6) Output: The optimized Bayesian neural network model has higher prediction accuracy and generalization ability.
[0057] S4. Based on the predicted energy density values of the positive electrode materials, a target positive electrode material is screened; the target positive electrode material is a positive electrode material whose predicted energy density exceeds an energy density threshold.
[0058] In actual application, the screening process of the target positive electrode material specifically includes: screening out the positive electrode materials that exceed the energy density threshold from the sorted positive electrode materials as the target positive electrode materials, and then forming a list of high energy density candidate materials.
[0059] Furthermore, after step S4, the following steps are also included.
[0060] S5. Synthesize the target positive electrode material.
[0061] Optionally, the method for synthesizing the target positive electrode material is a sol-gel method.
[0062] S6. Verify the energy density of the target cathode material to obtain the actual energy density of each target cathode material.
[0063] In actual application, the electrochemical performance of the synthesized target positive electrode material is tested to verify the energy density, cycle stability and rate performance of the material. The predicted results are compared with the experimental data to verify the accuracy and reliability of the model, providing reliable guidance for the research and development of positive electrode materials for sodium-ion batteries.
[0064] S7. Based on the difference between the predicted energy density value and the actual energy density of the target positive electrode material, optimize the parameters of the energy density prediction model and update the energy density prediction model.
[0065] The technical effects of this application are as follows:
[0066] This application improves the accuracy and interpretability of energy density prediction of positive electrode materials for sodium-ion batteries by introducing a systematic feature engineering method and optimizing feature parameters through correlation analysis and independence evaluation, thereby enhancing the characterization ability of energy density prediction values. At the same time, the Bayesian neural network is combined to quantify prediction uncertainty and enhance the reliability and generalization ability of the model. By constructing a model based on Bayesian neural networks, combining feature engineering and uncertainty quantification, performance optimization in complex component space is achieved. The prediction results are verified by experiments to ensure the practicality of the model. This application can quickly identify positive electrode materials with high energy density potential, which not only improves R&D efficiency and reduces experimental costs, but also provides new theoretical and technical support for the optimal design of positive electrode materials for sodium-ion batteries, and is expected to promote the rapid development and commercial application of sodium-ion battery technology.
[0067] Example 2: This application also provides a computer device, which can be a server or a terminal, and its internal structure diagram can be as follows: Figure 4 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above-mentioned methods are implemented.
[0068] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
[0069] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0070] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0071] This document uses specific examples to illustrate the principles and implementation methods of this application. The description of the above examples is only intended to help understand the method and core concept of this application. At the same time, for those skilled in the art, based on the concept of this application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for screening positive electrode materials for sodium ion batteries based on feature engineering, characterized in that: The method for screening positive electrode materials for sodium ion batteries based on feature engineering includes: Obtain parameter data of a sodium ion battery cathode material; the parameter data includes: chemical composition, crystal structure and electrochemical performance; Feature engineering is used to screen out characteristic parameters that affect the energy density of the sodium ion battery cathode material from parameter data to obtain a characteristic parameter set; the feature engineering includes: feature calculation, correlation analysis, independence assessment and feature selection; Inputting the characteristic parameter set into an energy density prediction model to predict the energy density of each cathode material to obtain a predicted energy density value of each cathode material; the energy density prediction model is constructed based on a Bayesian neural network; Based on the predicted energy density values of the positive electrode materials, a target positive electrode material is screened and obtained; the target positive electrode material is a positive electrode material whose predicted energy density value exceeds an energy density threshold.
2. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 1, characterized in that: Based on the predicted energy density values of each cathode material, after screening the target cathode material, the following steps are also required: synthesizing the target cathode material; Verify the energy density of the target cathode material to obtain the actual energy density of each target cathode material; Based on the difference between the predicted energy density value and the actual energy density of the target cathode material, the parameters of the energy density prediction model are optimized and the energy density prediction model is updated.
3. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 2, characterized in that: The method for synthesizing the target positive electrode material is a sol-gel method.
4. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 1, characterized in that: Feature engineering is used to screen out the characteristic parameters that affect the energy density of the sodium-ion battery cathode material in the parameter data, and obtain a characteristic parameter set, which specifically includes: Calculate characteristic parameters of parameter data to obtain an initial characteristic parameter set; Using the correlation analysis method, the correlation analysis is performed on each characteristic parameter in the initial characteristic parameter set and the energy density, the target feature is screened out, and the correlation coefficient mean and correlation coefficient standard deviation of the target feature are obtained; the target feature is a feature whose correlation coefficient exceeds a threshold; Use the Gini coefficient to evaluate the independence of the target feature and obtain the independence coefficient of the target feature; The importance of each target feature is calculated using the correlation coefficient mean, correlation coefficient standard deviation and independence coefficient of the target feature, and the target features are ranked from high to low according to their importance to obtain the ranking results; According to the sorting results, the feature selection method is used to perform error analysis on the target features in turn, and the target features are screened according to the error changes to obtain the feature parameter set.
5. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 4, characterized in that: Calculate the characteristic parameters of the parameter data to obtain the initial characteristic parameter set, which specifically includes: Cleaning the parameter data by using an abnormal data removal method to obtain cleaned parameter data; the abnormal data is an abnormal value or duplicate data; Performing feature calculation on the parameter data after cleaning to extract feature parameters, wherein the feature parameters include: relative molecular mass, average ion radius, electronegativity and entropy value feature parameters; Normalize the feature parameters to obtain the initial feature parameter set.
6. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 4, characterized in that: The calculation formula for the importance of the target feature is as follows: I=ρ·(1-σ)·α·(1-σ)·(1+G); Among them, I is the importance of the target feature; ρ is the mean of the correlation coefficient; σ is the standard deviation of the correlation coefficient; α is the independence coefficient; and G is the Gini coefficient.
7. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 4, characterized in that: The feature selection method is a forward selection method.
8. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 1, characterized in that: The energy density prediction model includes an input layer, a hidden layer and an output layer connected in sequence; The input layer is used to receive a feature parameter set; The hidden layer is used to extract features from feature parameters using an attention mechanism; The output layer is used to predict the energy density prediction value of each positive electrode material, and quantify the uncertainty of the energy density prediction value of each positive electrode material based on the Monte Carlo rejection method.
9. The method for screening positive electrode materials for sodium ion batteries based on feature engineering according to claim 1, characterized in that: The sample sampling method used in the analysis process of the correlation analysis method is the Bootstrap random sampling method.
10. A computer device comprising: A memory, a processor, and a computer program stored in the memory and runnable on the processor, characterized in that the processor executes the computer program to implement the sodium ion battery positive electrode material screening method based on feature engineering according to any one of claims 1 to 9.