Battery material combination method and equipment and computer storage medium

Abstract

The invention discloses a battery material combination method and device and a computer storage medium, and relates to the technical field of new energy materials, and the method comprises the steps: generating one or more groups of candidate material combinations based on battery material data; extracting attribute features of the candidate material combination, and converting the attribute features into feature vectors; and inputting the feature vector into a pre-trained prediction model to obtain a performance prediction result of the candidate material combination. According to the method, the candidate material combination is generated, and the performance of the candidate material combination is predicted by using the prediction model, so that efficient screening and safety performance prediction of the battery material combination are realized, the research and development efficiency of the battery material is remarkably improved, and meanwhile, the selection range of the battery material combination is also expanded.

Description

Technical Field

[0001] The present application relates to the technical field of new energy materials, and in particular to battery material combination methods, equipment and computer storage media. Background Art

[0002] As an emerging energy storage technology, the stability of sodium batteries in practical applications is a key factor restricting their widespread application.

[0003] In the design of sodium batteries, layered oxides are usually used as positive electrode materials, which have higher energy density. However, during the electrochemical cycle, layered oxide positive electrode materials are susceptible to irreversible phase changes caused by the relative sliding of transition metal layers, resulting in unstable battery cycles. Negative electrode materials generally use hard carbon materials with higher sodium storage capacity and good cycle stability. However, hard carbon materials may produce large volume expansion during the charge and discharge process, thus affecting the performance and service life of sodium batteries.

[0004] Therefore, the material combination used in sodium batteries will directly affect its performance and safety. The current evaluation of battery material combinations mainly relies on laboratory tests and empirical design principles. This approach requires long-term experimental studies on various material combinations to evaluate the stability and safety of various material combinations during battery cycling. However, this method requires a lot of time and resources, and cannot cover all possible material combinations, which limits the range of battery material choices. Summary of the invention

[0005] The main purpose of this application is to provide a battery material combination method, device and computer storage medium, aiming to solve the technical problems of low efficiency in battery material combination evaluation and limited coverage of combinations, which lead to limited battery material selection.

[0006] To achieve the above-mentioned purpose, an embodiment of the present application provides a battery material combination method, the battery material combination method comprising: Based on the battery material data, generate one or more groups of candidate material combinations; Extracting property features of the candidate material combination and converting the property features into feature vectors; The feature vector is input into a pre-trained prediction model to obtain a performance prediction result of the candidate material combination.

[0007] In one embodiment, the step of generating one or more groups of candidate material combinations based on the battery material data includes: determining an interaction rule of a material combination, wherein the interaction rule is used to guide pairing of the material combination; Based on the interaction rules, obtaining a candidate list of the material combinations; According to a preset material property standard, the candidate list of the material combination is screened to determine the candidate material combination.

[0008] In one embodiment, the step of extracting the property features of the candidate material combination and converting the property features into a feature vector comprises: Obtaining material properties of each battery material in the candidate material combination and property values ​​corresponding to the material properties; The property value of each battery material is mapped into a set of feature vectors, and each property value corresponds to an element in the feature vector.

[0009] In one embodiment, before the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the step includes: Acquiring safety performance data corresponding to material properties of battery materials, the safety performance data including impact test results of the battery materials under external impact conditions; Constructing a battery material data set according to the material properties and the safety performance data; Based on the battery material data set, training a prediction model; The trained prediction model is used to predict the safety performance of the candidate material combination, and the performance prediction result of the candidate material combination is output.

[0010] In one embodiment, before the step of training the prediction model based on the battery material data set, the step includes: Dividing the battery material data set into subsets of equal size, each subset containing the same amount of data; In each iterative training, a different subset is selected as a validation set, and the remaining subsets are selected as training sets. The prediction model is iteratively trained, and the prediction error is calculated using the validation set.

[0011] In one embodiment, after the step of iteratively training the prediction model and calculating the prediction error using the validation set, the method further includes: After each iterative training, based on the prediction error, the hyperparameter configuration of the prediction model is updated; Using the prediction model after updating the hyperparameter configuration, repeatedly execute the steps of selecting a different subset as a validation set and the remaining subsets as training sets in each iterative training, iteratively training the training sets, and calculating the prediction error using the validation set until the prediction error is reduced to a preset threshold.

[0012] In one embodiment, after the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the following steps are included: Determining a confidence interval for the safety performance of the battery material based on historical safety performance data of the battery material; comparing the performance prediction result of the candidate material combination with the confidence interval; According to the comparison result, the safety level of the candidate material combination is determined.

[0013] In one embodiment, after the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the step further includes: According to the performance prediction results, determine the material properties that need to be optimized, and regenerate a new candidate material combination; Inputting the feature vector of the new candidate material combination into the prediction model to obtain a new performance prediction result; Based on the new performance prediction result, the new candidate material combination is iteratively adjusted until a preset performance target is reached or the adjustment of the candidate material combination is completed.

[0014] An embodiment of the present application also provides a battery material combination device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the battery material combination method as described above.

[0015] An embodiment of the present application also provides a computer storage medium, which is a computer-readable storage medium. A computer program is stored on the computer storage medium. When the computer program is executed by a processor, the steps of the battery material combination method described above are implemented.

[0016] The embodiment of the present application discloses a battery material combination method, which generates one or more candidate material combinations based on battery material data; extracts the attribute characteristics of the candidate material combination and converts the attribute characteristics into a feature vector; and inputs the feature vector into a pre-trained prediction model to obtain a performance prediction result of the candidate material combination. The present application generates candidate material combinations and uses a prediction model to predict the performance of the candidate material combinations, thereby achieving efficient screening and safety performance prediction of battery material combinations, significantly improving the research and development efficiency of battery materials, and also expanding the selection range of battery material combinations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic flow chart of a first embodiment of a battery material combination method according to an embodiment of the present application; Figure 2 This is a schematic flow chart of a second embodiment of a battery material combination method according to an embodiment of the present application; Figure 3 This is a schematic flow chart of a third embodiment of a battery material combination method according to an embodiment of the present application; Figure 4 This is a schematic flow chart of a fourth embodiment of a battery material combination method according to an embodiment of the present application; Figure 5 This is a schematic flow chart of a fifth embodiment of a battery material combination method according to an embodiment of the present application; Figure 6 This is a schematic diagram of the structure of the battery material combination equipment for this application.

[0018] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0019] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0020] As an emerging energy storage technology, sodium batteries have attracted much attention due to their advantages in energy density, cost-effectiveness, and resource abundance. Despite this, the stability and reliability of sodium batteries in practical applications are still key factors restricting their widespread application. The performance and safety of sodium batteries depend largely on the material combination used, and the current evaluation of battery material combinations mainly relies on laboratory tests and experience-based design principles. This evaluation method requires long-term experimental studies on various material combinations to evaluate their stability and safety during battery use. However, this method not only consumes a lot of time and resources, but also finds it difficult to cover all possible material combinations, thereby limiting the range of battery material choices.

[0021] In order to solve the above-mentioned defects existing in the related art, the embodiment of the present application proposes a battery material combination method, which generates one or more groups of candidate material combinations based on battery material data; extracts the attribute characteristics of the candidate material combination and converts the attribute characteristics into a feature vector; inputs the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination. The present application generates candidate material combinations and uses the prediction model to predict the performance of the candidate material combinations, thereby achieving efficient screening and safety performance prediction of battery material combinations, significantly improving the research and development efficiency of battery materials, and also expanding the selection range of battery material combinations.

[0022] For ease of understanding, the technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0023] The battery material combination method of the first embodiment proposed in this application, please refer to Figure 1 The method comprises steps S10 to S30: Step S10: Generate one or more groups of candidate material combinations based on the battery material data.

[0024] In this embodiment, the existing battery material data are first collected and sorted, and the battery material data includes the positive electrode material, negative electrode material, electrolyte, and separator material data that constitute the sodium battery. The collected battery material data will serve as the basis for generating candidate material combinations. Then, the algorithm is used to generate random candidate material combinations.

[0025] Optionally, the battery material data can be automatically captured from academic databases, material performance databases, etc. on the Internet through crawler tools. The crawler tool can automatically access web pages and extract required data according to preset crawling rules and keywords, and save them to local data.

[0026] For example, the crawler tool was used to access the Materials Project (a material performance database) database, and all data entries related to the positive electrode materials of sodium batteries were retrieved through the API interface, including the chemical composition, crystal structure, electrochemical properties, etc. of the positive electrode materials. Subsequently, the crawler tool captured the abstracts and citation information of research papers on sodium-ion batteries published in recent years based on the set search conditions, such as "since 2015", "Sodium-ion battery" and "article". For electrolyte and diaphragm materials, the crawler tool accessed the ScienceDirect database and the Web of Science (WoS) database respectively, and extracted experimental data and performance evaluation related to electrolyte and diaphragm materials.

[0027] It should be noted that the ScienceDirect database is a comprehensive full-text literature database, and the WoS database is a multidisciplinary research results citation database.

[0028] Optionally, battery material data can be collected through experimental data acquisition and integration. Through laboratory testing, directly measure the material properties of battery materials, or cooperate with research institutions to share battery material experimental data. After implementing the battery material measurement experiment, record the experimental conditions and experimental results, including structural analysis of battery materials, electrochemical performance testing, etc. Then store and manage the battery material experimental data.

[0029] Exemplarily, an X-ray Diffraction Analysis (XRD) analysis is performed on a positive electrode material to determine the economic structure of the positive electrode material. The microscopic morphology of the positive electrode material is obtained by scanning electron microscopy (SEM) imaging. The charge transfer performance of the positive electrode material can be evaluated by electrochemical impedance spectroscopy (EIS) testing. The actual energy density and cycle life of the positive electrode material can be measured by charge and discharge cycle testing. After recording the above experimental data on the positive electrode material, the experimental data of the battery material is stored and managed through the Laboratory Information Management System (LIMS).

[0030] Optionally, a genetic algorithm (GA) is used to generate one or more candidate material combinations. First, a fitness function is defined to evaluate the performance of each material combination. In this embodiment, key performance indicators such as energy density and cycle life of battery materials can be used as input parameters of the fitness function. By calculating the performance of each material combination on these performance indicators, a fitness value is assigned to each material combination.

[0031] Then a population is initialized, which contains multiple randomly generated material combinations. A new generation of population is generated by performing selection, crossover and mutation operations on these material combinations. Specifically, in the selection operation, probability selection is performed based on the fitness value of the material combination, so that the material combination with a higher fitness value has a greater probability of being selected. In the crossover operation, some genes of the two selected material combinations are exchanged to generate a new material combination. In the mutation operation, some genes of the material combination are randomly changed to introduce new gene mutations.

[0032] By continuously iterating the above process, a material combination with a higher fitness value is gradually found. A termination condition can be set, such as reaching a preset maximum number of iterations or finding a solution (i.e., a material combination) that meets the requirements. The final material combination is used as a candidate material combination. The method of using a genetic algorithm to generate a candidate material combination can quickly find a potential optimal solution in a larger search space, thereby improving the efficiency and accuracy of battery design.

[0033] Optionally, a particle swarm optimization (PSO) algorithm is used to generate one or more sets of candidate material combinations. In this embodiment, key performance indicators such as energy density and cycle life of battery materials are input as adaptability functions to evaluate the performance of each material combination. Each particle represents a candidate material combination in the search space and has a position and speed. The position corresponds to the specific parameters of the material combination, and the speed determines the movement direction and step size of the particle in the search space.

[0034] At initialization, a group of particles are randomly generated, each particle represents a possible material combination solution. In each iteration, the particle updates its speed and position by tracking its own historical best position and the best position of the entire group. When updating the speed, the particle's own experience and the experience of neighboring particles are taken into account, so that the particle can move towards a better solution.

[0035] As the iterations proceed, the particle swarm gradually converges to the optimal area in the search space. After each iteration, the fitness value of each particle is recalculated, and the global optimal and individual optimal positions are updated based on the fitness value. This process continues until the termination condition is met, such as reaching the preset maximum number of iterations, the fitness value exceeds a certain threshold, or the improvement of the group's optimal position is no longer significant. Finally, one or more groups of candidate material combinations with higher fitness values ​​are output based on the positions of the particles.

[0036] Step S20: extracting the property features of the candidate material combination and converting the property features into a feature vector.

[0037] In this embodiment, the attribute characteristics of the candidate material combination are a set of attributes that have a decisive influence on the material combination. For example, for positive electrode materials, the main focus is on their crystal structure, electrochemical activity, ion diffusion rate, etc.; for negative electrode materials, the main focus is on their specific capacity, electronic conductivity, stability, etc.

[0038] Optionally, each material property of the candidate material combination is directly mapped to the dimension of the feature vector. Each material property is first quantized to ensure that its value can be processed by the model.

[0039] Specifically, the material properties of each battery material in the candidate material combination and the property values ​​corresponding to the material properties are obtained, and then the property values ​​of each battery material are mapped into a set of feature vectors, and each property value corresponds to an element in the feature vector.

[0040] Exemplarily, the crystal structure of the positive electrode material can be quantified by its lattice parameters, the electrochemical activity can be represented by its open circuit voltage, and the ion diffusion rate can be quantified by its conductivity. The specific capacity of the negative electrode material can be quantified by its theoretical capacity, the electronic conductivity can be measured by its conductivity, and the stability can be evaluated by the results of thermal stability tests. Then, the quantified property values ​​are arranged in a certain order to form a feature vector. For example, if a candidate material combination consists of two positive electrode materials and three negative electrode materials, then its feature vector contains the quantified values ​​of at least five material properties, and each property value corresponds to a dimension of the feature vector.

[0041] Optionally, the principal component analysis (PCA) technique is used to reduce the dimension of the attribute characteristics. In this embodiment, the material attribute characteristic data of all candidate material combinations are first collected to form a characteristic matrix. Then, the PCA algorithm is applied to decompose the characteristic matrix to extract several principal components, which can explain most of the variability in the material attribute characteristic data. Each principal component is a projection of the original eigenvector in the new coordinate space. The first few principal components are selected according to the contribution rate of the principal component and combined into a new eigenvector.

[0042] For example, assume that data for 100 candidate material combinations are collected, and each material combination is described by 10 different property characteristics, such as the crystal structure parameters, electrochemical activity, and ion diffusion rate of the positive electrode material, and the specific capacity, electronic conductivity, and stability of the negative electrode material.

[0043] The data of these candidate material combinations are organized into a feature matrix with 100 rows and 10 columns, where each row represents a candidate material combination and each column represents an attribute feature. A principal component analysis result is obtained by applying the PCA algorithm, showing that the first three principal components can explain more than 85% of the variability in the data. Therefore, the first three principal components are selected as new eigenvectors, and each principal component is a projection of the original 10-dimensional eigenvector in the new three-dimensional space. In the specific implementation, the feature matrix is ​​first standardized to ensure that the mean of each attribute feature is 0 and the standard deviation is 1. Then, the PCA algorithm is used to calculate the covariance matrix of the feature matrix, and the eigenvalues ​​and eigenvectors of the covariance matrix are found. According to the contribution rate of the eigenvalue, the eigenvectors corresponding to the first three largest eigenvalues ​​are selected, and these eigenvectors are the new principal components. Finally, the original feature matrix is ​​multiplied by the eigenvectors of the three principal components to obtain a new feature matrix with 100 rows and 3 columns. This new matrix is ​​the set of eigenvectors after dimensionality reduction.

[0044] Optionally, attribute features are processed using feature engineering, which includes feature construction, feature selection, and feature conversion.

[0045] During the feature construction phase, new features can be created by calculating ratios, differences, or interactions between material properties. For example, the ratio of the electrochemical activity of the cathode material to its ion diffusion rate may have an important impact on battery performance, so this ratio can be used as a new feature.

[0046] In the feature selection stage, statistical methods or model selection methods are used to determine which features are most useful for predicting battery performance. For example, recursive feature elimination (RFE) can be used to select the most influential features.

[0047] In the feature conversion stage, appropriate mathematical transformations are performed on the selected attribute features, such as standardization or normalization, to eliminate the dimensional effects between different attribute features. After completing the three steps of feature engineering, the transformed feature values ​​are combined into a feature vector.

[0048] Step S30: inputting the feature vector into a pre-trained prediction model to obtain a performance prediction result of the candidate material combination.

[0049] In this embodiment, a machine learning model or a deep learning model can be used for training, such as a support vector machine (SVM), a convolutional neural network (CNN), a long short-term memory network (LSTM) or a deep neural network (DNN), to establish a prediction model for predicting the safety performance of a material combination.

[0050] Optionally, a prediction model for predicting the safety performance of a material combination is constructed using an SVM model. In this embodiment, the optimal model parameters, including the kernel function type, penalty parameters, and kernel function parameters, are first determined by techniques such as grid search and cross-validation. Then, the SVM model is trained using these model parameters until the SVM model achieves a satisfactory accuracy rate on the training set, and the prediction model training is completed. After the training is completed, the feature vector of the candidate material combination is input into the prediction model, and the prediction model outputs the performance prediction result of the candidate material combination.

[0051] Exemplarily, the radial basis function (RBF) is selected as the kernel function, and the optimal penalty parameter C and kernel function parameter γ are found in the parameter space through grid search. After multiple iterations and cross-validation, C=10 and γ=0.1 are determined as the optimal parameters. Subsequently, the SVM model is trained on a material combination data set containing 100 samples using the optimal parameters. After about 20 rounds of iterations, the accuracy of the SVM model on the training set reaches more than 95%, indicating that the prediction model training is complete. Finally, the feature vectors of the 100 candidate material combinations are input into the trained prediction model to obtain the performance prediction results of the 100 candidate material combinations.

[0052] Optionally, a CNN model is used to construct a prediction model for predicting the safety performance of a material combination. In this embodiment, a multi-layer convolutional neural network is designed, which includes multiple convolutional layers and pooling layers to extract complex patterns and associations in feature vectors. The end of the neural network is connected to a fully connected layer and an output layer for performance prediction. During the training of the prediction model, the back propagation algorithm and the gradient descent method are used to optimize the network weights until the loss function of the prediction model on the validation set is minimized. The trained CNN model is used to receive the feature vector of the candidate material combination, and the performance prediction result of the candidate material combination is predicted through its deep network structure.

[0053] Exemplarily, a CNN model with three convolutional layers and two pooling layers is constructed, and each convolutional layer is followed by a ReLU (Rectified Linear Unit) activation function to introduce nonlinearity. At the end of the neural network, two fully connected layers are added, the first fully connected layer has 128 neurons, and the second fully connected layer directly outputs the performance prediction results. In the model training phase, the Adam (Adaptive Moment Estimation) optimizer and the cross entropy loss function were used, and 20 cycles of training were performed on a data set containing 100 samples. After training, the loss of the CNN model on the validation set dropped below 0.02, indicating that the prediction model obtained based on the CNN model training has good prediction ability. Subsequently, the reduced feature vector was input into the prediction model to obtain the performance prediction data of 100 candidate material combinations.

[0054] It should be noted that the Adam optimizer is an efficient stochastic optimization algorithm used in deep learning, which can adaptively adjust the learning rate for different parameters.

[0055] Please refer to Figure 2 In the battery material combination method of the second embodiment proposed in the present application, step S10 may further include the following steps S110-S130: Step S110: determining an interaction rule of a material combination, wherein the interaction rule is used to guide pairing of the material combination.

[0056] In this embodiment, the interaction rules define the possible interaction mechanisms between different battery materials, including but not limited to chemical compatibility, complementarity of electrochemical performance, and matching of physical properties. The interaction rules serve as guiding principles in the material combination pairing process, ensuring that the generated material combination has theoretical synergy, thereby improving the quality and prediction efficiency of the material combination prediction, while also improving the overall performance of the material combination.

[0057] Step S120: Based on the interaction rules, a candidate list of the material combinations is obtained.

[0058] In this embodiment, the candidate list of material combinations includes a plurality of pairing schemes of material combinations selected according to the interaction rules, and each pairing scheme may be a potential material combination that eventually forms a high-efficiency battery material. The generation of the candidate list takes into account all possible material pairs in the battery material data and the interactions between them.

[0059] Step S130: Screening the candidate list of material combinations according to a preset material property standard to determine the candidate material combination.

[0060] It should be noted that the material property standards may include the conductivity, ion diffusion rate, stability and cost of the battery material, etc. By evaluating the material properties of each material combination in the candidate list, the material combination that meets the preset standards can be screened out, thereby determining the final candidate material combination.

[0061] Please refer to Figure 3 The battery material combination method of the third embodiment proposed in the present application may include the following steps S310~S340 before step S30: Step S310: Acquire safety performance data corresponding to material properties of battery materials, wherein the safety performance data includes impact test results of the battery materials under external impact conditions.

[0062] In this embodiment, the material properties of the battery material include physical property data and chemical property data. Physical property data include density, electrical conductivity, ion diffusion rate, etc. Chemical property data include capacity, cycle life, rate performance, etc. Safety performance data corresponding to these material properties are obtained, especially the impact test results of the battery material under external impact conditions. The impact test results are used to evaluate the safety response of the battery material when subjected to mechanical impact, puncture, extrusion, falling or other forms of external force, such as whether thermal runaway, explosion or leakage occurs.

[0063] Specifically, before obtaining the safety performance data corresponding to the material properties of the battery material, an impact test is performed on the battery material and the impact test result is recorded. Then, the safety performance data of the battery material is determined according to the impact test result.

[0064] For example, the battery material is subjected to an impact test, such as by using a pendulum of a specified weight to hit the battery material sample at different speeds, and recording the deformation of the battery material sample after the impact; or by using a sharp object to puncture the battery material sample, and recording the safety response during the puncture process; or by simulating the performance of the battery material sample when it is squeezed and dropped, etc. Subsequently, the results of the corresponding impact test are recorded, including the shape change of the battery material sample after the impact, the temperature rise, and whether thermal runaway, explosion or leakage occurs, and other safety performance data.

[0065] After the impact test is completed, the safety performance data of the battery material is determined based on the impact test results. For example, the brittle fracture of a battery material in a mechanical impact test is related to its low density and high conductivity, while materials that are prone to thermal runaway in a puncture test usually have a lower ion diffusion rate, while a positive electrode material shows good safety in a drop test, and its chemical property data shows higher rate performance and cycle stability.

[0066] Step S320: constructing a battery material data set according to the material properties and the safety performance data.

[0067] Optionally, key attributes are screened out from the material attributes of the battery material, and the attribute features of these key attributes are extracted. The attribute features will be used as input to the prediction model to predict the safety performance of the battery material. The safety performance data obtained from the impact test results are then used as labels, corresponding to the attribute features of the key attributes, to form a battery material dataset for training. The battery material dataset contains the attribute features of each battery material and the labels of its corresponding safety performance data.

[0068] Specifically, the material properties and safety performance data are first preprocessed, including data cleaning to remove outliers and missing values ​​to ensure data quality and integrity; standardization or normalization of the data so that material properties of different dimensions can be compared at the same scale; data can also be transformed, such as linearizing nonlinear relationships, or using logarithmic transformations and other methods to improve the distribution characteristics of the data. When processing safety performance data, the classification labels need to be encoded, such as using One-Hot Encoding to convert classification data into a numerical form suitable for the model. The impact test results are then associated with the material properties of the corresponding battery materials.

[0069] Exemplarily, it is assumed that a set of material properties and safety performance data of battery materials are received, including material properties such as density, conductivity, ion diffusion rate, capacity, cycle life, and safety performance data such as impact test results and thermal runaway risk.

[0070] First, the data is cleaned. For example, an outlier in the conductivity data is detected using the box plot method and automatically removed. At the same time, the missing values ​​in the capacity data are automatically filled using the interpolation method. The data is then standardized, and the material properties such as density, conductivity, and ion diffusion rate are scaled to the range of [0, 1] using the minimum-maximum standardization method; the material properties such as capacity and cycle life are converted to a standard normal distribution using the Z-score standardization method. Then, nonlinear relationships such as capacity and cycle life are converted to linear relationships using logarithmic transformation to facilitate model training and prediction. Finally, the one-hot encoding method is used to convert classification labels such as impact test results and thermal runaway risks into binary vectors, with each category corresponding to a binary bit.

[0071] Exemplarily, after the preprocessing operation is completed, the preprocessed material properties are associated with the corresponding preprocessed safety performance data to form a battery material data set. For example, for a battery material sample, its ion diffusion rate is associated with the thermal runaway risk in the corresponding impact test result to form a data point containing material properties and safety performance data.

[0072] Step S330: training a prediction model based on the battery material data set.

[0073] It should be noted that in terms of model selection, a variety of deep learning architectures can be used, such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory Networks (LSTM) or Deep Neural Networks (DNN), etc. These network structures can learn complex features and nonlinear relationships from battery material data.

[0074] Exemplarily, based on the characteristics of battery material data, a deep neural network with multiple hidden layers is selected as the basic model because the deep neural network can effectively process high-dimensional data and capture the nonlinear characteristics in the data.

[0075] In the process of model construction, the structure of the deep neural network is first determined, including the number of nodes in the input layer, hidden layer, and output layer. The number of nodes in the input layer matches the material property dimension of the battery material, and the number of nodes in each hidden layer is determined according to empirical formulas or through experiments to balance the expressive power and computational complexity of the model. The design of the output layer depends on the type of safety performance data. If it is a classification problem, the number of nodes in the output layer is equal to the number of categories, and the softmax function is used to output the probability distribution; if it is a regression problem, the output layer has only one node and uses a linear activation function.

[0076] In this embodiment, a deep neural network model is used to train the material properties and safety performance data of battery materials to construct a prediction model that can predict the safety performance of battery material combinations.

[0077] Optionally, the battery material data set is divided into a training set and a test set, the training set is used for training the prediction model, and the test set is used to evaluate the generalization ability of the prediction model. During the training of the prediction model, the performance of the prediction model is optimized by adjusting the parameter settings of the prediction model on the training set. Subsequently, the test set is used to verify the generalization ability of the prediction model to obtain a verification result. Among them, the verification result will be evaluated by calculating the performance indicators of the prediction model, including calculating the accuracy, recall rate, F1 score, mean square error or confusion matrix, etc.

[0078] It is understandable that the test set does not participate in the training of the prediction model, so the test set can provide the performance of the prediction model on unknown data, that is, the generalization ability of the prediction model.

[0079] Specifically, during the data set division process, the samples in the battery material data set are allocated to the training set and the test set according to a preset ratio. Generally speaking, the training set contains most of the sample data, for example, 70%-80% of the sample data is used as the training set, and the remaining 20%-30% of the sample data constitutes the test set. This division ensures that the prediction model can fully learn the complex relationship between the material properties and safety performance of battery materials during the training process, and the prediction ability of the prediction model can be verified on an independent test set.

[0080] On the training set, the prediction model will learn the mapping relationship from feature data to safety performance labels by adjusting internal parameters. This learning process involves multiple iterations, and the prediction model gradually optimizes its prediction accuracy in each iteration. After training is completed, the test set is used to evaluate and verify the performance of the prediction model.

[0081] In order to evaluate the performance of the prediction model, the performance indicator data of the prediction model will be calculated, including accuracy, recall, etc. By calculating the performance indicator data, the comprehensive performance of the prediction model on the test set can be obtained, so as to determine whether the prediction model has achieved the expected performance level. If the prediction model performs well on the test set, it means that the prediction model has good generalization ability and can be used for actual battery material screening and risk assessment. If the prediction model performs poorly on the test set, it is necessary to further adjust the prediction model parameters and re-iterate the training process to improve its generalization ability.

[0082] It should be noted that the accuracy reflects the proportion of samples predicted to be positive by the prediction model that are actually positive.

[0083] For example, assume that there is a data set containing 100 battery material samples, 70% of the samples are divided into training sets, and the remaining 30% of the samples are used as test sets, that is, the training set contains 70 samples and the test set contains 30 samples.

[0084] During iterative training on the training set, the prediction model learns the mapping from feature data to safety performance labels by adjusting internal parameters.

[0085] After training, the test set is used to evaluate and verify the performance of the prediction model and calculate the accuracy of the prediction model. For example, on the test set, the prediction model predicted the safety performance of 30 samples, of which 26 were predicted to be positive and 4 were predicted to be negative. Of the 26 samples predicted to be positive, 24 were actually positive, so the accuracy was 24 / 26=92.3%, indicating that the prediction model has good generalization ability and can be used for actual battery material screening and risk assessment.

[0086] Optionally, a cross-validation method is used to divide the battery material data set into subsets of equal size, each containing the same amount of data. In each iteration of training, a different subset is selected as a validation set, and the remaining subsets are used as training sets. The prediction model is iteratively trained, and the prediction error is calculated using the validation set.

[0087] It should be noted that the battery material dataset is divided into several subsets of equal size in order to ensure that each subset can represent the characteristics of the entire battery material dataset, so that in the subsequent prediction model training and verification process, the prediction model can learn comprehensive and balanced features.

[0088] Specifically, the training set is divided into K subsets. In each iterative training, one subset is selected as the validation set, and the remaining K-1 subsets are combined as the training set. In each iteration, the prediction model is trained on the new training set, and the prediction error is calculated on the selected validation set, including comparing the difference between the prediction results of the prediction model and the labels of the actual safety performance data. The iterative training process will continue until each subset has been used as a validation set once. Through cross-validation, the prediction model is tested on multiple different training combinations, which helps to more accurately evaluate the generalization ability of the prediction model. After each iteration, the prediction error of the prediction model is recorded, which will be used to adjust the hyperparameter settings of the prediction model.

[0089] Optionally, after iteratively training the prediction model and calculating the prediction error using the validation set, after each iterative training, the hyperparameter configuration of the prediction model will be updated based on the change trend of the prediction error to reduce the risk of overfitting or underfitting and improve the prediction accuracy of the prediction model. Then, using the prediction model with updated hyperparameter configuration, repeat the steps of selecting a different subset as the validation set and the remaining subset as the training set in each iterative training, iteratively training the prediction model, and calculating the prediction error using the validation set until the prediction error is reduced to a preset threshold, or the prediction error is reduced below the preset threshold, or a predetermined number of iterations is reached.

[0090] It should be noted that hyperparameter configuration refers to the parameters that can be adjusted during the training process of the prediction model, such as learning rate, number of hidden layer nodes, batch size, regularization coefficient, etc. By analyzing the prediction error, it is possible to determine which hyperparameters have a significant impact on the performance of the prediction model, and then adjust the hyperparameter configuration to optimize the performance of the prediction model.

[0091] In this embodiment, by selecting different subsets as validation sets in each iterative training, it is possible to ensure that the prediction model is trained and validated on different data combinations, thereby improving the generalization ability of the prediction model. By iteratively training the training set and calculating the prediction error using the validation set, the performance of the prediction model on independent data can be monitored in real time.

[0092] For example, suppose there is a battery material data set containing 100 battery material samples, which is divided into 5 subsets, each containing 20 sample data. In each iteration of training, a different subset is selected as the validation set, and the remaining subsets are combined as the training set. For example, the first iteration of training selects the first subset as validation set 1, and the remaining 4 subsets are combined as training set 1. The prediction model is trained on training set 1, and the prediction error is calculated on validation set 1. Then the second iteration of training is performed, the second subset is selected as validation set 2, and the remaining 4 subsets are combined into training set 2, and training continues until all subsets have been used as validation sets once.

[0093] The prediction model was trained on 5 different combinations by cross-validation. After each iteration, the prediction error of the prediction model was recorded, which will be used to adjust the hyperparameter settings of the prediction model. By adjusting the hyperparameters of the prediction model, the prediction model can minimize the training error and optimize the performance of the prediction model, thereby ensuring the reliability and accuracy of the prediction model in practical applications.

[0094] Step S340: using the trained prediction model to predict the safety performance of the candidate material combination, and outputting the performance prediction result of the candidate material combination.

[0095] It should be noted that candidate material combinations refer to battery material combinations that have not yet been experimentally verified. Since the experimental verification process is time-consuming and costly, the predictive model can predict the safety performance of these material combinations to screen out potential high-performance material combinations, thereby guiding researchers to focus resources on those material combinations that are most likely to succeed.

[0096] Specifically, the material properties of the candidate material combination are preprocessed to ensure the quality and consistency of the model input data and improve the accuracy and reliability of the prediction model. Then, the feature vector of the preprocessed candidate material combination is extracted and input into the trained prediction model to obtain the performance prediction result of the candidate material combination.

[0097] For example, a set of candidate material combinations that have completed data preprocessing operations are selected for safety performance prediction. Assume that the candidate material combination includes a new positive electrode material and an existing negative electrode material, and its material properties include: density of 0.95 g / cm³, conductivity of 150 mS / cm, ion diffusion rate of 1.2×10 -4 cm 2 / s. The material properties of the battery material combination and the safety performance data corresponding to the material properties are feature extracted, and the feature vector is input into the prediction model, which will output the safety performance prediction result of the candidate material combination.

[0098] Please refer to Figure 4 The battery material combination method of the fourth embodiment proposed in the present application, after step S30, further includes the following steps S350~S370: Step S350: Determine the confidence interval of the safety performance of the battery material based on the historical safety performance data of the battery material.

[0099] It should be noted that the confidence interval is a statistical tool used to estimate the reliability of performance prediction results. Specifically, the confidence interval indicates the interval within which a population parameter (such as an average value or a proportion, etc.) may fall under a certain confidence level (usually 95%). In this embodiment, the confidence interval is calculated based on historical safety performance data, indicating that there is a 95% confidence that the safety performance indicators of battery materials (such as thermal stability, cycle life, safety test results, etc.) will fall within this interval.

[0100] Step S360: Compare the performance prediction result of the candidate material combination with the confidence interval.

[0101] Step S370: Determine the safety level of the candidate material combination according to the comparison result.

[0102] Exemplarily, historical safety performance data of five battery materials are selected, including indicators such as impact resistance, thermal stability, and electrochemical stability. Taking impact resistance as an example, the historical safety performance data shows that the average value is 85 Newtons and the standard deviation is 5 Newtons. The 95% confidence interval is calculated to be [75.4, 94.6] Newtons. The 95% confidence intervals of thermal stability and electrochemical stability indicators are [300, 350] degrees Celsius and [0.95, 1.05], respectively.

[0103] In step S360, for a set of candidate material combinations, the trained prediction model is used to predict performance. Assuming that the candidate combination consists of materials A, B, and C, the prediction model predicts that the impact resistance of the candidate material combination is 80 Newtons, the thermal stability is 320 degrees Celsius, and the electrochemical stability is 1.0. Comparing these predicted values ​​with the previously calculated confidence intervals, the results show that the predicted value of impact resistance of 80 Newtons is within the confidence interval of [75.4, 94.6] Newtons, the predicted value of thermal stability of 320 degrees Celsius is within the confidence interval of [300, 350] degrees Celsius, and the predicted value of electrochemical stability of 1.0 is equal to the lower limit of the confidence interval.

[0104] Based on the comparison results, the safety level of candidate material combination ABC is determined. Since all predicted performance values ​​are within the corresponding confidence intervals, the candidate material combination is classified as a high safety level. This indicates that based on the historical safety performance data and the prediction model, the candidate material combination has high safety under extreme conditions and is suitable for further use in battery research and development and production.

[0105] Please refer to Figure 5 The battery material combination method of the fifth embodiment proposed in the present application, after step S30, further includes the following steps S380~S382: Step S380: Determine the material properties that need to be optimized based on the performance prediction results, and regenerate a new candidate material combination.

[0106] In this embodiment, the performance prediction results output by the prediction model are statistically analyzed, and the correlation coefficient between each material attribute and the target safety performance indicator is calculated to identify which material attributes are most relevant to the improvement or reduction of safety performance. For example, if it is found that the cycle stability of the battery is highly correlated with the crystallinity of a particular material, then the crystallinity is the key attribute that needs to be optimized.

[0107] Optionally, the interaction rules of the material combinations can be adjusted based on the correlation coefficient between each material property and the target performance indicator, and optimized when randomly generating material combinations.

[0108] Optionally, the Pearson correlation coefficient is used to analyze the relationship between each material property and the safety performance index. For each material property, calculate its Pearson correlation coefficient with the battery cycle stability or other safety performance indicators. Sort the calculated correlation coefficients to find the material properties that are most relevant to the safety performance indicators. Set optimization goals for the top-ranked material properties, such as increasing crystallinity to enhance cycle stability. Finally, according to the optimization goals, adjust the range or property values ​​of these material properties to generate new candidate material combinations.

[0109] Optionally, identify key material properties that affect safety performance based on multivariate regression analysis. Use a multivariate linear regression model to analyze the relationship between all material properties and safety performance indicators and establish a regression equation. Identify the material properties that have the greatest impact on safety performance through the size and significance level of the coefficients in the regression equation. Select several properties with the largest and most significant coefficients as optimization priorities. Use the Design of Experiments (DOE) method to adjust the different levels of these key material properties and generate new candidate material combinations.

[0110] Step S381: input the feature vector of the new candidate material combination into the prediction model to obtain a new performance prediction result.

[0111] Step S382: Based on the new performance prediction result, iteratively adjust the new candidate material combination until a preset performance target is reached or the adjustment of the candidate material combination is completed.

[0112] In this embodiment, a performance target threshold is first set to define an acceptable safety performance range for battery materials. Then, based on the gap between the predicted results and the target safety performance indicators, the key properties in the candidate material combination are adjusted. The adjustment process includes increasing or decreasing the content of a specific battery material, changing the processing conditions of the battery material, or adjusting the combination of battery materials to optimize the overall performance of the battery. Then, iterative optimization is performed to adjust the material properties according to the results of the previous iteration, generate new candidate material combinations, and input the feature vectors of these combinations into the prediction model again. Until the design goal is achieved (i.e., the performance target threshold is achieved), such as maximizing the energy density while ensuring the safety of the material structure under various test conditions, a potential high-performance battery material combination is output.

[0113] The battery material combination method of the sixth embodiment proposed in the present application, after step S30, further includes: Finite Element Analysis (FEA) is performed on the battery design based on the optimized candidate material combination. Finite element analysis involves dividing the structural model of the candidate material combination into a finite number of small elements, each of which has certain geometric and physical properties. By applying external forces to these elements, that is, performing impact tests, and then calculating the stress and deformation of each element, the overall response of the battery structure based on the candidate material combination under a specific impact or pressure can be simulated. During the stress analysis process, the focus is on the stress distribution in the battery design and possible damage modes, such as the initiation and propagation of cracks.

[0114] Computer vision technology is then used to extract key information from FEA simulation results, including identifying and quantifying stress concentration areas and potential battery structural weaknesses. Computer vision technology can automatically identify key features in simulation results through image processing algorithms, reducing reliance on manual analysis.

[0115] Finally, an interactive visualization tool is developed to display FEA results in a graphical manner, converting complex stress data and deformation patterns into intuitive images, so that the behavior of the battery structure under impact based on the candidate material combination can be more deeply understood and analyzed, thereby identifying potential problems in the battery design based on the candidate material combination, and adjusting the candidate material combination accordingly to optimize the structural strength and safety of the battery. Through the application of FEA and visualization technology, the battery material combination method of this embodiment not only improves the accuracy of battery design, but also enhances the efficiency of the battery design process.

[0116] The present application provides a battery material combination device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the battery material combination method in the above-mentioned embodiment one.

[0117] Reference below Figure 6 , which shows a schematic diagram of the structure of a battery material assembly device suitable for implementing an embodiment of the present application. The battery material assembly device in the embodiment of the present application may include various hardware and software components for implementing a scheduling method for inspection tasks. Figure 6 The battery material combination device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0118] like Figure 6As shown, the battery material combination device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the battery material combination device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the battery material combination device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a battery material combination device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0119] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0120] The battery material combination device provided by the present application adopts the battery material combination device method in the above embodiment, which can solve the technical problem that the battery material combination evaluation efficiency is low and the coverage combination is limited, resulting in limited battery material selection. Compared with the prior art, the beneficial effects of the battery material combination device provided by the present application are the same as the beneficial effects of the battery material combination method provided by the above embodiment, and the other technical features in the battery material combination device are the same as the features disclosed in the previous embodiment method, which will not be repeated here.

[0121] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0122] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0123] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the battery material combination method in the above-mentioned embodiment.

[0124] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0125] The computer-readable storage medium may be included in the battery material combination device; or may exist independently without being assembled into the battery material combination device.

[0126] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the battery material combination device, the battery material combination device: generates one or more groups of candidate material combinations based on battery material data; extracts attribute characteristics of the candidate material combination and converts the attribute characteristics into feature vectors; inputs the feature vectors into a pre-trained prediction model to obtain performance prediction results of the candidate material combination.

[0127] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0129] The modules involved in the embodiments described in the present application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0130] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned battery material combination method, and can solve the technical problems of low efficiency of battery material combination evaluation and limited coverage of combinations, resulting in limited battery material selection. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the battery material combination method provided in the above-mentioned embodiment, and will not be repeated here.

[0131] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the battery material combination method as described above when executed by a processor.

[0132] The computer program product provided in this application can solve the technical problem that the battery material combination evaluation efficiency is low and the coverage combination is limited, resulting in limited battery material selection. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of this application are the same as the beneficial effects of the battery material combination method provided in the above embodiment, and will not be repeated here.

[0133] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

[0134] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system 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, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0135] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0136] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A battery material combination method, characterized in that: The battery material combination method comprises: Based on the battery material data, generate one or more groups of candidate material combinations; Extracting property features of the candidate material combination and converting the property features into feature vectors; The feature vector is input into a pre-trained prediction model to obtain a performance prediction result of the candidate material combination.

2. The battery material combination method according to claim 1, characterized in that: The step of generating one or more groups of candidate material combinations based on the battery material data comprises: determining an interaction rule of a material combination, wherein the interaction rule is used to guide pairing of the material combination; Based on the interaction rules, obtaining a candidate list of the material combinations; According to a preset material property standard, the candidate list of the material combination is screened to determine the candidate material combination.

3. The battery material combination method according to claim 1, characterized in that: The step of extracting the attribute features of the candidate material combination and converting the attribute features into a feature vector comprises: Obtaining material properties of each battery material in the candidate material combination and property values ​​corresponding to the material properties; The property value of each battery material is mapped into a set of feature vectors, and each property value corresponds to an element in the feature vector.

4. The battery material combination method according to claim 1, characterized in that: Before the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the method includes: Acquiring safety performance data corresponding to material properties of battery materials, the safety performance data including impact test results of the battery materials under external impact conditions; Constructing a battery material data set according to the material properties and the safety performance data; Based on the battery material data set, training a prediction model; The trained prediction model is used to predict the safety performance of the candidate material combination, and the performance prediction result of the candidate material combination is output.

5. The battery material combination method according to claim 4, characterized in that: Before the step of training the prediction model based on the battery material data set, the method includes: Dividing the battery material data set into subsets of equal size, each subset containing the same amount of data; In each iterative training, a different subset is selected as a validation set, and the remaining subsets are selected as training sets. The prediction model is iteratively trained, and the prediction error is calculated using the validation set.

6. The method according to claim 5, characterized in that After the step of iteratively training the prediction model and calculating the prediction error using the validation set, the method further includes: After each iterative training, based on the prediction error, the hyperparameter configuration of the prediction model is updated; Using the prediction model after updating the hyperparameter configuration, repeatedly execute the steps of selecting a different subset as a validation set and the remaining subsets as training sets in each iterative training, iteratively training the training sets, and calculating the prediction error using the validation set until the prediction error is reduced to a preset threshold.

7. The battery material combination method according to claim 1, characterized in that: After the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the method further comprises: Determining a confidence interval for the safety performance of the battery material based on historical safety performance data of the battery material; comparing the performance prediction result of the candidate material combination with the confidence interval; According to the comparison result, the safety level of the candidate material combination is determined.

8. The battery material combination method according to claim 1, characterized in that: After the step of inputting the feature vector into a pre-trained prediction model to obtain the performance prediction result of the candidate material combination, the method further includes: According to the performance prediction results, determine the material properties that need to be optimized, and regenerate a new candidate material combination; Inputting the feature vector of the new candidate material combination into the prediction model to obtain a new performance prediction result; Based on the new performance prediction result, the new candidate material combination is iteratively adjusted until a preset performance target is reached or the adjustment of the candidate material combination is completed.

9. A battery material assembly device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the battery material combination method according to any one of claims 1 to 8.

10. A computer storage medium, characterized in that: The computer storage medium is a computer-readable storage medium, and a computer program is stored on the computer storage medium. When the computer program is executed by a processor, the steps of the battery material combination method according to any one of claims 1 to 8 are implemented.

Images