Vehicle battery capacity retention rate evaluation method, device, equipment and medium

Abstract

The invention relates to the technical field of new energy vehicle batteries, aims to solve the problem of low evaluation precision of the battery capacity retention ratio in the prior art, and provides a vehicle battery capacity retention ratio evaluation method and device, equipment and a medium. The method comprises the steps that vehicle data are acquired, and the vehicle data comprise vehicle operation data and battery data; the battery data are preprocessed, preprocessed battery data are obtained, and the preprocessed battery data comprise multiple pieces of battery characteristic information; according to the preprocessed battery data, carrying out battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data; and obtaining a battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle data and a preset input model. According to the method, the battery capacity attenuation model is established for brand vehicle models, and the evaluation precision of the battery capacity retention ratio is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy vehicle batteries, and in particular to a method, device, equipment and medium for evaluating the capacity retention rate of a vehicle battery. Background Art

[0002] With the development of the times, the status of new energy vehicles is increasing day by day, and there is a trend of catching up with fuel-driven vehicles. In the production cost of new energy vehicles, power batteries account for a large proportion. However, as the power battery is used for a longer time, the battery performance will gradually decline, and the battery capacity will gradually decay. It is difficult for users to understand the detailed health status of the battery. At present, it is necessary to use a third-party agency to use specific equipment for inspection, and the inspection period is long, which makes it difficult for users to conveniently understand the battery capacity retention rate.

[0003] Patent No. CN 117991108 B discloses a data-based power battery damage detection method and system. The method constructs a damage diagnosis model by processing vehicle battery data, wherein the damage diagnosis model can construct multiple indicators according to the scenario definition, thereby predicting the abnormal state of the power battery cell level. The prediction method adopts a consistency correlation analysis method, and performs correlation analysis on various battery data in each charge and discharge cycle based on the time dimension, obtains the correlation of the current charge and discharge cycle change, calculates the battery performance evaluation index based on the correlation, and obtains the battery performance evaluation result. However, the correlation analysis of the battery in this method is limited to calculating the correlation and evaluation index, and does not make a detailed distinction between the actual relevant features. Therefore, the low-correlation features will affect the evaluation precision during the calculation process and reduce the evaluation accuracy.

[0004] Therefore, how to distinguish relevant features in detail and accurately evaluate the battery capacity retention rate is an urgent problem to be solved. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a vehicle battery capacity retention rate evaluation method, device, equipment and storage medium to solve the problem of low accuracy in battery capacity retention rate evaluation in the prior art.

[0006] In a first aspect, an embodiment of the present invention provides a method for evaluating a vehicle battery capacity retention rate, the method comprising:

[0007] Acquiring vehicle data, wherein the vehicle data includes vehicle operation data and battery data;

[0008] Preprocessing the battery data to obtain preprocessed battery data, wherein the preprocessed battery data includes: a number of battery characteristic information;

[0009] According to the pre-processed battery data, a battery capacity attenuation correlation analysis is performed to obtain battery capacity attenuation characteristic data;

[0010] A battery capacity retention rate evaluation result is obtained based on the battery capacity decay characteristic data, the vehicle operation data and a pre-constructed battery capacity decay model.

[0011] Preferably, the preprocessing of the battery data to obtain the preprocessed battery data includes:

[0012] Performing missing data processing on the battery data to obtain first battery data;

[0013] Performing anomaly detection according to the first battery data to obtain second battery data;

[0014] Performing data smoothing processing on the second battery data to obtain third battery data;

[0015] Performing data ratio balancing processing according to the third battery data and a preset first threshold to obtain fourth battery data;

[0016] Normalizing the data according to the fourth battery data to obtain fifth battery data;

[0017] Performing data dimension reduction according to the fifth battery data to obtain sixth battery data;

[0018] According to the sixth battery data, construct a metric space to obtain seventh battery data;

[0019] According to the seventh battery data, data standardization is performed to obtain pre-processed battery data.

[0020] Preferably, performing abnormality detection to obtain second battery data according to the first battery data includes:

[0021] Performing statistical analysis based on the first battery data to obtain an abnormality score;

[0022] If the abnormal score is greater than a fixed value, the abnormal score data point is obtained as a hidden abnormal point;

[0023] According to the hidden abnormal point, a first abnormal feature is obtained;

[0024] According to the first abnormal feature, obtaining first upper and lower limits based on the standard deviation;

[0025] According to the first abnormal feature and the first upper and lower limits, performing abnormal correction to obtain a second abnormal feature;

[0026] According to the second abnormal feature and the first battery data, abnormality detection is performed to obtain second battery data.

[0027] Preferably, performing data ratio balancing processing to obtain fourth battery data according to the third battery data and a preset first threshold value includes:

[0028] Classify and count the third battery data to obtain the data quantity of the third battery data of each data type;

[0029] If the amount of data of a data type is greater than or equal to a preset first threshold, deleting a first fixed amount of data from the third battery data corresponding to the data type to obtain first under-sampled data;

[0030] If the data quantity of a data type is less than a preset first threshold, adding a second fixed quantity of data to the third battery data corresponding to the data type to obtain first oversampled data;

[0031] Fourth battery data is acquired according to the first over-sampling data and the first under-sampling data.

[0032] Preferably, performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data based on the pre-processed battery data includes:

[0033] According to the preprocessed battery data, a Pearson correlation coefficient is calculated to obtain a correlation coefficient between the battery characteristic and the battery capacity attenuation;

[0034] According to the battery characteristic, the correlation coefficient and the preset second threshold, a battery capacity decay correlation analysis is performed to obtain first battery capacity decay characteristic data;

[0035] According to the battery characteristics and a pre-built battery capacity decay model, a feature importance assessment is performed to obtain second battery capacity decay feature data;

[0036] Battery capacity decay characteristic data is acquired according to the first battery capacity decay characteristic data and the second battery capacity decay characteristic data.

[0037] Preferably, performing battery capacity attenuation correlation analysis to obtain first battery capacity attenuation characteristic data according to the battery characteristic, the correlation coefficient and a preset second threshold value includes:

[0038] If the correlation coefficient is greater than the preset second threshold, the battery feature corresponding to the correlation coefficient is a battery capacity attenuation feature;

[0039] According to the battery capacity attenuation characteristic, first battery capacity attenuation characteristic data is obtained.

[0040] Preferably, obtaining the battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle data and a pre-built battery capacity attenuation model comprises:

[0041] Horizontally splicing the battery capacity attenuation feature data and the vehicle data by columns to obtain a fusion feature matrix;

[0042] According to the fused feature matrix and the pre-built battery capacity attenuation model, a battery capacity retention rate evaluation result is obtained.

[0043] In a second aspect, an embodiment of the present invention provides a vehicle battery capacity retention rate evaluation device, the device comprising:

[0044] A data acquisition module, which acquires vehicle data, wherein the vehicle data includes vehicle operation data and battery data;

[0045] A preprocessing module, preprocessing the battery data to obtain preprocessed battery data;

[0046] A correlation analysis module, performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data based on the pre-processed battery data;

[0047] The model analysis module obtains a battery capacity retention rate evaluation result according to the battery capacity decay characteristic data, the vehicle data and a pre-constructed battery capacity decay model.

[0048] In a third aspect, an embodiment of the present invention provides a vehicle networking system, comprising: at least one cloud server and at least one vehicle, wherein the vehicle is used to connect to the cloud server to obtain vehicle data, and the cloud server is used to implement the method described in any one of claims 1-7 when processing vehicle data.

[0049] In a fourth aspect, an embodiment of the present invention provides a storage medium having computer program instructions stored thereon, which implements the method of the first aspect in the above-mentioned embodiment when the computer program instructions are executed by a processor.

[0050] In summary, the beneficial effects of the present invention are as follows:

[0051] The embodiments of the present invention provide a vehicle battery capacity retention rate assessment method, device, equipment and storage medium, which obtain vehicle data and make more scientific and accurate decisions by analyzing the operating data. Monitoring the vehicle data can identify abnormal battery conditions, thereby improving the safety and health of the battery. The battery data is preprocessed to obtain preprocessed battery data. Data cleaning can improve the accuracy and reliability of the data and reduce errors. After processing missing values ​​and abnormal values, more accurate data information can be provided to help managers make more effective decisions. The standardized and processed data is easier to understand and interpret, providing high-quality basic data for subsequent analysis. Process the battery data, perform a battery capacity decay correlation analysis to obtain the battery capacity decay characteristic data, and obtain the first battery characteristic closely related to the battery capacity decay by performing a correlation analysis on the pre-processed battery data, which can not only provide key input for the subsequent battery capacity retention rate prediction model, but also be used for real-time monitoring and optimization of battery management; obtain the battery capacity retention rate evaluation result based on the battery capacity decay characteristic data, the vehicle data and the pre-built battery capacity decay model, and the battery capacity decay characteristic data alone is not enough to fully reflect the health status of the battery. By combining with vehicle data, it can provide more comprehensive information, thereby improving the model's prediction accuracy for the battery capacity retention rate. By combining a variety of vehicle data and battery characteristics, the model can not only adapt to a variety of types of vehicles, but also adapt to different battery types, improving the prediction performance in a variety of application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the drawings required for use in the embodiment of the present invention will be briefly introduced below. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work, and these are all within the protection scope of the present invention.

[0053] Figure 1 The present invention is a flowchart of a method for evaluating the capacity retention rate of a vehicle battery according to an embodiment of the present invention.

[0054] Figure 2 It is another flow chart of a method for evaluating the capacity retention rate of a vehicle battery according to an embodiment of the present invention.

[0055] Figure 3 It is another flow chart of a method for evaluating the capacity retention rate of a vehicle battery according to an embodiment of the present invention.

[0056] Figure 4 It is another flow chart of a method for evaluating the capacity retention rate of a vehicle battery according to an embodiment of the present invention.

[0057] Figure 5It is a structural schematic diagram of a vehicle battery capacity retention rate evaluation device according to an embodiment of the present invention.

[0058] Figure 6 It is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The features and exemplary embodiments of various aspects of the present invention will be described in detail below. In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and Examples. It should be understood that the specific embodiments described herein are only configured to explain the present invention and are not configured to limit the present invention. For those skilled in the art, the present invention can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present invention by illustrating examples of the present invention.

[0060] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0061] It should be noted that all actions of acquiring signals, information or data in the present invention are performed in compliance with the corresponding data protection laws and policies of the location and with the authorization given by the owner of the corresponding device.

[0062] Example 1

[0063] See also Figure 1 The embodiment of the present invention provides a vehicle battery capacity retention rate evaluation method, device, equipment and medium, the method comprising:

[0064] S1. Obtain vehicle data;

[0065] Real-time understanding of the vehicle's operating status and battery performance, timely detection of potential problems, and analysis of operating data can make more scientific and accurate decisions. Monitoring vehicle data can identify abnormal behavior, thereby improving safety and reducing the risk of accidents caused by battery degradation. Real-time battery data helps optimize battery efficiency and extend battery life. Through data analysis, vehicle energy efficiency can be further improved.

[0066] Specifically, sensors and data processing equipment are installed on the vehicle to collect required vehicle data in real time. The vehicle data includes vehicle operation data and battery data. The vehicle operation data at least includes vehicle type, user usage habits and vehicle maintenance records. The battery data at least includes number of cycles, charge and discharge rate, charge and discharge depth, ambient temperature, altitude changes in the environment, battery type, and SOH.

[0067] S2. preprocessing the battery data to obtain preprocessed battery data, wherein the preprocessed battery data includes a number of battery features;

[0068] Data cleaning preprocessing can improve the accuracy and reliability of data, lay a solid foundation for subsequent analysis, reduce errors, and provide more accurate data information after processing missing and outliers, helping managers make more effective decisions. Standardized and processed data is easier to understand and interpret, which helps to discover potential patterns in the data and provides high-quality basic data for subsequent analysis.

[0069] Preferably, see Figure 2 , performing data cleaning preprocessing on the battery data, and obtaining the preprocessed battery data includes:

[0070] S21, performing missing data processing on the battery data to obtain first battery data;

[0071] By processing missing values, we can ensure the integrity of the data set and reduce bias in analysis. A complete data set can provide more reliable analysis results and provide effective support for business decisions. The processed data is easier to analyze and model, reducing the complexity of subsequent steps. The unified processing of missing values ​​can maintain the consistency and comparability of the data.

[0072] Specifically, we traverse the battery data and use Python's Pandas library to check whether there are missing values ​​in each column, and record the number and proportion of missing values. If the proportion of missing values ​​in a record exceeds 30% of the total, it will be deleted. For records with a missing value proportion below 30%, the mean interpolation method is used to fill in the missing values. The missing values ​​are replaced with the mean of the column to which the missing values ​​belong. After filling, the data set is checked again to confirm that all missing values ​​have been processed.

[0073] S22, performing anomaly detection according to the first battery data to obtain second battery data;

[0074] By detecting the hidden anomalies that exist, eliminating data that does not meet expectations, and better capturing the real pattern of the data when training the model, the generalization ability of the model is improved, so that the processed first battery data is more in line with the actual situation, and can provide more effective support for subsequent analysis. These hidden anomalies may not be very obvious, but they may have a negative impact on the training and prediction results of the model. Identifying potential hidden anomalies can effectively reduce such negative impacts.

[0075] S22-1. Performing statistical analysis on the first battery data to obtain an abnormality score;

[0076] The Z-score method is a statistical technique used to detect outliers and standardize data. The Z-score represents the difference between a data point and the mean of a data set. Compared with other statistical techniques, the Z-score method is highly adaptable and can be used in various types of data analysis, including outlier detection, data standardization, and statistical analysis. It can also be used in combination with other statistical methods to improve the effect of the model.

[0077] Specifically, according to the first battery data, the Z-score method is used to calculate the Z-score of each data point to obtain the abnormality score. The specific calculation formula is as follows:

[0078]

[0079] Where Z is the Z-score, X is the sample value to be calculated, μ is the mean of the sample, and σ is the standard deviation of the sample.

[0080] S22-2, if the abnormal score is greater than a fixed value, obtaining the abnormal score data point as a hidden abnormal point;

[0081] Capturing hidden anomalies helps ensure the integrity of the data cleaning process and the stability of the model, especially when these hidden anomalies have a significant impact on subsequent analysis or machine learning models.

[0082] If the anomaly score of the implicit anomaly point is greater than a fixed value, the data point is obtained as a implicit anomaly point, wherein the fixed value is determined according to the actual situation and is not limited here.

[0083] S22-3. Obtain a first abnormal feature according to the implicit abnormal point;

[0084] Specifically, according to the implicit abnormal point, a corresponding battery feature is obtained and recorded as a first abnormal feature.

[0085] S22-4. Obtain first upper and lower limits based on the standard deviation according to the first abnormal feature;

[0086] Implicit outliers may cause data overflow or data instability, affecting the convergence of the model. By correcting the outliers, the overall quality of the data set can be improved. After correcting the outliers, the first abnormal feature is corrected to the second abnormal feature, providing a basis for subsequent data analysis and decision-making, and improving the computational stability of the model.

[0087] Specifically, upper and lower limits are set for the first abnormal feature. Based on the standard deviation, the upper and lower limits are set to the mean plus or minus a certain multiple of the standard deviation. The specific setting method is as follows:

[0088] Lower limit: L = μ-k·σ

[0089] Upper limit: U = μ + k·σ

[0090] Among them, k is a constant multiple, which depends on the actual situation and is not limited here. μ is the mean and σ is the standard deviation.

[0091] S22-5, performing anomaly correction according to the first abnormality feature and the first upper and lower limits to obtain a second abnormality feature;

[0092] When the actual data exceeds the first upper and lower limits, it means that the data is abnormal and exceeds the normal fluctuation range. Correcting this abnormal data can make the data more consistent with the actual situation and ensure the accuracy of data analysis or model training.

[0093] Specifically, actual data of the first abnormal feature is obtained. If the actual data exceeds the first upper and lower limits, the first abnormal feature corresponding to the data is obtained as the second abnormal feature.

[0094] S22-6, performing anomaly detection to obtain second battery data according to the second abnormal feature and the first battery data;

[0095] Specifically, an abnormality detection is performed on the first battery data, data meeting the second abnormality feature is identified and excluded, and the second battery data is acquired.

[0096] S23, performing data smoothing processing on the second battery data to obtain third battery data;

[0097] Smoothing can remove short-term fluctuations and noise, making data easier to understand and analyze. Smoothed data can improve the performance of the model. The weighted moving average method can make the model more robust and accurate in prediction. The weight distribution strategy can be adjusted according to specific needs to make the model more targeted and suitable for different data characteristics. By emphasizing recent data, the weighted moving average can more effectively capture short-term trends, thereby improving the accuracy of short-term predictions and making it easier to identify outliers in the data, thereby improving the evaluation effect of battery capacity retention rate.

[0098] Specifically, the window size k is selected, where k is a constant and is set according to the specific situation. It is not limited here. The average value of each second battery data point is calculated. The calculation formula is as follows:

[0099]

[0100] Wherein, w is the weight value corresponding to the data, and a weight value is assigned to each data point. The weight value is set according to the specific situation to obtain the third battery data after smoothing.

[0101] S24, performing data ratio balancing processing according to the third battery data and the preset first threshold to obtain fourth battery data;

[0102] In real-world applications, data sets often have class imbalances, where the number of data samples in some categories may be much larger than that in other categories. This imbalance can cause the model to favor the majority class and ignore the minority class during training, thus affecting the model's generalization ability and accuracy.

[0103] By balancing the proportions of the data set, the model’s bias towards the majority class is reduced, enabling it to better learn the characteristics of each category and improve the model’s predictive performance when processing unbalanced data sets, especially in its application in vehicle battery capacity retention rate evaluation models.

[0104] Specifically, for the third battery data, different methods are selected to perform data balancing processing based on characteristics of the data.

[0105] Optionally, the third battery data is classified and classified and counted to obtain the data quantity of the third battery data of each data type; if the data quantity of the third battery data category is greater than a preset first threshold, an undersampling method is used to delete a first fixed number of data from the data of the category so that its quantity is consistent with the number of category samples less than or equal to the preset first threshold within an error range, and the preset first threshold and the error range are determined according to the actual number of samples and are not limited here, and the first undersampling data is obtained; if the data quantity of the third battery data category is less than the preset first threshold, an oversampling method is used to add a second fixed number of data from the data of the category so that its quantity is consistent with the number of category samples greater than or equal to the preset first threshold within an error range, and the preset first threshold and the error range are determined according to the actual number of samples and are not limited here, and the first oversampling data is obtained, and finally the first undersampling data, the first oversampling data and the original sample are merged to obtain the fourth battery data.

[0106] S25, performing data normalization according to the fourth battery data to obtain fifth battery data;

[0107] Unnormalized data may lead to large scale differences between features, which in turn requires more iterations to find the optimal solution during the optimization process. If the range of some feature values ​​is particularly large, while the range of other features is very small, the model will tend to rely on features with larger values, resulting in an imbalance in feature weights. Normalization can make each feature in the same scale range, which can effectively avoid this deviation, prevent the model from overfitting on certain features, and improve the generalization ability of the model. In practical applications, the feature value ranges of cycle number, charge and discharge rate, charge and discharge depth, ambient temperature, altitude change of the environment, battery type and SOH are inconsistent. After normalization, the data is ensured to be in a unified range, the relative size relationship of the original data is retained, and the performance of the model is avoided from being reduced due to differences in the scale of different features. The Min-Max normalization method is suitable for data distributed within a fixed range and can maintain the relative proportion of the original data.

[0108] Specifically, the Min-Max normalization method scales all features to the range of [0,1] to eliminate the impact of numerical differences between features on the model. The calculation formula of this method is as follows:

[0109]

[0110] Among them, Xnorm is the normalized eigenvalue, X is the original data, and Xmin and Xmax are the minimum and maximum values ​​of the feature respectively.

[0111] Min-Max normalization is performed on each feature, and calculations are performed one by one according to the formula to finally obtain the fifth battery data.

[0112] S26. Performing data dimension reduction according to the fifth battery data to obtain sixth battery data;

[0113] By reducing the dimension of the data, the computational burden of subsequent processing is reduced. The PCA method can help remove the redundancy between features and retain the most important information. The reduced-dimensional data can be more easily visualized, helping to analyze the intrinsic structure of the data, and has good advantages in linear data processing. The t-SNE method is suitable for complex nonlinear data and can effectively reveal the intrinsic structure of the data. It retains the similarity between points in the high-dimensional space in the low-dimensional space, making the clustering structure more obvious. t-SNE provides excellent data visualization and is suitable for displaying classification and clustering results.

[0114] Specifically, the PCA method is first used for preliminary dimensionality reduction, and the covariance matrix of the fifth battery data is calculated to capture the linear relationship between the features; the covariance matrix is ​​decomposed to obtain the eigenvalues ​​and corresponding eigenvectors; the first k largest eigenvalues ​​and their corresponding eigenvectors are selected to form a new feature space, where k is a constant; the data is projected into the new feature space to obtain the reduced-dimensional data. Then, t-SNE is used to process nonlinear relationships, and the similarity between each point in the reduced-dimensional data and other points is calculated, represented by Gaussian distribution, and t distribution is used in low-dimensional space to retain the similarity between data points. The points in the low-dimensional space are optimized by gradient descent, so that similar points are closer in the low-dimensional space, and the results of classification and clustering are displayed.

[0115] S27, constructing a metric space according to the sixth battery data to obtain seventh battery data;

[0116] Choosing an appropriate distance metric can more accurately capture the similarities and differences between data points. Using a reasonable distance metric can make the analysis more flexible and provide more meaningful feature representation for subsequent model training.

[0117] Specifically, the Euclidean distance is used as the distance metric to calculate the distance matrix between data points. The Euclidean distance calculation formula is as follows:

[0118]

[0119] In the new metric space, each data point is represented by its eigenvector after dimensionality reduction, and the distance matrix is ​​used to reflect the similarity between data points to obtain the seventh battery data.

[0120] S28, performing data standardization according to the seventh battery data to obtain pre-processed battery data;

[0121] After limiting the feature range of the fifth battery data, data standardization is performed according to the different distribution characteristics of the features to ensure that the preprocessing of battery data with complex feature distribution and containing outliers can achieve the expected effect, which is helpful for subsequent data analysis. The RobustScaler method can scale and center using the median and quartile range of the data, and is insensitive to outliers. Therefore, it is more robust when there are outliers in the data and can effectively handle complex data distribution.

[0122] Specifically, the median of each feature of the data is subtracted to center the data to zero, and the data is scaled using the interquartile range (IQR) to obtain the preprocessed battery data. IQR refers to the difference between the 75th percentile (Q3) and the 25th percentile (Q1) of the data, and the standardized formula is as follows:

[0123]

[0124] Among them, quartiles means arranging the data from small to large and then dividing it into four equal parts. Each part contains 25% of the data. Q1 (first quartile) represents the points in the first 25% of the data set; Q2 (second quartile or median) represents the median of the data set, that is, the point at the 50% position; Q3 (third quartile) represents the points in the first 75% of the data set. IQR (X) = Q3-Q1 represents the interquartile range.

[0125] S3. Performing battery capacity attenuation correlation analysis based on the pre-processed battery data to obtain battery capacity attenuation characteristic data;

[0126] By performing correlation analysis on the pre-processed battery data, the correlation coefficient closely related to the battery characteristics and battery capacity decay can be effectively obtained. It can not only provide key input for the subsequent battery capacity retention rate prediction model, but also be used for real-time monitoring and optimization of battery management.

[0127] Preferably, see Figure 3 , performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data based on the pre-processed battery data includes:

[0128] S31. Calculating the Pearson correlation coefficient according to the preprocessed battery data to obtain a correlation coefficient between the battery characteristic and the battery capacity attenuation;

[0129] A commonly used correlation analysis method is the Pearson correlation coefficient, which is used to measure the linear correlation between two variables. The correlation coefficient associated therewith is determined by calculating the correlation between each of the battery characteristics and the battery attenuation capacity.

[0130] Specifically, the Pearson correlation coefficient calculation formula is as follows:

[0131]

[0132] Among them, r represents the Pearson correlation coefficient, Xi is the eigenvalue, Yi is the target variable, is the mean of the eigenvalues, is the battery attenuation capacity, n is the number of pre-processed battery data, the range of the correlation coefficient is [-1,1], r=1 indicates a completely positive correlation, r=0 indicates no linear correlation, and r=-1 indicates a completely negative correlation, thereby determining the correlation coefficient related to the battery attenuation capacity.

[0133] S32, performing battery capacity attenuation correlation analysis to obtain first battery capacity attenuation characteristic data according to the battery characteristic, the correlation coefficient and a preset second threshold value;

[0134] By using a preset threshold to filter out features that are highly correlated with battery capacity decay, and eliminating features that are irrelevant or weakly correlated with battery capacity decay, the model's prediction accuracy for battery capacity decay is improved. Retaining highly correlated features can avoid overfitting of the model, thereby improving the performance of the model on actual operating data and improving the robustness of the prediction.

[0135] Specifically, the preset second threshold is 0. If the correlation coefficient is greater than the preset second threshold, the battery feature corresponding to the correlation coefficient is a battery capacity attenuation-related feature, and battery capacity attenuation feature data is obtained.

[0136] S33, performing feature importance evaluation according to the battery feature and a pre-built battery capacity decay model to obtain second battery capacity decay feature data;

[0137] S33-1, inputting the battery characteristics into each preset model to obtain an evaluation result, wherein the preset models include a random forest model, a GBoost model, and a LightGBM model;

[0138] Specifically, the battery features are input into the random forest model, GBoost model and LightGBM model for training, and the mean square error (MSE) and determination coefficient (R 2)Evaluate the performance of each model and obtain the evaluation results. A low mean square error means that the model has a small difference from the actual value when predicting, indicating that the model has strong predictive ability. A high determination coefficient means that the model can explain most of the variability of the data, indicating the effectiveness of the model.

[0139] S33-2, determining a battery capacity attenuation model according to the evaluation result;

[0140] Specifically, from the perspective of data set size, the battery feature data set belongs to small and medium-sized data sets, while GBoost and LightGBM perform better when processing large-scale data sets. For small and medium-sized data, the simplicity and effectiveness of random forests make it a reasonable choice; from the perspective of computational efficiency, random forests use the parallelization capability of multi-core processors to achieve higher computational efficiency, which is higher than the other two models; random forests provide intuitive and easy-to-understand feature importance evaluation, such as the Gini coefficient evaluation index, which can identify key features related to the target variable. Although GBoost and LightGBM also provide feature importance, the model is more complex and requires more explanation and understanding; the random forest model is relatively simple, easy to implement and explain, and has low model complexity, while GBoost and LightGBM involve more parameter adjustments and learning processes, which may lead to more time for model tuning; the random forest model has a lower risk of overfitting by integrating multiple decision trees, can effectively handle high-dimensional features, and reduce overfitting. According to the evaluation results, random forests have obvious advantages over the other two models in performance evaluation, so the random forest model is selected as the battery capacity decay model.

[0141] S33-3, performing feature importance evaluation according to the battery feature and the battery capacity attenuation model to obtain second battery capacity attenuation feature data;

[0142] The Pearson correlation coefficient has a good effect on linear correlation features, but it cannot accurately capture nonlinear correlation features. Therefore, the random forest model is used to evaluate the importance of nonlinear correlation features.

[0143] Specifically, the battery features are evaluated for feature importance using a pre-built random forest model, which can effectively select the features most relevant to battery capacity attenuation. The battery features are trained. During the training process, the random forest will construct multiple decision trees, each of which is trained on a randomly selected feature subset. The Gini coefficient is used to evaluate the purity of the nodes in the classification model. In the decision tree, the smaller the Gini coefficient, the more uniform the distribution of sample categories in the node. The Gini coefficient is calculated for each tree, and the importance of the feature is evaluated by accumulating the contribution of each feature. The Gini coefficient calculation formula is:

[0144]

[0145] Where D is the data set, C is the number of categories, and pi is the proportion of samples belonging to the i-th category in the data set. The value range of the Gini coefficient is [0,1]. The smaller the value, the more likely the samples are to belong to the same category.

[0146] Calculate the Gini coefficient of the current node; then evaluate each feature and threshold and calculate the Gini coefficient after each split; finally select the split that results in the smallest Gini coefficient as the best split for the node.

[0147] At the same time, combined with the visualization results of t-SNE, the data points in the high-dimensional feature space are projected into the two-dimensional space to observe the distribution and clustering of the data points, and analyze the relationship between the features and clusters. According to the clustering results visualized by t-SNE, identify which features appear frequently in the same cluster and are highly important features in the random forest. Using the guidance provided by the importance of random forest features, further analyze the significance of different clusters in t-SNE, identify which features have a key impact on the formation of these clusters and the battery attenuation pattern, so as to more intuitively identify and understand the features that contribute most to battery capacity attenuation. t-SNE helps to reveal possible nonlinear relationships and complex interaction effects in the data, which may be ignored by linear models or traditional analysis methods. Combining the results of the two can explain the features that affect battery capacity attenuation, providing an improvement in model interpretability.

[0148] S34, acquiring battery capacity attenuation characteristic data according to the first battery capacity attenuation characteristic data and the second battery capacity attenuation characteristic data;

[0149] Specifically, the two sets of feature data are merged, and the Pandas library is used to complete the merging operation to obtain the battery capacity attenuation feature data.

[0150] S4. Obtaining a battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle data and a pre-built battery capacity attenuation model;

[0151] Battery capacity decay characteristic data alone is not enough to fully reflect the health of the battery. By combining it with vehicle data, it can provide more comprehensive information, thereby improving the model's prediction accuracy for battery capacity retention. By combining multiple vehicle data and battery characteristics, the model can not only adapt to multiple types of vehicles, but also different battery types, improving prediction performance in multiple application scenarios.

[0152] Preferably, see Figure 4 , obtaining a battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle data and a pre-built battery capacity attenuation model includes:

[0153] S41, horizontally splicing the battery capacity attenuation characteristic data and the vehicle data by column to obtain a fusion characteristic matrix;

[0154] The fused data can provide more comprehensive information, which can help the model better understand the battery degradation pattern, thereby improving the accuracy of predicting the battery capacity retention rate. It can capture the potential relationship between battery use and vehicle operation, and improve the model's prediction effect.

[0155] Specifically, the battery capacity attenuation feature data and the vehicle data are horizontally spliced ​​by column, and the data are used as different columns to obtain a complete fusion feature matrix.

[0156] S42. Obtain a battery capacity retention rate evaluation result based on the fused feature matrix and a pre-constructed battery capacity decay model.

[0157] By analyzing the vehicle's operating data and battery characteristics, the system establishes a battery capacity attenuation model for brand models, which can accurately predict the future battery capacity retention rate and effectively evaluate the battery health status. Random Forest is an ensemble learning machine learning model based on decision trees, which is used for classification and regression tasks. The model builds multiple decision trees for prediction and then integrates the results to improve the accuracy, robustness and generalization ability of the model, and can effectively handle data with nonlinear relationships. Since the training data and features of each tree are randomly sampled, the risk of overfitting is reduced. Based on the integration of multiple trees, even if some of the decision trees are affected by outliers, the prediction results of the overall model will not be significantly affected, especially in regression tasks. The Bootstrap sampling method is a resampling technique that is often used in statistical analysis and machine learning model training, especially in ensemble learning algorithms such as random forests. Its core idea is to sample with replacement from the original data set, generate multiple different sample sets, and train the model based on these sample sets. This method can effectively reduce the overfitting problem and improve the robustness of the model.

[0158] Specifically, the fused feature matrix data is divided into a training set and a test set with a data division ratio of 8:2, and is input into the random forest model for training. For each decision tree, the Bootstrap sampling method is used to randomly extract samples from the training set for training. Each node calculates the best split point based on the randomly selected features. In order to avoid overfitting, the maximum depth and number of leaf nodes of each tree are limited, which can prevent the tree from becoming too complex. The above steps are repeated until all decision trees are trained. The number of decision trees represents the number of trees to be built in the random forest. Increasing the number of trees can improve the stability of the model. The maximum depth limits the maximum depth of each tree, which can effectively prevent overfitting. Similarly, the fused feature matrix data is input into the GBoost model and the LightGBM model, and the stacked generalization method is used to take the output results of the three models as new features. The stacked generalization method is an integrated learning method that improves the overall prediction performance of a single model by taking the prediction results of multiple models as input. In order to avoid overfitting, K-fold cross validation is used to generate new features to improve the prediction and generalization capabilities of the random forest model. After feature fusion is completed, it is input into the random forest model. After training, the model outputs the battery capacity retention rate evaluation result, which indicates the capacity retention rate of the battery at present or in the future, indicating the remaining life of the battery. The evaluation index includes at least the mean square error (MSE), which is used to evaluate the gap between the predicted result and the true value. The mean square error calculation formula is as follows:

[0159]

[0160] Where n is the total number of samples, yi is the actual value of the ith sample, and y^i is the predicted value of the ith sample. For each sample, calculate the difference between the predicted value and the actual value, i.e., yi-y^i; square each residual, i.e., (yi-y^i) 2 ; Finally, sum all squared residuals and divide by the total number of samples n.

[0161] Example 2

[0162] See also Figure 5 The embodiment of the present invention provides a vehicle battery capacity retention rate evaluation device based on embodiment 1, the device comprising:

[0163] A data acquisition module, which acquires vehicle data;

[0164] A preprocessing module, preprocessing the battery data to obtain preprocessed battery data;

[0165] A correlation analysis module, performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data based on the pre-processed battery data;

[0166] The model analysis module obtains a battery capacity retention rate evaluation result according to the battery capacity decay characteristic data, the vehicle data and a pre-constructed battery capacity decay model.

[0167] The device can make more scientific and accurate decisions by analyzing the operation data. Monitoring the vehicle data can identify the abnormal state of the battery, thereby improving the safety and health of the battery. The battery data is preprocessed to obtain preprocessed battery data. Data cleaning can improve the accuracy and reliability of the data and reduce errors. After processing missing and abnormal values, more accurate data information can be provided to help managers make more effective decisions. The standardized and processed data is easier to understand and interpret, providing high-quality basic data for subsequent analysis. According to the preprocessed battery data, a battery capacity decay correlation analysis is performed to obtain battery capacity decay characteristic data. By performing a correlation analysis on the preprocessed battery data, a first battery feature closely related to the battery capacity decay is obtained, which can not only provide key inputs for the subsequent battery capacity retention rate prediction model, but also be used for real-time monitoring and optimization of battery management. According to the battery capacity decay characteristic data, the vehicle data and the pre-constructed battery capacity decay model, a battery capacity retention rate evaluation result is obtained. The battery capacity decay characteristic data alone is not enough to fully reflect the health status of the battery. By combining with the vehicle data, more comprehensive information can be provided, thereby improving the prediction accuracy of the model for the battery capacity retention rate. By combining multiple vehicle data and battery characteristics, the model can not only adapt to multiple types of vehicles, but also adapt to different battery types, improving the prediction performance in various application scenarios.

[0168] It should be noted that the modules and units in a vehicle battery capacity retention rate evaluation device in this embodiment correspond one-to-one to the steps in a vehicle battery capacity retention rate evaluation method in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned vehicle battery capacity retention rate evaluation method, which will not be repeated here.

[0169] Example 3

[0170] In addition, combined Figure 5 A vehicle battery capacity retention rate evaluation method described in an embodiment of the present invention may be implemented by an electronic device. Figure 6 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention is shown.

[0171] The electronic device may include a processor and a memory storing computer program instructions.

[0172] Specifically, the processor may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present invention.

[0173] The memory may include a large capacity memory for data or instructions. By way of example and not limitation, the memory may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. Where appropriate, the memory may include a removable or non-removable (or fixed) medium. Where appropriate, the memory may be inside or outside a data processing device. In a particular embodiment, the memory is a non-volatile solid-state memory. In a particular embodiment, the memory includes a read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM) or a flash memory or a combination of two or more of these.

[0174] The processor implements any one of the vehicle battery capacity retention rate evaluation methods in the above embodiments by reading and executing computer program instructions stored in the memory.

[0175] In one example, the electronic device may further include a communication interface and a bus. Figure 6 As shown, the processor 401, the memory 402, and the communication interface 403 are connected via a bus 410 and communicate with each other.

[0176] The communication interface is mainly used to implement communication between the modules, devices, units and / or equipment in the embodiments of the present invention.

[0177] Bus includes hardware, software or both, and the parts of electronic equipment are coupled to each other.For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In suitable cases, bus may include one or more buses. Although the embodiment of the present invention describes and shows a specific bus, the present invention considers any suitable bus or interconnection.

[0178] Example 4

[0179] In addition, combined Figure 5 The vehicle battery capacity retention rate evaluation method described in the embodiment of the present invention can be implemented by a vehicle networking system. The vehicle networking system includes: at least one cloud server and at least one vehicle.

[0180] Specifically, the vehicle is used to connect to the cloud server to obtain vehicle data, wherein the vehicle data includes vehicle operation data and battery data.

[0181] The cloud server reads and processes the vehicle data to implement any one of the vehicle battery capacity retention rate evaluation methods in the above embodiments.

[0182] Example 5

[0183] In addition, in combination with a vehicle battery capacity retention rate evaluation method in the above embodiment, an embodiment of the present invention can provide a computer-readable storage medium for implementation. The computer-readable storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any vehicle battery capacity retention rate evaluation method in the above embodiment is implemented.

[0184] In summary, the embodiments of the present invention provide a method, device, equipment and medium for evaluating the capacity retention rate of a vehicle battery.

[0185] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0186] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present invention are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0187] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0188] The above is only a specific implementation of the present invention. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the system, module and unit described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. It should be understood that the protection scope of the present invention is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be covered within the protection scope of the present invention.

Claims

1. A method for evaluating the capacity retention rate of a vehicle battery, characterized in that: The method comprises: Acquiring vehicle data, wherein the vehicle data includes vehicle operation data and battery data; Preprocessing the battery data to obtain preprocessed battery data, wherein the preprocessed battery data includes: a number of battery characteristic information; According to the pre-processed battery data, a battery capacity attenuation correlation analysis is performed to obtain battery capacity attenuation characteristic data; A battery capacity retention rate evaluation result is obtained based on the battery capacity decay characteristic data, the vehicle operation data and a pre-constructed battery capacity decay model.

2. The vehicle battery capacity retention rate evaluation method according to claim 1, characterized in that: The preprocessing of the battery data to obtain the preprocessed battery data includes: Performing missing data processing on the battery data to obtain first battery data; Performing anomaly detection according to the first battery data to obtain second battery data; Performing data smoothing processing on the second battery data to obtain third battery data; Performing data ratio balancing processing according to the third battery data and a preset first threshold to obtain fourth battery data; Normalizing the data according to the fourth battery data to obtain fifth battery data; Performing data dimension reduction according to the fifth battery data to obtain sixth battery data; According to the sixth battery data, construct a metric space to obtain seventh battery data; According to the seventh battery data, data standardization is performed to obtain pre-processed battery data.

3. The vehicle battery capacity retention rate evaluation method according to claim 2, characterized in that: The performing abnormality detection to obtain second battery data according to the first battery data includes: Performing statistical analysis based on the first battery data to obtain an abnormality score; If the abnormal score is greater than a fixed value, the abnormal score data point is obtained as a hidden abnormal point; According to the hidden abnormal point, a first abnormal feature is obtained; According to the first abnormal feature, obtaining first upper and lower limits based on the standard deviation; According to the first abnormal feature and the first upper and lower limits, abnormal correction is performed to obtain a second abnormal feature; According to the second abnormal feature and the first battery data, abnormality detection is performed to obtain second battery data.

4. The vehicle battery capacity retention rate evaluation method according to claim 2, characterized in that: The performing data ratio balancing processing to obtain fourth battery data according to the third battery data and the preset first threshold value includes: Classify and count the third battery data to obtain the data quantity of the third battery data of each data type; If the amount of data of a data type is greater than or equal to a preset first threshold, deleting a first fixed amount of data from the third battery data corresponding to the data type to obtain first under-sampled data; If the data quantity of a data type is less than a preset first threshold, adding a second fixed quantity of data to the third battery data corresponding to the data type to obtain first oversampled data; Fourth battery data is acquired according to the first over-sampling data and the first under-sampling data.

5. The vehicle battery capacity retention rate evaluation method according to claim 1, characterized in that: The performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data according to the pre-processed battery data comprises: According to the preprocessed battery data, a Pearson correlation coefficient is calculated to obtain a correlation coefficient between the battery characteristic and the battery capacity attenuation; According to the battery characteristic, the correlation coefficient and the preset second threshold, a battery capacity decay correlation analysis is performed to obtain first battery capacity decay characteristic data; According to the battery characteristics and a pre-built battery capacity decay model, a feature importance assessment is performed to obtain second battery capacity decay feature data; Battery capacity decay characteristic data is acquired according to the first battery capacity decay characteristic data and the second battery capacity decay characteristic data.

6. The vehicle battery capacity retention rate evaluation method according to claim 5, characterized in that: The performing battery capacity attenuation correlation analysis to obtain first battery capacity attenuation characteristic data according to the battery characteristic, the correlation coefficient and the preset second threshold value comprises: If the correlation coefficient is greater than the preset second threshold, the battery feature corresponding to the correlation coefficient is a battery capacity attenuation feature; According to the battery capacity attenuation characteristic, first battery capacity attenuation characteristic data is obtained.

7. The vehicle battery capacity retention rate evaluation method according to claim 1, characterized in that: The obtaining of the battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle data and the pre-built battery capacity attenuation model comprises: Horizontally splicing the battery capacity attenuation feature data and the vehicle data by columns to obtain a fusion feature matrix; According to the fused feature matrix and the pre-built battery capacity attenuation model, a battery capacity retention rate evaluation result is obtained.

8. A vehicle battery capacity retention rate evaluation device, characterized in that: The device comprises: A data acquisition module, which acquires vehicle data, wherein the vehicle data includes vehicle operation data and battery data; A preprocessing module, preprocessing the battery data to obtain preprocessed battery data; A correlation analysis module, performing battery capacity attenuation correlation analysis to obtain battery capacity attenuation characteristic data based on the pre-processed battery data; The model analysis module obtains a battery capacity retention rate evaluation result according to the battery capacity attenuation characteristic data, the vehicle operation data and a pre-constructed battery capacity attenuation model.

9. A vehicle networking system, characterized in that: include: At least one cloud server and at least one vehicle, the vehicle is used to connect to the cloud server to obtain vehicle data, and the cloud server is used to implement the method as described in any one of claims 1-7 when processing vehicle data.

10. A storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images