A lithium battery charging data monitoring method and system

Abstract

The present application relates to the field of data processing, and more particularly to a lithium battery charging data monitoring method and system. The method comprises the steps of: obtaining a feature sequence of historical charging data points, obtaining feature data of any dimension in all feature sequences of historical charging data points, calculating the category consistency of feature data of each dimension and feature data of other dimensions, determining the reference weight of feature data of each dimension according to the category consistency; obtaining the classification delicacy of feature data of each dimension, obtaining the time sequence weight adjustment coefficient of feature data of each dimension according to the classification delicacy; setting the weight of each clustering stage according to the time sequence weight adjustment coefficient and the reference weight, participating in the clustering iteration update with the weight of each clustering stage, completing the clustering processing, and realizing the anomaly detection of the charging data point at the current moment. By improving the accuracy of clustering, the accuracy of charging detection is further improved.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to a method and system for monitoring lithium battery charging data. Background Technology

[0002] With the rapid development of new energy technologies, lithium batteries, with their advantages of high energy density and long cycle life, are widely used in electric vehicles, energy storage systems, and portable electronic devices. The safe and efficient use of lithium batteries relies heavily on precise monitoring of the charging process. Timely detection of abnormal charging data points and subsequent corrective measures are crucial for extending battery life and ensuring the safe operation of equipment.

[0003] Currently, clustering analysis is a common technique for identifying abnormal data in lithium battery charging data point monitoring. Traditional clustering methods typically employ a single clustering strategy when processing lithium battery charging data points, either referencing only data points with a coarse classification dimension or relying solely on data points with a fine classification dimension. While referring only to data with a coarse classification dimension can quickly achieve macro-level clustering, the lack of comprehensive data feature information makes it difficult to accurately distinguish subtle differences, resulting in coarse clustering results that cannot effectively identify potential abnormal charging data points. On the other hand, relying solely on data with a fine classification dimension can achieve detailed clustering, but due to the high data dimensionality and computational complexity, it is prone to getting trapped in local optima during the clustering process, failing to obtain globally optimal clustering results and reducing the accuracy and reliability of abnormal data monitoring.

[0004] Furthermore, existing clustering analysis methods for lithium battery charging data points lack targeted processing of data characteristics at different stages of the clustering process, failing to fully leverage the advantages of data from different dimensions in clustering. This results in poor overall clustering performance, making it difficult to meet the growing demand for safe and efficient monitoring of lithium batteries. Therefore, there is an urgent need for a method that can improve clustering accuracy, effectively prevent getting trapped in local optima, and achieve comprehensive and accurate monitoring of lithium battery charging data points. Summary of the Invention

[0005] To address the problem of preventing getting trapped in local optima and achieving accurate classification, this invention provides a lithium battery charging data monitoring method and system.

[0006] In a first aspect, the present invention provides a method for monitoring lithium battery charging data, employing the following technical solution:

[0007] A method for monitoring lithium battery charging data, comprising the following steps:

[0008] Obtain the current charging data point of the lithium battery and the corresponding historical charging data point;

[0009] Obtain the feature sequence of historical charging data points, where the feature data in the feature sequence reflects the values ​​and trends of historical charging data points.

[0010] Feature data of any dimension is obtained from the feature sequence of all historical charging data points. The class consistency of the feature data of each dimension with the feature data of other dimensions is calculated. The baseline weight of the feature data of each dimension is determined based on the class consistency. The baseline weight is positively correlated with the class consistency. The classification detail of the feature data of each dimension is obtained. The classification detail represents the fineness of the classification. The temporal weight adjustment coefficient of the feature data of each dimension is obtained based on the classification detail. The weight adjustment coefficient is positively correlated with the classification detail.

[0011] The weights of each clustering stage are set according to the time-series weight adjustment coefficient and the baseline weight. The weights of each clustering stage are used in the clustering iteration update to complete the clustering process, so as to realize the anomaly detection of the charging data points at the current moment.

[0012] This invention introduces feature sequences to reflect the information on the value and trend of charging data points, thereby preventing situations where the value alone cannot accurately reflect the anomalies in lithium battery charging. Furthermore, it analyzes category consistency to reflect the contribution of each dimension of data to clustering, thus assigning greater weight to data with a high contribution, improving the accuracy of clustering. Further, it sets a time-series weight adjustment coefficient based on the level of classification detail to adjust the weights at each stage, thereby giving greater weight to data in dimensions with low classification detail in the early stages of clustering, achieving macro-level classification in the early stages; and giving greater weight to data in dimensions with high classification detail in the later stages of clustering, achieving more detailed classification in the later stages of clustering, preventing getting trapped in local optima, and improving classification accuracy.

[0013] Preferably, the step of obtaining the feature sequence of historical charging data points includes:

[0014] Obtain data for each dimension from historical charging data points, and obtain the historical charging data sequence corresponding to each dimension of the historical charging data points;

[0015] The Gaussian pyramid algorithm is used to process the historical charging data sequence to obtain several layers of processed data sequence. The data corresponding to each dimension of the historical charging data point in each layer of processed data sequence are obtained and recorded as the macro data of each dimension of the historical charging data point. The sequence of all dimensions of the historical charging data point, the slope of all dimensions of the historical charging data point, and the slope of all macro data of all historical charging data points is recorded as the feature sequence.

[0016] This invention utilizes Gaussian pyramids to process historical charging data sequences to obtain trend information under different macroscopic conditions, thereby extracting more comprehensive information on changes in charging data and providing an information basis for accurate anomaly detection.

[0017] Preferably, the calculation of the category consistency of feature data in each dimension with feature data in other dimensions includes:

[0018] Statistical histograms of the feature data for each dimension are obtained by performing statistical analysis on the feature data for each dimension.

[0019] The characteristic sequences corresponding to the characteristic data of each independent peak in the statistical histogram are recorded as the analytical characteristic sequences of each independent peak.

[0020] Calculate the consistency of the distribution of each independent peak with other dimensions;

[0021] The normalized value of the mean of the distribution consistency of all independent peaks in the statistical histogram of feature data in each dimension is used as the class consistency of feature data in each dimension with feature data in other dimensions.

[0022] This invention accurately reflects the consistency of data classification across different dimensions by introducing independent peaks and ensuring their distribution consistency with other dimensions.

[0023] Preferably, the calculation of the distribution consistency of each independent peak with other dimensions includes:

[0024]

[0025] in, This represents the probability that the analytical characteristic sequence of each independent peak is distributed in the statistical histogram of the characteristic data in the i-th dimension, and that it is located in the j-th independent peak. Represents the logarithmic function with base 10. This represents the number of independent peaks in the statistical histogram of the feature data in the i-th dimension. Indicates the number of dimensions of the feature sequence. This indicates that the distribution of each independent peak is consistent with that of other dimensions.

[0026] This invention reflects the central consistency of independent peaks with data from other dimensions by analyzing the information entropy of the distribution of characteristic sequences corresponding to characteristic data in other dimensions, thereby providing a basis for accurately analyzing the classification consistency across different dimensions.

[0027] Preferably, determining the baseline weights of feature data for each dimension based on category consistency includes:

[0028] The ratio of category consistency to the sum of category consistency across all dimensions is used as the baseline weight for the feature data of each dimension.

[0029] Preferably, the level of detail in obtaining the classification of feature data across each dimension includes:

[0030] The normalized value of the number of independent peaks in the statistical histogram of the feature data of each dimension is used as the classification detail of the feature data of each dimension.

[0031] Preferably, the step of obtaining the temporal weight adjustment coefficients for feature data of each dimension based on the level of classification detail includes:

[0032] The product of the classification detail and the preset adjustment coefficient is used as the temporal weight adjustment coefficient for the feature data of each dimension.

[0033] Preferably, the step of setting the weights for each clustering stage based on the time-series weight adjustment coefficient and the baseline weight, and participating the weights of each clustering stage in the clustering iteration update to complete the clustering process, includes:

[0034] The normalized value of the ratio of the baseline weight to the time-series weight adjustment coefficient is used as the weight of the first clustering stage. If the weight of the first clustering stage is less than the preset weight threshold, the feature data of that dimension is removed to obtain the adjusted feature sequence. The weight of the first clustering stage is used in the clustering iteration update. In response to the clustering stability being greater than the preset stability threshold, the normalized value of the product of the baseline weight and the time-series weight adjustment coefficient is used as the weight of the second clustering stage to participate in the clustering iteration update until the clustering cutoff condition is met, and the clustering is completed.

[0035] This invention utilizes a time-series weight adjustment coefficient to adjust the weights of data in different dimensions at each clustering stage. This results in data in dimensions with low classification detail having a larger weight in the early stages of clustering, thereby increasing the ability to grasp the macroscopic picture in the early stages of clustering and preventing getting trapped in local optima. Conversely, data in dimensions with high classification detail have a smaller weight in the later stages of clustering, thereby increasing the ability to grasp detailed information in the later stages of clustering and improving the accuracy of classification.

[0036] Preferably, the step of detecting anomalies in the charging data points at the current moment includes:

[0037] Obtain the feature sequence of the charging data point at the current moment, input the feature sequence of the charging data point at the current moment into the clustering result to obtain the category to which the charging data point at the current moment belongs, and judge the abnormal situation based on the distance between the charging data point at the current moment and the cluster center of the category to which it belongs.

[0038] Secondly, the present invention provides a lithium battery charging data monitoring system, which adopts the following technical solution:

[0039] A lithium battery charging data monitoring system includes a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned lithium battery charging data monitoring method is implemented.

[0040] By adopting the above technical solution, a computer program is generated from the above-mentioned lithium battery charging data monitoring method and stored in a memory so that it can be loaded and executed by a processor. In this way, a terminal device can be made based on the memory and the processor for convenient use.

[0041] The present invention has the following technical effects:

[0042] This invention introduces a feature sequence to reflect the value and trend of charging data points, thereby preventing situations where the value alone cannot accurately reflect abnormalities in lithium battery charging.

[0043] Furthermore, by analyzing category consistency to reflect the contribution of data from each dimension to clustering, greater weights are assigned to data with high contribution, thereby improving the accuracy of clustering;

[0044] Furthermore, a time-series weight adjustment coefficient is set based on the level of classification detail to adjust the weights at each stage. This results in data with less classification detail having a larger weight in the early stages of clustering, thus enabling a better grasp of macro information and preventing local optima. Conversely, data with more classification detail have a larger weight in the later stages of clustering, thus enabling a better grasp of detailed information and improving classification accuracy. Attached Figure Description

[0045] Figure 1 This is a flowchart of a lithium battery charging data monitoring method according to an embodiment of the present invention. Detailed Implementation

[0046] This invention discloses a method for monitoring lithium battery charging data, referring to... Figure 1 This includes steps S1-S4:

[0047] S1: Obtain the current charging data point of the lithium battery and the corresponding historical charging data point.

[0048] Specifically, acquire each type of charging data of the lithium battery at the current moment, and record all types of charging data of the lithium battery at the current moment as data points.

[0049] The charging data points corresponding to the current time in the historical time are obtained and recorded as historical charging data points.

[0050] The historical moment corresponding to the current moment is the historical moment that is in the same charging stage as the current moment.

[0051] The charging data can be of the following types: charging voltage, charging current, charging temperature, state of charge, and electrode material impedance, or other types. This embodiment does not impose any specific restrictions.

[0052] S2: Obtain the feature sequence of historical charging data points, wherein the feature data in the feature sequence reflects the value and trend of historical charging data points.

[0053] It should be noted that charging anomalies sometimes manifest as abnormal charging data, and sometimes as abnormal trends in charging data changes. Therefore, to accurately reflect the charging status of lithium batteries, it is necessary to consider both the values ​​and trends. Consequently, before cluster analysis, a feature index needs to be constructed that can reflect both the values ​​and trends of the charging data.

[0054] Preferably, as an example, the feature sequence of historical charging data points is obtained, including:

[0055] Obtain data for each dimension from historical charging data points, and obtain the historical charging data sequence corresponding to each dimension of the historical charging data points;

[0056] The Gaussian pyramid algorithm is used to process the historical charging data sequence to obtain several layers of processed data sequence. The data corresponding to each dimension of historical charging data in each historical charging data point are obtained from each layer of processed data sequence and recorded as the macro data of each dimension of historical charging data point. The sequence of all dimension data, the slope of all dimension data and the slope of all macro data of all historical charging data points is recorded as the feature sequence.

[0057] It is understandable that, due to the influence of noise and other factors on the collected data, the trend of charging data cannot be accurately reflected by only one level of change information. Therefore, Gaussian pyramid processing can reflect trend information at different levels.

[0058] It should be added that the methods for obtaining historical charging data sequences include:

[0059] Obtain any dimension of data from historical data points, obtain the time series data of the entire process in which that dimension of data is located, and arrange the obtained time series data in time sequence to obtain the historical charging data sequence.

[0060] Further details are needed regarding the methods for obtaining the slopes of each dimension of data and the slopes of each macroscopic data point, including:

[0061] The difference between each dimension's data and the previous data in the corresponding historical charging data sequence is used as the slope of each dimension's data.

[0062] The difference between each macro data point and the previous data point in the corresponding processed data sequence is used as the slope of each macro data point.

[0063] S3: Obtain feature data of any dimension from the feature sequence of all historical charging data points, calculate the class consistency of feature data of each dimension with feature data of other dimensions, determine the baseline weight of feature data of each dimension based on class consistency, the baseline weight is positively correlated with class consistency; obtain the classification detail of feature data of each dimension, the classification detail represents the fineness of classification, obtain the time-series weight adjustment coefficient of feature data of each dimension based on classification detail, the weight adjustment coefficient is positively correlated with classification detail.

[0064] It's important to note that, to avoid getting trapped in local optima and achieve accurate classification analysis, it's crucial to leverage data from dimensions with a broad macro-level of classification in the early stages of clustering. This allows for a better grasp of the macro-level information in the data, enabling macro-level cluster analysis. To prevent cluster analysis from focusing solely on the macro-level and failing to achieve more detailed classification, after identifying the macro-level clustering criteria, it's necessary to utilize data from dimensions with a higher level of detail to further refine the classification analysis based on the macro-level findings.

[0065] S30: Obtain feature data of any dimension from the feature sequence of all historical charging data points, calculate the class consistency of feature data of each dimension with feature data of other dimensions, and determine the benchmark weight of feature data of each dimension based on class consistency.

[0066] It should be noted that, because the contribution of each dimension of the historical charging data point feature sequence to classification varies, some dimensions can facilitate classification while others can interfere with it. To achieve accurate classification, it is necessary to consider the classification contribution of each dimension. Therefore, before analyzing the macro and micro aspects of each dimension's classification, it is essential to first analyze its classification contribution and assign a baseline weight to each dimension based on this contribution.

[0067] It should be further explained that the dimensions that can promote classification should be consistent with the classification of other dimensions. If the classification consistency of a dimension is low with other dimensions, it means that the data of that dimension is inaccurate or that the data of that dimension cannot promote classification analysis. Therefore, the benchmark weight can be set based on the classification consistency.

[0068] Preferably, as an example, feature data of any dimension is obtained from the feature sequence of all historical charging data points. The class consistency of feature data of each dimension with feature data of other dimensions is calculated, and the baseline weight of feature data of each dimension is determined based on class consistency, including:

[0069] From the feature sequence of all historical charging data points, feature data of any dimension is obtained, and statistical histograms of feature data of each dimension are obtained by statistical analysis of feature data of each dimension.

[0070] The statistical histogram is filtered to obtain the local minimum points in the filtered statistical histogram, and the region between every two local minimum points is taken as an independent peak.

[0071] Specifically, if there is no minimum point before the minimum point, the region between the first point and the extreme point is considered as an independent peak; if there is no minimum point after the minimum point, the region between the minimum point and the last point is considered as an independent peak.

[0072] The characteristic sequences corresponding to the characteristic data of each independent peak in the statistical histogram are recorded as the analytical characteristic sequences of each independent peak.

[0073] Calculate the consistency of the distribution of each independent peak with other dimensions:

[0074]

[0075] in, This represents the probability that the analytical characteristic sequence of each independent peak is distributed in the statistical histogram of the characteristic data in the i-th dimension, and that it is located in the j-th independent peak. Represents the logarithmic function with base 10. This represents the number of independent peaks in the statistical histogram of the feature data in the i-th dimension. Indicates the number of dimensions of the feature sequence. This indicates that the distribution of each independent peak is consistent with that of other dimensions.

[0076] Under normal circumstances, if the data in one dimension is consistent with the data in other dimensions, the clustering characteristics of the data in that dimension should also be reflected in other dimensions. For example, if the data in one dimension shows a clustering characteristic of an independent peak, even if it does not show a clustering characteristic of an independent peak in other dimensions, it should be concentrated in a few independent peaks, rather than being evenly distributed among all independent peaks. This clustering characteristic should be present in order to meet the requirements of classification consistency. This reflects the discreteness of the analytical characteristic sequence of independent peaks in other dimensions. The larger the value, the more evenly the analytical characteristic sequence of independent peaks is distributed in other dimensions. In other words, the clustering characteristics of independent peaks are not reflected in other dimensions, and therefore the classification consistency between independent peaks and other dimensions is poor.

[0077] The normalized value of the mean of the distribution consistency of all independent peaks in the statistical histogram of feature data in each dimension is used as the class consistency of feature data in each dimension with feature data in other dimensions.

[0078] The ratio of category consistency to the sum of category consistency across all dimensions is used as the baseline weight for the feature data of each dimension.

[0079] It is understandable that the greater the categorical consistency, the greater the consistency between the data in that dimension and the data in other dimensions. Therefore, the data in that dimension is more likely to promote classification, and the weight of the data in that dimension should be increased.

[0080] It should be added that the method for obtaining the probability of the analytical characteristic sequence distribution of each independent peak in the statistical histogram of the feature data in the i-th dimension being in the j-th independent peak includes:

[0081] Obtain the number of analytical feature sequences for each independent peak. Then, obtain the number of j-th independent peaks in the statistical histogram of the corresponding dimension of the feature data in the i-th dimension of each independent peak's analytical feature sequence. Divide the distribution number of the j-th independent peak by the number of analytical feature sequences for each independent peak to obtain the probability that the analytical feature sequence of each independent peak is distributed in the j-th independent peak of the statistical histogram of the feature data in the i-th dimension.

[0082] S31: Obtain the classification detail of feature data in each dimension, wherein the classification detail represents the fineness of classification, and obtain the temporal weight adjustment coefficient of feature data in each dimension based on the classification detail.

[0083] It should be noted that in order to focus more on data information of dimensions with higher macro-level classification in the early stage of clustering, and more on data information of dimensions with higher level of classification in the later stage of clustering, it is necessary to analyze the classification detail of data in each dimension.

[0084] It should be further explained that data of the same type have a high degree of similarity in values ​​on one dimension, and data of the same type have clustering characteristics on one dimension. Therefore, the higher the level of detail in the classification of data on one dimension, the more independent peaks there are in that dimension.

[0085] Preferably, as an example, the classification detail of the feature data in each dimension is obtained, and the temporal weight adjustment coefficients of the feature data in each dimension are obtained based on the classification detail, including:

[0086] The normalized value of the number of independent peaks in the statistical histogram of the feature data of each dimension is used as the classification detail of the feature data of each dimension.

[0087] The product of the classification detail and the preset adjustment coefficient is used as the temporal weight adjustment coefficient for the feature data of each dimension.

[0088] Understandably, the number of independent peaks reflects the number of data categories in that dimension; the more categories there are, the more detailed the data classification is in that dimension.

[0089] S4: Set the weights for each clustering stage according to the time-series weight adjustment coefficient and the baseline weight, and participate the weights of each clustering stage in the clustering iteration update to complete the clustering process, so as to realize the anomaly detection of the charging data point at the current moment.

[0090] S40: Set the weights for each clustering stage according to the time-series weight adjustment coefficient and the baseline weight, and participate the weights of each clustering stage in the clustering iteration update to complete the clustering process.

[0091] It's important to note that in the early stages of clustering, to focus more on high-dimensional data with broad classification scope (i.e., prioritizing data with lower classification granularity), the weights of these low-granularity data should be set relatively high. Conversely, in the later stages, to focus more on data with higher classification granularity, the weights of these high-granularity data should also be set relatively high. To reduce interference from high-granularity but low-contribution data in the early stages of clustering, and to reduce computational load, it's advisable to consider removing some high-granularity but low-contribution data before performing cluster analysis.

[0092] Preferably, as an example, the weights of each clustering stage are set according to the time-series weight adjustment coefficient and the baseline weight, and the weights of each clustering stage participate in the clustering iteration update to complete the clustering process, including:

[0093] The normalized value of the ratio of the baseline weight to the time-series weight adjustment coefficient is used as the weight in the first clustering stage. If the weight in the first clustering stage is less than a preset weight threshold, the feature data of that dimension is removed to obtain an adjusted feature sequence. Clustering iterations are performed based on the weights of the first clustering stage and the adjusted feature sequence. In response to the clustering stability being greater than a preset stability threshold, the normalized value of the product of the baseline weight and the time-series weight adjustment coefficient is used as the weight in the second clustering stage for iterative clustering updates, until the clustering cutoff condition is met, completing the clustering process. This embodiment uses the K-means clustering algorithm for clustering analysis. Other embodiments can utilize other clustering algorithms for clustering analysis; this embodiment does not impose specific limitations. The clustering cutoff condition can utilize the built-in clustering cutoff condition of the K-means clustering algorithm, which will not be elaborated here.

[0094] Understandably, in the early stages of clustering, dividing the baseline weights by a time-series weight adjustment factor allows for a lower weighting of dimensions with high granularity. This enables more focus on the information in these less granular dimensions, achieving macro-level clustering in the early stages. Conversely, in the later stages, multiplying the baseline weights by the time-series weight adjustment factor allows for a higher weighting of these granular dimensions. This again enables more focus on the information in these granular dimensions, achieving detailed clustering in the later stages. Removing data from low-weight dimensions reduces the computational burden.

[0095] It should be added that methods for obtaining cluster stability include:

[0096] Obtain the cluster centers at each clustering stage. The inverse of the distance between the cluster center of a clustering stage and the cluster center of the previous clustering stage is taken as the stability of the cluster center. The mean of the stability of all cluster centers is taken as the stability of the cluster.

[0097] It should be further added that the clustering iterative update is performed based on the weights and adjusted feature sequences from the first clustering stage, including:

[0098] The weights of the first clustering stage are used as the weights of each dimension of the feature data in the adjusted feature sequence to calculate the distance between the cluster center and each adjusted feature sequence. Based on the distance between the cluster center and each adjusted feature sequence, the cluster center closest to the adjusted feature sequence is obtained. The adjusted feature sequence is divided into the range of the cluster center. The mean of the adjusted feature sequence within the range of the cluster center is used as the updated cluster center.

[0099] The normalized value of the product of the baseline weight and the time-series weight adjustment coefficient is used as the weight in the second clustering stage to participate in the clustering iteration update, which is the same as the clustering iteration update process described above, and will not be repeated here.

[0100] S41: To achieve anomaly detection of the charging data point at the current moment.

[0101] Preferably, as an example, to implement anomaly detection of the charging data point at the current moment, the following is included:

[0102] The feature sequence of the charging data point at the current moment is obtained. This feature sequence is then input into the clustering result to determine the category to which the charging data point belongs. The distance between the charging data point at the current moment and the cluster center of its category is calculated. If the distance between the charging data point at the current moment and the cluster center of its category is greater than a preset distance threshold, it is determined that there is an anomaly in the charging at the current moment. This embodiment uses the upper quartile of the distances between the feature sequences of all charging data points in the category and the cluster center as the preset distance threshold. Other embodiments may use other methods to set the preset distance threshold; this embodiment does not impose specific limitations.

[0103] This invention also discloses a lithium battery charging data monitoring system, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a lithium battery charging data monitoring method according to the present invention.

[0104] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0105] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (DRAM), high-bandwidth memory, hybrid memory cube, etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.

Claims

1. A method for monitoring lithium battery charging data, characterized in that, Including the following steps: Obtain the current charging data point of the lithium battery and the corresponding historical charging data point; Obtain the feature sequence of historical charging data points, where the feature data in the feature sequence reflects the values ​​and trends of historical charging data points. Feature data of any dimension is obtained from the feature sequence of all historical charging data points. The class consistency of the feature data of each dimension with the feature data of other dimensions is calculated. The baseline weight of the feature data of each dimension is determined based on the class consistency. The baseline weight is positively correlated with the class consistency. The classification detail of the feature data of each dimension is obtained. The classification detail represents the fineness of the classification. The temporal weight adjustment coefficient of the feature data of each dimension is obtained based on the classification detail. The weight adjustment coefficient is positively correlated with the classification detail. The weights for each clustering stage are set according to the time-series weight adjustment coefficient and the baseline weight. These weights are then used in the iterative clustering update to complete the clustering process, enabling anomaly detection of charging data points at the current moment, including: Obtain the feature sequence of the charging data point at the current moment, input the feature sequence of the charging data point at the current moment into the clustering result to obtain the category to which the charging data point at the current moment belongs, and judge the abnormal situation based on the distance between the charging data point at the current moment and the cluster center of the category to which it belongs.

2. The lithium battery charging data monitoring method according to claim 1, characterized in that, The acquisition of the feature sequence of historical charging data points includes: Obtain data for each dimension from historical charging data points, and obtain the historical charging data sequence corresponding to each dimension of the historical charging data points; The Gaussian pyramid algorithm is used to process the historical charging data sequence to obtain several layers of processed data sequence. The data corresponding to each dimension of the historical charging data point in each layer of processed data sequence are obtained and recorded as the macro data of each dimension of the historical charging data point. The sequence of all dimensions of the historical charging data point, the slope of all dimensions of the historical charging data point, and the slope of all macro data of all historical charging data points is recorded as the feature sequence.

3. The lithium battery charging data monitoring method according to claim 1, characterized in that, The calculation of the category consistency between feature data in each dimension and feature data in other dimensions includes: Statistical histograms of the feature data for each dimension are obtained by performing statistical analysis on the feature data for each dimension. The characteristic sequences corresponding to the characteristic data of each independent peak in the statistical histogram are recorded as the analytical characteristic sequences of each independent peak. Calculate the consistency of the distribution of each independent peak with other dimensions; The normalized value of the mean of the distribution consistency of all independent peaks in the statistical histogram of feature data in each dimension is used as the class consistency of feature data in each dimension with feature data in other dimensions.

4. The lithium battery charging data monitoring method according to claim 3, characterized in that, The calculation of the consistency of the distribution of each independent peak with other dimensions includes: in, This represents the probability that the analytical characteristic sequence of each independent peak is distributed in the statistical histogram of the characteristic data in the i-th dimension, and that it is located in the j-th independent peak. Represents the logarithmic function with base 10. This represents the number of independent peaks in the statistical histogram of the feature data in the i-th dimension. Indicates the number of dimensions of the feature sequence. This indicates that the distribution of each independent peak is consistent with that of other dimensions.

5. The lithium battery charging data monitoring method according to claim 1, characterized in that, The determination of the baseline weights for feature data in each dimension based on category consistency includes: The ratio of category consistency to the sum of category consistency across all dimensions is used as the baseline weight for the feature data of each dimension.

6. The lithium battery charging data monitoring method according to claim 3, characterized in that, The level of detail in obtaining the feature data for each dimension includes: The normalized value of the number of independent peaks in the statistical histogram of the feature data of each dimension is used as the classification detail of the feature data of each dimension.

7. A lithium battery charging data monitoring method according to claim 3, characterized in that, The temporal weight adjustment coefficients for obtaining feature data of each dimension based on classification detail include: The product of the classification detail and the preset adjustment coefficient is used as the temporal weight adjustment coefficient for the feature data of each dimension.

8. The lithium battery charging data monitoring method according to claim 1, characterized in that, The process of setting weights for each clustering stage based on the time-series weight adjustment coefficient and the baseline weight, and then participating the weights of each clustering stage in the iterative clustering update to complete the clustering process includes: The normalized value of the ratio of the baseline weight to the time-series weight adjustment coefficient is used as the weight of the first clustering stage. If the weight of the first clustering stage is less than the preset weight threshold, the feature data of that dimension is removed to obtain the adjusted feature sequence. The weight of the first clustering stage is used in the clustering iteration update. In response to the clustering stability being greater than the preset stability threshold, the normalized value of the product of the baseline weight and the time-series weight adjustment coefficient is used as the weight of the second clustering stage to participate in the clustering iteration update until the clustering cutoff condition is met, and the clustering is completed.

9. A lithium battery charging data monitoring system, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement a lithium battery charging data monitoring method according to any one of claims 1-8.

Images