A battery charging optimization and fault diagnosis method

Abstract

This invention provides a battery charging optimization and fault diagnosis method, belonging to the field of battery charging technology. The method includes: acquiring historical charging data of similar objects; extracting features based on charging scenarios to form charging scenario feature data; performing comprehensive performance feature analysis based on historical charging data to form comprehensive performance change feature data; acquiring current charging performance data of the target object; combining the comprehensive performance change feature data with charging strategy analysis to form optimal recommended charging strategy data; acquiring real-time charging adjustment data; combining the optimal recommended charging strategy data with the data to form real-time charging data for the object; and acquiring real-time charging data of the target object; combining the current charging performance data and the comprehensive performance change feature data with fault early warning analysis to form real-time charging fault early warning data. This method effectively guides customers to achieve good battery charging management and significantly improves battery performance.

Description

Technical Field

[0001] This invention relates to the field of battery charging technology, and in particular to a method for battery charging optimization and fault diagnosis. Background Technology

[0002] With the advancement of global energy transition and carbon neutrality goals, the number of electric vehicles on the road continues to grow rapidly, and charging infrastructure is becoming increasingly widespread. As the primary way for electric vehicles to interact with the power grid, the economics, safety, and impact on battery life of charging have become a focus of attention for both users and operators.

[0003] As the core energy storage component of electric vehicles, the performance degradation of power batteries directly determines the vehicle's range and operating costs, while battery faults (such as internal short circuits, lithium plating, and inconsistency deterioration) can lead to serious safety accidents. Therefore, developing intelligent charging optimization methods and early fault diagnosis technologies is of great significance for improving the user experience of electric vehicles, reducing maintenance costs, and ensuring operational safety.

[0004] Therefore, designing a battery charging optimization and fault diagnosis method to guide customers in achieving good battery charging management through reasonable charging data segmentation and processing, and effectively improving battery performance, is an urgent problem to be solved. Summary of the Invention

[0005] This invention provides a method for optimizing battery charging and diagnosing faults.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: Firstly, a battery charging optimization and fault diagnosis method is provided. The method includes: acquiring historical charging data of similar objects, performing feature extraction based on charging scenarios to form charging scenario feature data; performing comprehensive performance feature analysis based on historical charging data to form comprehensive performance change feature data; acquiring current charging performance data of the target object, combining it with the comprehensive performance change feature data to perform charging strategy analysis to form optimal recommended charging strategy data; acquiring real-time charging adjustment data, combining it with the optimal recommended charging strategy data to form real-time charging data of the object; and acquiring real-time charging data of the target object, combining it with current charging performance data and comprehensive performance change feature data to perform fault early warning analysis to form real-time charging fault early warning data.

[0007] Therefore, this method extracts charging characteristic information for different charging scenarios by acquiring historical charging data of objects of the same type, thereby providing an accurate basis for judging whether the charging scenario of the object is abnormal. At the same time, it establishes characteristic data of charging performance changes based on the object, and provides a reference for making early warning judgments on whether faults will occur during subsequent real-time charging based on the changing characteristics of different faults under performance changes. In addition, it can also provide the best strategy for each charging based on performance change data, fully guiding customers to achieve good battery charging management, effectively improving the battery performance status, and ensuring the safety of battery use.

[0008] Optionally, historical charging data of objects of the same type can be obtained, and feature extraction based on charging scenarios can be performed to form charging scenario feature data. This includes: obtaining single charging information for each historical charging object based on historical charging data; performing scenario clustering analysis based on correlation for different single charging information to form scenario clustering data; and performing feature analysis based on scenario clustering data to form clustered scenario feature data.

[0009] Therefore, feature extraction for charging scenarios mainly involves acquiring charging parameter data that characterizes a specific charging scenario. This acquired feature data is then used for real-time scenario identification of electric vehicle charging, accurately providing reference data for determining whether a charging scenario is abnormal. Of course, different scenarios inherently have significant differences; therefore, clustering can quickly classify charging scenarios and clean the data. After clustering, feature analysis can be used to obtain the corresponding scenario's identification features.

[0010] Optionally, for different single-charge information, a correlation-based scenario clustering analysis is performed to form scenario clustering data, including: setting charging scenario analysis parameters, extracting secondary charging scenario analysis parameters corresponding to different charging scenario analysis parameters under different single-charge information, forming corresponding secondary charging scenario parameter vectors; determining the cosine similarity between any two adjacent secondary charging scenario parameters based on the secondary charging scenario parameter vectors corresponding to different single-charge information; determining the average similarity value based on all cosine similarities; arbitrarily extracting a secondary charging scenario parameter vector as the scenario parameter baseline vector, and clustering all secondary charging scenario parameter vectors that meet the following conditions to form the charging scenario baseline cluster number corresponding to the scenario parameter baseline vector. The criteria are as follows: The cosine similarity of the baseline vectors for the same scene is not less than the average similarity value, and the total number of secondary charging scene parameter vectors with cosine similarity not less than the average similarity value is not less than the minimum cluster size. For the remaining secondary charging scene parameter vectors after clustering, continue to extract scene parameter baseline vectors for clustering until the extracted scene parameter baseline vectors no longer meet any of the following conditions, at which point clustering stops: The cosine similarity of the baseline vectors for the same scene is not less than the average similarity value; the total number of secondary charging scene parameter vectors with cosine similarity not less than the average similarity value is not less than the minimum cluster size. Collect the different charging scene baseline clustering data to form scene clustering data.

[0011] Therefore, correlation-based scenario clustering analysis of single-charge information mainly involves rationally clustering single-charge information under similar charging scenarios, providing more accurate and effective basic big data for subsequent feature extraction for specific charging scenarios. The essence of clustering analysis is that different charging scenarios have significantly different charging scenario analysis parameters. This significant difference is not merely a comparison of single parameters, but a comparison of combinations of parameters that characterize the scenario. It should be noted that charging scenario analysis parameters should include two types of parameters: one is the electrical performance parameters of charging itself under the corresponding scenario, such as current, voltage, power, and internal resistance; the other is the environmental parameters provided by the charging scenario that can affect charging performance, such as temperature, humidity, and input power. Of course, the specific parameters selected for charging scenario analysis can be chosen according to actual needs. It is understandable that, regardless of the selection, the parameters will inevitably include those affecting battery performance changes, which provides the basic data for subsequent analysis of the comprehensive charging performance changes of electric vehicles using charging scenario analysis parameters. Here, regarding the clustering method, although the analysis parameters of individual charging scenarios may have inconsistent variability, there is significant synergy when these parameters are combined. Therefore, similarity-based analysis is reasonable and accurate. The analysis parameters of the corresponding charging scenarios under a single charging item are aggregated into vector data. The cosine similarity of the vectors is used to represent the magnitude of the correlation. Since the cosine similarity values ​​between vectors of the same scenario are high, to ensure the correctness of clustering, the mean cosine similarity is used as the test value, and the number of clusters is used as the test value for achieving the cluster density standard. The mean cosine similarity is mainly used to exclude data with weak correlation, as this data does not have a significant impact on subsequent scene feature extraction. Of course, although some abnormal charging scenarios may be numerous in large datasets, their cluster density is still significantly different from that of normal charging scenarios. Therefore, in actual clustering processing, adjustments can be made to the mean cosine similarity to achieve better clustering results. The required number of clusters can be used to exclude abnormal data and determine the data with truly stable clustering properties. The minimum cluster size can be set according to the actual clustering situation. It should also be noted that all the charging data for these individual items are data from the same type of object. After all, the battery performance of different types of electric vehicles varies greatly in its initial state and performance changes, which will affect the accuracy of the feature data, and thus affect the accuracy of using feature data for charging optimization and fault prediction.

[0012] Optionally, feature analysis is performed on the scene clustering data to form clustered scene feature data, including: scene anomaly labeling based on the charging scenes corresponding to different charging scene benchmark clustering data: if the charging scene corresponding to the charging scene benchmark clustering data is a normal charging scene, it is labeled as normal charging scene benchmark clustering data, and the corresponding scene parameter benchmark vector is determined as the corresponding charging scene clustering feature vector; if the charging scene corresponding to the charging scene benchmark clustering data is an abnormal charging scene, it is labeled as abnormal charging scene benchmark clustering data, and the corresponding scene parameter benchmark vector is determined as the corresponding charging scene clustering feature vector.

[0013] Therefore, the purpose of feature analysis is to determine the feature parameters that can represent this type of clustered data. Since the clustered data is formed by correlation clustering based on a baseline vector of arbitrary scene parameters, it is reasonable and effective to determine this baseline vector as the corresponding feature vector. Of course, the clustered charging scenarios include normal charging scenarios and abnormal charging scenarios, which need to be distinguished to provide a reference for subsequent judgment of scenario anomalies. This distinction can be based on charging analysis parameters, especially parameters that characterize the charging scenario environment, or it can be achieved by labeling when extracting single charging information from historical data. Any method that can quickly determine the scenario anomaly is acceptable.

[0014] Optionally, comprehensive performance characteristic analysis is performed based on historical charging data to form comprehensive performance change characteristic data, including: for different historical charging objects under historical charging data, different single charging information is collected in chronological order to form object-sub-item sequential charging datasets; comprehensive performance change analysis is performed on different object-sub-item sequential charging datasets to form corresponding comprehensive charging performance change data; feature analysis of fault events is performed by combining the object charging comprehensive performance change data corresponding to different historical charging objects to form fault event comprehensive characteristic data; and comprehensive performance change data and fault event comprehensive characteristic data are combined to form comprehensive performance change characteristic data.

[0015] Therefore, the feature analysis of comprehensive performance based on historical charging data mainly aims to obtain the relationship between charging scenario parameters and changes in electric vehicle charging performance. This provides a reasonable data analysis basis for subsequent performance prediction and a methodological path for optimizing charging strategies. Of course, comprehensive performance changes can also reflect performance changes before macro-level faults occur. Therefore, by mapping comprehensive performance change data with faults, we can establish basic data that can predict faults and provide a basis for reasonable charging fault prediction.

[0016] Optionally, a comprehensive performance change analysis is performed on different object-specific sequential charging datasets to form corresponding comprehensive charging performance change data, including: determining the corresponding object single-charge capacity, object single-charge internal resistance, and object single-charge energy efficiency from the single-charge information in different object-specific sequential charging datasets; setting normalized parameter values ​​corresponding to different charging scenario analysis parameters; determining the ratio of the secondary charging scenario analysis parameter to the normalized parameter value as the secondary charging normalized parameter corresponding to the corresponding charging scenario analysis parameter for different object-specific sequential charging datasets; for different single-charge information, combining the corresponding object single-charge capacity, object single-charge internal resistance, object single-charge energy efficiency, and different secondary charging normalized parameters to form a corresponding single-charge normalized dataset; arbitrarily extracting no less than K single-charge normalized datasets from different object-specific sequential charging datasets to form a performance impact dataset; and setting a battery capacity change impact feature model. Where n represents the number of the analysis parameters for different charging scenarios, This represents the charging normalization parameter corresponding to the charging scenario analysis parameter numbered n. This represents the correlation formula for the influence of the charging normalization parameter relative to the battery capacity for the charging scenario analysis parameter numbered n. This represents the capacity-influencing parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the capacity-time relationship that affects battery capacity. This represents the battery capacity value; for the performance impact dataset, the characteristic model of battery capacity change impact is analyzed based on different single-charge normalized datasets to determine the constants to be analyzed and form the characteristic formula of battery capacity change impact; the characteristic model of battery internal resistance change impact is set: ,in, This represents the correlation formula for the influence of the normalized charging parameter relative to the battery internal resistance, corresponding to the charging scenario analysis parameter numbered n. This represents the internal resistance influence parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between internal resistance and battery internal resistance. This represents the battery's internal resistance value; for the performance impact dataset, the characteristic model of the impact of battery internal resistance change is analyzed based on different single-charge normalized datasets to determine the constant to be analyzed, forming the characteristic formula of the impact of battery internal resistance change; the characteristic model of the impact of energy efficiency change is set: ,in, This represents the correlation formula for the influence of the charging normalization parameter on the relative energy efficiency of the charging scenario analysis parameter numbered n. This represents the efficiency-affecting parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between battery efficiency and efficiency. This represents the battery energy efficiency value; for the performance impact dataset, the energy efficiency change impact feature model is analyzed based on different single-charge normalized datasets to determine the constants to be analyzed and form the energy efficiency change impact feature formula.

[0017] Therefore, comprehensive performance change analysis needs to consider two aspects. First, the comprehensive performance change needs to comprehensively reflect the influence of charging scenario parameter data on the charging performance changes of electric vehicles of the same type. Based on this, basic data needs to be obtained from different objects of the same type to ensure the representativeness and comprehensiveness of the analysis results. Therefore, this application obtains the corresponding single charging information based on objects, and randomly extracts no less than a certain amount of information from each object when conducting influence relationship analysis. This ensures the randomness of the extracted data information and also makes the rationality and representativeness of the total performance change influence relationship established by the extracted data. Of course, under big data, the number of single charging normalized data extracted may exceed the total number of constants to be analyzed in the corresponding model. The relationship can be determined by grouping and analyzing, and then using the average of the same type of parameters to be analyzed to determine the relationship. Alternatively, an initial relationship can be obtained by analyzing the same amount of data first, and then the remaining data can be gradually and orderly substituted with the difference between the result obtained based on the relationship and the actual result as the target to form the relationship. As long as the normalized data can be fully utilized to achieve a reasonable relationship analysis, it is acceptable. The K value can be set according to the actual situation. On the other hand, it's crucial to determine parameters that comprehensively reflect changes in charging performance. For a battery, each charge provides the most representative parameters for performance changes caused by charging. Considering the magnitude of the impact, battery capacity, internal resistance, and energy efficiency are chosen as essential parameters. This ensures representativeness and, through a combined parameter approach, effectively avoids the one-sidedness of relying on a single representative parameter. It's important to note that historical data from a single charge may vary depending on the charging scenario, affecting the analysis parameters. However, performance changes are assessed based on the battery's inherent performance, so it's necessary to eliminate the influence of different charging scenarios. Therefore, by setting standard reference values ​​for each analysis parameter for data normalization, we can ensure that the relationships formed in subsequent analyses are unaffected by charging scenarios. This provides a reasonable and reliable basis for more accurate assessments of overall battery performance changes. Of course, the normalization parameter values ​​can be set according to actual conditions. Normalization uses the normalization parameter as the denominator and the corresponding parameter from the single charge data as the numerator, forming a ratio as the normalization value. In addition, the three parameters characterizing performance changes are also affected by time, so the time factor is also considered in the analysis.

[0018] Optionally, feature analysis of fault events can be performed by combining the comprehensive charging performance change data of different historical charging objects to form comprehensive fault event feature data. This includes: fitting the single-charge battery capacity of different objects in chronological order to form corresponding object battery capacity change curves for different object sequential charging datasets; fitting the single-charge internal resistance of different objects in chronological order to form corresponding object internal resistance change curves for different object sequential charging datasets; fitting the single-charge energy efficiency of different objects in chronological order to form corresponding object energy efficiency change curves for different objects; and calibrating the time of the first charging fault for different historical charging objects, and calibrating the corresponding object battery capacity change curve, object internal resistance change curve, and object energy efficiency change curve according to the calibrated first fault time point, and then truncating... Extract the initial fault experience graphs from the start of the curve to the time of the first fault, and form initial fault experience graphs for the battery capacity, internal resistance, and energy efficiency of the target battery. Compare the initial fault experience graphs for the battery capacity, internal resistance, and energy efficiency of the target battery from different historical charging targets with the same fault, and determine the corresponding minimum coverage graph. Based on the minimum coverage graphs corresponding to different types of curves, determine the proportion of the initial fault experience graphs on different historical charging targets, and combine the proportions corresponding to different historical charging targets to form the initial fault experience proportion range of the corresponding type of experience graph. For different fault types, combine the initial fault experience proportion ranges of the corresponding different types of experience graphs to form the comprehensive fault event feature data corresponding to the fault type.

[0019] Therefore, changes in overall charging performance will inevitably manifest when a charging failure occurs. Thus, feature analysis of past failure data can extract characteristic information about the failure, providing accurate data for future fault prediction. Here, capacity, internal resistance, and energy efficiency are considered as characterization parameters for fault analysis. Fault occurrence has a significant time-cumulative effect, so the changes in overall performance parameters need to be considered from a time perspective. The cumulative amount of performance parameters over time is used as the basis for assessing the probability of fault occurrence during fault analysis. Since the obtained data is a change curve, using the curve image before the fault occurs as the analysis object is reasonable and accurate. First, the analysis focuses on the initial fault; therefore, secondary or subsequent faults may be inapplicable due to modifications to battery performance after the initial fault. Also, given the high stability of current battery performance, the time of the initial fault is significantly delayed, generally occurring in the mid-to-late stages of electric vehicle use; therefore, considering only the initial fault is reasonable. Furthermore, the extraction of fault feature information is based on the feature information set of the three overall performance data sets to avoid reducing data accuracy due to a single feature. The main method of feature acquisition is to acquire the image before the fault, and compare the images of different objects corresponding to the same type of fault with the time axis to determine the smallest overlapping area. This area is the area where the fault must exist. Considering the smallest area of ​​object difference, it is also necessary to combine it with the range of individual differences under big data to make it reasonable. This range of individual differences is represented by the proportion of the smallest area in the image of the corresponding object. The range is determined based on the minimum and maximum values. The feature information of the three performance parameters are all collected in this way to form the feature data corresponding to this type of fault.

[0020] Optionally, the current charging performance data of the target object is obtained, and charging strategy analysis is performed in conjunction with comprehensive performance change characteristic data to form optimal recommended charging strategy data. This includes: determining the current battery capacity, current battery internal resistance, current battery energy efficiency, and current planned charging time based on the current charging performance data; extracting current charging scenario analysis parameters for known different charging scenario analysis parameters based on the current charging performance data; and combining the known current charging scenario analysis parameters and the current planned charging time with the characteristic formulas for the impact of battery capacity change, battery internal resistance change, and energy efficiency change, applying these parameters to unknown different charging scenarios. The parameters for scenario analysis are determined to ensure that the determined parameters meet the following conditions. Then, the optimal recommended charging strategy data is formed by combining all the parameters for charging scenario analysis: the battery capacity value determined by the determined parameters is close to and less than the current battery capacity value; the battery internal resistance value determined by the determined parameters is close to and greater than the current battery internal resistance value; the battery energy efficiency determined by the determined parameters is close to and less than the current battery energy efficiency; and the sum of the differences between the determined battery capacity and the current battery capacity value, the differences between the determined battery internal resistance value and the current battery internal resistance value, and the differences between the determined battery energy efficiency and the current battery energy efficiency is minimized among the selected parameter combinations.

[0021] Therefore, the current charging performance data includes the comprehensive battery performance parameters after previous charging, as well as some known analytical parameters for the charging scenario when charging is about to begin. These analytical parameters are mainly environmental parameters and the parameters of the corresponding charging pile in the specific scenario. To optimize charging, it is necessary to ensure that the charging to begin immediately can minimize the reduction in capacitance, the increase in internal resistance, and the reduction in energy efficiency. Therefore, it is reasonable and efficient to select the remaining position analysis parameters based on these constraints. Minimizing the changes in the three comprehensive performance parameters is the goal of the planning analysis. The resulting complete data can then be used as the best recommendation strategy to guide users to achieve ideal charging management.

[0022] Optionally, real-time charging data of the target object is acquired, and fault warning analysis is performed in conjunction with current charging performance data and comprehensive performance change characteristic data to form real-time charging fault warning data. This includes: obtaining the corresponding real-time charging scenario parameter vector based on the real-time charging data, and calculating the cosine similarity with different charging scenario clustering feature vectors to determine the charging scenario clustering feature vector corresponding to the maximum cosine similarity; forming a real-time charging scenario anomaly analysis result based on the scenario anomaly calibration result corresponding to the determined charging scenario clustering feature vector; if the real-time charging scenario anomaly analysis result is abnormal, then scenario anomaly information is output, and fault warning analysis is further performed in conjunction with real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data; otherwise, no information is output, and fault warning analysis is further performed in conjunction with real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data.

[0023] Therefore, it is essential to first confirm whether the charging scenario is abnormal before conducting fault warning analysis. Abnormal charging scenarios can cause significant damage to the battery, so providing early warnings based on these findings is both reasonable and necessary. Of course, fault warning analysis will also be based on the charging scenario conditions to better address faults.

[0024] Optionally, if the real-time charging scenario anomaly analysis result is abnormal, scenario anomaly information is output, and fault warning analysis is further performed by combining real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data. Otherwise, no information is output, and fault warning analysis is further performed by combining real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data, including: determining the real-time battery capacity experience graph, real-time internal resistance experience graph, and real-time energy efficiency experience graph at the expected completion time of real-time charging based on real-time charging data; comparing the real-time battery capacity experience graph, real-time internal resistance experience graph, and real-time energy efficiency experience graph with the corresponding minimum coverage graph respectively; if all three experience graphs contain the corresponding minimum coverage graph, and the area ratio of the minimum coverage graph on the corresponding experience graph is within the corresponding first fault experience ratio range, then charging fault warning information is formed; otherwise, charging normal information is formed.

[0025] Therefore, fault warning mainly determines whether the battery performance changes after the next charge, under the planned charging strategy, are mapped to fault characteristic information. If so, it means that the next charge will lead to a fault; otherwise, it means that the next charge is normal. Attached Figure Description

[0026] Figure 1 This is a schematic diagram illustrating an application scenario of a battery charging optimization and fault diagnosis method provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating a battery charging optimization and fault diagnosis method provided in an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the charging scenario and comprehensive performance change feature extraction process of a battery charging optimization and fault diagnosis method provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of the architecture of a battery charging optimization and fault diagnosis system provided in an embodiment of the present invention. Detailed Implementation

[0027] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0028] To facilitate understanding of the embodiments of the present invention, Figure 1 This is a schematic diagram illustrating an application scenario of a battery charging optimization and fault diagnosis method provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating a battery charging optimization and fault diagnosis method provided in an embodiment of the present invention. Figure 3 This diagram illustrates the process of extracting charging scenario and overall performance change features for a battery charging optimization and fault diagnosis method according to an embodiment of the present invention. The specific process of this battery charging optimization and fault diagnosis method is as follows: S1: Obtain historical charging data for objects of the same type, perform feature extraction based on the charging scenario, and form charging scenario feature data.

[0029] Historical charging data of similar objects is obtained, and features are extracted based on charging scenarios to form charging scenario feature data. This includes: obtaining single charging information for each historical charging object based on historical charging data; performing scenario clustering analysis based on correlation for different single charging information to form scenario clustering data; and performing feature analysis based on scenario clustering data to form clustered scenario feature data.

[0030] Feature extraction for charging scenarios primarily involves acquiring charging parameter data that characterizes a specific charging scenario. This acquired feature data is then used for real-time scene identification of electric vehicle charging, accurately providing reference data to determine whether a charging scenario is abnormal. Of course, different scenarios inherently differ significantly; therefore, clustering can quickly classify charging scenarios and clean the data. After clustering, feature analysis can be used to obtain the corresponding scenario's identification features.

[0031] For different single-charge information, a correlation-based scenario clustering analysis is performed to form scenario clustering data, including: setting charging scenario analysis parameters, extracting secondary charging scenario analysis parameters corresponding to different charging scenario analysis parameters under different single-charge information, forming corresponding secondary charging scenario parameter vectors; determining the cosine similarity between any two adjacent secondary charging scenario parameters based on the secondary charging scenario parameter vectors corresponding to different single-charge information; determining the average similarity value based on all cosine similarities; arbitrarily extracting a secondary charging scenario parameter vector as the scenario parameter baseline vector, and clustering all secondary charging scenario parameter vectors that meet the following conditions to form charging scenario baseline clustering data corresponding to the scenario parameter baseline vector: The total number of secondary charging scenario parameter vectors whose cosine similarity to the baseline vectors of the same scene is not less than the average similarity value, and whose cosine similarity to the baseline vectors of the same scene is not less than the average similarity value, is not less than the minimum cluster size. For the remaining different secondary charging scenario parameter vectors after clustering, continue to extract the scene parameter baseline vectors for clustering until the secondary charging scenario parameter vectors corresponding to the extracted scene parameter baseline vectors no longer meet any of the following conditions, at which point clustering stops: the cosine similarity of the baseline vectors of the same scene is not less than the average similarity value; the total number of secondary charging scenario parameter vectors whose cosine similarity to the baseline vectors of the same scene is not less than the minimum cluster size; and the different charging scenario baseline clustering data are combined to form scene clustering data.

[0032] Correlation-based scenario clustering analysis of single-charge information primarily involves rationally clustering single-charge information within similar charging scenarios. This provides a more accurate and effective foundation of big data for subsequent feature extraction targeting specific charging scenarios. The essence of clustering analysis lies in the significant differences in charging scenario analysis parameters across different charging scenarios. These differences are not merely comparisons of single parameters, but rather comparisons of combinations of parameters characterizing the scenario. It's important to note that charging scenario analysis parameters should include two types: electrical performance parameters inherent in the charging scenario, such as current, voltage, power, and internal resistance; and environmental parameters that influence charging performance, such as temperature, humidity, and input power. The specific parameters chosen for scenario analysis can be selected based on actual needs. Regardless of the selection, the parameters will inevitably include those affecting battery performance changes. This provides the foundational data for directly analyzing the overall charging performance changes of electric vehicles using these scenario analysis parameters. Here, regarding the clustering method, although the analysis parameters of individual charging scenarios may have inconsistent variability, there is significant synergy when these parameters are combined. Therefore, similarity-based analysis is reasonable and accurate. The analysis parameters of the corresponding charging scenarios under a single charging item are aggregated into vector data. The cosine similarity of the vectors is used to represent the magnitude of the correlation. Since the cosine similarity values ​​between vectors of the same scenario are high, to ensure the correctness of clustering, the mean cosine similarity is used as the test value, and the number of clusters is used as the test value for achieving the cluster density standard. The mean cosine similarity is mainly used to exclude data with weak correlation, as this data does not have a significant impact on subsequent scene feature extraction. Of course, although some abnormal charging scenarios may be numerous in large datasets, their cluster density is still significantly different from that of normal charging scenarios. Therefore, in actual clustering processing, adjustments can be made to the mean cosine similarity to achieve better clustering results. The required number of clusters can be used to exclude abnormal data and determine the data with truly stable clustering properties. The minimum cluster size can be set according to the actual clustering situation. It should also be noted that all the charging data for these individual items are data from the same type of object. After all, the battery performance of different types of electric vehicles varies greatly in its initial state and performance changes, which will affect the accuracy of the feature data, and thus affect the accuracy of using feature data for charging optimization and fault prediction.

[0033] Feature analysis is performed on the scene clustering data to form clustered scene feature data, including: scene anomaly labeling based on the charging scene corresponding to the benchmark clustering data of different charging scenes: if the charging scene corresponding to the benchmark clustering data of the charging scene is a normal charging scene, it is labeled as the benchmark clustering data of normal charging scene, and the corresponding scene parameter benchmark vector is determined as the corresponding charging scene clustering feature vector; if the charging scene corresponding to the benchmark clustering data of the charging scene is an abnormal charging scene, it is labeled as the benchmark clustering data of abnormal charging scene, and the corresponding scene parameter benchmark vector is determined as the corresponding charging scene clustering feature vector.

[0034] The purpose of feature analysis is to determine the characteristic parameters that can represent this type of clustered data. Since the clustered data is formed by correlation clustering based on a baseline vector of arbitrary scene parameters, it is reasonable and effective to determine this baseline vector as the corresponding feature vector. Of course, the clustered charging scenarios include normal charging scenarios and abnormal charging scenarios, which need to be distinguished to provide a reference for subsequent judgment of scenario anomalies. This distinction can be based on charging analysis parameters, especially parameters that characterize the charging scenario environment, or it can be achieved by labeling when extracting single charging information from historical data. Any method that can quickly determine the scenario anomaly is acceptable.

[0035] S2: Perform comprehensive performance characteristic analysis based on historical charging data to generate comprehensive performance change characteristic data.

[0036] Based on historical charging data, a comprehensive performance characteristic analysis is performed to form comprehensive performance change characteristic data. This includes: for different historical charging objects under the historical charging data, the corresponding single charging information is collected in chronological order to form object-specific sequential charging datasets; a comprehensive performance change analysis is performed on different object-specific sequential charging datasets to form corresponding comprehensive charging performance change data; a feature analysis of fault events is performed by combining the object charging comprehensive performance change data corresponding to different historical charging objects to form fault event comprehensive characteristic data; and the comprehensive performance change data and fault event comprehensive characteristic data are combined to form comprehensive performance change characteristic data.

[0037] The feature analysis of comprehensive performance based on historical charging data mainly aims to obtain the relationship between charging scenario parameters and changes in electric vehicle charging performance. This provides a reasonable data analysis basis for subsequent performance change prediction and a methodological path for optimizing charging strategies. Of course, comprehensive performance changes can also reflect performance changes before macro-level faults occur. Therefore, by mapping comprehensive performance change data with faults, a basic data for fault prediction can be established, providing a basis for reasonable charging fault prediction.

[0038] A comprehensive performance change analysis is performed on different object-specific sequential charging datasets to form corresponding comprehensive charging performance change data. This includes: determining the object's single-charge capacity, single-charge internal resistance, and single-charge energy efficiency from the single-charge information in different object-specific sequential charging datasets; setting normalized parameter values ​​for different charging scenario analysis parameters; determining the ratio of the analysis parameters to the normalized parameter values ​​for different charging scenario analysis parameters in the single-charge information of different object-specific sequential charging datasets as the corresponding charging scenario analysis parameter's normalized parameter; for different single-charge information, combining the object's single-charge capacity, single-charge internal resistance, single-charge energy efficiency, and different charging normalized parameters to form a corresponding single-charge normalized dataset; arbitrarily extracting no less than K single-charge normalized datasets from different object-specific sequential charging datasets to form a performance impact dataset; and setting a battery capacity change impact feature model. Where n represents the number of the analysis parameters for different charging scenarios, This represents the charging normalization parameter corresponding to the charging scenario analysis parameter numbered n. This represents the correlation formula for the influence of the charging normalization parameter relative to the battery capacity for the charging scenario analysis parameter numbered n. This represents the capacity-influencing parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the capacity-time relationship that affects battery capacity. This represents the battery capacity value; for the performance impact dataset, the characteristic model of battery capacity change impact is analyzed based on different single-charge normalized datasets to determine the constants to be analyzed and form the characteristic formula of battery capacity change impact; the characteristic model of battery internal resistance change impact is set: ,in, This represents the correlation formula for the influence of the normalized charging parameter relative to the battery internal resistance, corresponding to the charging scenario analysis parameter numbered n. This represents the internal resistance influence parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between internal resistance and battery internal resistance. This represents the battery's internal resistance value; for the performance impact dataset, the characteristic model of the impact of battery internal resistance change is analyzed based on different single-charge normalized datasets to determine the constant to be analyzed, forming the characteristic formula of the impact of battery internal resistance change; the characteristic model of the impact of energy efficiency change is set: ,in, This represents the correlation formula for the influence of the charging normalization parameter on the relative energy efficiency of the charging scenario analysis parameter numbered n. This represents the efficiency-affecting parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between battery efficiency and efficiency. This represents the battery energy efficiency value; for the performance impact dataset, the energy efficiency change impact feature model is analyzed based on different single-charge normalized datasets to determine the constants to be analyzed and form the energy efficiency change impact feature formula.

[0039] Comprehensive performance change analysis needs to consider two aspects. First, the comprehensive performance change needs to comprehensively reflect the influence of charging scenario parameter data on the charging performance changes of electric vehicles of the same type. Based on this, basic data needs to be obtained from different objects of the same type to ensure the representativeness and comprehensiveness of the analysis results. Therefore, this application obtains the corresponding single charging information based on objects, and randomly extracts no less than a certain amount of information from each object when conducting influence relationship analysis. This ensures the randomness of the extracted data information and also makes the rationality and representativeness of the total performance change influence relationship established by the extracted data. Of course, under big data, the number of normalized single charging data extracted may exceed the total number of constants to be analyzed in the corresponding model. The relationship can be determined by grouping and analyzing, and then using the average of the same type of parameters to be analyzed to determine the relationship. Alternatively, an initial relationship can be obtained by analyzing the same amount of data first, and then the remaining data can be gradually and orderly substituted with the difference between the result obtained based on the relationship and the actual result as the target to form the relationship. As long as the normalized data can be fully utilized to achieve a reasonable relationship analysis, it is acceptable. The K value can be set according to the actual situation. On the other hand, it's crucial to determine parameters that comprehensively reflect changes in charging performance. For a battery, each charge provides the most representative parameters for performance changes caused by charging. Considering the magnitude of the impact, battery capacity, internal resistance, and energy efficiency are chosen as essential parameters. This ensures representativeness and, through a combined parameter approach, effectively avoids the one-sidedness of relying on a single representative parameter. It's important to note that historical data from a single charge may vary depending on the charging scenario, affecting the analysis parameters. However, performance changes are assessed based on the battery's inherent performance, so it's necessary to eliminate the influence of different charging scenarios. Therefore, by setting standard reference values ​​for each analysis parameter for data normalization, we can ensure that the relationships formed in subsequent analyses are unaffected by charging scenarios. This provides a reasonable and reliable basis for more accurate assessments of overall battery performance changes. Of course, the normalization parameter values ​​can be set according to actual conditions. Normalization uses the normalization parameter as the denominator and the corresponding parameter from the single charge data as the numerator, forming a ratio as the normalization value. In addition, the three parameters characterizing performance changes are also affected by time, so the time factor is also considered in the analysis.

[0040] By combining the comprehensive charging performance change data of different historical charging objects, feature analysis is performed on fault events to form comprehensive fault event feature data. This includes: fitting the single-charge battery capacity of different objects in chronological order to form corresponding object battery capacity change curves; fitting the single-charge internal resistance of different objects in chronological order to form corresponding object internal resistance change curves; fitting the single-charge energy efficiency of different objects in chronological order to form corresponding object energy efficiency change curves; and calibrating the time of the first charging fault for different historical charging objects, and calibrating the corresponding object battery capacity change curve, object internal resistance change curve, and object energy efficiency change curve according to the calibrated first fault time point, and extracting the data. The curves from the start to the time of the first fault are plotted to form the initial fault experience graphs for the battery capacity, internal resistance, and energy efficiency of the target battery. These graphs are then compared with those for different historical charging targets with the same fault to determine the minimum coverage graph. Based on the minimum coverage graphs for different types of curves, the proportion of initial fault experience graphs for different historical charging targets is determined, and these proportions are aggregated to form the initial fault experience proportion range for each type of experience graph. For different fault types, the initial fault experience proportion ranges for different types of experience graphs are aggregated to form comprehensive fault event feature data corresponding to the fault type.

[0041] Changes in overall charging performance will inevitably manifest when a charging fault occurs. Therefore, feature analysis of past fault data can extract characteristic information about the fault occurrence, providing accurate data references for future reasonable fault prediction. Here, capacity, internal resistance, and energy efficiency are considered as characterization parameters for fault analysis. Fault occurrence has a significant time cumulative effect, so it is necessary to consider the changes in overall performance parameters from a time perspective, and use the cumulative amount of performance parameters over time as the basis for assessing the probability of fault occurrence during fault analysis. Since the obtained data is a change curve, it is reasonable and accurate to use the curve image before the fault occurs as the analysis object. First, the analysis focuses on the initial fault; therefore, secondary or subsequent faults may be inapplicable due to modifications and adjustments to battery performance after the initial fault. Furthermore, given the high stability of current battery performance, the time of the initial fault is significantly delayed, generally occurring in the middle to late stages of electric vehicle use; therefore, considering only the initial fault is reasonable. Additionally, the extraction of fault feature information is based on the feature information set of the three overall performance data sets to avoid reducing data accuracy due to a single feature. The main method of feature acquisition is to acquire the image before the fault, and compare the images of different objects corresponding to the same type of fault with the time axis to determine the smallest overlapping area. This area is the area where the fault must exist. Considering the smallest area of ​​object difference, it is also necessary to combine it with the range of individual differences under big data to make it reasonable. This range of individual differences is represented by the proportion of the smallest area in the image of the corresponding object. The range is determined based on the minimum and maximum values. The feature information of the three performance parameters are all collected in this way to form the feature data corresponding to this type of fault.

[0042] S3: Obtain the current charging performance data of the target object, combine it with the comprehensive performance change characteristic data to analyze the charging strategy and form the best recommended charging strategy data.

[0043] The process involves acquiring the current charging performance data of the target object, combining it with comprehensive performance change characteristic data to perform charging strategy analysis, and generating optimal recommended charging strategy data. This includes: determining the current battery capacity, current battery internal resistance, current battery energy efficiency, and current planned charging time based on the current charging performance data; extracting current charging scenario analysis parameters from known analysis parameters for different charging scenarios based on the current charging performance data; and combining the known current charging scenario analysis parameters and the current planned charging time with the characteristic formulas for the impact of battery capacity change, battery internal resistance change, and energy efficiency change on unknown charging scenarios. The parameters for analysis are determined to ensure that the determined parameters meet the following conditions. Then, the optimal recommended charging strategy data is formed by combining the parameters of all charging scenarios: the battery capacity value determined by the determined parameters is close to and less than the current battery capacity value; the battery internal resistance value determined by the determined parameters is close to and greater than the current battery internal resistance value; the battery energy efficiency determined by the determined parameters is close to and less than the current battery energy efficiency; and the sum of the differences between the determined battery capacity and the current battery capacity value, the differences between the determined battery internal resistance value and the current battery internal resistance value, and the differences between the determined battery energy efficiency and the current battery energy efficiency is minimized among the selected parameter combinations.

[0044] The current charging performance data includes the comprehensive battery performance parameters after previous charging, as well as some known analysis parameters for the upcoming charging scenario. These analysis parameters are mainly environmental parameters and the parameters of the corresponding charging pile in the specific scenario. To optimize charging, it is necessary to ensure that the upcoming charging can minimize the reduction of capacitance, the increase of internal resistance, and the reduction of energy efficiency. Therefore, it is reasonable and efficient to select the remaining location analysis parameters based on these constraints. Minimizing the changes in the three comprehensive performance parameters is the goal of the planning analysis. The resulting complete data will then serve as the best recommendation strategy to guide users to achieve ideal charging management.

[0045] S4: Obtain real-time charging adjustment data and combine it with the best recommended charging strategy data to form real-time charging data for the object.

[0046] Of course, the recommended optimal charging strategy may be adjusted by users for various reasons. By acquiring user adjustment data, the optimal charging strategy data is adjusted to form a real-time charging strategy that can start charging immediately.

[0047] S5: Obtain real-time charging data of the target object, combine it with current charging performance data and comprehensive performance change characteristic data to perform fault early warning analysis, and form real-time charging fault early warning data.

[0048] The system acquires real-time charging data for the target object and performs fault warning analysis by combining current charging performance data and comprehensive performance change characteristic data to form real-time charging fault warning data. This includes: obtaining the corresponding real-time charging scenario parameter vector based on the real-time charging data, calculating the cosine similarity between the parameter vector and different charging scenario clustering feature vectors, and determining the charging scenario clustering feature vector corresponding to the maximum cosine similarity; forming a real-time charging scenario anomaly analysis result based on the scenario anomaly calibration result corresponding to the determined charging scenario clustering feature vector; if the real-time charging scenario anomaly analysis result is abnormal, then the scenario anomaly information is output, and fault warning analysis is continued to be performed by combining real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data; otherwise, no information is output, and fault warning analysis is continued to be performed by combining real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data.

[0049] Before conducting fault warning analysis, it's essential to first confirm whether the charging scenario is abnormal. Abnormal charging scenarios can cause significant damage to the battery, so providing early warnings based on these findings is both reasonable and necessary. Of course, the fault warning analysis will also be based on the charging scenario conditions to better address potential faults.

[0050] If the real-time charging scenario anomaly analysis result is abnormal, scenario anomaly information will be output, and fault warning analysis will continue to be performed in conjunction with real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data. Otherwise, no information will be output, and fault warning analysis will continue to be performed in conjunction with real-time charging data and comprehensive performance change characteristic data to form real-time charging fault warning data, including: based on real-time charging data, determining the real-time battery capacity experience graph, real-time internal resistance experience graph, and real-time energy efficiency experience graph at the expected completion time of real-time charging; comparing the real-time battery capacity experience graph, real-time internal resistance experience graph, and real-time energy efficiency experience graph with the corresponding minimum coverage graph respectively: if all three experience graphs contain the corresponding minimum coverage graph, and the area ratio of the minimum coverage graph on the corresponding experience graph is within the corresponding first fault experience ratio range, then charging fault warning information will be formed; otherwise, charging normal information will be formed.

[0051] The fault warning mainly determines whether the battery performance changes after the next charge, under the planned charging strategy, are mapped to fault characteristic information. If so, it means that the next charge will lead to a fault; otherwise, it means that the next charge is normal.

[0052] Figure 4 This is a schematic diagram of the architecture of the battery charging optimization and fault diagnosis system provided in an embodiment of the present invention.

[0053] The system includes a data acquisition unit for acquiring historical charging data, real-time charging adjustment data, and current charging performance data for similar objects; a feature extraction unit for extracting features based on charging scenarios from the historical charging data acquired by the data acquisition unit to form charging scenario feature data, and performing comprehensive performance feature analysis based on the historical charging data to form comprehensive performance change feature data; a charging optimization unit for analyzing charging strategies based on the current charging performance data acquired by the data acquisition unit and the comprehensive performance change feature data formed by the feature extraction unit to form optimal recommended charging strategy data; and an early warning analysis unit for generating real-time charging data for the object based on the real-time charging adjustment data acquired by the data acquisition unit and the optimal recommended charging strategy data, and performing fault early warning analysis based on the real-time charging data, current charging performance data, and comprehensive performance change feature data to form real-time charging fault early warning data.

[0054] It should be understood that the functional unit in the embodiments of the present invention can be a central processing unit (CPU), which can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0055] It should also be understood that the memory included in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0056] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0057] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0058] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0059] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0060] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0061] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0062] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0063] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0064] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A battery charging optimization and fault diagnosis method, characterized in that, The method includes: Obtain historical charging data for objects of the same type, perform feature extraction based on charging scenarios, and form charging scenario feature data; Based on the historical charging data, a comprehensive performance characteristic analysis is performed to form comprehensive performance change characteristic data; Obtain the current charging performance data of the target object, combine it with the comprehensive performance change characteristic data to perform charging strategy analysis, and form the best recommended charging strategy data; Real-time charging adjustment data is acquired and combined with the optimal recommended charging strategy data to form real-time charging data for the object. Real-time charging data of the target object is acquired, and fault early warning analysis is performed by combining the current charging performance data and the comprehensive performance change characteristic data to form real-time charging fault early warning data.

2. The method according to claim 1, characterized in that, The process of acquiring historical charging data for similar objects, performing feature extraction based on charging scenarios, and forming charging scenario feature data includes: Based on the historical charging data, obtain the single charging information for each historical charging object; For different single charging information, a correlation-based scenario clustering analysis is performed to form scenario clustering data; Feature analysis is performed on the clustered data of the described scenarios to form clustered scenario feature data.

3. The method according to claim 2, characterized in that, The step of performing correlation-based scenario clustering analysis on different single-charge information to form scenario clustering data includes: Set charging scenario analysis parameters, extract sub-item charging scenario analysis parameters corresponding to different charging scenario analysis parameters under different single charging information, and form a corresponding sub-item charging scenario parameter vector. Based on the parameter vectors of the secondary charging scenarios corresponding to different single charging information, the cosine similarity between any two adjacent parameters of the secondary charging scenarios is determined. Based on all the cosine similarities, the average similarity value is determined; Arbitrarily extract one of the sub-item charging scenario parameter vectors as the scenario parameter baseline vector, and cluster all the sub-item charging scenario parameter vectors that satisfy the following conditions to form the charging scenario baseline clustering data corresponding to the scenario parameter baseline vector: The total number of the sub-item charging scene parameter vectors that have a cosine similarity to the scene reference vector that is not less than the average similarity value, and whose cosine similarity to the scene reference vector is not less than the average similarity value, is not less than the minimum clustering size. For the remaining different sub-item charging scenario parameter vectors after clustering, continue to extract scenario parameter baseline vectors for clustering until the sub-item charging scenario parameter vectors corresponding to the extracted scenario parameter baseline vectors no longer satisfy any of the following conditions, at which point clustering stops: There exists a scenario whose cosine similarity to the baseline vector of the scene is not less than the average similarity value; The total number of the second-order charging scene parameter vectors whose cosine similarity to the baseline vector of the scene is not less than the average similarity value is not less than the minimum clustering size; The different benchmark clustering data of the charging scenarios are combined to form the scenario clustering data.

4. The method according to claim 3, characterized in that, The step of performing feature analysis based on the scene clustering data to form clustered scene feature data includes: Based on the different charging scenario benchmark clustering data, scenario anomaly identification is performed: If the charging scenario corresponding to the charging scenario benchmark clustering data is a normal charging scenario, then it is labeled as normal charging scenario benchmark clustering data, and the corresponding scenario parameter benchmark vector is determined as the corresponding charging scenario clustering feature vector. If the charging scenario corresponding to the charging scenario benchmark clustering data is an abnormal charging scenario, it is labeled as abnormal charging scenario benchmark clustering data, and the corresponding scenario parameter benchmark vector is determined as the corresponding charging scenario clustering feature vector.

5. The method according to claim 4, characterized in that, The step of performing comprehensive performance characteristic analysis based on the historical charging data to form comprehensive performance change characteristic data includes: For different historical charging objects under the historical charging data, the different single charging information corresponding to them are collected in time order to form an object-sub-item sequential charging dataset; A change analysis of the overall performance is performed on different sequential charging datasets of the aforementioned objects to generate corresponding overall charging performance change data. By combining the comprehensive charging performance change data of different historical charging objects, feature analysis is performed on the fault events to form comprehensive fault event feature data. The comprehensive performance change data is formed by combining the comprehensive charging performance change data and the comprehensive fault event characteristic data.

6. The method according to claim 4, characterized in that, The step of performing a comprehensive performance change analysis on different sequential charging datasets of the object items to generate corresponding comprehensive charging performance change data includes: For the single charging information in different object sequential charging datasets, determine the corresponding object single battery capacity, object single internal resistance, and object single energy efficiency. Set normalization parameter values ​​corresponding to different charging scenario analysis parameters, and for different sub-item charging scenario analysis parameters in the single charging information of different object sub-item sequential charging datasets, determine the ratio of the sub-item charging scenario analysis parameter to the normalization parameter value as the sub-item charging normalization parameter corresponding to the charging scenario analysis parameter. For different single-charge information, the corresponding single-charge battery capacity, single-charge internal resistance, single-charge energy efficiency, and different sub-item charging normalization parameters are collected to form a corresponding single-charge normalization dataset. From the different object-sub-item sequential charging datasets, at least K single-charge normalized datasets are randomly extracted to form a performance impact dataset; Define the characteristic model of the impact of battery capacity changes: Where n represents the number of the different charging scenario analysis parameters, This represents the charging normalization parameter corresponding to the charging scenario analysis parameter numbered n. This represents the correlation formula for the influence of the charging normalization parameter corresponding to the charging scenario analysis parameter numbered n on the battery capacity. This represents the capacity influence parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the capacity-time relationship that affects battery capacity. Indicates the battery capacity value; For the performance impact dataset, the battery capacity change impact feature model is analyzed according to different single-charge normalized datasets to determine the constants to be analyzed and form the battery capacity change impact feature formula; Establish a characteristic model for the influence of changes in battery internal resistance: ,in, The expression represents the correlation formula for the influence of the charging normalization parameter corresponding to the charging scenario analysis parameter numbered n on the battery internal resistance. This represents the internal resistance influence parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between internal resistance and battery internal resistance. Indicates the internal resistance of the battery; For the performance impact dataset, the battery internal resistance change impact feature model is analyzed according to different single-charge normalized datasets to determine the constant to be analyzed and form the battery internal resistance change impact feature formula. Define the characteristic model of the impact of changes in energy efficiency: ,in, This represents the correlation formula for the influence of the charging normalization parameter on the relative energy efficiency corresponding to the charging scenario analysis parameter numbered n. This represents the efficiency impact parameter factor corresponding to the charging scenario analysis parameter numbered n. This represents the time-dependent relationship between battery efficiency and efficiency. This indicates the battery energy efficiency value; For the performance impact dataset, the energy efficiency change impact feature model is analyzed according to different single-charge normalized datasets to determine the constants to be analyzed and form the energy efficiency change impact feature formula.

7. The method according to claim 6, characterized in that, The method involves combining the comprehensive charging performance change data of different historical charging objects to perform feature analysis on fault events, forming comprehensive fault event feature data, including: For different sequential charging datasets of the objects, the single battery capacity of the objects is fitted in chronological order to form the corresponding object battery capacity change curve; For different sequential charging datasets of the objects, the single internal resistance of the objects is fitted in chronological order to form the corresponding internal resistance change curves of the objects. For different sequential charging datasets of the objects, the single energy efficiency of the different objects is fitted in chronological order to form the corresponding object energy efficiency change curves; The time of the first charging failure is calibrated for different historical charging objects, and the corresponding object's battery capacity change curve, object's internal resistance change curve and object's energy efficiency change curve are calibrated according to the calibrated first failure time point. The first failure experience graphs from the beginning of the curve to the first failure time point are extracted to form object battery capacity first failure experience graphs, object internal resistance first failure experience graphs and object energy efficiency first failure experience graphs respectively. By comparing the first fault experience graphs of the battery capacity, the first fault experience graph of the internal resistance, and the first fault experience graph of the energy efficiency of different historical charging objects with the same fault, the corresponding minimum coverage graph is determined. Based on the minimum coverage pattern corresponding to different types of curves, the proportion of the first fault experience pattern corresponding to different historical charging objects is determined, and the proportions corresponding to different historical charging objects are combined to form the first fault experience proportion range of the corresponding type of experience pattern. For different fault types, the proportion range of the first fault experience in the corresponding different type experience graphs is collected to form comprehensive fault event feature data corresponding to the fault type.

8. The method according to claim 7, characterized in that, The process of acquiring the current charging performance data of the target object, combining it with the comprehensive performance change characteristic data to perform charging strategy analysis, and forming optimal recommended charging strategy data includes: Based on the current charging performance data, the current battery capacity, current battery internal resistance, current battery energy efficiency, and current planned charging time are determined. Based on the current charging performance data, extract the current charging scenario analysis parameters for different known charging scenario analysis parameters; Based on the known current charging scenario analysis parameters and the current planned charging time point, and combining the characteristic formulas for the influence of battery capacity change, battery internal resistance change, and energy efficiency change, the parameters of the unknown different charging scenario analysis parameters are determined. The determined parameters are guaranteed to meet the following conditions. Then, the parameters of all the charging scenario analysis parameters are used to form the optimal recommended charging strategy data: The battery capacity value determined based on the established parameters is close to but less than the current battery capacity value; The battery internal resistance value determined based on the established parameters is close to and greater than the current battery internal resistance value. The battery energy efficiency determined based on the established parameters is close to and less than the current battery energy efficiency. Among the selected parameter combinations, the sum of the difference between the determined battery capacity and the current battery capacity, the difference between the determined battery internal resistance and the current battery internal resistance, and the difference between the determined battery energy efficiency and the current battery energy efficiency is minimized.

9. The method according to claim 8, characterized in that, The process involves acquiring real-time charging data of the target object, combining it with current charging performance data and comprehensive performance change characteristic data to perform fault early warning analysis, and forming real-time charging fault early warning data, including: The corresponding real-time charging scenario parameter vector is obtained based on the real-time charging data, and the cosine similarity is calculated with different charging scenario clustering feature vectors to determine the charging scenario clustering feature vector with the largest cosine similarity. Based on the determined scene anomaly calibration results corresponding to the clustering feature vectors of the charging scene, real-time charging scene anomaly analysis results are generated: If the real-time charging scenario anomaly analysis result is abnormal, scenario anomaly information is output, and fault warning analysis is further performed by combining the real-time charging data and the comprehensive performance change characteristic data to form real-time charging fault warning data. Otherwise, no information is output, and fault warning analysis is further performed by combining the real-time charging data and the comprehensive performance change characteristic data to form real-time charging fault warning data.

10. The method according to claim 9, characterized in that, If the real-time charging scenario anomaly analysis result is abnormal, then scenario anomaly information is output, and fault warning analysis is further performed by combining the real-time charging data and the comprehensive performance change characteristic data to form real-time charging fault warning data; otherwise, no information is output, and fault warning analysis is further performed by combining the real-time charging data and the comprehensive performance change characteristic data to form real-time charging fault warning data, including: Based on the real-time charging data, determine the real-time battery capacity experience graph, real-time internal resistance experience graph, and real-time energy efficiency experience graph at the expected real-time charging completion time point. The real-time battery capacity graph, the real-time internal resistance graph, and the real-time energy efficiency graph are compared with their corresponding minimum coverage graphs: If all three experience patterns contain the corresponding minimum coverage pattern, and the area ratio of the minimum coverage pattern on the corresponding experience pattern is within the range of the corresponding first fault experience ratio, then a charging fault warning message is generated; otherwise, a normal charging message is generated.

Images