A method for evaluating consistency defects of automobile battery monomers based on cluster analysis

Abstract

The application provides a battery cell consistency defect evaluation method based on cluster analysis, which fully utilizes the operation big data of the same specification battery used on different vehicles, provides a comprehensive inconsistency defect detection system including cell abnormal frequency, aging speed, voltage divergence and the like, can early discover and alarm potential cell defect risks in the whole life cycle of a specific target vehicle, overcomes the limitation that the prior art can only meet single vehicle diagnosis, and significantly improves the detection accuracy and provides wide adaptability.

Description

Technical Field

[0001] This invention belongs to the field of automotive power battery defect assessment technology, specifically involving a method for assessing the consistency defects of automotive battery cells based on cluster analysis. Background Technology

[0002] Currently, the power battery packs used in electric vehicles inevitably exhibit inconsistencies in the performance of individual battery cells due to the large number of cells within the pack. This inconsistency can cause the battery pack performance to fail to meet design requirements and lead to excessively rapid lifespan degradation. In severe cases, it can even cause overheating and thermal runaway, posing safety risks. However, existing technologies for detecting battery inconsistencies are not yet perfect. Large-scale sensing devices targeting each cell cannot achieve rapid consistency assessment, hindering early detection of inconsistencies and the implementation of necessary measures to address potential faults or safety risks. Furthermore, current consistency testing is limited to the single-vehicle level, lacking effective means to achieve more accurate assessments using a wider range of samples for cells of the same specifications used in different vehicle models and operating conditions. Therefore, providing a more accurate and real-time method for assessing the consistency of automotive battery cells is a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0003] In view of this, the present invention provides a method for evaluating the consistency defects of automotive battery cells based on cluster analysis, specifically including the following steps:

[0004] Step 1: The new energy vehicle big data platform collects voltage, current, temperature, and corresponding time data of battery cells of the same specifications in each electric vehicle; preprocesses this data to divide the vehicle into charging segments and driving segments.

[0005] Step 2: For any segment M, calculate the standardized correlation coefficient between the voltage of cell i and the voltage of other cells, calculate the absolute value of the difference between the voltage of cell i and the mean voltage of all cells in the charging segment, and calculate the Hausdorff distance between the voltage of cell i and the mean voltage of all cells in the driving segment.

[0006] Step 3: Use the results obtained in Step 2 to perform K-means clustering on the battery cells to obtain two clusters: abnormal cells and normal cells. Then, determine the specific cluster to which each battery cell belongs based on the cluster centers of the two clusters.

[0007] Step 4: Based on the two clusters and cluster centers obtained in Step 3, calculate the Dunn index between normal and abnormal single-unit clusters in segment M. Compare this index with a preset Dunn index threshold, and record and alarm the single-unit numbers in abnormal single-unit clusters that exceed the threshold. Statistically analyze the recording of abnormal single units across all segments throughout the entire lifecycle of vehicle q, and calculate the proportion f of alarm segments for single-unit i in the total number of segments. i ;

[0008] Step 5: Calculate the average distance S from all cells in the normal cluster to the cluster center for the charging segment, and record the corresponding driving mileage L, average charging temperature and temperature change range for the segment; select the average distance S and driving mileage L corresponding to different average charging temperatures and perform linear fitting, define the slope of the fitted line as the cell aging coefficient, and the selected maximum value of S as the cell voltage divergence coefficient.

[0009] Step Six: Periodically perform the aforementioned steps on the same vehicle model, and according to the stated f i Assess the quantity and frequency risks of monomer defects, assess the risk of abnormal aging of monomers based on the monomer aging coefficient, and assess the risk of abnormal dispersion of monomers based on the monomer voltage divergence coefficient.

[0010] Furthermore, the preprocessing performed in step one specifically includes: deduplicating multiple frames of data collected at the same time point and retaining only one frame; deleting the data frame if the battery cell voltage is missing more than two cells; and determining the charging segment and driving segment based on the flag bits of the relevant data items. If the time interval between adjacent segments of the same type is less than five minutes, they are merged into the same segment.

[0011] Furthermore, step two specifically includes:

[0012] First, extract the voltage sequence of unit i from frame m in segment M: [v i1 v i2 …v im ];

[0013] Perform moving average filtering on the voltage vector, calculate the first-order difference of the filtering result, and construct the voltage difference matrix ΔV for all n individual cells. M :

[0014]

[0015] For any single cell i, calculate the average value across all frames of its voltage differential sequence:

[0016]

[0017] The correlation coefficient between the voltage of cell i and the voltage of other cells j is calculated using the following formula.

[0018]

[0019] The average correlation coefficient between the voltage of cell i and the voltages of other cells is taken. Then, through standardization, the standardized correlation coefficient vector of all monomers is obtained:

[0020] Extract the individual cell voltage sequences [v] from the last 5% SOC of the charging segment. i1 v i2 …v im ] and calculate its mean.

[0021]

[0022] calculate The absolute value d of the difference from the average voltage of all individual cells i :

[0023]

[0024] The above calculation results were standardized to obtain the corresponding...

[0025] For the driving segment, the average voltage sequence is first obtained by averaging the voltage of each individual unit in each frame j.

[0026]

[0027] Hausdorf distance:

[0028] d i =H(v) i ,v)

[0029] The above calculation results are standardized to obtain the corresponding d. i std .

[0030] Furthermore, the specific process of K-means clustering performed in step three is as follows:

[0031] Using the results obtained in step two, construct a set S = {p1, p2, ..., p...} n}, where p i =[ρ i std ,d i std Clustering is performed to obtain abnormal single-monitor cluster u1 and normal single-monitor cluster u2, and the cluster number h is set to 2;

[0032] Based on the centers c1 and c2 of the two clusters, select the corresponding two battery cells p. i and p j Calculate the Euclidean distances from all monomers to the centers of the two clusters and determine the cluster to which a monomer belongs using the following criteria:

[0033]

[0034] After completing the above determination for all individuals, recalculate the cluster centers c1 and c2 as follows:

[0035]

[0036] Repeat the above calculation of Euclidean distance and determination of the cluster to which the individual belongs until the cluster center no longer changes.

[0037] Furthermore, in step four, the Dunn index (DVI) between the two clusters is calculated based on the following formula:

[0038]

[0039] Where d(u1,u2) represents the distance between the centers of clusters u1 and u2, and max 1≤k≤2 d′(u k () represents the maximum distance between any two points within cluster u1 or u2;

[0040] Set the threshold DVI of the Dunn index. s And compare it with the calculated DVI. If DVI > DVI s If h=2, record the individual cluster number and alarm in the abnormal individual cluster; otherwise, set h to 1 and wait to execute the next clustering.

[0041] The vector obtained by statistically analyzing the recorded abnormal individuals across all segments of the entire lifecycle of vehicle q is: [f1, f2, ..., f n ].

[0042] Furthermore, the specific process for determining the monomer aging coefficient and monomer voltage divergence coefficient in step five includes:

[0043] For any charging segment, calculate the average distance S from all cells in a normal cell cluster to the cluster center, record the corresponding driving mileage L, and calculate the average charging temperature T using the sensing data from each temperature probe. M and temperature variation range ΔT M :

[0044]

[0045]

[0046] Where m is the number of segments and n is the number of temperature probes. This indicates that the j-th probe in the i-th frame measures the temperature;

[0047] BMS system fault diagnosis alarm temperature threshold and Temperature range T serves as the lower and upper bounds. d Intervals, defining temperature ranges: And create the corresponding sets A0, A1, ..., A x ; Traverse all charging segments, if the ΔT of a certain segment M >0.7T d If the segment is skipped, the next segment is traversed; if the segment's ΔT... M ≤0.7T d Then, based on the average charging temperature T of this segment... M Temperature range The corresponding (L) i ,S i Store in set A i middle;

[0048] Choose set A, which has the most data points from all sets. max And using all the data points (L1,S1), (L2,S2),...(L n ,S n Perform a linear fit based on the least squares method, define the slope k obtained from the fit as the monomer aging coefficient, and record the set A. max Maximum distance S from the center of a normal single cluster max As the voltage divergence coefficient of a single unit.

[0049] Furthermore, the specific process for assessing the quantity and frequency risks of individual defects for a certain vehicle model in step six includes:

[0050] Based on the following matrix obtained through step four for m vehicles to be evaluated:

[0051]

[0052] Among them, f i j This represents the ratio of the number of alarm segments for individual vehicle i to the total number of segments for vehicle j.

[0053] Calculate the Quantitative Risk Index (QR) using the following formulas. i Frequency risk indicator FR i :

[0054]

[0055]

[0056] Wherein, num(f i j ≠0) indicates the ratio f i The number of vehicles that is not zero, QR i Indicates the ratio f i The percentage of vehicles with a non-zero number of vehicles out of the total number of vehicles, FR i f represents the ratio that is not zero. i Mean;

[0057] Set quantity risk thresholds separately and in and frequency risk threshold FR s Assess the corresponding risks according to the following rules:

[0058] If for any single entity i, we have There are no abnormal single-unit defects;

[0059] If a monomer i exists, And FR i >FR s If the defect is found, then there is an abnormal monomer defect, and monomer i is an abnormal monomer; otherwise, there is no abnormal monomer defect.

[0060] If a monomer i exists, Then there is an abnormal monomer defect, and monomer i is an abnormal monomer;

[0061] The process of assessing the risk of abnormal aging of monomers includes:

[0062] Based on the individual aging coefficient vector of m vehicles obtained in step five: [k1,k2,…,k m ], and the corresponding threshold k s Compare and calculate the aging rate risk factor (ARR):

[0063]

[0064] Set the ARR (Aging Rate Risk Threshold) s When ARR ≥ ARR s It was determined that these m vehicles had a risk of individual abnormal aging;

[0065] The process of assessing the risk of monomeric anomalous divergence includes:

[0066] For each temperature range, set a corresponding threshold for the voltage divergence coefficient: S s1 ,S s2 ,…,S sx And based on the individual voltage divergence coefficient vector of m vehicles obtained in step five: The threshold values ​​corresponding to the temperature ranges of each element are used to construct a threshold vector in sequence: The individual cell voltage divergence risk factor DR is calculated using the following formula:

[0067]

[0068] Set the individual voltage divergence risk threshold DR s When DR≥DR s It was determined that these m vehicles posed a risk of individual abnormal dispersion.

[0069] The cluster analysis-based method for assessing the consistency defects of automotive battery cells provided by the present invention fully utilizes the big data of batteries of the same specifications used in different vehicles to provide a comprehensive inconsistency defect detection system, including the frequency of abnormality of cells, aging rate, voltage dispersion, etc. It can detect and warn of potential cell defect risks throughout the entire life cycle of a specific target model at an early stage, overcome the limitation of existing technologies that can only meet the diagnosis of single vehicles, significantly improve the accuracy of detection and provide broad adaptability. Attached Figure Description

[0070] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 A flowchart for evaluating the consistency defects of automotive battery cells based on cluster analysis, provided by this invention;

[0072] Figure 2 The flowchart for data preprocessing provided by this invention. Detailed Implementation

[0073] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0074] like Figure 1 As shown, this invention provides a method for evaluating the consistency defects of automotive battery cells based on cluster analysis, specifically including the following steps:

[0075] Step 1: The new energy vehicle big data platform collects voltage, current, temperature, and corresponding time data of battery cells of the same specifications in each electric vehicle; preprocesses this data to divide the vehicle into charging segments and driving segments.

[0076] Step 2: For any segment M, calculate the standardized correlation coefficient between the voltage of cell i and the voltage of other cells, calculate the absolute value of the difference between the voltage of cell i and the mean voltage of all cells in the charging segment, and calculate the Hausdorff distance between the voltage of cell i and the mean voltage of all cells in the driving segment.

[0077] Step 3: Use the results obtained in Step 2 to perform K-means clustering on the battery cells to obtain two clusters: abnormal cells and normal cells. Then, determine the specific cluster to which each battery cell belongs based on the cluster centers of the two clusters.

[0078] Step 4: Based on the two clusters and cluster centers obtained in Step 3, calculate the Dunn index between normal and abnormal single-unit clusters in segment M. Compare this index with a preset Dunn index threshold, and record and alarm the single-unit numbers in abnormal single-unit clusters that exceed the threshold. Statistically analyze the recording of abnormal single units across all segments throughout the entire lifecycle of vehicle q, and calculate the proportion f of alarm segments for single-unit i in the total number of segments. i ;

[0079] Step 5: Calculate the average distance S from all cells in the normal cluster to the cluster center for the charging segment, and record the corresponding driving mileage L, average charging temperature and temperature change range for the segment; select the average distance S and driving mileage L corresponding to different average charging temperatures and perform linear fitting, define the slope of the fitted line as the cell aging coefficient, and the selected maximum value of S as the cell voltage divergence coefficient.

[0080] Step Six: Periodically perform the aforementioned steps on the same vehicle model, and according to the stated f i Assess the quantity and frequency risks of monomer defects, assess the risk of abnormal aging of monomers based on the monomer aging coefficient, and assess the risk of abnormal dispersion of monomers based on the monomer voltage divergence coefficient.

[0081] In a preferred embodiment of the present invention, the specific preprocessing procedure performed in step one is as follows: Figure 2 As shown, this includes: deduplicating multiple frames of data collected at the same time point and retaining only one frame; deleting a data frame if the battery cell voltage is missing more than two cells; and determining charging segments and driving segments based on the flag bits of relevant data items, and merging adjacent segments of the same type into the same segment if the time interval is less than five minutes.

[0082] In a preferred embodiment of the present invention, step two specifically includes:

[0083] First, extract the voltage sequence of unit i from frame m in segment M: [v i1 v i2 …v im ];

[0084] The following formula can be used to perform moving average filtering on the voltage vector:

[0085]

[0086] v[τ-i] represents the data point at time τ-i, v MAF [τ] is the current output, and Ψ represents the number of data points selected during the calculation process, which can be 10 in this embodiment.

[0087] Calculate the first-order difference of the filtering result: Δv MAF [τ]=v MAF [τ+1]-v MAF [τ], and construct all n individual unit voltage difference matrices ΔV M :

[0088]

[0089] For any single cell i, calculate the average value across all frames of its voltage differential sequence:

[0090]

[0091] The correlation coefficient between the voltage of cell i and the voltage of other cells j is calculated using the following formula.

[0092]

[0093] The average correlation coefficient between the voltage of cell i and the voltages of other cells is taken. Then, through standardization, the standardized correlation coefficient vector of all monomers is obtained:

[0094] Standardization can be performed using methods well-known in the art, applicable to any dataset X = {x1, x2, ..., x...} n Each element in} is defined as follows:

[0095]

[0096] in,

[0097] mean

[0098] Standard deviation

[0099] Extract the individual cell voltage sequences [v] from the last 5% SOC of the charging segment. i1 v i2 …v im ] and calculate its mean.

[0100]

[0101] calculate The absolute value d of the difference from the average voltage of all individual cells i :

[0102]

[0103] The above calculation results were standardized to obtain the corresponding...

[0104] For the driving segment, the average voltage sequence is first obtained by averaging the voltage of each individual unit in each frame j.

[0105]

[0106] Based on voltage sequence [v i1 v i2 ... v im ]and The Hausdorff distance for each individual i is calculated from the two sequences:

[0107]

[0108] Hausdorff distance is a measure of the similarity between two sets of points. It is a definition of the distance between two sets of points: Suppose we have two sets A = {a1 a2 ... a...} q}, B={b1 b2 ... b q}, then the Hausdorff distance between these two sets of points is defined as:

[0109] H(A,B)=max(h(A,B),h(B,A))

[0110] Where h(A,B) = max a∈A min b∈B ||ab||, h(B,A)=max b∈B min a∈A ||ba||, ||·|| is the Euclidean distance paradigm between point sets A and B.

[0111] h(A,B) first considers each point a in the point set A. i to a distance from this point a i The nearest midpoint of set B, bj The distance between ||a i -b j Sort the data and then take the maximum value among the distances as the value of h(A,B). h(B,A) can be obtained similarly. The Hausdorff distance H(A,B) represents the larger of h(A,B) and h(B,A).

[0112] The Hausdorff distance d between the voltage vector of the i-th individual and the average voltage vector can be calculated. i The calculation results are standardized to obtain the corresponding...

[0113] In a preferred embodiment of the present invention, the specific process of K-means clustering performed in step three is as follows:

[0114] Using the results obtained in step two, construct a set S = {p1, p2, ..., p...} n},in Perform clustering to obtain abnormal single-monitor cluster u1 and normal single-monitor cluster u2, and set the cluster number h to 2;

[0115] Based on the centers c1 and c2 of the two clusters, select the corresponding two battery cells p. i and p j Calculate the Euclidean distances from all monomers to the centers of the two clusters and determine the cluster to which a monomer belongs using the following criteria:

[0116]

[0117] After completing the above determination for all individuals, recalculate the cluster centers c1 and c2 as follows:

[0118]

[0119] Repeat the above calculation of Euclidean distance and determination of the cluster to which the individual belongs until the cluster center no longer changes.

[0120] Furthermore, in step four, the Dunn index (DVI) between the two clusters is calculated based on the following formula:

[0121]

[0122] Where d(u1,u2) represents the distance between the centers of clusters u1 and u2, and max 1≤k≤2 d′(u k () represents the maximum distance between any two points within cluster u1 or u2;

[0123] Set the threshold DVI of the Dunn index. s And compare it with the calculated DVI. If DVI > DVI sIf h=2, record the individual cluster number and alarm in the abnormal cluster; otherwise, set h to 1 and wait for the next clustering. This invention preferably uses DVI. s =5, but not limited to this.

[0124] The vector obtained by statistically analyzing the recorded abnormal individuals across all segments of the entire lifecycle of vehicle q is: [f1, f2, ..., f n ].

[0125] In a preferred embodiment of the present invention, the specific process of determining the monomer aging coefficient and the monomer voltage divergence coefficient in step five includes:

[0126] For any charging segment, calculate the average distance S from all cells in a normal cell cluster to the cluster center, record the corresponding driving mileage L in kilometers, and calculate the average charging temperature T using the sensing data from each temperature probe. M and temperature variation range ΔT M :

[0127]

[0128]

[0129] Where m is the number of segments and n is the number of temperature probes. This indicates that the j-th probe in the i-th frame measures the temperature;

[0130] BMS system fault diagnosis alarm temperature threshold and Temperature range T serves as the lower and upper bounds. d Intervals, defining temperature ranges: And create the corresponding sets A0, A1, ..., A x ; Traverse all charging segments, if the ΔT of a certain segment M >0.7T d If the segment is skipped, the next segment is traversed; if the segment's ΔT... M ≤0.7T d Then, based on the average charging temperature T of this segment... M Temperature range The corresponding (L) i ,S i Store in set A i In this invention, T is preferred. d =3℃;

[0131] Choose set A, which has the most data points from all sets. max And using all the data points (L1,S1), (L2,S2),...(L n ,S nPerform a linear fit based on the least squares method, define the slope k obtained from the fit as the monomer aging coefficient, and record the set A. max Maximum distance S from the center of a normal single cluster max As the voltage divergence coefficient of a single unit.

[0132] In a preferred embodiment of the present invention, the specific process of assessing the quantity risk and frequency risk of individual defects for a certain vehicle model in step six includes:

[0133] Based on the following matrix obtained through step four for m vehicles to be evaluated:

[0134]

[0135] Among them, f i j This represents the ratio of the number of alarm segments for individual vehicle i to the total number of segments for vehicle j.

[0136] Calculate the Quantitative Risk Index (QR) using the following formulas. i Frequency risk indicator FR i :

[0137]

[0138]

[0139] Wherein, num(f i j ≠0) indicates the ratio f i The number of vehicles that is not zero, QR i FR represents the percentage of vehicles with a ratio fi that is not zero out of the total number of vehicles. i f represents the ratio that is not zero. i Mean;

[0140] Set quantity risk thresholds separately and in and frequency risk threshold FR s Assess the corresponding risks according to the following rules:

[0141] If for any single entity i, we have There are no abnormal single-unit defects;

[0142] If a monomer i exists, And FR i >FR s If the defect is found, then there is an abnormal monomer defect, and monomer i is an abnormal monomer; otherwise, there is no abnormal monomer defect.

[0143] If a monomer i exists, Then there is an abnormal monomer defect, and monomer i is an abnormal monomer;

[0144] Preferred examples of the present invention FR s =0.5, but not limited to this.

[0145] The process of assessing the risk of abnormal aging of monomers includes:

[0146] Based on the individual aging coefficient vector of m vehicles obtained in step five: [k1,k2,…,k m ], and the corresponding threshold k s Compare and calculate the aging rate risk factor (ARR):

[0147]

[0148] Set the ARR (Aging Rate Risk Threshold) s When ARR ≥ ARR s It was determined that these m vehicles had a risk of individual abnormal aging;

[0149] Example k of the present invention s The upper quartile of the aging coefficient k for all vehicle models, ARR s =0.2, but not limited to this.

[0150] The process of assessing the risk of monomeric anomalous divergence includes:

[0151] For each temperature range, set a corresponding threshold for the voltage divergence coefficient: S s1 ,S s2 ,…,S sx And based on the individual voltage divergence coefficient vector of m vehicles obtained in step five: The threshold values ​​corresponding to the temperature ranges of each element are used to construct a threshold vector in sequence: The individual cell voltage divergence risk factor DR is calculated using the following formula:

[0152]

[0153] Set the individual voltage divergence risk threshold DR s When DR≥DR s It was determined that these m vehicles posed a risk of individual abnormal dispersion.

[0154] Preferred example S of the present invention s1 ,S s2 ,…,S sx DR is the upper quartile of the divergence coefficient S for all vehicle models within a specific temperature range. s =0.2, but not limited to this.

[0155] It should be understood that the sequence number of each step in the embodiments of the present invention 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.

[0156] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for evaluating the consistency defects of automobile battery cells based on cluster analysis, characterized in that: Specifically, the following steps are included: Step 1: The new energy vehicle big data platform collects voltage, current, temperature, and corresponding time data of battery cells of the same specifications in each electric vehicle; preprocesses this data to divide the vehicle into charging segments and driving segments. Step two, for any one segment M of the individual cells i , respectively, the normalized correlation coefficient between the voltage of the individual cells i and other individual cell voltages, the absolute value of the difference between the voltage of the individual cells i and the mean of all individual cell voltages in the charging segment, and the Hausdorff distance between the voltage of the individual cells i and the mean of all individual cell voltages in the driving segment; Step 3: Using the results obtained in Step 2, perform K-means clustering on the battery cells to obtain two clusters: abnormal cells and normal cells. Then, determine the specific cluster to which each battery cell belongs based on the cluster centers of the two clusters. Step four, calculate the segment according to the two clusters and the cluster center obtained in step three M The Dunn index between normal monomer clusters and abnormal monomer clusters is compared with a preset Dunn index threshold value, and when the Dunn index is greater than the threshold value, the number of monomers in the abnormal monomer cluster is recorded and an alarm is given; the number of vehicles q The abnormal monomers in all segments in the whole life cycle are recorded, and the number of monomers i The alarm segment accounts for a ratio in the total segment f i ; Step five, calculate the average distance from all monomers in the normal cluster to the cluster center for the charging segment S and record the corresponding driving mileage of the segment L , the average temperature of charging and the temperature variation range Select the distance average value corresponding to different average charging temperatures S With the driving range L Linear fitting, define the slope of the fitted line as the monomer aging coefficient, the selected S The maximum value is the monomer voltage divergence coefficient; Step six, periodically performing the previous steps for the same vehicle model and updating the model based on the results f i evaluating a quantity risk and a frequency risk of occurrence of a monomer defect, evaluating a risk of occurrence of a monomer abnormal aging based on the monomer aging coefficient, and evaluating a monomer abnormal divergence risk based on the monomer voltage divergence coefficient.

2. The method of claim 1, wherein: The preprocessing performed in step one specifically includes: deduplicating multiple frames of data collected at the same time point and retaining only one frame; deleting the frame if the battery cell voltage is missing more than two cells; and determining the charging segment and driving segment based on the flag bits of the relevant data items. If the time interval between adjacent segments of the same type is less than five minutes, they are merged into the same segment.

3. The method of claim 2, wherein: Step two specifically includes: First, extract the snippet M monomers i of m frame voltage sequence: ; A moving average filter is performed on the voltage vectors, the first difference of the filtered result is calculated and a total n matrix of single cell voltage differences Δ V M : For any monomer i the average value of all its frames is calculated: The monomers were calculated using the following equation i Voltage vs. other monomers j Correlation coefficient between voltages : for the monomers i The correlation coefficient between the voltage and the voltage of the other monomers is averaged The normalized correlation coefficient vector of all monomers is obtained by normalization processing: ; Extracting the last 5% SOC of each monobloc from the charging segment i Voltage sequence and computing their mean : Computations Absolute value of the difference from the average of all monomer voltages d i : The above calculation results are subjected to standardization to obtain corresponding ; For the driving segment, first the average voltage sequence is obtained for each frame j by averaging the individual voltages in each frame ; Based on the voltage sequence With The Hausdorff distance of each monomer i is calculated from the two sequences: The above calculation results are subjected to standardization to obtain corresponding .

Images