Battery pack abnormal monomer detection method and system based on local outlier factor algorithm

Abstract

The invention discloses a battery pack abnormal monomer detection method and system based on a local outlier factor algorithm. The method comprises the following steps: sampling and collecting voltage data and current data of each battery monomer in a battery pack by adopting a sliding window method to obtain a monomer voltage matrix; respectively obtaining a first voltage characteristic F1 and a second voltage characteristic F2 according to the monomer voltage matrix U; calculating a voltage change consistency feature F3 of each cell in the sliding window according to the cell voltage matrix U and the current data of the cell; integrating the first voltage feature F1, the second voltage feature F2 and the voltage change consistency feature F3 into feature points, and integrating and mapping the feature points into data points; and performing anomaly detection on the data point set by adopting a local outlier factor (LOF) algorithm to obtain an abnormal data point which is an abnormal single battery in the battery pack. According to the invention, the accuracy of the abnormal cells in the battery pack can be improved, and the abnormal condition type can be judged.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicle battery management, and in particular relates to a method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm. Background Art

[0002] With the development of new energy technologies, power battery packs are being used more and more widely, such as in electric vehicles, energy storage equipment, power tools, and so on. Power battery packs include multiple battery cells connected in series or in parallel. Normally, a power battery cell inconsistency failure refers to the situation in which, in the same battery pack, each cell has significant differences in parameters such as voltage, capacity, internal resistance, and temperature characteristics. This difference exceeds the normal range, resulting in a decrease in the overall performance of the battery pack and posing a safety hazard. The intuitive manifestation of power battery cell inconsistency failure is in the following two aspects: first, the power battery's internal resistance, capacity and other cell performance parameters are inconsistent; second, the power battery's operating voltage, state of charge and other working states are inconsistent. At present, the methods for identifying cell inconsistencies in electric vehicle battery packs are mainly divided into machine learning methods, data clustering analysis methods, outlier detection methods, etc.

[0003] The basic idea of ​​identifying inconsistent power battery cells through machine learning is to extract features, train the machine learning model through relevant features, and then use the trained model to identify abnormal cells. Some scholars use wavelet analysis to extract diagnostic feature vectors, involving fault diagnosis BP neural network, which is used to diagnose inconsistent battery cell faults. The disadvantage of machine learning related algorithms is that the amount of calculation is large, and the battery management system (BMS) of electric vehicles is not suitable for algorithms with large amounts of data calculation, so it is difficult to deploy in the power battery BMS system. At the same time, it is often necessary to set different hyperparameters for machine learning algorithms for different data sets. Different hyperparameter settings often bring completely different results, so it is difficult to apply between different models.

[0004] The main idea of ​​the method based on data clustering analysis is to extract feature points and then perform cluster analysis on each feature point to find outliers and identify abnormal cells. However, in some general clustering algorithms, the number of clusters must be specified first. This method requires predicting the types of inconsistent battery cells before clustering. Based on the K-means clustering algorithm, some scholars have proposed a power battery voltage inconsistency identification algorithm based on Kmeans++ clustering with dynamic k value by optimizing the number of clusters and initial center selection. Although this method can identify potential abnormal cells to a certain extent, it is still necessary to specify the number of classes before clustering.

[0005] At present, algorithms based on outlier detection usually use mathematical statistics methods, such as calculating Z scores and using the 3-σ criterion for judgment, or making box plots and identifying outliers through quartiles, to identify data points that are significantly different from most data. Such methods cannot have too high requirements on the dimension of the data. Traditional outlier detection methods are often unable to effectively process high-dimensional data. At the same time, noise and abnormal fluctuations in the data may affect the calculation results, resulting in a large amount of normal data being misjudged as abnormal. Summary of the invention

[0006] In order to solve the problems in the prior art, the present invention proposes a method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm.

[0007] In a first aspect, the present invention proposes a method for detecting abnormal cells in a battery pack based on a local outlier factor algorithm, the method comprising:

[0008] S1: Using the sliding window method, sample the voltage data and current data of each battery cell in the battery pack to obtain a cell voltage matrix;

[0009] S2: Calculate the average Z score of each battery cell in the sliding window according to the cell voltage matrix U to obtain the first voltage feature F1;

[0010] S3: Calculate the second voltage feature F2 of each cell in the sliding window according to the cell voltage matrix U;

[0011] S4: Calculate the voltage variation consistency feature F3 of each cell within the sliding window according to the cell voltage matrix U and the current data of the battery cell;

[0012] S5: Integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points;

[0013] S6: Use the local outlier factor (LOF) algorithm to perform anomaly detection on the data point set to obtain an abnormal data point, which is an abnormal battery cell in the battery pack.

[0014] In a second aspect, the present invention proposes a battery pack abnormal monomer detection system based on a local outlier factor algorithm, the system comprising:

[0015] A data acquisition module, used to respond to a fault detection request and collect voltage data and current data of each cell in the battery pack by sampling in real time;

[0016] A feature extraction module, used to process the voltage data and current data of each monomer, extract features, and obtain a first voltage feature F1, a second voltage feature F2, and a voltage change consistency feature F3;

[0017] A feature integration module, used to integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points to obtain a data point set of each battery cell of the battery pack;

[0018] A detection module, in which a local outlier factor LOF algorithm program is set to perform anomaly detection on a set of data points and determine the type of abnormal situation of the abnormal point;

[0019] The detection module is used to output and display abnormal battery cells and their abnormal condition types.

[0020] In a third aspect, the present invention provides an electronic device, comprising:

[0021] one or more processors;

[0022] A memory for storing one or more instructions, wherein when the one or more instructions are executed by the one or more processors, the one or more processors implement the method proposed in the first aspect of the present invention.

[0023] In a fourth aspect, the present invention proposes a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enables the processor to implement the method proposed in the first aspect of the present invention.

[0024] The beneficial effects of the present invention are as follows: by combining the sliding window method to calculate three different features (F1, F2 and F3), each battery can be well scored in the previous time window. After the features are normalized, the features of different scales are well normalized to the same dimension for subsequent local outlier factor LOF calculation. By selecting a suitable k value and LOF threshold, outliers can be found more accurately for further analysis, which not only overcomes the dependence of traditional clustering methods on the number of specified clusters, but also does not require a large amount of calculation related to machine learning algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flowchart of steps of an embodiment of the present invention;

[0026] Figure 2 It is a schematic diagram of the overall process of an embodiment of the present invention;

[0027] Figure 3 Schematic diagram of the structure of a battery pack abnormal monomer detection system in an embodiment of the present invention;

[0028] Figure 4 This is a graph showing the LOF scores of each cell in the battery pack during the simulation experiment of the present invention;

[0029] Figure 5 This is a diagram showing the difference between the voltage of each cell and the average voltage in the battery pack during the simulation experiment of the present invention. DETAILED DESCRIPTION

[0030] The terms "first", "second", "third", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances. This is just a way of distinguishing objects with the same attributes when describing the embodiments of this application.

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The embodiment of the present invention proposes a method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm. The method includes:

[0033] S1: Using the sliding window method, sample the voltage data and current data of each battery cell in the battery pack to obtain a cell voltage matrix.

[0034] For example, there are N battery cells (also referred to as cells) in the target battery pack (also referred to as a battery group). The size of the sliding window is set to collect M frames of voltage data of each battery cell in the battery pack.

[0035] Through the investigation and analysis of the existing electric vehicle battery management system BMS data set, it is found that when the electric vehicle is running, its BMS data has certain distortion, sampling random errors, etc. If such short-term error fragments cannot be effectively identified, it will be easy to mistakenly mark normal battery cells as abnormal battery cells. The embodiment of the present invention adopts a sliding window filtering method, which can effectively filter abnormal data and retain the original characteristics of the data.

[0036] The monomer voltage matrix U is specifically:

[0037]

[0038] Among them, u 1,1 Indicates the voltage data of the first battery cell in the battery pack in the sliding window in the first frame, u 1,N Indicates the voltage data of the Nth battery cell in the battery pack in the sliding window in the first frame, u M,1 Indicates the voltage data of the first battery cell in the battery pack in the sliding window at the Mth frame, uM,N Represents the voltage data of the Nth battery cell in the battery pack in the sliding window in the Mth frame.

[0039] Through the investigation and analysis of abnormal battery cell data in the battery packs of a large number of electric vehicles, it is found that the voltage data and voltage mean, voltage median, and current sensitivity of abnormal battery cells are different from those of ordinary battery cells. Therefore, this invention is designed to extract the Z score, consistency coefficient, and product of the single cell voltage standard deviation and median absolute deviation PVMD of the battery cell in the sliding window. These three features are used as three-dimensional space points to calculate the local outlier factor of all cells. Finally, the outlier factor is obtained as a further inconsistency discrimination standard.

[0040] S2: According to the cell voltage matrix U, calculate the average Z score of each battery cell in the sliding window to obtain the first voltage feature F1.

[0041] Specifically, the Z score of each monomer voltage data in the sliding window is calculated through the monomer voltage matrix U. First, the Z score of each frame is calculated, and then the average Z score of each monomer in the segment is calculated. The specific calculation formula is:

[0042]

[0043] Where i represents the i-th moment, j represents the j-th monomer; Z i,j Indicates the Z score of the jth monomer in the i-th frame, u i,j represents the voltage data of the jth cell at the i-th moment, represents the average voltage data of the battery pack at the i-th moment, std(u i,1 ,…,u i,N )express, represents the average Z score of each battery cell in the time window, and M represents the number of frames in the time window. The first voltage feature F1 is used to represent Right now:

[0044] It should be noted that the Z-score, also known as the standard score, is used to describe the relative position of a value to the mean. It is calculated by dividing the difference between the value and the mean by the standard deviation. In the field of data outlier detection, the Z-score can be used to detect outliers in a data set. Usually, if the absolute value of the Z-score is greater than a certain threshold (such as 2 or 3), the data point is considered an outlier.

[0045] The embodiment of the present invention calculates the average Z score of the cell voltage based on the entire segment in the sliding window, which plays a role in filtering errors and avoids the problem of false alarm caused by distortion of a certain frame.

[0046] S3: According to the cell voltage matrix U, the second voltage feature F2 of each cell in the sliding window is calculated.

[0047] Specifically, according to the cell voltage matrix U, the cell voltage standard deviation and median absolute deviation of each cell in the sliding window are calculated respectively, and the product PVMD of the cell voltage standard deviation and median absolute deviation is performed to obtain the second voltage feature F2.

[0048] The specific calculation formula of the product of the single cell voltage standard deviation and the median absolute deviation PVMD is:

[0049]

[0050] PVMD j =std(u ,j )×median(|u ,j -median(U,axis=1)|);

[0051] F2=PVMD j ;

[0052] Where, PVMD j The product of the standard deviation of the cell voltage of the jth cell and the median absolute deviation is represented by the second voltage feature F2 to represent PVMD j ;u ,j represents the column vector of all voltage data of the jth cell in the time window, u i, Indicates the cell voltage of all battery cells at the i-th frame, u i,j represents the voltage data of the jth cell at the i-th moment, std(.) represents the calculation of the standard deviation of ., median(.) represents the median of ., if there is an axis parameter, it means taking the median of the corresponding axis direction, such as median(U,axis=1) represents taking the row mean of matrix U, axis=1 means calculating the rows to obtain an M×1 column vector, where each median(|u i,j -median(u i )|) represents the median absolute deviation of the jth monomer in the sliding window.

[0053] The standard deviation of the monomer voltage indicates the change range of the monomer within this time window. If the change range is large, the standard deviation of the monomer voltage will become larger.

[0054] Since individual cells in the battery pack have large voltage deviations, they will affect the voltage mean. Therefore, the battery cell represented by the median value often reflects the normal state of the battery pack better than the average value. The median absolute deviation indicates the deviation of the cell from the median value of the voltage in the battery pack within the time window.

[0055] The product of the monomer voltage standard deviation and the median absolute deviation. This feature can reflect both the change amplitude of the monomer within the time window and the deviation from the median value. Adding this feature helps to enhance the sensitivity of the subsequent local outlier factor algorithm to abnormal monomers.

[0056] S4: Calculate the voltage variation consistency feature F3 of each cell within the sliding window according to the cell voltage matrix U and the current data of the battery cell.

[0057] In a preferred embodiment, the calculation process of the voltage variation consistency feature F3 is:

[0058] S401: Remove the first frame of the monomer voltage matrix from the monomer voltage matrix U to obtain a first voltage matrix U1.

[0059] Specifically, the first voltage matrix U1 is:

[0060]

[0061] S402: Remove the last frame of the monomer voltage matrix from the monomer voltage matrix U to obtain a second voltage matrix U2.

[0062] Specifically, the second voltage matrix U2 is:

[0063]

[0064] S403: Subtract the first voltage matrix U1 from the second voltage matrix U2 to obtain a voltage difference matrix U between every two frames of each cell in the sliding window. diff .

[0065] Specifically, the voltage difference matrix U diff The calculation formula is:

[0066] U diff =U2-U1;

[0067]

[0068] In the formula, u diff i,j Represents the voltage difference of the jth cell from the i-th to the i+1-th time.

[0069] S404: Based on the voltage difference matrix U diff As well as the current data of the battery cells, the Pearson correlation coefficient r of each cell in the sliding window is calculated.

[0070] In some preferred embodiments, according to the voltage difference matrix U diff , calculate the voltage difference matrix Udiff The average value of each row in the data is used to obtain the average voltage difference U of each frame. diff_mean ; According to the current data of each battery cell in the battery pack, the total current of the entire battery pack is calculated; according to the voltage difference matrix U diff The average voltage difference U of each column and frame of data diff_mean As well as the total current of the entire battery pack, calculate the Pearson correlation coefficient of each monomer in the sliding window.

[0071] Specifically, the calculation formula of the Pearson correlation coefficient is:

[0072]

[0073] In the formula, I j+1 represents the total current of the battery pack at the i+1th moment, u diff i,j It represents the voltage difference of the jth cell from the i-th to the i+1-th time.

[0074] It should be noted that the Pearson correlation coefficient, also known as the Pearson product-moment correlation coefficient, is a statistic used to measure the degree of linear correlation between two variables.

[0075] S405: Calculate the relative deviation RD of the Pearson correlation coefficient of each battery cell according to the Pearson correlation coefficient r of each cell in the sliding window to obtain a consistency coefficient feature F3.

[0076] Specifically, according to the Pearson correlation coefficient r of each monomer in the sliding window, the average value r of the Pearson correlation coefficients of all monomers in the entire battery pack is calculated. mean , and then calculate the relative deviation RD of the Pearson correlation coefficient of each battery cell.

[0077] Calculate the relative deviation of the Pearson correlation coefficient of each battery cell, and the calculation formula is:

[0078]

[0079] F3=RD j

[0080] Where, RD j Represents the relative deviation of the Pearson correlation coefficient of the jth monomer, RD j Use F3 to represent; r j Represents the Pearson correlation coefficient of the jth monomer in the sliding window.

[0081] In the embodiment of the present invention, the Pearson correlation coefficient is used to study the linear relationship between the battery cell voltage and the battery pack current. After the above processing from S401 to S405, the voltage change consistency feature is extracted, which reflects the consistency of the change of each cell in the battery pack and the battery pack current. By calculating its relative deviation and comparing the differences between different cells, this feature can more sensitively capture the inconsistency of cells in the battery pack caused by the increase of the internal resistance of the battery cell.

[0082] In a preferred embodiment, the first voltage feature F1, the second voltage feature F2, and the consistency coefficient feature F3 are respectively normalized, and the normalized calculation formula is:

[0083]

[0084] In the formula, F scaled represents the normalized features, F represents the input features, which are specifically F1, F2 or F3, F min Represents the minimum value of the input feature, F max Represents the maximum value among the input features.

[0085] S5: Integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points.

[0086] Specifically, in three-dimensional Euclidean coordinates, the first voltage feature F1, the second voltage feature F2, and the consistency coefficient feature F3 are integrated into a feature point, (F1, F2, F3), and the feature point (F1, F2, F3) is integrated and mapped into a data point.

[0087] The same method is used to obtain the data points corresponding to each cell in the target battery pack to form a data point set. In order to study the regularity of the data point set, the embodiment of the present invention uses a local outlier factor (LOF) algorithm to detect abnormal data points, thereby detecting abnormal cells in the battery pack.

[0088] S6: Use the local outlier factor (LOF) algorithm to perform anomaly detection on the data point set to obtain an abnormal data point, which is an abnormal battery cell in the battery pack.

[0089] In some preferred embodiments, a local outlier factor (LOF) algorithm is used to perform anomaly detection on a data point set (i.e., detect abnormal data points), and the specific process includes:

[0090] S601: Determine the number of neighbors k of each data point p, that is, determine the k value.

[0091] S602: Calculate the reachable distance reach_dist(p,o) between each data point p and a point o in its k-neighborhood.

[0092] Specifically, for each data point p, calculate the points in its k-neighborhood, k-neighborhood N k (p) refers to the k data points closest to data point p.

[0093] The calculation formula of the reachable distance reach_dist(p,o) is:

[0094] reach_dist(p,o)=max(dist(p,o),dist(p,N k (p)))

[0095] Among them, N k (p) represents the k-neighborhood of data point p, dist(p,o) represents the Euclidean distance from data point p to any of its neighboring points o, and dist(p,N k (p)) represents the data point p to its k-neighborhood N k (p) is the Euclidean distance.

[0096] S603: Calculate the local reachable density lrd(p) of each data point, and the calculation formula is:

[0097]

[0098] Among them, |N k (p)| represents the size of the k-neighborhood of any data point p, that is, the k value.

[0099] Similarly, calculate the local reachability density lrd(o) of any point o in the k-neighborhood of the data point p.

[0100] S604: Obtain a local outlier factor LOF score of the data point by comparing the local reachability density lrd(p) of the data point with the local reachability density lrd(o) of the data point o in its k-neighborhood.

[0101] Specifically, the calculation formula of the local outlier factor LOF score is:

[0102]

[0103] Where LOF(p) represents the local outlier factor LOF score of any data point p, lrd(p) represents the local reachability density of any data point p, lrd(o) represents the local reachability density of any neighbor data point o in the k-neighborhood of any data point p, o∈N k (p) represents any neighbor data point o in the k-neighborhood of any data point p.

[0104] S605: Setting a threshold, comparing the local outlier factor LOF score of the data point p with the threshold, and obtaining an abnormal data point, which is an abnormal battery cell.

[0105] Specifically, the LOF score threshold is set, and the local outlier factor LOF score of any data point p is compared with the LOF score threshold, and the data points exceeding the LOF score threshold are further judged; then, combined with the monomer voltage matrix U, the inconsistent state is classified and graded, which can be divided into voltage inconsistency fault and internal resistance inconsistency fault. Similarly, in order to avoid sampling anomalies, the data needs to be processed by sliding window.

[0106] In a preferred embodiment, when the local outlier factor LOF score of data point p exceeds a threshold, combined with the monomer voltage matrix U, the data point p is detected as an abnormal data point, and its abnormal conditions are divided into voltage abnormality (i.e., voltage inconsistency abnormality) and internal resistance abnormality (i.e., internal resistance inconsistency abnormality).

[0107] Specifically,

[0108] For voltage inconsistency anomalies (or faults): When the local outlier factor LOF score of data point p exceeds the threshold, the average value of the difference between the single cell voltage and the average voltage in the sliding window is calculated. The calculation formula is:

[0109]

[0110] V diff The (i) value reflects the degree to which the voltage of the i-th cell deviates from the average voltage within the time window.

[0111] Specifically: When V diff (i) When ≥30mV, it should be considered that the monomer voltage is inconsistent;

[0112] When V diff (i) ≥50mV, it should be regarded as moderate monomer voltage inconsistency;

[0113] When V diff (i) ≥80mV, it should be regarded as severe single cell voltage inconsistency.

[0114] For internal resistance inconsistency anomalies (or faults): When the local outlier factor LOF score of data point p exceeds the threshold, the calculation sliding window shows that for monomers with larger internal resistance, when the current changes, the voltage consumed by the internal resistance part also changes greatly, which is reflected in the monomer voltage, and the change of the monomer voltage will be more drastic. Therefore, it is necessary to first intercept the time window segment where the time is continuous and the current changes, and calculate the standard deviation of the monomer voltage in the time window. The calculation formula is:

[0115]

[0116] Then, the average standard deviation of each battery cell in the battery pack within its time window is calculated, and the calculation formula is:

[0117]

[0118] When std(u j )>3×std(u mean ), should be regarded as inconsistent monomer internal resistance;

[0119] When std(u j )>4×std(u mean ), should be regarded as moderate monomer internal resistance inconsistency;

[0120] When std(u j )>5×std(u mean ), should be regarded as severe monomer internal resistance inconsistency.

[0121] Based on the same inventive concept, the embodiment of the present invention proposes a battery pack abnormal monomer detection system based on a local outlier factor algorithm. In this system, the system has the same or similar technical features as the above-mentioned battery pack abnormal monomer detection method, which will not be described in detail below.

[0122] Reference Figure 3 As shown, the battery pack abnormal monomer detection system based on the local outlier factor algorithm includes:

[0123] A data acquisition module, used to respond to a fault detection request and collect voltage data and current data of each cell in the battery pack by sampling in real time;

[0124] A feature extraction module, used to process the voltage data and current data of each monomer, extract features, and obtain a first voltage feature F1, a second voltage feature F2, and a voltage change consistency feature F3;

[0125] A feature integration module, used to integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points to obtain a data point set of each battery cell of the battery pack;

[0126] A detection module, in which a local outlier factor LOF algorithm program is set to perform anomaly detection on a set of data points and determine the type of abnormal situation of the abnormal point;

[0127] The detection module is used to output and display abnormal battery cells and their abnormal condition types.

[0128] An embodiment of the present invention provides an electronic device, including:

[0129] one or more processors;

[0130] A memory for storing one or more instructions, wherein when the one or more instructions are executed by the one or more processors, the one or more processors implement the method proposed in the first aspect of the present invention.

[0131] An embodiment of the present invention provides a computer-readable storage medium having executable instructions stored thereon. When the executable instructions are executed by a processor, the processor implements the method provided in the first aspect of the present invention.

[0132] Simulation verification:

[0133] Based on computers and Python programming language, the present invention uses the actual inconsistent battery data of battery cells for simulation verification, constructs the relevant algorithm designed by the present invention, reads the battery pack data, and finally obtains the following experimental results:

[0134] Figure 4 The LOF score diagram of each cell in the battery pack during the simulation experiment of the present invention. Figure 4 In this test, the k value in the LOF local outlier factor is set to 5 and the threshold is set to 10. Figure 4 It can be seen that at this moment, the local outlier factor of monomer No. 24 reached 16.86.

[0135] from Figure 4 In the method, three features are automatically extracted and the LOF local outlier factor is calculated, and the battery cells with inconsistent cells are automatically located.

[0136] Figure 5 This is a diagram showing the difference between the voltage of each cell and the average voltage in the battery pack during the simulation experiment of the present invention. Figure 5 In the example, when the algorithm is used to monitor the operation data of electric vehicles, when the battery cell LOF is abnormal, the difference between the battery cell voltage and the average voltage in the sliding window is obtained. Figure 5 It can be seen that during the operation of cell No. 24, there is a large voltage drop, there is inconsistency in the cell voltage reduction, and the voltage fluctuates greatly, and there is also an inconsistency in the increase of the cell internal resistance.

[0137] from Figure 5 It can be seen that the No. 24 battery cell is indeed inconsistent, and different from the method of directly detecting the battery cell voltage, this method is not only sensitive to the drop in battery voltage, but also sensitive to the increase in the internal resistance of the battery, which proves the sensitivity of this algorithm to battery cell imbalance.

[0138] A person skilled in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the relevant hardware through a program, and the program can be stored in a computer-readable storage medium, which can include: ROM, RAM, disk or CD, etc.

[0139] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting abnormal cells in a battery pack based on a local outlier factor algorithm, characterized in that: include: The sliding window method is used to sample and collect the voltage and current data of each battery cell in the battery pack to obtain the cell voltage matrix; According to the cell voltage matrix U, the average Z score of each battery cell in the sliding window is calculated to obtain the first voltage feature F1; According to the monomer voltage matrix U, the second voltage feature F2 of each monomer in the sliding window is calculated; According to the cell voltage matrix U and the current data of the battery cell, the voltage variation consistency feature F3 of each cell in the sliding window is calculated; Integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points; The local outlier factor (LOF) algorithm is used to perform anomaly detection on the data point set to obtain an abnormal data point, which is an abnormal battery cell in the battery pack.

2. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 1, characterized in that: According to the cell voltage matrix U, the cell voltage standard deviation and median absolute deviation of each cell in the sliding window are calculated respectively, and the product PVMD of the cell voltage standard deviation and median absolute deviation is performed to obtain the second voltage feature F2.

3. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 1, characterized in that: The calculation process of the voltage variation consistency feature F3 is as follows: Remove the first frame of the monomer voltage matrix from the monomer voltage matrix U to obtain a first voltage matrix U1; Remove the last frame of the single voltage matrix from the single voltage matrix U to obtain a second voltage matrix U2; Subtract the first voltage matrix U1 from the second voltage matrix U2 to obtain the voltage difference matrix U between every two frames of each monomer in the sliding window: diff ; According to the voltage difference matrix U diff And the current data of the battery cells, calculate the Pearson correlation coefficient r of each cell in the sliding window; According to the Pearson correlation coefficient r of each cell in the sliding window, the relative deviation RD of the Pearson correlation coefficient of each battery cell is calculated to obtain the consistency coefficient feature F3.

4. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 3, characterized in that: According to the voltage difference matrix U diff , calculate the voltage difference matrix U diff The average value of each row in the data is used to obtain the average voltage difference U of each frame. diff_mean ; According to the current data of each battery cell in the battery pack, the total current of the entire battery pack is calculated; According to the voltage difference matrix U diff The average voltage difference U of each column and frame of data diff_mean As well as the total current of the entire battery pack, calculate the Pearson correlation coefficient of each monomer in the sliding window.

5. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 1, characterized in that: Also includes: The first voltage feature F1, the second voltage feature F2, and the consistency coefficient feature F3 are respectively normalized.

6. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 1, characterized in that: The local outlier factor LOF algorithm is used to detect anomalies in a data point set. The specific process includes: Determine the number of neighbors k of the data point p, that is, determine the k value; Calculate the reachable distance reach_dist(p,o) between data point p and data point o in its k-neighborhood; Calculate the local reachability density of data point p and data point o in its k-neighborhood respectively; By comparing the local reachability density lrd(p) of data point p and the local reachability density lrd(o) of data point o in its k-neighborhood, the local outlier factor LOF score of the data point is obtained; A threshold is set, and the local outlier factor LOF score of the data point p is compared with the threshold to obtain an abnormal data point, which is an abnormal battery cell.

7. The method and system for detecting abnormal cells in a battery pack based on a local outlier factor algorithm according to claim 6, characterized in that: When the local outlier factor LOF score of the data point p exceeds the threshold, combined with the monomer voltage matrix U, the data point p is detected as an abnormal data point, and its abnormal situation is divided into voltage abnormality and internal resistance abnormality.

8. The battery pack abnormal monomer detection system based on the local outlier factor algorithm is characterized by: The system includes: A data acquisition module, used to respond to a fault detection request and collect voltage data and current data of each cell in the battery pack by sampling in real time; A feature extraction module, used to process the voltage data and current data of each monomer, extract features, and obtain a first voltage feature F1, a second voltage feature F2, and a voltage change consistency feature F3; A feature integration module, used to integrate the first voltage feature F1, the second voltage feature F2, and the voltage change consistency feature F3 into feature points, and map the feature points into data points to obtain a data point set of each battery cell of the battery pack; A detection module, in which a local outlier factor (LOF) algorithm is set to detect anomalies in a data point set and determine the type of anomaly of anomaly points; The detection module is used to output and display abnormal battery cells and their abnormal condition types.

9. An electronic device, comprising: one or more processors; A memory for storing one or more instructions, wherein when the one or more instructions are executed by the one or more processors, the one or more processors are enabled to implement the method according to any one of claims 1 to 7. 10 . A computer-readable storage medium having executable instructions stored thereon, wherein when the executable instructions are executed by a processor, the processor is enabled to implement the method according to any one of claims 1 to 7.

Images