Battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy combination

Abstract

The invention discloses a battery fault diagnosis method based on the combination of DBSCAN, an empirical distribution function and information entropy, and the method comprises the steps: carrying out the preprocessing of obtained battery voltage data, obtaining the battery voltage data with a normal timestamp and no obvious abnormal value, and dividing the data; the information entropy and the empirical distribution data are adopted to count the battery threshold state, and self-adaptive abnormal early warning thresholds of different battery packs are determined; calculating a voltage outlier degree by using a DBSCAN algorithm, and calculating an outlier weight; and performing joint early warning according to the self-adaptive abnormal early warning threshold and the discrete weight. According to the method, the self-adaptive abnormal early warning threshold values of different battery packs are determined through the empirical distribution function to adapt to individual differences of different batteries, then whether the batteries are in the abnormal state or not is judged through the information entropy and the abnormal coefficient, fault early warning is carried out according to the combination of the DBSCAN and the information entropy, the fault early warning accuracy is improved, and the fault early warning efficiency is improved. And safe operation of the battery system is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of battery diagnosis, and in particular to a battery fault diagnosis method based on DBSCAN, an empirical distribution function and information entropy. Background Art

[0002] During operation, battery systems may experience failures due to electrochemical aging, thermal runaway, mechanical damage, overcharge / over-discharge, and other factors. These can lead to performance degradation at best, or even safety incidents such as fire and explosion at worst. Therefore, fault diagnosis of battery systems is necessary. Model-based battery anomaly diagnosis methods have very limited model transferability during the battery anomaly diagnosis process. Furthermore, the computing power requirements and data quality of machine learning-based battery anomaly diagnosis methods restrict their further development. Therefore, statistical-based battery anomaly diagnosis methods remain a hot topic of research. Among the many statistical algorithms, information entropy algorithms stand out due to their strong interpretability and robustness.

[0003] Battery fault diagnosis and early warning methods based on information entropy often combine standard normalization and the 3σ method to calculate the anomaly coefficient and determine battery abnormality. Existing research considers an anomaly coefficient exceeding 3 to be a low-probability event, indicating an abnormal battery state. However, due to differences in battery manufacturing and operating conditions, the anomaly coefficient may still exceed 3 during normal operation. A constant threshold cannot meet the warning requirements under various battery operating conditions, making it necessary to determine an adaptive anomaly threshold.

[0004] The calculation of information entropy is limited by its calculation principle. As battery anomalies become more severe, multiple outliers may appear, and the information entropy calculation result will decrease. Relying solely on the anomaly coefficient to determine battery anomaly status will cause the coefficient to decrease as the anomaly worsens, which is inconsistent with the actual effect. It is necessary to improve the information entropy anomaly diagnosis method for the entire battery life cycle.

[0005] The early warning effect achieved by calculating battery inconsistency information based on information entropy is based on the consistency of battery voltage. However, although the consistency of battery voltage is strong, the battery voltage value has its own strong changes during the alternation of charging conditions and discharging conditions and when the lithium iron phosphate battery passes through the plateau period, and the inconsistency between batteries fluctuates. Therefore, if the battery voltage state is not divided and the information entropy is directly calculated for the battery voltage, it will cause false alarms of battery abnormality information. Summary of the Invention

[0006] The purpose of the present invention is to provide a battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy. The adaptive abnormal warning threshold of different battery packs is determined by the empirical distribution function, and whether the battery is in an abnormal state is judged by information entropy and abnormal coefficient. Fault warning is performed based on the combination of DBSCAN and information entropy, which improves the accuracy of fault warning and ensures the safe operation of the battery system.

[0007] To achieve the above objectives, the present invention provides a battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy, comprising the following steps:

[0008] S1. Acquire battery voltage data, preprocess the battery voltage data to obtain battery voltage data with normal timestamps and no obvious abnormal values, and divide the battery voltage data into a threshold training data set and an algorithm application data set;

[0009] S2. Use information entropy and empirical distribution data to statistically analyze the battery threshold status in the threshold training data set and determine the adaptive abnormal warning thresholds for different battery packs;

[0010] S3, using the DBSCAN algorithm to calculate the voltage outlier degree in the same calculation window in the algorithm application data set, and calculate the outlier weight;

[0011] S4. Perform joint warning based on adaptive abnormal warning threshold and discrete weight.

[0012] Preferably, step S2 specifically includes:

[0013] S21, calculating the information entropy and anomaly coefficient of the threshold training data set;

[0014] S22. Observe the empirical distribution of the abnormal coefficient using the empirical distribution function;

[0015] S23. Select the abnormal coefficient corresponding to the observation value of 0.998 as the adaptive abnormal warning threshold for different battery packs.

[0016] Preferably, in step S3, the DBSCAN algorithm is used to calculate the voltage outlier degree in the same calculation window in the algorithm application data set, including:

[0017] The DBSCAN algorithm is used to perform cluster analysis on battery voltage data, identify battery charge and discharge intervals, and determine whether there are obvious outliers in the battery voltage under the calculation window. If outlier voltage information exists, the voltage is counted and the difference between the average value of abnormal battery voltage data and the average value of normal battery voltage data is calculated. Based on the DBSCAN clustering results, different information entropy calculation windows are divided.

[0018] Preferably, in step S3, the discrete weight α is calculated based on the difference, and the formula is:

[0019] α=e 1+x1 ;

[0020] Where x1 represents the absolute value of the difference.

[0021] Preferably, step S4 specifically includes:

[0022] S41, calculating the information entropy of different information entropy calculation windows in the application algorithm data set;

[0023] S42. Calculating an abnormality coefficient of the battery voltage based on the calculated information entropy;

[0024] S43: Multiply the abnormal coefficient of the outlier battery voltage by the outlier weight. When the calculated result exceeds the adaptive abnormal warning threshold, it is determined that the battery in the current window is in an abnormal state.

[0025] Preferably, step S41 specifically includes:

[0026] Considering all battery cell voltages in the same calculation window as a whole, construct the approximate number matrix, the formula is:

[0027]

[0028] Where l is the information entropy resolution, c i,j represents the total number of battery cells of the jth battery in the ith equal division interval under the information entropy time window, and n represents the number of battery cells;

[0029] Calculate the value range of the i-th partition in the information entropy matrix. The formula is:

[0030]

[0031] Where x max and x min They are the maximum and minimum values within the calculation window CW respectively;

[0032] Based on the probability matrix, the probability matrix is constructed as follows:

[0033]

[0034] Where p i,j Represents c i,j Relative to the probability of battery cell i under the calculation window, the information entropy h of the jth battery cell in the information entropy sequence is j The calculation method is shown in the formula:

[0035]

[0036] Calculate the information entropy sequence H under the information entropy window based on the probability matrixm , calculate the information entropy h according to the j-th battery cell in the information entropy sequence j , the formula is:

[0037] H m =[h1,h2,......,h j ......h n ]

[0038]

[0039] The calculation window is traversed through all battery voltage data under the batch window according to the sliding step size to obtain the information entropy matrix H, which is:

[0040] H=[H1,H2,......,H m ......H batch_end ] T ;

[0041] Where H batch_end Represents the last value of the information entropy sequence obtained after traversing the batch window.

[0042] Preferably, step S42 specifically includes:

[0043] According to the information entropy h j Calculate the coefficient of variation z for each battery cell j , and the coefficient of variation of all battery cells under the information entropy calculation window constitutes the coefficient of variation Z m , the formula is:

[0044]

[0045] Z m =[z1,z2,......,z j ......z n ];

[0046] Where h ave With h stdj In the same information entropy calculation window, the battery information entropy sequence H is m The average value and standard deviation of the information entropy of the j-th battery cell are expressed as follows:

[0047]

[0048] Calculate the variation coefficient sequence of each information entropy sequence in the information entropy matrix to obtain the variation coefficient matrix Z, the formula is:

[0049] Z=[Z1,Z2,......,Z m ......Z batch_end ]T ;

[0050] Where Z batch_end Represents the last value of the coefficient of variation sequence obtained after traversing the batch window, represented by H batch_end Calculated.

[0051] Therefore, the present invention adopts the above-mentioned battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy, which has the following beneficial effects:

[0052] (1) Adaptively determining the abnormal warning threshold through the empirical distribution function can adapt to the individual differences of different batteries, reduce the false alarm rate, and improve the accuracy of diagnosis;

[0053] (2) The voltage-based DBSCAN clustering algorithm and the outlier weight α are introduced to further amplify the information entropy calculation results, effectively solving the problem that the information entropy cannot be properly identified in the later stage of battery failure, further optimizing the fault diagnosis results, and clustering the battery voltage fluctuations to avoid the problem of inaccurate entropy calculation results caused by the battery voltage fluctuations themselves;

[0054] (3) By combining information entropy and cluster analysis, the early abnormal state and late fault state of the battery can be identified quickly and accurately, improving the diagnostic efficiency and ensuring the safe operation of the battery system.

[0055] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Flowchart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings herein can be arranged and designed in various different configurations. In the description of the present invention, it should be noted that the orientation or position relationship indicated by the terms "upper", "lower", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, or is the orientation or position relationship in which the product of the invention is usually placed when in use. It is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0058] Example

[0059] like Figure 1 As shown, the present invention provides a battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy, the steps comprising:

[0060] S1. Obtain battery voltage data, preprocess the battery voltage data, set the voltage data, remove outliers, and eliminate abnormal sampling points. Obtain battery voltage data with normal timestamps and no obvious outliers. Sort the battery voltage data with normal timestamps in ascending order. Use the battery voltage data from the week before the battery leaves the factory as the threshold training dataset, and divide the remaining battery voltage data into the algorithm application dataset.

[0061] S2. Use information entropy and empirical distribution data to statistically analyze the battery threshold status in the threshold training data set and determine the adaptive abnormal warning thresholds for different battery packs. Specifically, it includes:

[0062] S21, calculating the information entropy and anomaly coefficient of the threshold training data set;

[0063] S22. Observe the empirical distribution of the abnormal coefficient using the empirical distribution function;

[0064] S23. Select the abnormal coefficient corresponding to the observation value of 0.998 as the adaptive abnormal warning threshold for different battery packs.

[0065] S3. Use the DBSCAN algorithm to calculate the voltage outlier degree within the same calculation window in the algorithm application data set and calculate the outlier weight. The outlier weight amplifies the calculation results of the abnormal coefficient of the outlier battery cell. When multiple outliers occur in the battery, the abnormal coefficient calculated based on information entropy can still well reflect the abnormal state of the battery.

[0066] Specifically:

[0067] The DBSCAN algorithm is used to perform cluster analysis on battery voltage data, identify battery charge and discharge intervals, and determine whether there are obvious outliers in the battery voltage under the calculation window. If outlier voltage information exists, the voltage is counted and the difference between the average value of abnormal battery voltage data and the average value of normal battery voltage data is calculated. Based on the DBSCAN clustering results, different information entropy calculation windows are divided.

[0068] The discrete weight α is calculated based on the difference, and its formula is:

[0069] α=e 1+x1 ;

[0070] Where x1 represents the absolute value of the difference.

[0071] S4. Joint warning is performed based on the adaptive abnormal warning threshold and discrete weight, specifically:

[0072] S41. Calculate the information entropy of different information entropy calculation windows in the algorithm application data set. For a battery pack with n battery cells, the detailed steps of the information entropy calculation process are as follows:

[0073] Constructing a probable matrix: Considering the voltages of all battery cells in the same calculation window as a whole, construct a l×n probable matrix as shown below:

[0074]

[0075] Where l is the information entropy resolution, c i,j represents the total number of battery cells of the jth battery in the ith equal division interval under the information entropy time window, and n represents the number of battery cells;

[0076] Calculate the value range of the i-th partition in the information entropy matrix. The formula is:

[0077]

[0078] Where x max and x min They are the maximum and minimum values within the calculation window CW respectively;

[0079] Construct a probability matrix: Based on the probability matrix, construct a probability matrix, which is expressed as:

[0080]

[0081] Where p i,j Represents c i,j Relative to the probability of battery cell i in the calculation window, the formula is:

[0082]

[0083] Information entropy calculation: Calculate the information entropy sequence H under the information entropy calculation window based on the probability matrix m , the information entropy h of the jth battery cell in the information entropy sequence j The calculation method is shown in the formula:

[0084] H m =[h1,h2,......,h j ......h n ]

[0085]

[0086] The calculation window is traversed through all battery voltage data under the batch window according to the sliding step size to obtain the information entropy matrix H, which is:

[0087] H=[H1,H2,......,H m ......H batch_end ] T ;

[0088] Where H batch_end Represents the last value of the information entropy sequence obtained after traversing the batch window.

[0089] S42, calculating the abnormal coefficient of the battery voltage based on the calculated information entropy, specifically: according to the obtained information entropy h j Calculate the coefficient of variation z for each battery cell j , and the coefficient of variation of all battery cells under the information entropy calculation window constitutes the coefficient of variation Z m , the formula is:

[0090]

[0091] Z m =[z1,z2,......,z j ......z n ];

[0092] Where h ave With h stdj In the same information entropy calculation window, the battery information entropy sequence H is m The average value and standard deviation of the information entropy of the j-th battery cell are expressed as follows:

[0093]

[0094] Calculate the variation coefficient sequence of each information entropy sequence in the information entropy matrix to obtain the variation coefficient matrix Z, the formula is:

[0095] Z=[Z1,Z2,......,Z m ......Z batch_end ] T ;

[0096] Where Z batch_end Represents the last value of the coefficient of variation sequence obtained after traversing the batch window, represented by H batch_end Calculated.

[0097] S43. In the process of calculating the abnormal coefficient of step S42, the abnormal coefficient of the outlier battery voltage is multiplied by the outlier weight. When the calculated result exceeds the adaptive abnormal warning threshold, it is considered that the battery in the current window is in an abnormal state.

[0098] Therefore, the present invention adopts the above-mentioned battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy, determines the adaptive abnormal warning threshold of different battery packs through the empirical distribution function, judges whether the battery is in an abnormal state through information entropy and abnormal coefficient, and performs fault warning based on the combination of DBSCAN and information entropy, thereby improving the accuracy of fault warning and ensuring the safe operation of the battery system.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy, characterized in that the steps include: S1. Acquire battery voltage data, preprocess the battery voltage data to obtain battery voltage data with normal timestamps and no obvious abnormal values, and divide the battery voltage data into a threshold training data set and an algorithm application data set; S2. Use information entropy and empirical distribution data to statistically analyze the battery threshold status in the threshold training data set and determine the adaptive abnormal warning thresholds for different battery packs; S3, using the DBSCAN algorithm to calculate the voltage outlier degree in the same calculation window in the algorithm application data set, and calculate the outlier weight; S4. Perform joint warning based on adaptive abnormal warning threshold and discrete weight.

2. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 1 is characterized in that: Step S2 specifically includes: S21, calculating the information entropy and anomaly coefficient of the threshold training data set; S22. Observe the empirical distribution of the abnormal coefficient using the empirical distribution function; S23. Select the abnormal coefficient corresponding to the observation value of 0.998 as the adaptive abnormal warning threshold for different battery packs.

3. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 1 is characterized in that: In step S3, the DBSCAN algorithm is used to calculate the voltage outlier degree in the same calculation window in the algorithm application data set, including: The DBSCAN algorithm is used to perform cluster analysis on battery voltage data, identify battery charge and discharge intervals, and determine whether there are obvious outliers in the battery voltage under the calculation window. If outlier voltage information exists, the voltage is counted and the difference between the average value of abnormal battery voltage data and the average value of normal battery voltage data is calculated. Based on the DBSCAN clustering results, different information entropy calculation windows are divided.

4. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 3 is characterized in that: In step S3, the discrete weight α is calculated based on the difference, and the formula is: α＝e 1+x1 ； Where x1 represents the absolute value of the difference.

5. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 4 is characterized in that: Step S4 specifically includes: S41, calculating the information entropy of different information entropy calculation windows in the application algorithm data set; S42. Calculating an abnormality coefficient of the battery voltage based on the calculated information entropy; S43: Multiply the abnormal coefficient of the outlier battery voltage by the outlier weight. When the calculated result exceeds the adaptive abnormal warning threshold, it is determined that the battery in the current window is in an abnormal state.

6. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 5 is characterized in that: Step S41 specifically includes: Considering all battery cell voltages in the same calculation window as a whole, construct the approximate number matrix, the formula is: Where l is the information entropy resolution, c i,j represents the total number of battery cells of the jth battery in the ith equal division interval under the information entropy time window, and n represents the number of battery cells; Calculate the value range of the i-th partition in the information entropy matrix. The formula is: Where x max and x min They are the maximum and minimum values within the calculation window CW respectively; Based on the probability matrix, the probability matrix is constructed as follows: Where p i,j Represents c i,j Relative to the probability of battery cell i in the calculation window, the formula is: Calculate the information entropy sequence H under the information entropy window based on the probability matrix m , the information entropy h of the jth battery cell in the information entropy sequence j The calculation formula is: H m =[h1,h2,......,h j ......h n ] The calculation window is traversed through all battery voltage data under the batch window according to the sliding step size to obtain the information entropy matrix H, which is: H=[H1,H2,......,H m ......H batch_end ] T ; Where H batch_end Represents the last value of the information entropy sequence obtained after traversing the batch window.

7. The battery fault diagnosis method based on DBSCAN, empirical distribution function and information entropy according to claim 6 is characterized in that: Step S42 specifically includes: According to the information entropy h j Calculate the coefficient of variation z for each battery cell j , and the coefficient of variation of all battery cells under the information entropy calculation window constitutes the coefficient of variation Z m , the formula is: WITH m =[z1,z2,......,z j ......With n ]; Where h ave With h stdj In the same information entropy calculation window, the battery information entropy sequence H is m The average value and standard deviation of the information entropy of the j-th battery cell are expressed as follows: Calculate the variation coefficient sequence of each information entropy sequence in the information entropy matrix to obtain the variation coefficient matrix Z, the formula is: Z=[Z1,Z2,......,Z m ......WITH batch_end ] T ; Where Z batch_end Represents the last value of the coefficient of variation sequence obtained after traversing the batch window, represented by H batch_end Calculated.

Images