A method for identifying abnormal internal resistance of a power battery cell

By constructing a sliding voltage matrix and feature vectors to identify abnormal cell internal resistance, the problem of poor identification effect and insufficient accuracy in existing technologies is solved, realizing efficient and accurate identification of cell internal resistance, early detection of cell abnormalities, and ensuring battery pack safety.

CN117907863BActive Publication Date: 2024-11-15LIGOO (SHAN DONG) NEW ENERGY TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311356037.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-19
Publication Date
2024-11-15
Estimated Expiration
2043-10-19

AI Technical Summary

Technical Problem

Existing technologies are not effective in identifying abnormal internal resistance of power battery cells, especially polarization resistance. Problems such as non-convergence of calculation results, high data accuracy requirements, and asynchronous data acquisition affect the identification effect.

Method used

By acquiring and preprocessing battery data, a sliding voltage matrix is ​​constructed and the maximum sliding correlation coefficient is calculated. Eigenvalues ​​and eigenvectors are used to identify abnormal cell internal resistance. The sliding window method and a preset detection algorithm are used to determine the covariance between cell voltages, eliminate the influence of SOC inconsistency, and compare the relative relationships between cells in a reduced dimension.

Benefits of technology

It enables accurate identification of abnormal internal resistance of battery cells, reduces computational complexity, improves identification accuracy, and can identify battery cell abnormalities in advance, avoiding battery pack performance degradation and safety hazards.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117907863B_ABST
    Figure CN117907863B_ABST
Patent Text Reader

Abstract

The present invention belongs to the technical field of electric vehicle power batteries, and specifically relates to a method for identifying abnormal internal resistance of a power battery cell, comprising acquiring battery data and generating target data including charging segments and discharging segments, judging the size relationship between the data volume N of each discharge segment in the target data and a preset data volume threshold, constructing a sliding voltage matrix for the discharge segment of each cell based on a sliding window method and a preset sliding window threshold r, and calculating the sliding maximum correlation coefficient between adjacent cells according to the sliding voltage matrix to form a sliding maximum correlation coefficient matrix, calculating the eigenvalues ​​of the sliding maximum correlation coefficient matrix and their corresponding eigenvectors, and determining whether the internal resistance of the cell is abnormal according to the eigenvector corresponding to a certain cell; the present invention can effectively identify cells with abnormal internal resistance through cell features constructed through eigenvectors, and the implicit discrimination method for identifying the covariance between cell voltages has a smaller error than directly calculating the internal resistance value.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of electric vehicle power batteries, and in particular relates to a method for identifying abnormal internal resistance of a power battery cell. Background Art

[0002] As the core component of new energy vehicles, power batteries are directly related to the safe and stable driving of vehicles. According to national requirements, new energy vehicles need to send data to the national big data platform in accordance with the protocol format of the national standard GBT32960. Based on these data, we can use big data technology to monitor the safety hazards of power batteries in real time.

[0003] Normally, the battery pack of a power battery is composed of multiple modules, and a module is composed of multiple cells. As one of the many basic indicators of power batteries, the internal resistance of the cell is a potential culprit that induces safety hazards in the battery pack. In detail, the long-term difference between the internal resistance of the cells will reduce the consistency level of the battery pack, and the performance of the battery pack will be affected due to the short board effect. More seriously, a larger internal resistance of the cell may cause the cell to overcharge, bloating, rupture, and even leakage, fire, and thermal runaway.

[0004] Therefore, it is extremely valuable to monitor the internal resistance of battery cells through big data and identify abnormal internal resistance of battery cells in a timely manner.

[0005] Research shows that the internal resistance of a battery cell can be divided into two parts: ohmic internal resistance and polarization internal resistance. Ohmic resistance is the resistance generated by the battery system structure characteristics and the materials that make up the battery, including the contact resistance between electrodes, electrolyte solutions, diaphragms, etc. and various parts, the resistance generated by the oxide film on the electrode surface, and the diaphragm resistance. Under normal circumstances, ohmic internal resistance can be approximately regarded as a constant; polarization resistance is the resistance generated by the polarization effect of the electrode, which is greatly affected by current and temperature and is almost time-varying.

[0006] There are two main ways in the industry to identify abnormal cell internal resistance through big data technology: one is to identify the cell internal resistance through Ohm's law and its variations; the other is to build an equivalent circuit model (commonly used are second-order RC network, Thevenin model, etc.) and identify the cell internal resistance through parameter identification. However, the existing technologies have the following problems:

[0007] (1) The identification effect of polarization internal resistance is not good. (2) The calculated internal resistance of the battery cell often cannot converge directly. (3) The above methods have high requirements on data accuracy, and the data received by the cloud big data often cannot be directly used for identification. (4) Problems such as asynchronous data collection will have a great impact on the identification effect, especially the method of Ohm's law and its variants. Summary of the invention

[0008] The purpose of the present invention is to provide a method for identifying abnormal internal resistance of a power battery cell to solve the problems raised in the background technology.

[0009] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0010] A method for identifying abnormal internal resistance of a power battery cell comprises the following steps:

[0011] S1: Obtain battery data and perform preprocessing to generate target data including charging segments and discharging segments, and preferentially determine the relationship between the data volume N of each discharging segment in the target data and a preset data volume threshold. If the data volume N>data volume threshold is satisfied, then execute step S2 for the discharging segment that meets the condition, otherwise terminate the identification process;

[0012] S2: constructing a sliding voltage matrix for the discharge segment of each battery cell based on a sliding window method and a preset sliding window threshold r, and calculating the sliding maximum correlation coefficient between adjacent battery cells according to the sliding voltage matrix to form a sliding maximum correlation coefficient matrix;

[0013] S3: Calculate the eigenvalue of the sliding maximum correlation coefficient matrix and its corresponding eigenvector, and determine whether the internal resistance of the battery cell is abnormal based on a preset detection algorithm and the eigenvector corresponding to a certain battery cell.

[0014] As a further optimization scheme of the present invention, the preprocessing in step S1 includes: sorting the battery data according to the collection time, cleaning duplicate data and empty data, and then dividing the cleaned data into target data including charging segments and discharging segments according to the charging and discharging status.

[0015] As a further optimization scheme of the present invention, in step S1, the data volume threshold is set according to the number of sliding windows in the sliding window method, and the preset sliding window threshold r in the sliding window method is a custom parameter.

[0016] As a further optimization solution of the present invention, the discharge segment in the target data at least includes real-time voltage data that is not zero under the discharge condition.

[0017] As a further optimization scheme of the present invention, step S2 includes:

[0018] S2.1: Take the voltage of the i-th cell in the window to form an N-dimensional column vector, and get:

[0019] cell_volt i =(v1,v2,…,v N ),i=1,…,n

[0020] Where N is the number of data items, and n is the number of battery cells;

[0021] S2.2: For the voltage vector of the i-th cell, construct a sliding voltage matrix sliding_volt of (Nr)×(r+1) order through r sliding results i ,get:

[0022]

[0023] S2.3: Divide the sliding voltage matrix of the i-th cell and the j-th cell into blocks by columns:

[0024]

[0025] where k = 1,…,r+1, Represents the kth column vector of the sliding voltage matrix of the i-th cell:

[0026]

[0027] S2.4: Calculate the covariance matrix of the above block matrix and get:

[0028]

[0029] S2.5: Take the maximum value in the lower triangle of the above matrix as the final correlation coefficient between cell i and cell j:

[0030] max_covariance ij =max(covariance ij );

[0031] S2.6: Use the above method to calculate the maximum sliding correlation coefficient between the two cells, and form the maximum sliding correlation coefficient matrix between the cells:

[0032]

[0033] As a further optimization solution of the present invention, in step S3, the preset detection algorithm includes:

[0034] (1) Obtain the maximum eigenvalue and the minimum eigenvalue and denote them as λ respectively. max and λ min , and their corresponding eigenvectors are recorded as and

[0035] (2) Using the eigenvector pair (α t ,β t ) represents the abnormal characteristics of the internal resistance of the t-th cell for fault identification and visualization;

[0036] (3) For the i-th cell, if |α i -μα |>3σ α or |β i -μ β |>3σ β , it indicates that there is a cell abnormality in the i-th cell;

[0037] Among them, μ α , σ α Respectively represent ξ max The mean and standard deviation of the elements in μ β , σ β Respectively represent ξ min The mean and standard deviation of the elements in .

[0038] The beneficial effects of the present invention are:

[0039] (1) The internal resistance of the battery cell is an internal manifestation, and the battery cell voltage is an external manifestation of the internal resistance. The implicit discrimination method that identifies the covariance between the battery cell voltages has smaller errors than directly calculating the internal resistance value and is more accurate and effective.

[0040] (2) The purpose of using the sliding voltage matrix to calculate the sliding maximum correlation coefficient between two battery cells is to eliminate the influence of SOC inconsistency on internal resistance identification. Other internal resistance abnormality identification methods do not consider the influence of SOC inconsistency. By sliding comparison correlation and taking the maximum value, the difference between the charge and discharge curves can be compared to the greatest extent, eliminating the influence of the initial SOC.

[0041] (3) Experiments have shown that the cell features constructed by feature vectors can effectively identify cells with abnormal internal resistance.

[0042] (4) The high-dimensional covariance matrix is ​​reduced in dimension through the eigenvalue and eigenvector methods to facilitate the comparison of the relative relationships between battery cells.

[0043] (5) The calculation of the correlation coefficient can be vectorized, which solves the problem of high time complexity of some optimization algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is an execution flow chart of the present invention;

[0045] Figure 2 is a cell data diagram of the first charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0046] Figure 3 is a cell data diagram of the second charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0047] Figure 4 is a cell data diagram of the third charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0048] Figure 5 is a cell data diagram of the fourth charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0049] Figure 6 is a cell data diagram of the fifth charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0050] Figure 7 is a cell data diagram of the sixth charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0051] Figure 8 is a cell data diagram of the seventh charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0052] Fig. 9 is a cell data diagram of the eighth charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0053] Fig.10 is a cell data diagram of the 9th charge and discharge cycle identification of the case vehicle in a specific implementation manner;

[0054] Fig.11 This is a graph of battery cell data identified during the 10th charge and discharge cycle of the example vehicle in a specific implementation manner. DETAILED DESCRIPTION

[0055] The present application is further described in detail below in conjunction with the accompanying drawings. It is necessary to point out here that the following specific implementation methods are only used to further illustrate the present application and cannot be understood as limiting the scope of protection of the present application. Technical personnel in this field can make some non-essential improvements and adjustments to the present application based on the above application content.

[0056] Example 1

[0057] like Figure 1 As shown, this embodiment provides a method for identifying abnormal internal resistance of a power battery cell, comprising the steps of:

[0058] S1: Obtain battery data and perform preprocessing to generate target data including charging segments and discharging segments, and preferentially determine the relationship between the data volume N of each discharging segment in the target data and a preset data volume threshold. If the data volume N>data volume threshold is satisfied, then execute step S2 for the discharging segment that meets the condition, otherwise terminate the identification process;

[0059] The battery data is obtained by parsing the message according to the data format and requirements of the national standard GB / T32960.

[0060] In this embodiment, the preprocessing in step S1 includes: sorting the battery data according to the collection time, cleaning the duplicate data and the empty data, and then dividing the cleaned data into target data including the charging segment and the discharging segment according to the charging and discharging status; in addition, the discharging segment in the target data at least includes the real-time voltage data that is not 0 under the discharging condition.

[0061] S2: constructing a sliding voltage matrix for the discharge segment of each battery cell based on the sliding window method and a preset sliding window threshold r, and calculating the sliding maximum correlation coefficient between adjacent batteries according to the sliding voltage matrix to form a sliding maximum correlation coefficient matrix;

[0062] Step S2 specifically includes:

[0063] S2.1: Take the voltage of the i-th cell in the window to form an N-dimensional column vector, and get:

[0064] cell_volt i =(v1,v2,…,v N ),i=1,…,n

[0065] Where N is the number of data items, and n is the number of battery cells;

[0066] S2.2: For the voltage vector of the i-th cell, construct a sliding voltage matrix sliding_volt of (Nr)×(r+1) order through r sliding results i ,get:

[0067]

[0068] S2.3: Divide the sliding voltage matrix of the i-th cell and the j-th cell into blocks by columns:

[0069]

[0070] where k = 1,…,r+1, Represents the kth column vector of the sliding voltage matrix of the i-th cell:

[0071]

[0072] S2.4: Calculate the covariance matrix of the above block matrix and get:

[0073]

[0074] S2.5: Take the maximum value in the lower triangle of the above matrix as the final correlation coefficient between cell i and cell j:

[0075] max_covariance ij =max(covarianceij );

[0076] S2.6: Use the above method to calculate the maximum sliding correlation coefficient between the two cells, and form the maximum sliding correlation coefficient matrix between the cells:

[0077]

[0078] S3: Calculate the eigenvalue of the sliding maximum correlation coefficient matrix and its corresponding eigenvector, and determine whether the internal resistance of the battery cell is abnormal based on a preset detection algorithm and the eigenvector corresponding to a certain battery cell.

[0079] In step S3, the preset detection algorithm includes:

[0080] (1) Obtain the maximum eigenvalue and the minimum eigenvalue and denote them as λ respectively. max and λ min , and their corresponding eigenvectors are recorded as and

[0081] (2) Using the eigenvector pair (α t ,β t ) represents the abnormal characteristics of the internal resistance of the t-th cell for fault identification and visualization;

[0082] (3) For the i-th cell, if |α i -μ α |>3σ α or |β i -μ β |>3σ β , it indicates that there is a cell abnormality in the i-th cell;

[0083] Among them, μ α , σ α Respectively represent ξ max The mean and standard deviation of the elements in μ β , σ β Respectively represent ξ min The mean and standard deviation of the elements in .

[0084] Preferably, in step S1, the data volume threshold is set according to the number of sliding windows in the sliding window method, and the preset sliding window threshold r in the sliding window method is a custom parameter.

[0085] The following is an explanation of the above identification method in combination with the actual processing routine:

[0086] Combination Figure 2-Figure 11, the figure shows an actual case of identifying abnormal internal resistance of a battery cell using the identification method provided in this embodiment. The data of this case is divided into 10 charge and discharge cycles (numbered 1-10), where a charge and discharge cycle is defined as a segment from the start of the first charge to the end of the second charge. The cycle includes a complete charging segment and one or more discharge or static segments. Non-charging data in each charge and discharge cycle is selected for internal resistance identification. The left picture in each picture is a two-dimensional feature vector identified by the above-mentioned algorithm proposed in this embodiment, and is visualized with a single feature vector as the horizontal and vertical coordinates. The right picture in each picture is the battery cell voltage data of the battery pack in the selected segment, and is visualized according to the voltage data of each battery cell.

[0087] from Figure 2-11 It can be seen that since cycle 1, the abnormal internal resistance of battery cell No. 9 can be identified (outliers in the feature vector). Continuous attention in chronological order found that the battery ran to segment 9 with significant abnormalities (the vehicle could not drive). After unpacking and testing, the internal resistance of battery cell No. 9 was significantly high. It can be found that in the 9th cycle, the case vehicle broke down due to the problem of the internal resistance of battery cell No. 9 and entered the station for maintenance. The case proves that the method proposed in this embodiment can identify battery cell abnormalities at least 8 cycles in advance.

[0088] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying abnormal internal resistance of a power battery cell. It is characterized in that Includes steps: S1: Obtain battery data and perform preprocessing to generate target data including charging segments and discharging segments, and preferentially determine the relationship between the data volume N of each discharging segment in the target data and a preset data volume threshold. If the data volume N>data volume threshold is satisfied, then execute step S2 for the discharging segment that meets the condition, otherwise terminate the identification process; S2: constructing a sliding voltage matrix for the discharge segment of each battery cell based on a sliding window method and a preset sliding window threshold r, and calculating the sliding maximum correlation coefficient between adjacent battery cells according to the sliding voltage matrix to form a sliding maximum correlation coefficient matrix; Step S2 includes: S2.1: Take the voltage of the i-th cell in the window to form an N-dimensional column vector, and get: cell_volt i =(v 1 ,v 2 ,…,v N ),i=1,…,n Where N is the number of data items, and n is the number of battery cells; S2.2: For the voltage vector of the i-th cell, construct a sliding voltage matrix sliding_volt of (Nr)×(r+1) order through r sliding results i ,get: S2.3: Divide the sliding voltage matrix of the i-th cell and the j-th cell into blocks by columns: where k = 1,…,r+1, represents the kth column vector of the sliding voltage matrix of the i-th cell, Represents the kth column vector of the sliding voltage matrix of the jth cell; S2.4: Calculate the covariance matrix of the above block matrix and get: S2.5: Take the maximum value in the lower triangle of the above matrix as the final correlation coefficient between cell i and cell j: max_covariancce ij =max(covariance ij ); S2.6: Use the above method to calculate the maximum sliding correlation coefficient between the two cells, and form the maximum sliding correlation coefficient matrix between the cells: S3: Calculate the eigenvalue of the sliding maximum correlation coefficient matrix and its corresponding eigenvector, and determine whether the internal resistance of the battery cell is abnormal based on a preset detection algorithm and the eigenvector corresponding to a certain battery cell; In step S3, the preset detection algorithm includes: (1) Obtain the maximum eigenvalue and the minimum eigenvalue and denote them as λ respectively. max and λ min , and their corresponding eigenvectors are recorded as and (2) Using the eigenvector pair (α t ,β t ) represents the abnormal characteristics of the internal resistance of the t-th cell for fault identification and visualization; (3) For the i-th cell, if |α i -μ α |>3σ α or |β i -μ β |>3σ β , it indicates that there is a cell abnormality in the i-th cell; Among them, μ α , σ α Respectively represent ξ max The mean and standard deviation of the elements in μ β , σ β Respectively represent ξ min The mean and standard deviation of the elements in .

2. A method for identifying abnormal internal resistance of a power battery cell according to claim 1, Features: The preprocessing in step S1 includes: sorting the battery data according to the collection time, cleaning duplicate data and empty data, and then dividing the cleaned data into target data including charging segments and discharging segments according to the charging and discharging status.

3. A method for identifying abnormal internal resistance of a power battery cell according to claim 1, Features: In step S1, the data volume threshold is set according to the number of sliding windows in the sliding window method, and the preset sliding window threshold r in the sliding window method is a custom parameter.

4. A method for identifying abnormal internal resistance of a power battery cell according to claim 1, Features: The discharge segment in the target data at least includes real-time voltage data that is not zero under the discharge condition.