A method for detecting internal short-circuit monomers in a power battery pack

Abstract

The present invention provides a method for detecting internal short-circuited cells in a power battery pack, which makes full use of the voltage deviation characteristics of the battery cells during the evolution of the internal short circuit. The internal short-circuit detection method is developed in a data-driven manner, and can timely detect the internal short-circuited battery cells in the battery module. The method is applicable to all vehicle operating conditions and the entire life cycle of the battery, and is helpful for early warning of thermal runaway accidents of the power battery.

Description

Technical Field

[0001] The present invention belongs to the technical field of power battery safety early warning monitoring, and in particular relates to a method for detecting internal short-circuit monomers in a power battery pack. Background Art

[0002] At present, thermal runaway accidents of power batteries occur from time to time, which not only seriously threatens the life and property safety of drivers and passengers, but also creates considerable obstacles to the popularization and promotion of new energy vehicles. "Thermal runaway" refers to the abnormal thermal effect of a sharp change in the battery temperature rise rate caused by a chemical chain exothermic reaction inside the battery. Power batteries operating under extreme conditions such as mechanical abuse, electrical abuse, and thermal abuse may induce thermal runaway accidents.

[0003] Since there are many causes of thermal runaway and many phenomena are intertwined, it is very difficult to warn against thermal runaway. However, no matter whether the thermal runaway is caused by mechanical abuse, electrical abuse or thermal abuse, the internal short circuit of the battery cell will occur during the development of thermal runaway, and thus can serve as an important indicator of thermal runaway warning. In the prior art, there are still the following problems in detecting internal short circuits in power battery packs:

[0004] (1) There are many ways to trigger an internal short circuit, but most of them are triggering methods with a short time scale. At present, there is still a lack of equivalent alternative experiments that can show the long-term evolution characteristics of internal short circuits, and the existing internal short circuit detection algorithms can only target specific triggering methods.

[0005] (2) Most internal short-circuit detection methods are limited to the single-cell level, and there is a lack of internal short-circuit detection methods for single-cells in battery modules. Summary of the invention

[0006] In view of this, in view of the technical problems existing in the art, the present invention provides a method for detecting an internal short-circuited monomer in a power battery pack, which specifically includes the following steps:

[0007] Step 1: Perform a separate offline test on the battery cells in the power battery pack, including constant current charging and constant current discharging, obtain the open circuit voltage data of the battery cells, and divide multiple voltage intervals with different open circuit voltage OCV values ​​as boundaries;

[0008] Step 2: acquiring the terminal voltage data of each battery cell in the power battery pack in real time online at a fixed data frame period;

[0009] Step 3: Calculate the improved Shannon entropy of each monomer terminal voltage data in the time sliding window to reflect the distribution and disorder degree of the terminal voltage data in different voltage intervals in each data frame time;

[0010] Step 4: Pair the terminal voltage data of each monomer corresponding to each data frame with the calculated improved Shannon entropy as a data sample point, and the data sample points together constitute a two-dimensional data sample group;

[0011] Step 5: Using a clustering algorithm to preliminarily screen the two-dimensional data sample group to screen out outlier data sample points;

[0012] Step 6: Calculate the local density outlier factor of each outlier data sample point; when one or more local density outlier factors are greater than or equal to the preset fault threshold, an internal short circuit fault alarm is issued and the serial number index of the internal short circuit battery cell is returned.

[0013] Furthermore, the specific process of constant current charging and constant current discharging performed in step 1 includes:

[0014] First, a constant current charge of a certain rate is applied to the battery cell until the upper cut-off voltage set by the manufacturer stops charging; then a constant current discharge of the same rate is applied to the cell until the lower cut-off voltage set by the manufacturer stops discharging;

[0015] The cell terminal voltage data obtained under the same SOC during charge and discharge are averaged to obtain an OCV-SOC curve, and the OCV corresponding to each fixed SOC interval is used as the limit.

[0016] Furthermore, the improved Shannon entropy of the voltage at each cell terminal at a certain data frame in the time sliding window in step 3 [SE 1 ,SE 2 ,…,SE n ] is calculated using the following formula:

[0017]

[0018] Among them, P i,j is the probability that the terminal voltage data of battery cell j in this sliding window is distributed in the i-th voltage interval among the m voltage intervals divided, and n represents the total number of cells in the battery pack.

[0019] Furthermore, the specific process of step 5 to filter outlier data sample points includes:

[0020] 1) Select K data sample points as the initial clustering centers [C 1 ,C 2 ,…,C K ];

[0021] 2) Calculate the distance between each data sample point in the data sample group and the K cluster centers, and classify each sample point into the class with the smallest distance to the center;

[0022] 3) Recalculate the cluster centers C of each category i(i=1,2,…,K):

[0023]

[0024] And the cluster radius R i (i=1,2,…,K):

[0025]

[0026] Where N(i) is the total number of sample points contained in the i-th class, X ij is the data sample point corresponding to the jth monomer in the i-th category;

[0027] 4) Re-execute step 2) until the maximum number of iterations is exceeded or the class to which each sample point belongs no longer changes;

[0028] If the distance from a sample point to the cluster center is ≥ R i , it will be screened out and put into the outlier candidate set; otherwise, the sample point will not participate in subsequent calculations.

[0029] Furthermore, the specific calculation process of the local density outlier factor of each outlier data sample point in step 6 is as follows:

[0030] First, calculate the kth reachable distance of any battery cell sample point P relative to the target battery cell sample point O:

[0031] d k (P,O)=max{d k (O),d(P,O)}

[0032] Among them, d k (O) is the kth neighborhood N k The reachable distance from each point in (O) to point O, d(P, O) is the distance between point P and point O;

[0033] Then calculate the local reachable density lrd of the target battery cell sample point O k (O):

[0034]

[0035] Finally, the local density outlier factor value LOF of the target battery cell sample point O is calculated k (O) is:

[0036]

[0037] The internal short-circuit single cell detection method in the power battery pack provided by the present invention makes full use of the voltage deviation characteristics of the battery cell during the evolution of the internal short circuit, and develops the internal short-circuit detection method in a data-driven manner. It can timely detect the internal short-circuited battery cells in the battery module, is applicable to all vehicle operating conditions and the entire life cycle of the battery, and is helpful for early warning of thermal runaway accidents of the power battery. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flow chart of the method provided by the present invention;

[0039] Figure 2 It is a schematic diagram of the detection result of short circuit in a monomer according to an example of the present invention. DETAILED DESCRIPTION

[0040] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. 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.

[0041] The present invention provides a method for detecting an internal short-circuit monomer in a power battery pack, such as Figure 1 As shown, the specific steps include:

[0042] Step 1: Perform a separate offline test on the battery cells in the power battery pack, including constant current charging and constant current discharging, obtain the open circuit voltage data of the battery cells, and divide multiple voltage intervals with different open circuit voltage OCV values ​​as boundaries;

[0043] Step 2: acquiring the terminal voltage data of each battery cell in the power battery pack in real time online at a fixed data frame period;

[0044] Step 3: Calculate the improved Shannon entropy of each monomer terminal voltage data in the time sliding window to reflect the distribution and disorder degree of the terminal voltage data in different voltage intervals in each data frame time;

[0045] Step 4: Pair the terminal voltage data of each monomer corresponding to each data frame with the calculated improved Shannon entropy as a data sample point, and the data sample points together constitute a two-dimensional data sample group;

[0046] Step 5: Using a clustering algorithm to preliminarily screen the two-dimensional data sample group to screen out outlier data sample points;

[0047] Step 6: Calculate the local density outlier factor of each outlier data sample point; when one or more local density outlier factors are greater than or equal to the preset fault threshold, an internal short circuit fault alarm is issued and the serial number index of the internal short circuit battery cell is returned.

[0048] In a preferred embodiment of the present invention, the specific process of constant current charging and constant current discharging performed in step 1 includes:

[0049] First, a constant current charge of a certain rate is applied to the battery cell until the upper cut-off voltage set by the manufacturer stops charging; then a constant current discharge of the same rate is applied to the cell until the lower cut-off voltage set by the manufacturer stops discharging;

[0050] The cell terminal voltage data obtained under the same SOC during charge and discharge are averaged to obtain an OCV-SOC curve, and the OCV corresponding to each fixed SOC interval is used as the limit.

[0051] In a preferred embodiment of the present invention, the time sliding window in step 3 is located at a certain data frame at each monomer terminal voltage. 1 ,SE 2 ,…,SE n ] is calculated using the following formula:

[0052]

[0053] Among them, P i,j is the probability that the terminal voltage data of battery cell j in this sliding window is distributed in the i-th voltage interval among the m voltage intervals divided, and n represents the total number of cells in the battery pack.

[0054] In a preferred embodiment of the present invention, the specific process of step 5 of screening outlier data sample points includes:

[0055] 1) Select K data sample points as the initial clustering centers [C 1 ,C 2 ,…,C K ];

[0056] 2) Calculate the distance between each data sample point in the data sample group and the K cluster centers, and classify each sample point into the class with the smallest distance to the center;

[0057] 3) Recalculate the cluster centers C of each category i (i=1,2,…,K):

[0058]

[0059] And the cluster radius R i (i=1,2,…,K):

[0060]

[0061] Where N(i) is the total number of sample points contained in the i-th class, X ij is the data sample point corresponding to the jth monomer in the i-th category;

[0062] 4) Re-execute step 2) until the maximum number of iterations is exceeded or the class to which each sample point belongs no longer changes;

[0063] If the distance from a sample point to the cluster center is ≥ R i , it will be screened out and put into the outlier candidate set; otherwise, the sample point will not participate in subsequent calculations.

[0064] In a preferred embodiment of the present invention, the specific calculation process of the local density outlier factor of each outlier data sample point in step 6 is as follows:

[0065] First, calculate the kth reachable distance of any battery cell sample point P relative to the target battery cell sample point O:

[0066] d k (P,O)=max{d k (O),d(P,O)}

[0067] Among them, d k (O) is the kth neighborhood N k The reachable distance from each point in (O) to point O, d(P, O) is the distance between point P and point O;

[0068] Then calculate the local reachable density lrd of the target battery cell sample point O k (O):

[0069]

[0070] Finally, the local density outlier factor value LOF of the target battery cell sample point O is calculated k (O) is:

[0071]

[0072] In a preferred embodiment of the present invention, the detection and positioning of the internal short-circuit monomer is completed by performing the following steps:

[0073] Step 1: Connect the battery charging and discharging equipment to the battery cells in this power battery pack, apply a constant current charging condition of 0.05C (1 / 20 of the nominal capacity), and stop charging when the upper cut-off voltage given by the manufacturer is reached; then apply a constant current discharging condition of 0.05C (1 / 20 of the nominal capacity) until the discharge stops at the lower cut-off voltage given by the manufacturer. Take the average of the charging terminal voltage and the discharging terminal voltage at the same SOC during the charging and discharging process, obtain the OCV-SOC curve, and use the OCV corresponding to each 10% SOC interval as the partition boundary voltage, for a total of 11 partition boundary voltages [V lim1 ,V lim2 ,…,V lim11 ].

[0074] Step 2: Use the data platform to obtain the terminal voltage data of 98 cells in the power battery pack sampled by the BMS. The data sampling frequency is 0.1Hz, that is, every 10s is regarded as a data frame, and the terminal voltage data is composed of [V 1 ,V 2 ,…,V 98 ].

[0075] Step 3: Calculate the improved Shannon entropy value [SE 1 ,SE 2 ,…,SE 98 ].

[0076] Step 4: Obtain the terminal voltage data of each monomer at each data frame point [V 1 ,V 2 ,…,V 98 ] and the voltage-improved Shannon entropy data of each monomer obtained by the above steps [SE 1 ,SE 2 ,…,SE 98 ]First, the normalization operation is performed, and then a two-dimensional data set is formed. In the case of missing or abnormal voltage data, the data of this data frame will no longer participate in the subsequent model calculation.

[0077] Step 5: Input the two-dimensional data set constructed from each data frame into the clustering-based outlier initial screening model. Select three samples in the two-dimensional data set as the initial cluster centers [C 1 ,C 2 ,C 3 ]; Calculate the distance between each sample in the data set and the three cluster centers, and classify each sample into the class with the smallest distance to the center. Recalculate the cluster center and radius R of each class i(i=1,2,3), repeat the above operation until the maximum number of iterations is exceeded or the class to which each sample belongs does not change. If the distance from each sample point to the cluster center to which it belongs is ≥R i , then it is screened out and put into the outlier candidate set. For the distance from the sample point to the cluster center to which it belongs < R i , then it does not need to participate in subsequent calculations.

[0078] Step 6: Input the selected data sample points into the local density outlier factor calculation model to obtain the local density outlier factor [LOF 1 ,LOF 2 ,…,LOF 98 ]. When one or more local density outlier factors are greater than or equal to the preset fault threshold 5, an internal short circuit fault alarm is issued and the serial number index of the internal short circuit battery cell is returned.

[0079] The model calculation is performed using the actual vehicle operation data with a data frame length of 50000. One of the battery cells is selected so that its voltage data has a deviation of 0.02V from the original data since data frame 25000, and it is regarded as an internal short-circuited cell. Figure 2 The calculation results of the local density outlier factor of the normal monomer and the simulated internal short-circuit monomer in this embodiment are shown. The results show that the local density outlier factor of the internal short-circuit monomer exceeds the preset fault threshold of 5, and the internal short-circuit monomer can be detected, verifying the effectiveness of the method.

[0080] It should be understood that the size of the serial numbers of the steps in the embodiment of the present invention does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0081] 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 internal short-circuit monomers in a power battery pack, Features: The specific steps include: Step 1: Perform a separate offline test on the battery cells in the power battery pack, including constant current charging and constant current discharging, obtain the open circuit voltage data of the battery cells, and divide multiple voltage intervals with different open circuit voltage OCV values ​​as boundaries; Step 2: acquiring the terminal voltage data of each battery cell in the power battery pack in real time online at a fixed data frame period; Step 3: Calculate the improved Shannon entropy of each monomer terminal voltage data in the time sliding window to reflect the distribution and disorder degree of the terminal voltage data in different voltage intervals in each data frame time; Improved Shannon entropy [SE 1 ,SE 2 ,…,SE n ] is calculated using the following formula: where P i,j is the probability that the terminal voltage data of battery cell j within this sliding window is distributed in the i-th voltage interval among a total of m divided voltage intervals, and n represents the total number of cells in the battery pack; Step 4: Pair the terminal voltage data of each monomer corresponding to each data frame with the calculated improved Shannon entropy as a data sample point, and the data sample points together constitute a two-dimensional data sample group; Step 5: Using a clustering algorithm to preliminarily screen the two-dimensional data sample group to screen out outlier data sample points; Step 6: Calculate the local density outlier factor of each outlier data sample point. The specific calculation process is as follows: First, calculate the kth reachable distance of any battery cell sample point P relative to the target battery cell sample point O: d k (P,O)=max{d k (O),d(P,O)} Among them, d k (O) is the kth neighborhood N k The reachable distance from each point in (O) to point O, d(P, O) is the distance between point P and point O; Then calculate the local reachable density lrd of the target battery cell sample point O k (O): Finally, the local density outlier factor value LOF of the target battery cell sample point O is calculated k (O) is: When one or more local density outlier factors are greater than or equal to the preset fault threshold, an internal short circuit fault alarm is issued and the serial number index of the internal short circuit battery cell is returned.

2. The method according to claim 1, Features: The specific process of constant current charging and constant current discharging performed in step 1 includes: First, a constant current charge of a certain rate is applied to the battery cell until the upper cut-off voltage set by the manufacturer stops charging; then a constant current discharge of the same rate is applied to the cell until the lower cut-off voltage set by the manufacturer stops discharging; The cell terminal voltage data obtained under the same SOC during charge and discharge are averaged to obtain an OCV-SOC curve, and the OCV corresponding to each fixed SOC interval is used as the limit.

3. The method according to claim 1, Features: The specific process of step 5 to filter outlier data sample points includes: 1) Select K data sample points as the initial clustering centers [C 1 ,C 2 ,…,C K ]; 2) Calculate the distance between each data sample point in the data sample group and the K cluster centers, and classify each sample point into the class with the smallest distance to the center; 3) Recalculate the cluster centers C of each category i (i=1,2,…,K): And the cluster radius R i (i=1,2,…,K): Where N(i) is the total number of sample points contained in the i-th class, X ij is the data sample point corresponding to the jth monomer in the i-th category; 4) Re-execute step 2) until the maximum number of iterations is exceeded or the class to which each sample point belongs no longer changes; If the distance from a sample point to the cluster center is ≥ R i , it will be screened out and put into the outlier candidate set; otherwise, the sample point will not participate in subsequent calculations.

Images