Method and device for determining battery fault category

Abstract

The present invention discloses a method and device for determining a battery fault category. The method comprises: obtaining current operating information of the battery, the current operating information including at least one information category; merging pre-set fault sample data containing multiple fault categories with the current operating information to obtain new fault sample data; searching for at least one corresponding sub-sample data from the new fault sample data based on the at least one information category based on a predetermined algorithm; determining the number of data for each fault category in each sub-sample data based on pre-determined weights for each fault category; and determining the battery fault category based on the number of data for each fault category in each sub-sample data. The present invention can identify potential risks of energy storage lithium batteries and improve the safety of energy storage power stations.

Description

Technical Field

[0001] The present invention relates to the field of battery energy storage systems, and in particular to a method and device for determining a battery fault category. Background Art

[0002] Global industrial development has led to energy shortages and increased environmental pollution, ushering in large-scale growth in the new energy industry. Battery energy storage is a crucial component of energy utilization. With the booming development of electrochemical energy storage, market demand for lithium-ion batteries continues to grow. However, safety concerns about lithium-ion batteries have also led to numerous safety incidents. Battery safety is a prerequisite and guarantee for the application of energy storage technology. Scientific and effective evaluation and management of lithium-ion batteries will significantly improve safety.

[0003] Because lithium-ion batteries are crucial to the safe operation of energy storage power plants, real-time prediction of potential lithium-ion battery failures is essential. However, current research by scholars at home and abroad focuses on battery material improvements, condition monitoring, fault diagnosis and early warning, thermal management, circuit balance management, and fire prevention, explosion prevention, and fire extinguishing technologies.

[0004] Currently, there is no real-time prediction solution for potential failures of lithium-ion batteries, which is a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The present invention provides a method and apparatus for determining a battery fault category to solve at least one of the above-mentioned problems.

[0006] According to a first aspect of the present invention, a method for determining a battery fault category is provided, the method comprising:

[0007] Acquiring current operating information of the battery, the current operating information including: at least one information category;

[0008] Merging the pre-set fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data;

[0009] Based on a predetermined algorithm, searching for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category;

[0010] Determine the number of data for each fault category in each sub-sample data based on the predetermined weight of each fault category;

[0011] The fault category of the battery is determined according to the number of data of each fault category in each sub-sample data.

[0012] According to a second aspect of the present invention, a device for determining a battery fault category is provided, the device comprising:

[0013] an information acquisition unit, configured to acquire current operating information of the battery, the current operating information including: at least one information category;

[0014] A data merging unit, configured to merge the preset fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data;

[0015] a sub-sample data searching unit, configured to search for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category based on a predetermined algorithm;

[0016] a data quantity determining unit, configured to determine the data quantity of each fault category in each subsample data based on a predetermined weight of each fault category;

[0017] A fault category determination unit is configured to determine the fault category of the battery according to the number of data of each fault category in each sub-sample data.

[0018] At the same time, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0019] At the same time, the present invention also provides a computer-readable storage medium on which a computer program is stored, and the computer program implements the above method when executed by a processor.

[0020] It can be seen from the above technical solution that by merging the acquired current operating information of the battery with the pre-set fault sample data, new fault sample data is generated, and then based on a predetermined algorithm, the corresponding sub-sample data is searched from the new fault sample data according to each information category, and the number of sample data of each fault category in each sub-sample data is calculated. Then, the fault category of the battery is determined according to the number of sample data of each fault category, thereby realizing real-time judgment of the potential fault category of the battery, identifying the potential risks of the energy storage lithium battery, and improving the safety of the energy storage power station.

[0021] In order to make the above and other objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 is a flowchart of a method for determining a battery fault category according to an embodiment of the present invention;

[0024] Figure 2 is a detailed flow chart of a method for determining a battery fault category according to an embodiment of the present invention;

[0025] Figure 3 FIG. 1 is a structural block diagram of a device for determining a battery fault category according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention 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 making creative efforts are within the scope of protection of the present invention.

[0027] Given that there is no real-time prediction solution for potential lithium-ion battery (lithium battery) faults in the related art, embodiments of the present invention provide a solution for determining battery fault categories. This solution can obtain real-time lithium battery operating information and determine the potential fault category of the lithium battery based on the acquired operating information. This allows for timely inspection and repair of potential risks in energy storage lithium batteries, thereby improving the safety of energy storage power stations. The following describes embodiments of the present invention in detail with reference to the accompanying drawings.

[0028] Figure 1 FIG. 1 is a flow chart of a method for determining a battery fault category according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0029] Step 101: Acquire current operating information of the battery, where the current operating information includes at least one information category.

[0030] The information categories here may include: voltage, current, temperature, combustible gas concentration, strain (for example, battery deformation caused by uneven temperature), etc.

[0031] In actual operation, the current operating information of the battery may include: time, current battery voltage (eg, dynamic and static voltage), current battery current, current battery temperature, combustible gas concentration, strain, etc.

[0032] Step 102 : merging the preset fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data.

[0033] The fault categories here can include battery faults, battery actuator faults, battery sensor faults, etc. Battery faults include: internal short circuit faults, battery overcharge / overdischarge faults, battery thermal management faults, etc. Battery actuator faults include: BMS hardware faults, contactor faults, etc. Battery sensor faults include: voltage sensor faults, current sensor faults, temperature sensor faults, etc.

[0034] Before executing step 102 , the current operating information of the battery acquired in step 101 may be preprocessed (eg, data cleaning operation), and feature extraction may be performed on the preprocessed current operating information to obtain feature information of the battery.

[0035] In one embodiment, the fault sample data and the characteristic information of the battery may be merged to obtain new fault sample data.

[0036] In actual operation, the characteristic information of the battery can be called a feature vector, which is an n-dimensional data set.

[0037] The preset fault sample data containing multiple fault categories in step 102 may also be an n-dimensional data set.

[0038] The fault sample data here can be pre-set based on historical information of the battery. Based on the prior knowledge of the historical information, labels are set for the fault samples in the historical information, that is, fault categories are set for the fault samples, thereby forming fault sample data.

[0039] Step 103 : Based on a predetermined algorithm, search for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category (corresponding to the characteristic information).

[0040] Preferably, the predetermined algorithm may be a binary tree algorithm.

[0041] During the specific implementation process, different search termination rules can be set based on the binary tree algorithm; then, according to the at least one information category (feature information), each search termination rule is used to search for multiple sub-sample data corresponding to the information category and the search termination rule from the new fault sample data.

[0042] By searching for a specific sample through a binary tree, while completing downsampling, the retained sub-sample data is the adjacent sample of the search sample, which is actually a directional downsampling strategy that can improve the efficiency of searching for samples.

[0043] Step 104 : determining the number of data of each fault category in each sub-sample data based on the predetermined weight of each fault category.

[0044] Step 105 : Determine the fault category of the battery according to the number of data of each fault category in each sub-sample data.

[0045] Specifically, the number of sample data of each fault category in each sub-sample data is first determined; then, the fault category with the largest number of sample data is determined as the fault category of the battery.

[0046] In an embodiment of the present invention, new fault sample data is generated by merging the acquired current operating information of the battery with pre-set fault sample data. Subsequently, based on a binary tree algorithm, corresponding sub-sample data are searched from the new fault sample data according to each information category, and the number of sample data of each fault category in each sub-sample data is calculated. Then, the fault category of the battery is determined according to the number of sample data of each fault category, thereby realizing real-time judgment of the potential fault category of the battery, realizing potential risk identification and maintenance of energy storage lithium batteries, and improving the safety of energy storage power stations.

[0047] Figure 2 FIG. 1 is a detailed flow chart of a method for determining a battery fault category according to an embodiment of the present invention. Figure 2 As shown, the process includes:

[0048] Step 201: Data collection. The collected battery data includes original battery characteristics, such as time (timestamp), voltage, current, temperature, combustible gas concentration, and strain.

[0049] Step 202: Preprocess the data, specifically, perform a data cleaning operation. In an embodiment of the present application, the data cleaning rules may include: for missing values or unreasonable values that significantly exceed a threshold range, replace the value with the previous value, the next value, or the average value within a time interval calculated using a sliding window.

[0050] Step 203: Construct the battery feature vector. After the data cleaning is completed in step 202, the feature vector can be constructed based on the original feature data of the battery. x =[ x 1, x 2,……, x n ],in xn Refers to a specific battery characteristic parameter, such as battery voltage, current, temperature, combustible gas concentration and strain, etc. x As a complete sample in the subsequent battery fault diagnosis process.

[0051] Step 204: merge the battery feature vector constructed in step 203 into the pre-set training sample set (i.e., the above-mentioned fault sample data) to form a new sample set. X (i.e., the new fault sample data mentioned above).

[0052] For example, the fault categories in the training sample set are N categories, namely A , B , C … N Types of faults, such as battery body fault, actuator fault or battery sensor fault, etc. Each type of fault includes a , b , c ,…, n Samples will be tested P (That is, the above x=[x1, x2,……, x n ]) is combined with the training sample set to form a new sample set X= [ P , A , B , C ,… N ].

[0053] Then, based on the binary tree method, a feature is randomly selected (i.e., the above x n ) from the new sample set X Search for data corresponding to the feature until the stopping condition is reached.

[0054] Step 205: When the stopping condition is reached, a final sample subset is obtained.

[0055] Specifically, for a n Dimensional sample set X First, randomly select a feature dimension (for example, voltage feature dimension, current feature dimension, etc.) and use the binary tree method to find the sample set X The data corresponding to the feature dimension is found in the dataset. Each time a search is performed, a new subset is generated from the remaining sample set including the sample. The search is repeated until the samples are completely separated or the preset maximum search depth or termination condition is reached. The samples found based on the randomly selected features are used to obtain a new sample subset.

[0056] That is, the binary tree method search process will start with random features, and each search will search for a new sample set on a specific feature dimension. X When performing a search split, the resulting sample subset will have a different distribution and will always contain samples that are closer to the searched sample.

[0057] Step 206: Calculate the number of each type of fault samples in the sample subset.

[0058] In actual operation, the number of each type of fault samples in the sample subset can be calculated before the sample is found or in the last subset that reaches the termination condition. In order to avoid the influence of uneven distribution of the number of fault types, weighted measurement can be used to calculate the number of each type of fault samples in the subset.

[0059] It should be noted that the termination conditions here can be: 1. The current sample is found, that is, only the sample itself is left, then the number of each type of fault samples is calculated from the subset found in the previous step; or 2. The preset number of searches (such as 30 times) is reached, then the number of each type of fault samples is calculated from the sample subset found last time.

[0060] In specific implementation, in order to eliminate the influence of randomness, the idea of ensemble learning can be adopted to perform steps 204 to 206 on the sample loop. For example, different feature dimensions and termination conditions can be selected, and multiple search calculations can be performed to obtain the average value and determine the fault category.

[0061] In one embodiment, each loop process is independent of each other, and the use of distributed computing technology will further improve the efficiency of the algorithm.

[0062] Step 207 : Determine the fault category based on the number of fault samples of each type calculated in step 206 and mark the abnormal points.

[0063] Step 208: When the determined fault type is not a fault, the battery continues to operate; otherwise, the battery is maintained or replaced.

[0064] For easier understanding, an example is given below.

[0065] In this example, the number of samples in each sample subset is set to , calculate the number of each type of fault samples in each sample subset using the following formula: :

[0066] (1)

[0067] (2)

[0068] (3)

[0069] (4)

[0070] in, is the weighting coefficient of each fault category, For the sample set X The number of all samples in For the sample set X The number of various fault samples in .

[0071] Then the potential failure type of the battery is the current sample max( ) corresponds to the fault type.

[0072] Specifically, when a feature dimension is selected, the list of fault types in a sample subset can be represented as f1 = [0, 0, 0, ... 1, 0 ...], where the position with a value of 1 represents the fault type with the largest number in the sample subset. In this sample subset, when the number of multiple fault types is equal, the corresponding fault type list can have multiple values of 1.

[0073] Execute the above steps 204-206 in a loop, perform t search calculations on the sample set, and obtain multiple And its corresponding fault type list is shown below:

[0074] f 1=[0, 0, 0, …1, 0…] (5)

[0075] f 2=[0, 0, 1, …1, 0…] (6)

[0076] …

[0077] ft =[0, 0, 0, …1, 0…] (7)

[0078] but, f = f 1+ f 2+… ft (8)

[0079] in, f The location with the largest value can be determined as the potential fault type of the battery.

[0080] As can be seen from the above description, the embodiment of the present application pre-generates a training sample set with fault samples and labels based on prior knowledge, and then combines the current battery's test sample with the training sample set to form a new sample set. Then, the binary tree method is used to search for corresponding samples in the new sample set based on random features, and the sample subsets corresponding to the test sample are retained in different feature dimensions until the search condition is terminated, and the final sample subset is obtained. The number of each type of abnormal fault sample in the final sample subset is calculated to determine the potential fault category. At the same time, in order to avoid uneven distribution of the number of fault samples in the training sample set, different weights are set for fault types with different sample numbers. In addition, in order to eliminate the influence of randomness, the embodiment of the present application adopts an integrated learning strategy to independently perform multiple judgments, so that the potential fault type of the battery can be accurately determined by voting.

[0081] The embodiment of the present application integrates the idea of ensemble learning, is applicable to fault diagnosis scenarios with unbalanced samples, and does not require model training. It can quickly, accurately and effectively determine the potential fault type of the battery.

[0082] Based on similar inventive concepts, an embodiment of the present application further provides a device for determining a battery fault category, which can preferably be used to implement the process of the above-mentioned method for determining a battery fault category.

[0083] Figure 3 This is a structural block diagram of the battery fault category determination device, such as Figure 3 As shown, the device includes: an information acquisition unit 1, a data merging unit 2, a subsample data search unit 3, a data quantity determination unit 4 and a fault category determination unit 5, wherein:

[0084] An information acquisition unit 1 is configured to acquire current operating information of the battery, wherein the current operating information includes: at least one information category;

[0085] A data merging unit 2 is configured to merge the pre-set fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data;

[0086] The sub-sample data search unit 3 is configured to search for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category based on a predetermined algorithm;

[0087] The data quantity determining unit 4 is used to determine the data quantity of each fault category in each sub-sample data based on the predetermined weight of each fault category;

[0088] The fault category determination unit 5 is configured to determine the fault category of the battery according to the number of data of each fault category in each sub-sample data.

[0089] In an embodiment of the present invention, the current operating information of the battery obtained by the information acquisition unit 1 is merged with the pre-set fault sample data through the data merging unit 2 to generate new fault sample data. Then, the sub-sample data search unit 3 searches for corresponding sub-sample data from the new fault sample data according to each information category based on the budget algorithm. The data quantity determination unit 4 calculates the number of sample data of each fault category in each sub-sample data. Then, the fault category determination unit 5 determines the fault category of the battery according to the number of sample data of each fault category, thereby realizing real-time judgment of the potential fault category of the battery, realizing potential risk identification and maintenance of the energy storage lithium battery, and improving the safety of the energy storage power station.

[0090] In one embodiment, the above apparatus further includes: a pre-processing unit and a feature information generating unit, wherein:

[0091] A preprocessing unit, configured to preprocess the current operation information;

[0092] The feature information generating unit is used to perform feature extraction processing on the pre-processed current operation information to obtain the feature information of the battery.

[0093] After obtaining the characteristic information of the battery, the data merging unit may merge the fault sample data with the characteristic information of the battery.

[0094] In one embodiment, the predetermined algorithm may be a binary tree algorithm. The subsample data search unit 3 includes: a rule setting module and a subsample data search module, wherein:

[0095] The rule setting module is used to set different search termination rules based on the binary tree algorithm;

[0096] The sub-sample data search module is used to search for a plurality of sub-sample data corresponding to the information category and the search termination rule from the new fault sample data according to the at least one information category and each search termination rule.

[0097] The fault type determination unit 5 specifically includes: a quantity data determination module and a fault type determination module, wherein:

[0098] A quantity data determination module, configured to determine the quantity of data of each fault category in each sub-sample data;

[0099] The fault category determination module is configured to determine the fault category with the largest amount of data as the fault category of the battery.

[0100] The specific execution process of the above-mentioned units and modules can be found in the description of the above-mentioned method embodiment, which will not be repeated here.

[0101] In actual operation, the above-mentioned units and modules can be provided in combination or individually, and the present invention is not limited thereto.

[0102] This embodiment further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. The electronic device may be a desktop computer, a tablet computer, a mobile terminal, etc., but this embodiment is not limited thereto. In this embodiment, the electronic device may be implemented with reference to the aforementioned method embodiment and the embodiment of the apparatus for determining a battery fault category, the contents of which are incorporated herein, and any repetitions are omitted.

[0103] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the above-mentioned method for determining a battery fault category are implemented.

[0104] Specific embodiments are used in the present invention to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present invention.

Claims

1. A method for determining a battery fault category, characterized in that: The method comprises: Acquiring current operating information of the battery, the current operating information including: at least one information category; Merging the pre-set fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data; Based on a predetermined algorithm, searching for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category; Determine the number of data for each fault category in each sub-sample data based on the predetermined weight of each fault category; Determining the fault category of the battery according to the number of data of each fault category in each sub-sample data; The predetermined algorithm is a binary tree algorithm. Based on the predetermined algorithm, searching for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category includes: Based on the binary tree algorithm, set different search termination rules; According to the at least one information category, searching for a plurality of sub-sample data corresponding to the information category and the search termination rule from the new fault sample data using respective search termination rules; Determining the fault category of the battery according to the number of data of each fault category in each sub-sample data includes: Determining the number of data of each fault category in each sub-sample data; The fault category with the largest amount of data is determined as the fault category of the battery.

2. The method for determining a battery fault type according to claim 1, wherein: After obtaining the current operating information of the battery, the method further includes: The current operation information is preprocessed, and feature extraction is performed on the preprocessed current operation information to obtain feature information of the battery.

3. The method for determining a battery fault type according to claim 2, wherein: Merging the pre-set fault sample data containing multiple fault categories with the current operation information includes: The fault sample data and the characteristic information of the battery are combined and processed.

4. A device for determining a battery fault category, characterized in that: The device comprises: an information acquisition unit, configured to acquire current operating information of the battery, the current operating information including: at least one information category; A data merging unit, configured to merge the preset fault sample data containing multiple fault categories with the current operation information to obtain new fault sample data; a sub-sample data searching unit, configured to search for corresponding at least one sub-sample data from the new fault sample data according to the at least one information category based on a predetermined algorithm; a data quantity determining unit, configured to determine the data quantity of each fault category in each subsample data based on a predetermined weight of each fault category; a fault category determination unit, configured to determine a fault category of the battery according to the number of data of each fault category in each sub-sample data; The predetermined algorithm is a binary tree algorithm, and the subsample data search unit includes: The rule setting module is used to set different search termination rules based on the binary tree algorithm; The sub-sample data search module is used to search for a plurality of sub-sample data corresponding to the information category and the search termination rule from the new fault sample data according to the at least one information category and each search termination rule.

5. The device for determining a battery fault type according to claim 4, wherein: The device further comprises: A preprocessing unit, configured to preprocess the current operation information; a feature information generating unit, configured to perform feature extraction processing on the pre-processed current operation information to obtain feature information of the battery; The data merging unit is specifically configured to perform data merging processing on the fault sample data and the characteristic information of the battery.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method for determining a battery fault category according to any one of claims 1 to 3 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the computer program implements the steps of the method for determining a battery fault category according to any one of claims 1 to 3.

Images