Battery Fault Identification Method, Device, Electronic Equipment and Storage Medium

Abstract

The present invention discloses a battery fault identification method, device, equipment and storage medium. The method includes: performing variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components characterizing the battery cell voltage data; superimposing the intrinsic mode components of the same type among the plurality of intrinsic mode components to obtain a target mode combination component characterizing fault information; extracting a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component; and determining a fault identification result of the battery cell characterized by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor. The technical solution provided by the present invention can, to a certain extent, improve the accuracy of early battery fault diagnosis.

Description

Technical Field

[0001] The present invention relates to the field of battery fault diagnosis, and particularly to a battery fault identification method, device, electronic device and storage medium. Background Art

[0002] Extreme operating conditions and harsh environments may cause changes in the internal structure of the battery, generating excessive heat, which may ultimately lead to thermal runaway. In addition, during the manufacturing process, there may be defects in material distribution and structural design, which may further increase the risk of thermal runaway. The thermal runaway of power batteries has high harmfulness, multiple triggering incentives and influencing factors, complex evolution laws, and strong concealment before accidents occur. In the prior art, most of the fault determination and identification of battery fault diagnosis methods focus on a relatively short time range before the battery fails or thermal runaway occurs, and the accuracy of early battery fault diagnosis is relatively low. Summary of the Invention

[0003] In view of this, embodiments of the present invention provide a battery fault identification method, device, electronic device and storage medium, which can improve the accuracy of early battery fault diagnosis to a certain extent.

[0004] One aspect of the present invention provides a battery fault identification method, the method comprising: performing variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components representing the battery cell voltage data; superimposing the intrinsic mode components of the same type among the plurality of intrinsic mode components to obtain a target mode combination component representing fault information; extracting a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component; and determining a fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor.

[0005] In one embodiment, performing variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components representing the battery cell voltage data includes: determining target parameters by using a particle swarm optimization algorithm; the target parameters include a target penalty coefficient and a target number of modes; iteratively solving a variational expression based on the target penalty coefficient and the target number of modes, and determining a plurality of intrinsic mode components representing the battery cell voltage when an iterative improvement value is less than a preset convergence tolerance; the iterative improvement value is the square of the absolute average of the convergence to the intrinsic mode components in two adjacent iterative processes.

[0006] In one embodiment, superimposing the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain a target mode combination component representing fault information includes: superimposing the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain multiple mode combination components; determining the mode combination component of the target type among the multiple mode combination components as the target mode combination component representing fault information.

[0007] In one embodiment, before the step of determining the mode combination component of the target type among the multiple mode combination components as the target mode combination component representing fault information, the method further includes: obtaining historical voltage data of a plurality of historical battery cells; the types of the historical battery cells include safe battery cells and faulty battery cells; respectively performing variational mode decomposition on the historical voltage data of the plurality of historical battery cells to obtain a plurality of historical intrinsic mode components representing the historical voltage data; superimposing the historical intrinsic mode components of the same type among the multiple historical intrinsic mode components representing the same historical voltage data to obtain multiple historical mode combination components; calculating the correlation coefficient between each historical mode combination component and the type of the historical battery cell; determining the type of the historical mode combination component corresponding to the maximum value of the correlation coefficient as the target type.

[0008] In one embodiment, extracting a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component includes: calculating the ratio of the effective value of the cell voltage of the target mode combination component to the rectified average value of the voltage data of each sampling point in the target mode combination component to obtain the target differential waveform factor; calculating the ratio of the third-order central moment of the cell voltage of the target mode combination component to the cube of the standard deviation of the voltage data of each sampling point in the target mode combination component to obtain the target differential skewness factor.

[0009] In one embodiment, determining a fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor includes: generating two-dimensional feature data of the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; calculating the average spatial distance between the two-dimensional feature data and the preset two-dimensional feature data corresponding to the fault type; determining the fault type corresponding to the minimum value of the average spatial distance as the fault identification result of the battery cell represented by the battery cell voltage data.

[0010] In one embodiment, the number of the battery cell voltage data includes a plurality. Based on the target differential waveform factor and the target differential skewness factor, determining a fault identification result of the battery cell characterized by the battery cell voltage data includes: generating two-dimensional feature data of each battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; performing clustering analysis on the two-dimensional feature data of multiple battery cells, and determining the battery cells characterized by the two-dimensional feature data that cannot be classified into the same cluster as faulty battery cells.

[0011] On the other hand, the present invention also provides a battery fault identification device. The battery fault identification device includes: a voltage decomposition unit configured to perform variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components characterizing the battery cell voltage data; a mode component superposition unit configured to superpose the intrinsic mode components of the same type among the plurality of intrinsic mode components to obtain a target mode combination component characterizing fault information; a feature extraction unit configured to extract a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component; and a fault identification unit configured to determine a fault identification result of the battery cell characterized by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor.

[0012] On the other hand, the present invention also provides an electronic device. The electronic device includes a processor and a memory. The memory is configured to store a computer program. When the computer program is executed by the processor, the battery fault identification method described above is implemented.

[0013] On the other hand, the present invention also provides a computer-readable storage medium. The computer-readable storage medium is configured to store a computer program. When the computer program is executed by a processor, the battery fault identification method described above is implemented.

[0014] By performing variational mode decomposition on the acquired battery cell voltage data, then superposing the intrinsic mode components of the same type among the decomposed intrinsic mode components to obtain a plurality of mode combination components, then selecting a mode combination component characterizing fault information as the target mode combination component from the plurality of mode combination components, and extracting a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component, and then inputting these two dimensionless feature factors into a trained fault diagnosis model to obtain a fault identification result of the battery cell characterized by the battery cell voltage data, the accuracy of early battery fault diagnosis can be improved to a certain extent. Description of the Drawings

[0015] The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings. The drawings are schematic and should not be construed as imposing any limitation on the present invention. In the drawings:

[0016] Figure 1 It shows a schematic diagram of the steps of a battery fault identification method in an embodiment of the present disclosure;

[0017] Figure 2 It shows a schematic diagram of multiple intrinsic mode components obtained after variational mode decomposition of the battery voltage signal in an embodiment of the present disclosure;

[0018] FIG. 3(a) shows a schematic diagram of a differential waveform factor extracted based on a target mode combination component in an embodiment of the present disclosure;

[0019] FIG. 3(b) shows a schematic diagram of a differential skewness factor extracted based on a target mode combination component in an embodiment of the present disclosure;

[0020] Figure 4 It shows a schematic diagram of the result of clustering two-dimensional features of multiple battery cells in an embodiment of the present disclosure;

[0021] Figure 5 It shows a schematic diagram of the flow of a battery fault identification method in an embodiment of the present disclosure;

[0022] Figure 6 It shows a schematic diagram of a battery fault identification device in an embodiment of the present disclosure;

[0023] Figure 7 It shows a schematic diagram of the structure of an electronic device in an embodiment of the present disclosure. Detailed Embodiment

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present disclosure.

[0025] Please refer to Figure 1 , a battery fault identification method provided by an embodiment of the present disclosure, which may include the following multiple steps.

[0026] S110: Perform variational mode decomposition on the acquired battery cell voltage data to obtain multiple intrinsic mode components representing the battery cell voltage data.

[0027] In this embodiment, the method of variational mode decomposition based on signal decomposition can effectively remove the noise interference in the battery operating state while retaining the fault information of the original signal; by screening and combining modes, the early trend of faults is expanded, the timeliness of lithium-ion battery fault diagnosis is improved, and it helps to realize real-time online diagnosis of lithium-ion batteries.

[0028] In this embodiment, in signal processing, variational mode decomposition is a signal decomposition estimation method. In the process of obtaining the decomposition components, this method determines the frequency center and bandwidth of each component by iteratively searching for the optimal solution of the variational model, so as to be able to adaptively achieve the frequency-domain dissection of the signal and the effective separation of each component. The signal decomposition method based on parameter optimization can decompose the signal collected by the sensor into intrinsic mode functions of different frequencies. When a fault occurs, the phase-frequency characteristics of the battery also change. By analyzing the corresponding frequency band, the faulty battery can be identified.

[0029] In this embodiment, the multiple intrinsic mode components are subsequences that are relatively stable and represent different frequency scales obtained by performing variational mode decomposition on the battery voltage signal acquired by the sensor. Among them, the number of intrinsic mode components depends on the number of modes that need to be decomposed set by the user.

[0030] S120: Superimpose the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain a target mode combination component representing the fault information.

[0031] In this embodiment, the intrinsic mode components of the same type can all be used to represent the fault type of the battery cell, but specifically, it cannot be determined which intrinsic mode component should be used to extract the fault characteristics of the battery cell. Therefore, the intrinsic mode components of the same type can be superimposed to obtain a superimposed mode combination component, and then the mode combination component with the strongest correlation with the fault information in the superimposed mode combination component is determined as the target mode combination component.

[0032] In this embodiment, the superimposition of the intrinsic mode components of the same type among the multiple intrinsic mode components can be obtained by adding the signal data of the intrinsic mode components of the same type and at the same sampling time. Of course, there may be only one intrinsic mode component of the same type, and in this case, no superimposition operation is required, and this intrinsic mode component can be used as the combined mode component.

[0033] S130: Extract the target differential waveform factor and the target differential skewness factor of the battery cell voltage data based on the target mode combination component.

[0034] In this embodiment, by extracting two dimensionless features, namely the differential waveform factor and the differential skewness factor, if the dimensionless characteristic parameters of a certain cell voltage during the charge and discharge process are significantly different from those of other charged cells, it is determined that there is an abnormality in the cell voltage, thereby realizing the abnormal diagnosis of lithium-ion batteries.

[0035] The fault feature of the differential waveform factor is the ratio of the root mean square (RMS) value of the cell voltage to the rectified average value. Its physical meaning in the electronic field can be understood as the ratio of direct current to alternating current with equal power, and its value is greater than or equal to 1. The fault feature of the differential skewness factor is the ratio of the third central moment to the cube of the standard deviation. Skewness describes the distribution. Its physical meaning can be simply understood as follows: for a unimodal distribution, negative skewness means that the "head" of the distribution curve is on the right side and the "tail" is on the left side; positive skewness is the opposite. The waveform factor and the skewness factor are sensitive to the faults of lithium-ion battery cells. Especially when a fault occurs in the early stage, they have obvious abnormalities, indicating their high sensitivity to early faults.

[0036] S140: Based on the target differential waveform factor and the target differential skewness factor, determine the fault identification result of the battery cell represented by the battery cell voltage data.

[0037] In this embodiment, different dimensionless features have different sensitivities to the faults of battery cells. By diagnosing the faults of lithium battery cells through these two dimensionless features, namely the target differential waveform factor and the target differential skewness factor, it is possible to compare whether there are false alarms or missed detections, and comprehensively judge to obtain a more accurate and timely fault identification result.

[0038] In this embodiment, based on the two-dimensional feature formed by the target differential waveform factor and the target differential skewness factor, and the spatial distance between the two-dimensional feature formed by the differential waveform factor and the differential skewness factor of the faulty cells of different fault types in the preset database, the fault type with the shortest spatial distance is used as the fault type of the battery cell represented by the battery cell voltage data.

[0039] By inputting battery data in real time through an online window and extracting fault features for fault diagnosis, safety warning information can be displayed on the platform, detecting faults earlier than the BMS to indicate the trend of faults in some battery cells and taking corresponding protective measures.

[0040] In one embodiment, variational mode decomposition is performed on the acquired battery cell voltage data to obtain multiple intrinsic mode components characterizing the battery cell voltage data, which may include: determining target parameters using a particle swarm optimization algorithm; the target parameters include a target penalty coefficient and a target number of modes; iteratively solving a variational expression based on the target penalty coefficient and the target number of modes, and when the iterative improvement value is less than a preset convergence tolerance, determining multiple intrinsic mode components characterizing the battery cell voltage; the iterative improvement value is the square of the absolute average of the convergence to the intrinsic mode components in two adjacent iterative processes.

[0041] In this embodiment, variational mode decomposition (VMD) requires subjective definition of parameters and is extremely sensitive to parameter selection for different signals. Therefore, by using a parameter optimization algorithm to select a set of parameters suitable for the characteristics of the lithium-ion battery voltage signal for decomposition, a relatively ideal decomposition effect can be obtained, which is beneficial to subsequent feature extraction and fault information clustering.

[0042] Variational mode decomposition (VMD) assumes that any signal is composed of a series of sub-signals with specific center frequencies and finite bandwidths (i.e., IMFs). Based on classical Wiener filtering, by solving a variational problem, the center frequency and bandwidth limitations are obtained, the effective components corresponding to each center frequency in the frequency domain are found, and the mode functions are obtained. Its model construction involves Wiener filtering, Hilbert transform, and analytic signals, etc. The decomposition process of VMD is the process of solving the variational problem, and its algorithm mainly includes the construction of the variational problem and the solution of the variational problem. The solution process of VMD mainly includes two constraints: (1) requiring the sum of the bandwidths of the center frequencies of each mode component to be the smallest; (2) the sum of all mode components is equal to the original signal.

[0043] The so-called variational problem is to find the extreme value of a functional. In VMD, the functional refers to the VMD constrained variational model, and the extreme value to be found is that "the sum of the bandwidths of the center frequencies of each mode component is the smallest".

[0044] The VMD constrained variational model is as follows:

[0045]

[0046] where, u k ={u1, u2,..., u k} are the mode functions, and ω k ={ω1, ω2,..., ω k} are the center frequencies of each mode. f represents the original voltage signal; δ represents the Dirac distribution, t represents the sampling time; * represents the convolution operator.

[0047] The constrained problem in the VMD (Variational Mode Decomposition) constrained variational model is transformed into an unconstrained variational problem. By introducing a quadratic penalty factor and a Lagrange multiplier operator, the extended Lagrangian expression is as follows:

[0048]

[0049] Among them, α represents the penalty coefficient; λ represents the undetermined multiplier. The multiplier algorithm alternating direction method is used for solving.

[0050] Regarding the parameter settings of VMD, there are mainly three parameters: the penalty coefficient α, the number of modes k, and the convergence tolerance tol (tolerant).

[0051] The penalty coefficient α is used to determine the bandwidth of the IMF (Intrinsic Mode Function) components. The smaller the penalty coefficient, the larger the bandwidth of each IMF component. An overly large bandwidth will cause some components to contain signals of other components. The larger the penalty coefficient, the smaller the bandwidth of each IMF component. An overly small bandwidth will cause some signals in the decomposed signal to be lost. The common value range of the penalty coefficient α is usually 1000 - 3000.

[0052] The number of modes k is the number of IMF components specified for decomposition. If the set k is less than the number of useful components in the signal to be decomposed (under - decomposition), it will cause insufficient decomposition and lead to mode mixing; if the set value of K is greater than the number of useful components in the signal to be decomposed (over - decomposition), it will result in the generation of some useless false components. Therefore, the determination of the k value is very important for VMD.

[0053] The convergence tolerance tol is one of the optimization stopping criteria. In two consecutive iterations, when the absolute average square improvement towards the IMF convergence is less than tol, the optimization stops. Usually, it can be taken as 1e - 6 to 5e - 6.

[0054] The particle swarm optimization algorithm is a swarm intelligence optimization algorithm with good global optimization ability. Using the particle swarm algorithm to perform parallel optimization on the two parameters: the penalty coefficient α and the number of modes k can avoid the intervention of human subjective factors and automatically screen out the best combination of influencing parameters. Decompose the lithium - ion battery voltage signal with the optimized parameters to obtain different intrinsic mode components.

[0055] In one embodiment, superimposing the intrinsic mode components with the same type among the multiple intrinsic mode components to obtain a target mode combination component representing fault information may include: superimposing the intrinsic mode components with the same type among the multiple intrinsic mode components to obtain multiple mode combination components; determining the mode combination component with the target type among the multiple mode combination components as the target mode combination component representing fault information.

[0056] Please refer to Figure 2, in this embodiment, a series of sub-signals (IMFs, intrinsic mode components) with a specific center frequency and limited bandwidth are obtained by decomposing the battery cell voltage signal. Then, different components are split and combined to obtain multiple modal combination components. Then, a target modal combination component that characterizes the fault information is determined from the multiple modal combination components. Specifically, for example, the battery cell voltage data is decomposed into 5 intrinsic mode components (IMF1 - IMF8), where IMF1 represents the low-frequency component, and IMF2 - 7 characterize the high-frequency components. Then, IMF2 - 7 are superimposed to obtain two modal combination components. Since the high-frequency components can characterize the fault information, the high-frequency components are used as the target modal combination components.

[0057] In one embodiment, before the step of determining the modal combination component of the target type among the multiple modal combination components as the target modal combination component that characterizes the fault information, the method further includes: obtaining historical voltage data of a plurality of historical battery cells; the types of the historical battery cells include safe battery cells and faulty battery cells; respectively performing variational mode decomposition on the historical voltage data of the plurality of historical battery cells to obtain a plurality of historical intrinsic mode components that characterize the historical voltage data; superimposing the historical intrinsic mode components of the same type among the plurality of historical intrinsic mode components that characterize the same historical voltage data to obtain a plurality of historical modal combination components; calculating the correlation coefficient between each historical modal combination component and the fault type of the historical battery cell; determining the type of the historical modal combination component corresponding to the maximum value of the correlation coefficient as the target type that characterizes the fault information.

[0058] In this embodiment, since it is necessary to determine which modal combination component among the multiple modal combination components can best characterize the fault information. Therefore, the historical voltage data of the historical battery cells can be obtained. The historical battery cells include safe battery cells and faulty battery cells. Then, variational mode decomposition is performed on these historical voltage data, and the modal components of the same type in the decomposition results are combined to obtain the historical combination components corresponding to each historical voltage data.

[0059] Then, the correlation coefficient between the historical combination components of the same type and the type of the historical battery cell can be calculated. The type of the historical modal combination component corresponding to the maximum value of the correlation coefficient is determined as the modal combination component that can best characterize the fault information.

[0060] In one embodiment, extracting the target differential waveform factor and the target differential skewness factor of the battery cell voltage data based on the target modal combination component may include: calculating the ratio of the effective value of the cell voltage of the target modal combination component to the rectified average value of the voltage data of each sampling point in the target modal combination component to obtain the target differential waveform factor; calculating the ratio of the third-order central moment of the cell voltage of the target modal combination component to the cube of the standard deviation of the voltage data of each sampling point in the target modal combination component to obtain the target differential skewness factor.

[0061] Please refer to FIGS. 3(a) and 3(b). In this embodiment, the extracted target differential waveform factor is the ratio of the effective value (RMS) of each cell voltage to the rectified average value.

[0062]

[0063] Wherein, U rms is the effective value of the cell voltage within the charging segment sliding window, is the rectified average value of the cell voltage within the charging segment sliding window, u i is the cell voltage value of each sampling point within the window, n is the number of sampling points within the window, and p(u i ) is the occurrence probability of u i .

[0064] The extracted target differential skewness factor is the ratio of the third-order central moment of each cell voltage to the cube of the standard deviation.

[0065]

[0066] Wherein, u i is the cell voltage value of each sampling point within the window, μ i is the average value of the cell voltage within the window. n is the number of sampling points within the window, and p(u i ) is the occurrence probability of u i .

[0067] If the dimensionless characteristic parameter of a certain cell voltage during the charge and discharge process is significantly different from the data of other charging cells, it is determined that there is an abnormality in the cell voltage, and the abnormal diagnosis of the lithium-ion battery is realized.

[0068] In one embodiment, determining the fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor may include: generating two-dimensional feature data of the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; calculating the average spatial distance between the two-dimensional feature data and the preset two-dimensional feature data corresponding to different fault types; and determining the type corresponding to the minimum value of the average spatial distance as the fault identification result of the battery cell represented by the battery cell voltage data.

[0069] In this embodiment, generating the two-dimensional feature data of the battery cell voltage data by the target differential waveform factor and the target differential skewness factor helps to improve the accuracy of battery fault identification. The preset two-dimensional feature data is the two-dimensional feature data generated from the single-cell voltage data of preset battery cells of each type. First, the spatial distance between the two-dimensional feature data and each preset two-dimensional feature data can be calculated. Then, the average value of the spatial distances between different types of battery cells can be calculated respectively. Finally, the fault type corresponding to the minimum spatial distance is used as the fault type of the battery cell.

[0070] In one embodiment, the number of the battery cell voltage data includes multiple. Determining the fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor may include: generating two-dimensional feature data of each battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; performing clustering analysis on the two-dimensional feature data of multiple battery cells, and determining the battery cells represented by the two-dimensional feature data that cannot be grouped into the same cluster as the faulty battery cells.

[0071] In this embodiment, when it is necessary to determine whether there are faults in multiple battery cells, the number of battery cells with abnormal faults occurring simultaneously in the battery pack is extremely small. Therefore, the few abnormal features in the extracted fault feature sequence can automatically form clusters. The two-dimensional clustering method can be used to diagnose the faults of lithium battery cells, and the effects of false alarms and missed alarms can be compared to comprehensively obtain a more accurate and timely fault diagnosis conclusion.

[0072] DBSCAN defines a cluster as the largest set of density-connected points, which can divide the area with sufficient high density into clusters and can discover clusters of any shape in a spatial database with noise.

[0073] In the DBSCAN algorithm, there are two parameters, Radius ε and minPts. minPts is set according to experience, usually twice the dimension, that is, 4. ε is set manually or obtained through training according to actual requirements.

[0074] DBSCAN is a density-based clustering algorithm. The principle is that as long as any two sample points are directly density-reachable or density-reachable, then these two sample points are grouped into the same cluster. The sample points ABCE in the above figure belong to the same cluster. Therefore, the DBSCAN algorithm randomly selects a core point from the dataset D as the "seed", determines the corresponding clustering cluster starting from this seed, and when all core points have been traversed, the algorithm ends. Therefore, the samples that cannot be grouped into the same cluster can be determined as outliers.

[0075] Please refer to Figure 4 , it is difficult to determine the threshold for extracting a single feature to judge the fault information, which is prone to false alarm problems. The two-dimensional fault features of the comprehensive differential waveform factor and skewness factor are comprehensively clustered to judge the deviation of the fault monomer. Using the DBSCAN clustering method, by changing the size of the eps radius, accurate identification is achieved, without mis-screening or missing screening, and the outlier fault monomers are accurately identified. It realizes the extraction of fault-sensitive characteristic parameters from the lithium battery operation data to detect faults, without considering the complex battery fault mechanism, simplifies the fault detection process, and realizes the lithium battery fault diagnosis and safety warning to ensure the safe and stable operation of the battery.

[0076] Please refer to Figure 5 , the embodiment of this specification provides a scenario example of a battery fault identification method. First, pre-clean the lithium battery data, integrate, eliminate, screen, and correct the data. Decompose the voltage signal of the battery monomer after pre-cleaning. The signal decomposition method selects the variational mode decomposition method after optimizing the parameters to obtain different intrinsic mode components after decomposition. Then, stack and process the different intrinsic mode components according to their types to obtain different types of modal combination components. Select the modal combination component with the highest correlation with the battery fault among multiple modal combination components as the target modal combination component, and then extract the fault features of the target modal combination component. The selected features include the differential waveform factor and the differential skewness factor. Based on the differential waveform factor and the differential skewness factor, two-dimensional feature clustering is performed to identify the fault monomers, realizing the early fault diagnosis and safety warning of lithium-ion batteries.

[0077] Fuse the fault diagnosis principle and advanced signal processing methods to identify the characteristic information in different time domain and frequency domain scales under various fault states. The described fault features are to extract fault-sensitive characteristic parameters from the battery operation data to detect faults, without considering the complex battery fault mechanism, simplifies the fault detection process, and is widely used in practice.

[0078] Generally, in actual situations, the number of battery cells with abnormal faults occurring simultaneously in a battery pack is extremely small. Therefore, clustering can be automatically formed by a few abnormal features in the extracted fault feature sequence. Since different features have different sensitivities to faults, a two-dimensional clustering method is used to diagnose the faults of lithium battery cells, and the effects of false alarms and missed alarms can be compared to comprehensively judge and obtain a more accurate and timely fault diagnosis conclusion.

[0079] The voltage signal of a lithium battery cell is decomposed through signal decomposition, and fault features are extracted from the decomposed structure. Two-dimensional feature clustering is performed using dimensionless features such as the differential waveform factor and the differential skewness factor. By setting the clustering threshold, a few faulty battery cells are identified, effectively diagnosing and warning of the occurrence of faults. This method is targeted at the characteristics of the voltage information of lithium battery cells and has strong adaptability to different working conditions of the battery. It can effectively diagnose the faults of the battery in the charging, static, and discharging operating states, effectively warn of the occurrence of thermal runaway, and improve the safety of battery applications.

[0080] In the fault diagnosis of a battery pack, there are unknown abnormal signal patterns and corresponding specific features, and it is difficult to achieve without complete threshold setting information. Regarding the problem of threshold setting, the DBSCAN clustering method can greatly reduce the difficulty of threshold setting and obtain superior anomaly detection performance. DBSCAN clustering can accurately identify outlier faulty cells by changing the size of eps: radius.

[0081] The voltage signal of a lithium battery cell is decomposed through signal decomposition, and fault features are extracted from the decomposed structure. Two-dimensional feature clustering is performed using dimensionless features such as the differential waveform factor and the differential skewness factor. By setting the clustering threshold, a few faulty battery cells are identified, effectively diagnosing and warning of the occurrence of faults. This method is targeted at the characteristics of the voltage information of lithium battery cells and has strong adaptability to different working conditions of the battery. It can effectively diagnose the faults of the battery in the charging, static, and discharging operating states, effectively warn of the occurrence of thermal runaway, and improve the safety of battery applications.

[0082] Please refer to Figure 6 , an embodiment of the present disclosure further provides a battery fault identification device, and the battery fault identification device may include: a voltage decomposition unit, a modal component superposition unit, a feature extraction unit, and a fault identification unit.

[0083] The voltage decomposition unit is configured to perform variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components representing the battery cell voltage data.

[0084] A modal component superposition unit is configured to superpose the eigenmode components of the same type among the multiple eigenmode components to obtain a target modal combination component characterizing the fault information.

[0085] A feature extraction unit is configured to extract a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target modal combination component.

[0086] A fault identification unit is configured to determine a fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor.

[0087] For the specific functions and effects achieved by the battery fault identification device, reference may be made to other embodiments of this specification for explanation, which will not be elaborated here. Each module in the battery fault identification device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or stored in a memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above respective modules.

[0088] Please refer to Figure 7 , an embodiment of the present disclosure further provides an electronic device, where the electronic device includes a processor and a memory. The memory is used to store a computer program, and when the computer program is executed by the processor, the above battery fault identification method is implemented.

[0089] Among them, the processor can be a central processing unit (CPU). The processor can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., or a combination of the above various chips.

[0090] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs, non-transitory computer-executable programs, and modules, such as program instructions / modules corresponding to the method in the embodiments of the present invention. The processor executes various functional applications and data processing of the processor by running the non-transitory software programs, instructions, and modules stored in the memory, that is, the method in the above method embodiments is implemented.

[0091] The memory may include a program storage area and a data storage area. The program storage area may store an operating system and application programs required for at least one function. The data storage area may store data created by the processor and the like. In addition, the memory may include a high-speed random access memory and may also include a non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0092] An embodiment of the present disclosure also provides a computer-readable storage medium for storing a computer program, which when executed by a processor, implements the above-mentioned battery fault identification method.

[0093] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it may include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in this specification may include at least one of non-volatile and volatile memories. Non-volatile memories may include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, or optical memories, etc. Volatile memories may include random access memory (RAM) or external cache memories. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0094] It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0095] Among the various embodiments of this specification, a progressive approach is adopted for description. Different embodiments focus on describing the parts that are different from other embodiments. After reading this specification, those skilled in the art can learn about the various embodiments in this specification and the various technical features disclosed by the embodiments, and can make more combinations. For the sake of brevity of description, not all possible combinations of the various technical features in the embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as within the scope described in this specification.

[0096] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent in such process, method, commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of additional identical elements in the process, method, commodity or device including the said element.

[0097] Each of the various embodiments in this specification itself emphasizes the parts that are different from other embodiments, and the embodiments can be mutually interpreted. Any combination of the various embodiments in this specification by those skilled in the art based on general technical knowledge is covered by the disclosure scope of this specification.

[0098] The above are only the embodiments of this case and are not used to limit the protection scope of the claims of this case. For those skilled in the art, various changes and modifications can be made to this case. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this case shall be included within the scope of the claims of this case.

Claims

1. A method for identifying battery faults, characterized in that, The method includes: Performing variational mode decomposition on the acquired battery cell voltage data to obtain multiple intrinsic mode components characterizing the battery cell voltage data; wherein, optimizing two parameters, namely the penalty coefficient and the number of modes, in the variational mode decomposition relation, and using the optimized parameters to perform signal decomposition on the battery cell voltage data to obtain the multiple intrinsic mode components; the penalty coefficient and the number of modes are two parameters optimized in parallel and applicable to the characteristics of the lithium-ion battery voltage signal selected by the particle swarm optimization algorithm; the multiple intrinsic mode components are the intrinsic mode components characterizing the battery cell voltage determined by iteratively solving the variational expression based on the penalty coefficient and the number of modes, when the iterative improvement value is less than the preset convergence tolerance; the iterative improvement value is the square of the absolute average value converging to the intrinsic mode component in two adjacent iterative processes; the multiple intrinsic mode components are subsequences that are relatively stable and characterize different frequency scales obtained by performing variational mode decomposition on the battery voltage signal acquired by the sensor; Superposing the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain a target mode combination component characterizing the fault information; the target mode combination component is the mode combination component with the strongest correlation with the fault information among the superposed mode combination components; Extracting a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target mode combination component; the fault feature of the target differential waveform factor is the ratio of the effective value of the cell voltage to the rectified average value; the fault feature of the target differential skewness factor is the ratio of the third central moment to the cube of the standard deviation; the target differential waveform factor and the target differential skewness factor are eigenvalue sensitive to the faults of lithium-ion battery cells; Determining the fault identification result of the battery cell characterized by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; Determining the fault identification result of the battery cell characterized by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor includes: Generating two-dimensional feature data of the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; Calculating the average spatial distance between the two-dimensional feature data and the preset two-dimensional feature data corresponding to different fault types; Determining the fault type corresponding to the minimum value of the average spatial distance as the fault identification result of the battery cell characterized by the battery cell voltage data.

2. The method according to claim 1, wherein Superposing the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain a target mode combination component characterizing the fault information, including: Superposing the intrinsic mode components of the same type among the multiple intrinsic mode components to obtain multiple mode combination components; Determining the mode combination component of the target type among the multiple mode combination components as the target mode combination component characterizing the fault information.

3. The method according to claim 2, characterized in that, Before the step of determining the mode combination component of the target type among the multiple mode combination components as the target mode combination component characterizing the fault information, the method further includes: Obtain historical voltage data of a number of historical battery cells; the types of the historical battery cells include safe battery cells and faulty battery cells; Perform variational mode decomposition on the historical voltage data of a number of the historical battery cells respectively to obtain a plurality of historical intrinsic mode components characterizing the historical voltage data; Superimpose the historical intrinsic mode components of the same type among the plurality of historical intrinsic mode components characterizing the same historical voltage data to obtain a plurality of historical mode combination components; Calculate the correlation coefficients between each historical mode combination component and the type of the historical battery cell; Determine the type of the historical mode combination component corresponding to the maximum value of the correlation coefficient as the target type; 4. The method according to claim 1, wherein Extract the target differential waveform factor and the target differential skewness factor of the battery cell voltage data based on the target mode combination component, including: Calculate the ratio of the effective value of the cell voltage of the target mode combination component to the rectified average value of the voltage data of each sampling point in the target mode combination component to obtain the target differential waveform factor; Calculate the ratio of the third-order central moment of the cell voltage of the target mode combination component to the cube of the standard deviation of the voltage data of each sampling point in the target mode combination component to obtain the target differential skewness factor; 5. The method according to claim 1, wherein The number of the battery cell voltage data includes a plurality. Based on the target differential waveform factor and the target differential skewness factor, determine the fault identification result of the battery cell characterized by the battery cell voltage data, including: Generate two-dimensional feature data of each battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; Perform clustering analysis on the two-dimensional feature data of a plurality of battery cells, and determine the battery cells characterized by the two-dimensional feature data that cannot be classified into the same cluster as faulty battery cells; 6. A battery fault identification device, characterized in that, The battery fault identification device includes: A voltage decomposition unit, configured to perform variational mode decomposition on the acquired battery cell voltage data to obtain a plurality of intrinsic mode components characterizing the battery cell voltage data; wherein, optimize the two parameters of the penalty coefficient and the number of modes in the variational mode decomposition relation expression, and use the optimized parameters to perform signal decomposition on the battery cell voltage data to obtain the plurality of intrinsic mode components; the penalty coefficient and the number of modes are two parameters optimized in parallel that are selected by a particle swarm optimization algorithm and are applicable to the characteristics of the lithium-ion battery voltage signal; the plurality of intrinsic mode components are intrinsic mode components characterizing the battery cell voltage determined by iteratively solving the variational expression based on the penalty coefficient and the number of modes, when the iterative improvement value is less than a preset convergence tolerance; the iterative improvement value is the square of the absolute average value of the convergence to the intrinsic mode component in two adjacent iterative processes; the plurality of intrinsic mode components are subsequences that are relatively stable and characterize different frequency scales obtained by performing variational mode decomposition on the battery voltage signal acquired by the sensor; A mode component superposition unit, configured to superimpose the intrinsic mode components of the same type among the plurality of intrinsic mode components to obtain a target mode combination component characterizing the fault information; the target mode combination component is the mode combination component with the strongest correlation with the fault information among the superimposed mode combination components; A feature extraction unit, configured to extract a target differential waveform factor and a target differential skewness factor of the battery cell voltage data based on the target modal combination component; the fault feature of the target differential waveform factor is the ratio of the effective value of the cell voltage to the rectified average value; the fault feature of the target differential skewness factor is the ratio of the third-order central moment to the cube of the standard deviation; the target differential waveform factor and the target differential skewness factor are eigenvalue sensitive to the faults of lithium-ion battery cells. A fault identification unit, configured to determine a fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor. Determining a fault identification result of the battery cell represented by the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor includes: Generating two-dimensional feature data of the battery cell voltage data based on the target differential waveform factor and the target differential skewness factor; Calculating the average spatial distance between the two-dimensional feature data and the preset two-dimensional feature data corresponding to different fault types; Determining the fault type corresponding to the minimum value of the average spatial distance as the fault identification result of the battery cell represented by the battery cell voltage data.

7. An electronic device, characterized in that, The electronic device includes a processor and a memory, and the memory is used to store a computer program, and when the computer program is executed by the processor, the method according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images