Battery consistency early warning method, system, device and medium based on modal decomposition

Abstract

The present application belongs to the technical field of power battery fault monitoring, and particularly relates to a battery consistency early warning method, system, device and medium based on modal decomposition, which first acquires battery pack charging and discharging time series data and constructs a battery cell voltage matrix; adopts a frequency-time domain coupled modal decomposition method to extract multi-modal features of each battery cell through discrete Fourier transform and a variational modal decomposition algorithm of iterative optimization; then performs abnormality detection on the row vectors of the modal matrix using a box plot method, and determines target abnormal battery cells by taking the intersection of the detection results of each modal component; and finally issues a warning according to the voltage and temperature values of the abnormal battery cells. The present application can effectively identify early consistency degradation of the battery, significantly reduce the false positive rate and the missed detection rate, and is suitable for new energy vehicle battery safety monitoring and fault early warning.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power battery fault monitoring, and particularly relates to a battery consistency early warning method, system, device and medium based on modal decomposition. BACKGROUND

[0002] As a core component of new energy vehicles, the health status of power batteries directly affects the safety and reliability of vehicles, and great differences may occur between different battery groups due to factors such as actual temperature, aging rate, internal resistance, etc. At present, the monitoring of battery group consistency mainly adopts statistical methods or single-dimensional signal analysis techniques based on threshold values. For example, the invention patent with publication number CN115856692A discloses a power battery risk identification and tracing system based on range voltage, which further identifies abnormal battery cells through the range voltage of power batteries. The invention patent with publication number CN119479219A discloses a new energy booster abnormality early warning system and method, which selects feature vectors by using a wavelet transform algorithm and determines the abnormal position by establishing a booster abnormality positioning model using an improved artificial neural network. In addition, some researches use clustering algorithms or machine learning models to analyze battery parameters, such as dividing battery cell states by K-means clustering or predicting battery performance degradation trends using LSTM networks. Although these methods can achieve a certain degree of anomaly detection, they still face problems such as insufficient adaptability and single feature extraction in actual application.

[0003] The above methods have significant deficiencies in actual application. First, the traditional threshold method has a high false alarm rate for complex working conditions (such as fast charging and low temperature), and cannot distinguish between normal fluctuations and real faults. Second, single modal analysis (such as only frequency domain decomposition) loses time domain dynamic characteristics, leading to missed detection of early minor abnormalities. For example, although wavelet decomposition can extract high-frequency noise, it ignores the interaction of different frequency band modes. The clustering algorithm is sensitive to data distribution and is prone to failure in the later stage of battery group aging. These problems directly affect the safety and maintenance efficiency of new energy vehicles, and there is an urgent need for a consistency early warning method that can integrate multi-dimensional features and adapt to working conditions. SUMMARY

[0004] The purpose of the present application is to provide a battery consistency early warning method, system, device and medium based on modal decomposition, which aims to identify inconsistent battery cells by analyzing the voltage data of each battery cell in the power battery group, thereby early warning potential faults. The core idea is to extract the frequency domain and time domain coupling characteristics of the battery cell voltage by modal decomposition, and to detect abnormal battery cells by combining the box plot method.

[0005] The application achieves the above-mentioned purposes through the following technical solutions:

[0006] The first aspect, the application proposes a battery consistency early warning method based on modal decomposition, the method comprises:

[0007] Obtain the charge-discharge time sequence segment of the battery pack, including the voltage array and the temperature array;

[0008] Obtain the cell voltage matrix constructed by multiple charge-discharge time sequence segments;

[0009] Perform modal decomposition on each cell time sequence segment in the cell voltage matrix in the frequency-time domain coupling, extract the modal features of each cell, and obtain the modal matrix;

[0010] Wherein, the modal decomposition separates the modal components of different frequency bands through iterative optimization, and extracts the maximum modal feature to represent the cell state;

[0011] The row vectors of the modal matrix are identified by the box plot method to obtain the target abnormal cell number, and the intersection of the identification results of each component is obtained to obtain the target abnormal cell;

[0012] Determine the voltage value and temperature value of the target abnormal cell, and issue a warning signal.

[0013] Further, the charge-discharge time sequence segment of the battery pack is obtained, comprising:

[0014] Parse the battery message data issued by the supervision platform;

[0015] After sorting and deduplicating the battery message data according to the collection time, the voltage array and the temperature array of the cell are obtained;

[0016] Based on the charge, discharge and static state of the battery pack, the array is divided to obtain the charge-discharge time sequence segment of the battery pack.

[0017] Further, the cell voltage matrix is constructed based on the following method:

[0018] For a single charge-discharge time sequence segment, construct the cell voltage matrix :

[0019] ;

[0020] Wherein, represents the number of cells, represents the time sequence data of the th cell.

[0021] Further, the modal decomposition comprises the following steps:

[0022] Set the number of decomposition modes , the maximum number of iterations , and the tolerance error ;

[0023] Based on cell voltage matrix Time series data Initialize the modal component matrix ,in The timing length;

[0024] Using Discrete Fourier Transform to transform time series data Transform from the time domain to the frequency domain to obtain the transformed time-series data. ;

[0025] The modal component matrix is ​​updated iteratively. The modal components are gradually optimized until the termination condition is met.

[0026] From the iterated modal component matrix Extracting the maximum modal feature yields the modal feature vector of the battery cell. , , Indicates the first The first energy-saving cell One modal component;

[0027] Based on the modal feature vector Constructing the mode matrix , Indicates the first The maximum mode vector of the energy-saving cell.

[0028] Furthermore, the iterative update of the modal component matrix As shown in the following formula:

[0029] ;

[0030] in, Indicates the first One modal component, Represents the original frequency domain signal. Indicates the first One modal component, This indicates the last update number. One modal component.

[0031] Furthermore, the termination condition for the mode decomposition is:

[0032] The number of iterations exceeds the maximum number of iterations. ;

[0033] Or meet the error condition , Indicates the first Modal component results of the next iteration Indicates the first Modal component results of the sub-iteration.

[0034] Further, the box plot method is used to identify inconsistent battery cell numbers for row vectors of the modal matrix, and an intersection of identification results of each component is obtained to obtain a target abnormal battery cell, including:

[0035] For each row vector of the modal matrix , upper and lower quartiles and interquartile range are calculated.

[0036] If data points satisfy , the battery cell is marked as an abnormal battery cell, and an abnormal battery cell set is obtained.

[0037] A logical intersection of abnormal battery cell sets of modal components is taken to obtain an intersection , which is used as a target abnormal battery cell, as follows:

[0038] .

[0039] In a second aspect, the application provides a battery consistency early warning system based on modal decomposition, which is applied to the steps of the battery consistency early warning method described above, and the system includes:

[0040] A first acquisition module is configured to acquire a charge-discharge time sequence segment of a battery pack, including a voltage array and a temperature array.

[0041] A second acquisition module is configured to acquire a battery cell voltage matrix constructed by a plurality of charge-discharge time sequence segments.

[0042] A modal decomposition module is configured to perform modal decomposition of each battery cell time sequence segment in the battery cell voltage matrix in the frequency-time domain coupling mode, extract modal features of each battery cell, and obtain a modal matrix.

[0043] The modal decomposition separates modal components of different frequency bands through iterative optimization, and extracts maximum modal features to represent the state of the battery cell.

[0044] An abnormality identification module is configured to use the box plot method to identify inconsistent battery cell numbers for row vectors of the modal matrix, and take an intersection of identification results of each component to obtain a target abnormal battery cell.

[0045] An early warning module is configured to determine voltage and temperature values of the target abnormal battery cell, and issue an early warning signal.

[0046] In a third aspect, the present application provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the battery consistency early warning method.

[0047] In a fourth aspect, the present application provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the steps of the battery consistency early warning method.

[0048] The present application has the following beneficial effects:

[0049] The present application proposes a frequency-time domain coupling analysis method, realizes deep feature extraction of the battery pack consistency state, and can effectively capture early micro abnormalities that are difficult to find by traditional time domain analysis methods. The present application organically combines discrete Fourier transform and an improved variational modal decomposition algorithm, and significantly improves the sensitivity and accuracy of fault detection. Secondly, the present application adopts an abnormality recognition strategy combining multi-modal joint analysis and statistical testing, and takes the intersection after independent detection of each modal component by the box plot method, thereby avoiding misjudgment caused by a single detection standard, and making the early warning result more reliable. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 FIG. 1 is a flowchart of a battery consistency early warning method based on modal decomposition in an embodiment of the present application;

[0051] Figure 2 FIG. 2 is another flowchart of a battery consistency early warning method based on modal decomposition in an embodiment of the present application;

[0052] Figure 3 FIG. 3 is a flowchart of modal decomposition in an embodiment of the present application;

[0053] Figure 4 FIG. 4 is a structural block diagram of a battery consistency early warning system based on modal decomposition in an embodiment of the present application;

[0054] Figure 5 FIG. 5 is a structural diagram of an electronic device in an embodiment of the present application;

[0055] Figure 6 FIG. 6 is a cell voltage curve diagram of a cell with poor cell consistency in the case part of the specific embodiment of the present application;

[0056] Figure 7 FIG. 7 is a cell modal thermal diagram in the case part of the specific embodiment of the present application. DETAILED DESCRIPTION

[0057] The application will be described in further detail below with reference to the drawings. It is necessary to point out here that the following specific embodiments are only used to further illustrate the application and cannot be understood as limiting the scope of protection of the application. Those skilled in the art can make some non-essential improvements and adjustments to the application according to the above application content.

[0058] Please refer to Figures 1-3 In one embodiment, a battery consistency early warning method based on modal decomposition is proposed, and the method comprises the following steps:

[0059] S1, obtaining a charge-discharge time sequence segment of a battery pack, including a voltage array and a temperature array.

[0060] In the above step S1, specifically including: parsing the battery message data issued by the supervision platform (or uploaded by the vehicle terminal), such as parsing the message according to the data format of GBT32960 protocol and real-time parsing, to obtain the original battery data; after sorting and deduplicating the battery message data according to the collection time, the voltage array and the temperature array of the battery cell are obtained; the deduplication includes cleaning duplicate data and empty data; based on the charging, discharging and standing state of the battery pack, the array is divided to obtain the charge-discharge time sequence segment of the battery pack.

[0061] In the above array division, according to the charging state code and the current direction characteristics, the continuous time sequence data is divided into charge-discharge segments with clear physical meaning. Specifically, when the charging pile connection signal is detected and the current is positive, it is marked as a charging segment; when the vehicle is in driving state and the current is negative, it is marked as a discharging segment; when the absolute value of the current is less than a threshold and lasts for a set time length, it is determined as a standing state.

[0062] S2, obtaining a battery cell voltage matrix constructed by a plurality of charge-discharge time sequence segments; the battery cell voltage matrix is constructed based on the following manner: for a single charge-discharge time sequence segment, constructing a battery cell voltage matrix :

[0063] ;

[0064] Wherein, represents the number of battery cells, represents the time sequence data of the i-th battery cell.

[0065] In this embodiment, after obtaining the battery cell voltage matrix, the subsequent key modal decomposition step is executed, and this process realizes multi-dimensional feature extraction of the battery health state through a frequency-time domain coupling analysis method.

[0066] ​S3. Mode decomposition of each cell time series segment in the cell voltage matrix is performed by frequency-time domain coupling to extract the modal characteristics of each cell, and a modal matrix is obtained; wherein the mode decomposition separates the modal components of different frequency bands by iterative optimization, and extracts the maximum modal characteristics to represent the state of the cell.

[0067] More specifically, please refer to Figure 3 , the mode decomposition includes the following steps:

[0068] S3.1, set the number of decomposition modes , the maximum number of iterations , the tolerance error .

[0069] It can be understood that the mode decomposition process first performs parameter initialization, wherein the number of decomposition modes is dynamically configured according to the battery type, and the typical value is 3-5 modes; the maximum number of iterations is generally set to 100-200 times to ensure convergence; the tolerance error is recommended to be 0.001-0.01 to balance accuracy and efficiency.

[0070] S3.2, initialize the modal component matrix based on the time series data in the cell voltage matrix , wherein is the time series length.

[0071] It can be understood that the modal component matrix with dimensions is created for the time series data of each cell, and is initialized to zero.

[0072] S3.3, use discrete Fourier transform to convert the time series data from time domain to frequency domain to obtain transformed time series data ; wherein the discrete Fourier transform formula is as follows:

[0073] ;

[0074] wherein represents the value of the th time sampling point, is the transformed time series data, is the value of the th time sampling point after conversion, is the imaginary unit.

[0075] Understandably, in the frequency domain conversion stage, an optimized discrete Fourier transform algorithm is used to achieve efficient computation through the fast Fourier transform (FFT) to convert the time-domain voltage signal into a frequency-domain representation containing both real and imaginary parts.

[0076] S3.4, Update the modal component matrix iteratively. The modal components are gradually optimized until the termination condition is met.

[0077] In a preferred embodiment, the termination condition for mode decomposition is: the number of iterations exceeds the maximum number of iterations. Or, the error condition is met. , Indicates the first Modal component results of the next iteration Indicates the first The modal component results of the next iteration.

[0078] In a preferred embodiment, the modal component matrix is ​​updated iteratively. As shown in the following formula:

[0079] ;

[0080] in, These are the initial values ​​for each modality, and their dimensions are the same as the number of modalities. Indicates the first One modal component, Represents the original frequency domain signal. Indicates the first One modal component, This indicates the last update number. One modal component.

[0081] Understandably, the iterative update process in step S3.4 employs an improved variational mode decomposition algorithm to achieve mode separation. Specifically, firstly, a Hilbert transform is applied to the current mode component U[i,:] to extract its instantaneous frequency characteristics as the center frequency of the bandpass filter. Next, a Gaussian window filter centered at this frequency is applied to the original signal û in the frequency domain to obtain the updated frequency domain representation. Then, the time-domain signal is reconstructed through an inverse Fourier transform, completing a single mode update. During the iteration process, the system monitors the relative change in the Euclidean distance between mode components in two adjacent iterations in real time. Optimization terminates when this value is less than the preset tolerance error tol or reaches the maximum number of iterations iter. The resulting mode components accurately reflect the battery state characteristics of different frequency bands.

[0082] S3.5, From the iterated modal component matrix Extracting the maximum modal feature yields the modal feature vector of the battery cell. , representing the first modal component of the first battery cell.

[0083] It can be understood that, in the modal feature extraction stage, in specific implementation, the envelope line extreme value of each component is extracted as a feature point through time domain analysis on the converged modal components.

[0084] In particular, the dominant modal component with the largest energy proportion is selected for each battery cell, and the amplitude maximum value thereof is recorded as a feature vector .

[0085] The feature vectors of all battery cells are arranged in rows to form a modal matrix The row vectors of the matrix reflect the global state features of the battery cells, and the column vectors show the distribution of different modal components in the entire battery pack.

[0086] S3.6, based on the modal feature vectors a modal matrix is constructed :

[0087] ;

[0088] wherein, is the number of decomposed modal components, is the number of battery cells, is the maximum modal vector of the first battery cell, is the real part maximum value of the first modal component of the first battery cell. .

[0089] The above embodiment can effectively extract the feature performance of each battery cell in different frequency domains through modal decomposition. For the time series voltage data of each battery cell, after discrete Fourier transform, k modal components are separated by using an iterative optimization method. In each iteration process, the relative error between the current modal component and the last iteration result is calculated in real time, and the optimization is terminated when the error is less than a preset tolerance value tol or the maximum iteration number iter is reached. This dynamic adjustment mechanism ensures the convergence and stability of the decomposition process.

[0090] S4, the row vectors of the modal matrix are identified by the box plot method to identify inconsistent battery cell numbers, and the intersection of the identification results of each component is obtained to obtain the target abnormal battery cell.

[0091] In a preferred embodiment, the row vectors of the modal matrix are identified by the box plot method to identify inconsistent battery cell numbers, and the intersection of the identification results of each component is obtained to obtain the target abnormal battery cell, comprising: for each row vector of the modal matrix , calculating the upper and lower quartiles and interquartile range If data points satisfy or If the cell is marked as abnormal, a set of abnormal cells is obtained. ;right The logical intersection of the set of abnormal cells with each modal component is taken to obtain intersection A, which is then used as the target abnormal cell, as shown in the following formula:

[0092] .

[0093] In the specific implementation of step S4 above, statistical analysis is performed on each row of the modal matrix M (i.e., all cell characteristic values ​​under each modal component). First, the upper quartile Q3, lower quartile Q1, and interquartile range (IQR) of all cell characteristic values ​​under that modal component are calculated. By setting a threshold range of 1.5 times IQR, cells exhibiting abnormal behavior under that modal component can be effectively identified. Considering the potential risk of misjudgment under a single modal component, this scheme further requires that the target abnormal cell must be judged as abnormal in the detection of all k modal components, i.e., the logical intersection of the abnormal sets of each modality is taken, effectively reducing the false alarm rate. In practical applications, this method can accurately identify cell consistency degradation caused by aging, uneven temperature, or internal short circuits.

[0094] S5. Determine the voltage and temperature values ​​of the target abnormal cell and issue an early warning signal; in practice, the early warning signal may include a complete report containing the abnormality level, possible causes, and maintenance recommendations.

[0095] Understandably, the generation of the warning signal is based on a comprehensive diagnosis of the abnormal battery cell. This method not only records the abnormal battery cell's number, but also analyzes the historical trends of its voltage and temperature.

[0096] In one specific embodiment, this invention proposes a battery consistency early warning system based on modal decomposition, applied to the steps of executing the battery consistency early warning method as proposed in any of the above embodiments. In specific applications, the battery consistency early warning system of this invention can be implemented as an intelligent monitoring subsystem in a new energy vehicle battery management system (BMS), or as a core analysis module of a cloud-based battery health management platform. This system integrates onboard data acquisition terminals and cloud computing resources to construct a complete battery state monitoring solution. In practical applications, the system can first collect the voltage and temperature data of each cell in the battery pack in real time via the onboard CAN bus, and then transmit the data to the processing unit after standardized encapsulation according to the GBT32960 protocol. For vehicles with edge computing capabilities, the system can directly complete modal decomposition and anomaly detection in the onboard computing unit; for vehicles with limited computing power, the raw data can be uploaded to a cloud server for centralized processing.

[0097] Referring to Figure 4 , the system specifically comprises:

[0098] The first acquisition module is configured to acquire a charge-discharge time sequence segment of the battery pack, including a voltage array and a temperature array.

[0099] The second acquisition module is configured to acquire a cell voltage matrix constructed by a plurality of charge-discharge time sequence segments.

[0100] The modal decomposition module is configured to perform modal decomposition of each cell time sequence segment in the cell voltage matrix in the frequency-time domain coupling manner, extract modal features of each cell, and obtain a modal matrix.

[0101] The modal decomposition separates modal components of different frequency bands through iterative optimization, and extracts maximum modal features to represent the state of the cell.

[0102] The anomaly identification module is configured to identify inconsistent cell numbers by using the box plot method on the row vectors of the modal matrix, and obtain a target abnormal cell by taking the intersection of the identification results of each component.

[0103] The early warning module is configured to determine the voltage value and the temperature value of the target abnormal cell, and send an early warning signal.

[0104] In specific application, in the daily operation process of the vehicle, the system can monitor the consistency state of the battery pack in real time, and when an abnormal cell is detected, the system can send a hierarchical early warning (such as a prompt warning, a power limitation, or a suggestion for immediate repair) to the driver through the vehicle-mounted display screen. In the charging scenario, the system can be linked with the charging pile to take protective measures such as adjusting the charging strategy for the abnormal battery.

[0105] Referring to Figure 5 , a computer device 21 is proposed, which can be a terminal or a server, and the computer device 21 comprises a processor 23, a memory 24 and a network interface 25 connected through a system bus 22. The processor 23 of the computer device 21 is configured to provide computing and control capabilities. The memory of the computer device 24 comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface 26 of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement the steps of the battery consistency early warning method in any of the above embodiments.

[0106] In specific application, the electronic device of the present application can be specifically implemented as a vehicle-mounted computing unit (VCU), a main control module of a battery management system (BMS), or a dedicated battery health monitoring terminal device of a new energy vehicle.

[0107] In one specific embodiment, the present application provides a computer readable storage medium storing a computer program, which when executed by a processor, implements the steps of the battery consistency early warning method according to any of the above embodiments.

[0108] In order to more clearly illustrate the application and its advantages, the following will further explain the battery consistency early warning method provided by the present application in combination with specific experimental cases and related drawings.

[0109] Figure 6 is a voltage curve of a battery cell with poor consistency, according to the GBT32960 protocol, the voltage array of each time point can be obtained, and after a period of time, the voltage curve of each battery cell can be drawn according to each battery cell, and each line in the figure represents the voltage time series data of a certain battery cell.

[0110] Figure 7 is a battery cell modal thermal map, wherein modal 1, 2 and 3 are intermediate results calculated according to the modal decomposition algorithm proposed in the above embodiments.

[0111] According to the above two experimental drawings, from Figure 7 modal 1, it can be seen that battery cell #74 has significant abnormalities, which correspond to Figure 6 significant pressure drop of battery cell #74, which can indicate that the modal decomposition algorithm proposed in the above embodiments can accurately identify the inconsistency between battery cells, and the identification effect can be verified from the battery cell voltage curve.

[0112] The above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement to part of the technical features; and these modifications or replacements do not drive the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A battery consistency early warning method based on modal decomposition, characterized in that, The method comprises: obtaining a charge-discharge time sequence segment of a battery pack, including a voltage array and a temperature array; obtaining a cell voltage matrix constructed by a plurality of charge-discharge time sequence segments; performing modal decomposition on each cell time sequence segment in the cell voltage matrix in a frequency-time domain coupling manner, extracting modal features of each cell, and obtaining a modal matrix; wherein the modal decomposition separates modal components of different frequency bands through iterative optimization, and extracts maximum modal features to represent the state of the cell; using a box plot method to identify inconsistent cell numbers for row vectors of the modal matrix, and taking an intersection of identification results of each component to obtain a target abnormal cell; determining a voltage value and a temperature value of the target abnormal cell, and issuing a warning signal; the cell voltage matrix is constructed based on the following manner: For a single charge-discharge timing segment, a cell voltage matrix is constructed : ； wherein, represents the number of cells, represents the first timing data of the section of the cell; the modal decomposition comprises the following steps: set the number of decomposition modes , maximum number of iterations , tolerance error ; Based on the timing data in the cell voltage matrix Initializing the modal component matrix wherein is the timing length;​ using a discrete fourier transform to convert the time series data from the time domain to the frequency domain to obtain transformed time series data from the time domain to the frequency domain to obtain transformed time series data ; updating the modal component matrix by iteration optimizing the modal components step by step until a termination condition is met extract the maximum modal feature from the iteration modal component matrix , to obtain the modal feature vector of the battery cell , represents the first modal component of the first battery cell based on the modal feature vector constructing a modal matrix representing the first maximum modal vector of the battery cell the iteratively updated modal component matrix as follows: ； wherein, represents the th modal component, represents the original frequency domain signal, represents the th modal component, represents the th modal component updated last time. 2.The modal decomposition based battery consistency early warning method of claim 1, wherein, the method comprises: parsing battery message data issued by a supervision platform; after sorting and deduplicating the battery message data according to collection time, obtaining a voltage array and a temperature array of the cell; dividing the arrays based on charge, discharge and static states of the battery pack to obtain the charge-discharge time sequence segment of the battery pack. 3.The modal decomposition based battery consistency early warning method of claim 1, wherein, the termination condition of the modal decomposition is: the number of iterations exceeds the maximum number of iterations ; or the error condition is satisfied , denotes the modal component result of the i-th iteration, denotes the modal component result of the i-th iteration, denotes the modal component result of the i-1-th iteration. 4.The modal decomposition based battery consistency early warning method of claim 1, wherein, the target abnormal cell is obtained by taking a logical intersection of abnormal cell sets of k modal components, and taking an intersection A as the target abnormal cell, as follows: For each row vector of the modal matrix the upper and lower quartiles and the interquartile range are calculated; If the data point satisfies or , the battery cell is marked as an abnormal battery cell, and an abnormal battery cell set is obtained ; the system comprises: 。 5. A battery consistency warning system based on modal decomposition, applied to perform the steps of the battery consistency warning method according to any one of claims 1-4, characterized in that, a first obtaining module for obtaining a charge-discharge time sequence segment of a battery pack, including a voltage array and a temperature array; a second obtaining module for obtaining a cell voltage matrix constructed by a plurality of charge-discharge time sequence segments; a modal decomposition module for performing modal decomposition on each cell time sequence segment in the cell voltage matrix in a frequency-time domain coupling manner, extracting modal features of each cell, and obtaining a modal matrix; wherein the modal decomposition separates modal components of different frequency bands through iterative optimization, and extracts maximum modal features to represent the state of the cell; an abnormality identification module for using a box plot method to identify inconsistent cell numbers for row vectors of the modal matrix, and taking an intersection of identification results of each component to obtain a target abnormal cell; a warning module for determining a voltage value and a temperature value of the target abnormal cell, and issuing a warning signal. a memory and a processor, the memory stores a computer program, and the processor implements the steps of the battery consistency warning method of any one of claims 1-4 when executing the computer program.

6. An electronic device, comprising: a memory stores a computer program, and the computer program is executed by a processor to implement the steps of the battery consistency warning method of any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, ​

Images