Energy storage lithium ion battery capacity anomaly identification method, system and device

Abstract

The application discloses a kind of energy storage lithium ion battery capacity anomaly identification method, system and equipment, and it is related to lithium ion battery technical field.In order to solve the defects that the existing battery capacity diagnosis method needs to rely on complete charge-discharge data, complex battery model or large-scale historical operation data support, it is difficult to meet the capacity abnormality discrimination demand under the condition of fast, disturbance, data sparse in actual operation, obtains the static voltage curve after the charge-discharge of energy storage lithium ion battery stops, constructs static voltage time series;Select a group of wavelet functions that meet the multi-scale orthogonality, carry out wavelet decomposition to static voltage sequence, extract low-frequency approximation component;Feature extraction is carried out to the obtained low-frequency approximation component, and the standardized three-dimensional features of each battery monomer are obtained;Capacity abnormal battery is clustered using clustering algorithm, to obtain normal battery class and abnormal battery class.The application is mainly used for identifying the abnormality of the capacity of energy storage lithium ion battery.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lithium ion batteries, in particular to a capacity anomaly identification method, system and device for energy storage lithium ion batteries. BACKGROUND

[0002] Lithium ion batteries are widely used in energy storage systems due to their high energy density, long cycle life, high output voltage, no memory effect, and environmental friendliness. At the same time, driven by the increasing demand for clean energy and policy support, the application range of battery energy storage systems is continuously expanding, from new energy vehicles to large-scale energy storage projects, and its role in energy transformation is becoming increasingly important. The energy storage battery system is composed of thousands of lithium ion battery monomers in series and parallel connection. Due to manufacturing differences, use environment and different aging rates, each battery cell inevitably produces capacity deviation during use. When the capacity is inconsistent, the low-capacity battery will be filled first or emptied first, triggering voltage protection limits, causing the entire battery to end the charging and discharging process prematurely; this not only limits the overall energy utilization of the system, but also accelerates the aging of low-capacity batteries due to overcharging and overdischarging, further widening the gap, forming a vicious cycle.

[0003] A lithium ion battery pack short circuit fault diagnosis method based on relaxation time and adaptive clustering algorithm is disclosed in Chinese patent CN 120214621 A. The method constructs a relaxation time curve through an equivalent first-order circuit model and extracts statistical features as fault criteria. It combines adaptive DBSCAN algorithm for clustering analysis of single battery, uses critical neighborhood radius to quantify the degree of outlying, and then calculates fault score to realize the identification and positioning of battery internal short circuit fault. This method relies on charging segments that meet certain charging time and power conditions, and requires total current, total voltage, SOC, internal resistance, and single battery voltage as a prerequisite for extracting relaxation segments. The data selection is complex and limited by operating conditions, limiting the applicability of the algorithm to incomplete data conditions in actual scenarios. This method has certain advantages in fully utilizing the static relaxation behavior to reflect internal abnormalities, but its feature extraction dimension is limited. The time constant of the first-order equivalent circuit model is the analysis object, and it mainly relies on a small number of statistical features, which is difficult to describe the complex fault evolution process. At the same time, the algorithm relies on the setting of diagnostic threshold parameters, and the robustness and generalization ability are poor under different types of faults, and it cannot effectively identify capacity degradation and other non-short-circuit abnormalities.

[0004] A capacity anomaly diagnosis method based on wavelet transform and wavelet reconstruction is proposed in Chinese invention patent CN 113253142 B. The method obtains a group of reconstructed signal curves by wavelet transform and wavelet reconstruction of the electrical signals during the operation of the lithium battery pack, and judges the consistency problem of the lithium battery pack according to the curve group. The method does not establish an effective signal screening mechanism, and processes the electrical signals generated during the entire operation, resulting in that the diagnosis target only stays at the general consistency evaluation level, and cannot further identify specific abnormal types such as capacity attenuation, so that the diagnosis result is difficult to guide the battery maintenance. In addition, in terms of technical implementation, the method relies on manual observation of the distribution of the reconstructed curve group, and does not give how to quantify or distinguish the outlier curve, which may lead to strong subjectivity in the judgment process and difficulty in realizing standardized and automated diagnosis.

[0005] There are some capacity anomaly diagnosis methods in the prior art. Traditional capacity anomaly diagnosis methods mainly include Coulomb counting method and model method, such as equivalent circuit model (ECM). The first-order RC model is simple but has limited accuracy and cannot accurately describe the dynamic response. Although the multi-order RC has high accuracy, the number of parameters is large and the identification is unstable; machine learning methods based on full life cycle data (such as SVR support vector regression, RF random forest, and LSTM long short-term memory network) require a large number of real labels for model training, and the data labels in real running scenarios are often difficult to obtain, and the data of offline testing is difficult to reproduce the real working conditions, resulting in a decline in generalization performance; feature analysis methods based on impedance spectrum (EIS) or pulse response require special equipment for high-precision EIS testing, which is expensive and difficult to integrate into a lightweight embedded system. In addition, equivalent circuit fitting (such as ECM parameter extraction) or feature mapping (such as Nyquist figure analysis and spectrum analysis) are required. Pulse response testing requires specific temperature and sleep time, and is difficult to perform continuously in real operation, which is time-consuming and fixed in rhythm, making it difficult to deploy in practical applications. The above methods generally have the following problems: they rely on complete charge and discharge data, complex battery models, or large-scale historical operation data support, making it difficult to meet the needs of capacity anomaly identification under the conditions of rapid, disturbance, and sparse data in actual operation.

[0006] Therefore, there is a need for a lithium-ion battery capacity anomaly identification method, system and device for lithium-ion battery capacity anomaly identification that can only rely on the static voltage curve after charge and discharge to accurately identify battery capacity anomalies. SUMMARY

[0007] This invention addresses the shortcomings of existing battery capacity diagnosis methods, which rely on complete charge-discharge data, complex battery models, or large-scale historical operating data, making it difficult to meet the needs of capacity anomaly identification under rapid, disturbed, and data-sparse conditions in actual operation. It provides a method, system, and device for identifying capacity anomalies in energy storage lithium-ion batteries that can accurately identify battery capacity anomalies based solely on the static voltage curve after charging and discharging has stopped.

[0008] The present invention provides a method for identifying abnormal capacity in energy storage lithium-ion batteries, comprising the following steps:

[0009] S1. Obtain the static voltage curve of the energy storage lithium-ion battery after charging and discharging has stopped, and construct the static voltage time series.

[0010] S2. Select a set of wavelet functions ψ(t) that satisfy multi-scale orthogonality, and then apply this to the resting voltage sequence. Perform wavelet decomposition to extract low-frequency approximate components;

[0011] S3. Extract features from the obtained low-frequency approximation components to obtain the standardized three-dimensional features of each battery cell.

[0012] S4. Use a clustering algorithm to cluster the three-dimensional features of all batteries to obtain normal battery categories and abnormal battery categories.

[0013] Further: In S4, the density clustering DBSCAN algorithm is used. The standardized three-dimensional features of each battery cell are used as input to the density clustering DBSCAN algorithm to obtain the clustering results. It is determined whether there are three-dimensional features in the clustering results that cannot be assigned to any cluster. If so, the battery cell to which the three-dimensional feature belongs is a cell with abnormal capacity.

[0014] Further: In S2, the static voltage sequence The formula is:

[0015] ;

[0016] In the formula, Let be the voltage value at time t, and J be the preset maximum number of decomposition layers. The minimum number of decomposition layers is given by k, where k is the translation parameter. Let be the wavelet coefficients of the j-th layer, representing the high-frequency detail components of that layer; Let be the wavelet function of the j-th layer, used to extract high-frequency details of that layer; For the first The scale factor of a layer represents the low-frequency approximate component of that layer; For the first The scaling function at the layer scale is used to extract the low-frequency approximation components of that layer.

[0017] Further: in S2, the wavelet decomposition process is represented as:

[0018] ;

[0019] In the formula, is the low-frequency approximation component of the jth layer; is the high-frequency detail component sequence of the jth layer; is the low-frequency approximation component of the j-1th layer, is the wavelet low-pass filter coefficient, is the wavelet high-pass filter coefficient; n is the sampling point, and b is the sampling point of the filter coefficient.

[0020] Further: in S3, the feature extraction calculation method is as follows:

[0021] ;

[0022] ;

[0023] ;

[0024] In the formula, E is the energy feature, is the low-frequency approximation component of the Jth layer obtained after wavelet decomposition, H is the entropy feature, is the standard deviation feature, and N is the number of samples, is the mean value of the low-frequency approximation component of the Jth layer, n=1, 2,..., N.

[0025] Further: in S4, the clustering is based on "density accessibility" to divide the samples into three categories:

[0026] Core point: at least contains M points in its ε neighborhood;

[0027] Boundary point: less than M in the neighborhood, but in the neighborhood of a certain core point;

[0028] Noise point: neither a core point nor a boundary point, and is regarded as an abnormal point.

[0029] The recognition system for implementing the capacity abnormality recognition method of the energy storage lithium ion battery provided by the application comprises a battery static data acquisition module, a wavelet decomposition module, a feature extraction module and a clustering diagnosis module.

[0030] The battery static data acquisition module is used to acquire the static voltage curve of the energy storage lithium ion battery after the charge and discharge stop, and construct a static voltage time sequence.

[0031] The wavelet decomposition module is used for multi-scale wavelet decomposition of the static voltage time sequence to extract a low-frequency approximate component thereof;

[0032] The feature extraction module is used for obtaining the normalized three-dimensional features of each battery monomer;

[0033] The clustering diagnosis module adopts a clustering algorithm to perform abnormal diagnosis on the extracted three-dimensional features to the battery capacity state.

[0034] The electronic device comprises a memory, a processor, a computer program stored on the memory and capable of running on the processor, and the processor is used for executing the computer program to realize the energy storage lithium ion battery capacity abnormality identification method.

[0035] The energy storage lithium ion battery capacity abnormality identification method has the following beneficial effects:

[0036] The energy storage lithium ion battery capacity abnormality identification method provided by the application can accurately identify the battery capacity abnormality without complete and continuous charge-discharge curves.

[0037] The energy storage lithium ion battery capacity abnormality identification method uses wavelet transform to extract energy, entropy and standard deviation of low-frequency approximate components in different scales of the static voltage curve, fully describes the capacity state change information contained in the static behavior of the battery cell, and quickly identifies the battery with obvious capacity degradation, mutation or deviation from the average level of the group. Compared with the traditional representation method using charge-discharge data, the static voltage signal is easy to obtain and is not easily affected by working condition disturbance, and has higher engineering practicability. At the same time, the application does not require preconditions such as total current, total voltage, SOC, internal resistance and monomer voltage data, avoids the introduction of front-stage errors, simplifies the diagnosis process, enhances the adaptability to non-standard operation data, and is suitable for capacity abnormality screening tasks under complex working conditions. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 It is a practical application flowchart of the energy storage lithium ion battery capacity abnormality identification method;

[0039] Figure 2 It is a voltage curve schematic diagram of an energy storage battery module in operation; wherein the horizontal coordinate is time and the vertical coordinate is voltage value;

[0040] Figure 3 It is a process schematic diagram of extracting low-frequency components by discrete wavelet transform;

[0041] Figure 4 It is a wavelet transform result schematic diagram of the static voltage curve of the battery monomer in the battery pack;

[0042] Figure 5 is a schematic diagram of an abnormal capacity identification result of an energy storage lithium ion battery. DETAILED DESCRIPTION

[0043] The following merely illustrates the preferred embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. The following examples are used to explain the present application, but should not be interpreted as limiting the present application, and the protection scope of the present application should be subject to the protection scope of the claims. The embodiments of the present application are described in detail below, in order to facilitate the description of the present application and simplify the description, the technical terms used in the specification of the present application should be interpreted broadly, including but not limited to the conventional replacement schemes not mentioned in the present application, and including direct implementation and indirect implementation.

[0044] Embodiment 1

[0045] In combination Figures 1-5 In this embodiment, the present application discloses an abnormal capacity identification method of an energy storage lithium ion battery, comprising the following steps:

[0046] S1, obtaining the static voltage curve of the energy storage lithium ion battery after the charge and discharge is stopped, and constructing a static voltage time sequence.

[0047] Figure 2 As shown in the voltage curve schematic diagram of the energy storage battery module operation, it includes the charge and discharge operation process and the static process, in the complex charge and discharge working condition, the voltage characteristics of the low capacity battery are difficult to be directly distinguished from the normal battery, and the charge and discharge working condition of the battery is complex and changeable, there are disturbances and sparse data, which is difficult to obtain completely. This embodiment starts from the relatively easy to obtain static period, first obtains the static voltage curve of the energy storage lithium ion battery after the charge and discharge is stopped when there is no dispatching instruction, and constructs a static voltage sequence , w is the length of the voltage sequence.

[0048] Discrete wavelet transform is a decomposition technology that decomposes time domain signals into multiple frequency bands, which has both time domain and frequency domain localization analysis capabilities, and is very suitable for processing non-stationary and local mutation signals. Wavelet transform can obtain good time resolution in high frequency band and good frequency resolution in low frequency band, and is particularly suitable for signals such as voltage which slowly changes but contains local fluctuations.

[0049] S2, decomposing the static voltage sequence by using discrete wavelet transform.

[0050] Discrete wavelet transform (DWT) is a decomposition technique that breaks down a time-domain signal into multiple frequency bands. It possesses both time-domain and frequency-domain localization analysis capabilities, making it highly suitable for processing non-stationary signals with local abrupt changes. Wavelet transform achieves good time resolution in the high-frequency band and good frequency resolution in the low-frequency band, thus making it particularly applicable to slowly changing signals with local fluctuations, such as stationary voltage.

[0051] Specifically, a set of wavelet functions ψ(t) that satisfy multi-scale orthogonality are selected, and the static voltage sequence is... Perform layer-by-layer decomposition.

[0052] ;

[0053] In the formula, Let be the voltage value at time t, and J be the preset maximum number of decomposition layers. The minimum number of decomposition layers is given by k, where k is the translation parameter (lateral offset, representing the position in the time domain). Let be the wavelet coefficients of the j-th layer, representing the high-frequency detail components of that layer. Let be the wavelet function of the j-th layer, used to extract high-frequency details of that layer. For the first The scale factor of a layer represents the low-frequency approximate component of that layer. For the first The scaling function at the layer scale is used to extract the low-frequency approximation components of that layer.

[0054] The multi-scale orthogonality refers to the fact that the selected wavelet function and its scaled and translated family of functions are orthogonal to each other in different frequency ranges and at different time positions. That is, the basis functions in the same layer are uncorrelated with each other, and the basis functions in different layers are also uncorrelated with each other, thus forming a complete orthogonal basis in the entire signal space.

[0055] In discrete implementations, wavelet transform decomposes and reconstructs the signal using a filter bank, namely a pair of low-pass and high-pass filters. The wavelet decomposition process is represented as:

[0056] ;

[0057] In the formula, This is the low-frequency approximation component of the j-th layer; In the high-frequency detail component sequence of the j-th layer; For the low-frequency approximation component of layer j-1, These are the coefficients of a wavelet low-pass filter. denoted as wavelet high-pass filter coefficients; n represents the sampling points, and b represents the sampling points of the filter coefficients.

[0058] The "low frequency" and "high frequency" are frequency intervals automatically divided by wavelet decomposition. Wavelet decomposition separates the signal via a low-pass filter and a high-pass filter at each layer: the low-pass part forms a low-frequency approximation component, mainly reflecting the slow trend of the static voltage signal; the high-pass part forms a high-frequency detail component, mainly reflecting the rapid fluctuations of the voltage signal. Taking a static voltage signal with a sampling frequency of f as an example, the frequency range of the approximation component of the jth layer after wavelet decomposition is , and the frequency range of the detail component is .

[0059] The Daubechies series function can ensure that the components of each layer after decomposition are independent and do not interfere with each other, so that the low-frequency approximation component can truly represent the slow trend of the voltage. In this embodiment, the wavelet function Daubechies-6 is used to perform discrete wavelet decomposition on the static voltage signal V(t), and the sampling frequency of V(t) is 1 Hz, and the maximum decomposition layer number J is taken as an example.

[0060] Figure 3 is a schematic diagram of the discrete wavelet decomposition process. The original static voltage sequence V(t) is decomposed layer by layer, and the low-frequency approximation component and the high-frequency detail component are obtained at the first layer, and then the low-frequency approximation component obtained is continuously decomposed layer by layer until the maximum decomposition layer number is reached, where A1-A6 are the low-frequency approximation components of each layer, used to depict the slow voltage change of the battery during the static process, and D1-D6 are the high-frequency detail components in each layer, mainly containing noise and short-term disturbances of each layer. The low-frequency approximation component of the first layer corresponds to a frequency band of , and the high-frequency detail component corresponds to a frequency band of , the low-frequency approximation component of the second layer decomposition corresponds to a frequency band of , and the high-frequency detail component corresponds to a frequency band of , and so on to the frequency range of the low-frequency approximation component of the sixth layer , which belongs to an extremely low frequency range, and the high-frequency component corresponds to a frequency interval of .

[0061] Figure 4 The low-frequency approximation component curves of the static voltage sequences of different battery monomers in a battery module in this embodiment are shown. The capacity of the battery monomer 8 is lower than that of other battery monomers in the same group, the sampling frequency of V(t) is 1 Hz, and the maximum decomposition layer number is 6. At the beginning of the fourth layer decomposition, the low-frequency approximation component curve of the battery monomer 8 is separated from other normal battery monomers, and at this time the frequency range corresponding to the low-frequency approximation component curve is , with the increase of the decomposition layer, the low-frequency approximation curve of the capacity abnormal monomer deviates more significantly. During the standing process, the electrochemical reaction of the battery is not completely stopped, but there are slow side reactions (such as SEI film repair, electrolyte decomposition, ion concentration gradient diffusion, etc.), and the time constant is mainly in the low frequency range (<0.1Hz). These processes are closely related to capacity attenuation, making the standing voltage show a slow change trend gradually tending to balance. Therefore, extracting the low-frequency approximation component of the standing voltage can effectively retain the long-term trend reflecting the real electrochemical evolution and capacity change of the battery, while filtering out irrelevant interference, thereby improving the stability of capacity feature extraction. It should be noted that the maximum decomposition layer in the present application is not the only limitation, and the decomposition layer needs to be determined according to the sampling frequency of the standing voltage and the calculation resources.

[0062] S3, extracting the three-dimensional feature of the Jth layer low-frequency approximation component.

[0063] Specifically, the three-dimensional feature includes energy feature, entropy feature, and standard deviation feature, and the specific calculation method is as follows:

[0064] ;

[0065] ;

[0066] ;

[0067] In the formula, E is the energy feature, is the Jth layer low-frequency approximation component obtained after wavelet decomposition, H is the entropy feature, is the standard deviation feature, N is the number of samples, is the mean value of the Jth layer low-frequency approximation component, n=1, 2,..., N.

[0068] S4, after the three-dimensional features of the standing voltage curve are extracted by wavelet transform, a clustering algorithm is used to cluster the three-dimensional features of all batteries to obtain normal battery categories and abnormal battery categories.

[0069] To avoid manual setting of the number of categories and improve the sensitivity to isolated abnormal points, the DBSCAN algorithm of density clustering is used in this embodiment. DBSCAN algorithm is a density-based clustering algorithm, which has the ability to automatically discover clusters of arbitrary shape and identify outliers. DBSCAN algorithm does not depend on the preset number of categories, can identify irregularly shaped clusters, and can automatically mark noise points, and is particularly suitable for situations where there are a small number of abnormal values, non-spherical structures or uneven distribution density within the cluster in battery state monitoring. DBSCAN is based on "density accessibility" to divide samples into three categories:

[0070] Core point: contains at least M points in its ε neighborhood;

[0071] Border points: not dense enough to be core points, but within the neighborhood of some core point;

[0072] Noise points: not core points nor border points, considered as outliers; the clustering process starts from a core point, recursively expands the density-connected region, forms a cluster, until it can't expand any more.

[0073] Density reachability means: for a battery sample p, sample q falls in the ε neighborhood of sample p, then q is density reachable from p. Further, if there is a series of sample where each sample point is density reachable from the previous sample point is density reachable from is density reachable from DBSCAN is to recursively expand the region by this density reachability to cluster the samples.

[0074] Specifically:

[0075] First, the z-transform method is used to normalize the three-dimensional features respectively:

[0076] ;

[0077] wherein, is the normalized feature value.

[0078] The normalized three-dimensional features of each battery monomer are taken as the input of the density clustering DBSCAN algorithm to obtain the clustering result; it is judged whether there is a three-dimensional feature that cannot be divided into any cluster in the clustering result, if there is, the battery monomer to which the three-dimensional feature belongs is the capacity abnormal monomer.

[0079] Taking a 16-cell series battery module as an example, the capacity abnormal diagnosis execution steps are S1-S4.

[0080] First, the static voltage sequence of the 16 cells is obtained according to step S1, and the sampling frequency is 1 Hz;

[0081] According to step S2, discrete wavelet transform is performed on the 16 static voltage sequences, wherein the maximum decomposition layer is set to 6, and the 6th layer low-frequency approximation component of the 16 cells is obtained respectively; Figure 4

[0082] According to step S3, the three-dimensional features of the 6th layer low-frequency approximation components of the 16 cells are extracted respectively;

[0083] ​Finally, according to step S4, 16 groups of three-dimensional features are standardized and used as the input of the density clustering algorithm for diagnostic recognition, the neighborhood is set to 2, the minimum number of samples in the cluster is set to 10, and the battery cell that cannot be divided into any cluster is the abnormal capacity battery cell. The judgment result is as shown in Figure 5 Figure 5 is the result of the abnormal capacity diagnosis of the battery module composed of 16 battery cells, the capacity of the battery cell numbered 8 is lower than that of other battery cells in the battery module, the battery cell 8 is far away from the normal cluster and cannot be divided into any cluster, and is marked as an abnormal capacity point. The battery cell numbered 10 and the battery cell numbered 11 are located at the same position. Although the battery cells numbered 3, 4, 5 and 7 are also away from the remaining battery cells, they are normal battery cells because the distance does not exceed the threshold of the neighborhood setting.

[0084] Example 2

[0085] An abnormal capacity recognition system of energy storage lithium ion battery, comprising:

[0086] A battery static data acquisition module is used to acquire the voltage of the battery in the static process of the energy storage system and to construct a voltage time sequence.

[0087] A wavelet decomposition module is used to perform multi-scale wavelet decomposition on the static voltage time sequence and to extract the low-frequency approximate component thereof.

[0088] A feature extraction module is used to obtain the standardized three-dimensional features of each battery monomer.

[0089] A clustering diagnosis module is used to perform abnormal diagnosis on the capacity state of the battery by using a clustering algorithm on the extracted three-dimensional features.

[0090] Example 3

[0091] An electronic device comprising a memory, a processor, a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the abnormal capacity recognition method of the energy storage lithium ion battery.​

Claims

1. A method for identifying abnormal capacity of an energy storage lithium ion battery, characterized in that, The method comprises the following steps: S1, obtaining a static voltage curve of the energy storage lithium ion battery after the charge and discharge is stopped, and constructing a static voltage time sequence; S2, select a set of wavelet functions ψ(t) that satisfy the multi-scale orthogonality, and decompose the static voltage sequence Wavelet decomposition is performed to extract low-frequency approximation components; In S2, the sequence of resting voltages The formula is: ； In the formula, Vt is the voltage value at time t, J is the preset maximum decomposition layer number, is the minimum decomposition layer number, k is the translation parameter; is the wavelet coefficient of the jth layer, indicating the high-frequency detail component of the layer; is the wavelet function under the jth layer, used to extract the high-frequency details of the layer; is the scale coefficient of the jth layer, indicating the low-frequency approximate component of the layer; is the scale function under the scale of the jth layer, used to extract the low-frequency approximate component of the layer; The wavelet decomposition process is represented as: ； wherein is the low frequency approximation component of the jth layer; is the high frequency detail component sequence of the jth layer; is the low frequency approximation component of the j-1th layer, is the low pass wavelet filter coefficient, is the high pass wavelet filter coefficient; n is the sampling point, and b is the sampling point of the filter coefficient. S3, performing feature extraction on the obtained low-frequency approximation component to obtain a standardized three-dimensional feature of each battery monomer; In S3, the feature extraction is calculated as follows: ； ； ； wherein E is an energy feature, is a low frequency approximation component of the jth layer obtained after wavelet decomposition, H is an entropy feature, is a standard deviation feature, and N is a sampling number, is a mean value of the low frequency approximation component of the jth layer, and n = 1, 2,..., N. S4, using a clustering algorithm to cluster the three-dimensional features of all batteries to obtain a normal battery category and an abnormal battery category. 2.The energy storage lithium ion battery capacity abnormality identification method of claim 1, wherein, In S4, the density clustering DBSCAN algorithm is used, the standardized three-dimensional feature of each battery monomer is taken as the input of the density clustering DBSCAN algorithm, and a clustering result is obtained; it is judged whether there is a three-dimensional feature that cannot be divided into any cluster in the clustering result, if there is, the battery monomer to which the three-dimensional feature belongs is the capacity abnormal monomer. 3.The energy storage lithium ion battery capacity abnormality identification method of claim 1, wherein, In S4, the clustering is based on "density accessibility", and the samples are divided into three categories: Core point: at least contains M points in its ε neighborhood; Boundary point: insufficient M in the neighborhood, but in the neighborhood of a certain core point; Noise point: neither core point nor boundary point, regarded as an abnormal point.

4. A recognition system for implementing a capacity anomaly recognition method of an energy storage lithium ion battery as claimed in any one of claims 1 to 3, characterized in that, The method comprises a battery static data acquisition module, a wavelet decomposition module, a feature extraction module, and a clustering diagnosis module. The battery static data acquisition module is used to obtain a static voltage curve of the energy storage lithium ion battery after the charge and discharge is stopped, and to construct a static voltage time sequence. The wavelet decomposition module is used to perform multi-scale wavelet decomposition on the static voltage time sequence, and to extract a low-frequency approximation component thereof. The feature extraction module is used to obtain a standardized three-dimensional feature of each battery monomer. The clustering diagnosis module uses a clustering algorithm to perform abnormal diagnosis on the battery capacity state based on the extracted three-dimensional feature.

5. An electronic device, comprising: The computer program is stored in the memory and can run on the processor, and the processor is used to execute the computer program to realize the energy storage lithium ion battery capacity abnormality identification method of any one of claims 1-3.

Images