A method and system for diagnosing multi-dimensional inconsistency of energy storage battery modules based on low-frequency EIS

Abstract

The application discloses a kind of based on low-frequency EIS's energy storage battery module multidimensional inconsistency diagnosis method and system, the method includes: the EIS data obtained to same measurement batch is clustered, and the single battery corresponding to outlier data is abnormal battery;The EIS data of the same single battery in the module of different measurement batches is difference, and the change of EIS data in twice measurement before and after is obtained;EIS data change is clustered, and the single battery corresponding to outlier data is change abnormal battery;The EIS data of abnormal battery and change abnormal battery is analyzed with ECM fitting with the EIS data of normal single battery DRT, from multiple dimensions assess the reason of EIS data abnormal performance, and combined with mechanism analysis, diagnose the fault type of single battery.The application is helpful to realize the reliable inconsistency diagnosis of battery, to minimum intervention cost suppress inconsistency spread, prolong the overall service period of module.

Description

Technical Field

[0001] This invention belongs to the field of energy storage battery module diagnostic technology, and in particular, it is a multi-dimensional inconsistency diagnosis method and system for energy storage battery modules based on low-frequency EIS. Background Technology

[0002] Lithium-ion batteries are widely used in electric vehicles and energy storage systems. However, due to differences in materials and processes during manufacturing, as well as variations in operating conditions, inconsistencies inevitably exist between individual cells within a battery pack. In GW-level energy storage power stations, battery modules are tightly coupled in series and parallel, meaning that even minor differences in the electrochemical characteristics of any individual cell can gradually increase during use, ultimately manifesting as voltage dispersion across the entire module, increased temperature gradients, and a sharp drop in usable energy. Therefore, diagnosing and identifying the causes of inconsistencies among individual cells within a module has always been a pressing need in the industry.

[0003] Current mainstream methods generally rely on high-current charge-discharge or long-term resting to infer the health status of individual cells through capacity comparison or open-circuit voltage rebound. However, deep charge-discharge itself accelerates material fatigue, the resting process interrupts energy conversion, and the electro-thermal-aging coupling mechanism is complex, making it difficult to provide a unified explanation within the same framework. This results in diagnostic results lacking mechanistic support, and subsequent intervention strategies can only rely on experience, failing to form a closed-loop management system. Therefore, it is necessary to propose a rapid diagnostic method that covers inconsistencies in individual cells. As a complex combination of electrochemical devices, batteries can reflect internal changes very well through impedance. Numerous studies have shown that for large-scale lithium-ion batteries, impedance can be used to estimate their internal SOC, aging state, temperature, and other states. Therefore, impedance can also be used to diagnose inconsistencies in individual cells within a module. Currently, the commonly used method compares the inconsistencies of EIS data within the same test batch and the same module, but only compares the total EIS data. This may lead to the neglect of local anomalies in EIS, and it also fails to effectively identify EIS data showing abnormal trends within the normal range. Meanwhile, EIS (Electrochemical Indices) is composed of multiple complex kinetic processes coupled together. Therefore, understanding EIS data requires mechanistic analysis. Common methods include ECM fitting and DRT (Digital Thermochemical Traceability) analysis. By decoupling overlapping regions of EIS data into multiple independent electrochemical processes, the separation and quantification of electrode processes can be effectively achieved, thereby enabling mechanistic analysis of battery inconsistencies. Commonly used methods for identifying anomalous EIS data primarily employ clustering algorithms, which can be further divided into supervised and unsupervised clustering. Supervised clustering methods require extensive parameter training and exhibit poor transfer learning, limiting their applicability when dealing with diverse battery types. Summary of the Invention

[0004] The technical problem to be solved by this invention is to overcome the defects of the existing technology and provide a multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low- and medium-frequency EIS. It adopts a multi-dimensional electrochemical impedance spectroscopy diagnostic route, which has a short test time and covers impedance information covering ohmic, charge transfer and diffusion processes, corresponding to three scales: ion migration, interface reaction and solid-phase transport. At the same time, the impedance has a quantifiable response law to changes in temperature and state of charge, which helps to achieve reliable inconsistency diagnosis of batteries and can further provide interpretable and predictable decision-making basis, thereby suppressing the spread of inconsistency with minimal intervention cost and extending the overall service life of the module.

[0005] Therefore, the present invention adopts the following technical solution: a multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low-to-medium frequency EIS, comprising the following steps: 1) Measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed. 2) Cluster the EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells. 3) The difference between the EIS data of the same single cell in the modules of different measurement batches is calculated to obtain the change of EIS data between the two measurements. 4) Cluster the EIS data changes obtained in step 3) to obtain the central data and outlier data in the data changes. The individual cells corresponding to the outlier data are the individual cells with abnormal changes. 5) Perform DRT analysis and ECM fitting on the EIS data of abnormal cells and cells with abnormal changes, and compare them with the EIS data of normal cells to evaluate the reasons for the abnormal performance of EIS data from multiple dimensions. 6) Combine the causes and mechanisms of abnormal EIS data to diagnose the types of faults in individual cells.

[0006] Further, step 1) includes: Step 11), continuously and periodically measure the EIS data of each individual cell in the energy storage battery module to be diagnosed; Step 12) Perform KK verification on the EIS data and preprocess the data to remove obvious abnormal data and remeasure the EIS data of the corresponding single cell.

[0007] Further, step 2) includes: Step 21), obtain the 100-10Hz phase angle, 0.1-0.01Hz real part, and 100-0.01Hz imaginary part of the EIS data from the same measurement batch; Step 22) Perform adaptive DBSCAN-LoAD clustering on the EIS data features to obtain the center data in the dataset and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are abnormal individual cells.

[0008] Further, step 3) includes: Step 31) The EIS data of the latest measurement batch of the same single cell are compared with the EIS data of a certain historical measurement batch at the same frequency points to obtain the change of EIS data of each single cell in the module. Step 32) Perform interval mapping normalization on the EIS data changes obtained after subtraction to amplify the differences between EIS data.

[0009] Further, step 4) includes: Step 41), calculate and obtain the real and imaginary features of the EIS data changes in the low-to-medium frequency region; Step 42) Perform adaptive DBSCAN-LoAD clustering on the EIS data features obtained in Step 41) to obtain the central data that best represents the changes in EIS data of this measurement batch, and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are individual cells with abnormal changes.

[0010] Furthermore, the steps of the adaptive DBSCAN-LoAD clustering algorithm are as follows: Data preprocessing: The features of the input EIS data are standardized to eliminate the influence of different dimensions on subsequent density calculations and to construct a feature matrix suitable for density clustering. Parameter initialization: Set the basic algorithm parameters neighborhood radius ε and core point determination threshold minPts to provide initialization conditions for subsequent local density calculation and core point determination; Local density calculation: The local adaptive density (LoAD) index is used to quantify the local clustering degree of each sample point. The formula is as follows:

[0011] in: , Let i and j be the i-th and j-th sample points, respectively. for and The Euclidean distance between them; LoAD(x) is the kernel function bandwidth, used to control the range of influence of the neighborhood; i The higher the value, the greater the density of the area where that point is located; Core point determination: Based on DBSCAN density clustering rules, core points are determined by combining the LoAD density value with preset parameters; if If an ε-neighborhood contains no fewer than minPts samples, it is determined to be a core point, and all sample points within its neighborhood are assigned to the same density reachable region. Clustering formation and optimization: Initial clusters are formed by expanding the density reachability relationship with the core point as the center; low-density noise points are marked, and boundary points are assigned to a secondary location. Finally, the optimized clustering results are output to achieve accurate division of homogeneous working condition clusters.

[0012] This invention employs the adaptive DBSCAN-LoAD clustering algorithm for outlier data labeling, enabling adaptive search of shape and density parameters.

[0013] Further, step 5) includes: Step 51) Obtain EIS data of abnormal and normal single cells in one batch through Step 2), and obtain EIS data of abnormal and normal single cells in two batches through Step 4). Step 52) Perform DRT analysis and ECM fitting on the EIS data of abnormal and normal single cells, and obtain the parameters of each component in the ECM model through ECM fitting. Step 53) Determine the reasons for the discrepancies between the EIS data of normal cells and the EIS data of abnormal cells and cells with abnormal changes in the two dimensions of ECM and DRT.

[0014] Furthermore, in step 52), the analysis using DRT comprehensively considers the changes in peak area and position at different time scales, where 10 -2 The three peaks in the ~10Hz range are the focus of consideration and analysis.

[0015] Furthermore, in step 6), the ECM fitting is combined with DRT analysis and battery internal mechanism analysis to determine the mechanism cause of the discrepancy in EIS data. Based on the mechanism analysis, the abnormal causes of increased inconsistency in energy storage battery modules include abnormal temperature, abnormal SOC, abnormal aging, and abnormal internal short circuit.

[0016] This invention also provides a multi-dimensional inconsistency diagnosis system for energy storage battery modules based on low- and medium-frequency EIS, used to implement the above-mentioned multi-dimensional inconsistency diagnosis method for energy storage battery modules, comprising: EIS data measurement unit: used to measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed. First clustering unit: used to cluster EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells. EIS data change acquisition unit: used to calculate the difference between the EIS data of the same single cell in different batches of modules to obtain the change of EIS data between two consecutive measurements. The second clustering unit: clusters the EIS data change obtained by the EIS data change acquisition unit to obtain the central data and outlier data in the data change. The individual cells corresponding to the outlier data are the individual cells with abnormal changes. Anomaly Cause Assessment Unit: Used to perform DRT analysis and ECM fitting on the EIS data of abnormal cells and cells with abnormal changes, and the EIS data of normal cells, to assess the causes of abnormal EIS data performance from multiple dimensions. Fault Type Diagnosis Unit: Used to combine the causes and mechanisms of abnormal EIS data to diagnose the fault types of individual cells.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention creatively proposes to compare EIS data within the module with EIS data on a time scale, thereby detecting battery anomalies from the perspective of inconsistencies in EIS changes.

[0018] 2. The present invention adopts the adaptive DBSCAN-LoAD clustering algorithm, which has lower complexity and computational cost compared with the existing supervised algorithms, is more robust, and has excellent outlier identification effect. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the multi-dimensional inconsistency diagnosis method for energy storage battery modules of the present invention; Figure 2 This is a graph of battery EIS data obtained from the first test in a specific embodiment of the present invention; Figure 3 This is a schematic diagram of the adaptive DBSCAN-LoAD clustering algorithm in a specific embodiment of the present invention; Figure 4 This is a graph of EIS data obtained from the second test in a specific embodiment of the present invention; Figure 5 This is a graph showing the difference between the real and imaginary parts at different frequencies in two tests during a specific embodiment of the present invention. Figure 5 (a) in the diagram is the difference plot of the real part. Figure 5 (b) in the diagram is the difference graph of the imaginary part; Figure 6 This is a graph showing the DRT analysis results of EIS data in a specific embodiment of the present invention; Figure 7 This is a graph showing the ECM analysis results of EIS data in a specific embodiment of the present invention, wherein, Figure 7 (a) in the diagram is the structure diagram of the ECM model. Figure 7 (b) in the figure is the parameter identification result diagram; Figure 8 This is a diagram illustrating the structure of the multi-dimensional inconsistency diagnosis system for energy storage battery modules according to the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0022] Example 1 This embodiment provides a multi-dimensional inconsistency diagnosis method for energy storage battery modules based on mid-to-low frequency (1000-0.1Hz) EIS, such as... Figure 1 As shown, the steps are as follows: 1) Measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed. 2) Cluster the EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells. 3) The difference between the EIS data of the same single cell in the modules of different measurement batches is calculated to obtain the change of EIS data between the two measurements. 4) Cluster the EIS data changes obtained in step 3) to obtain the central data and outlier data in the data changes. The individual cells corresponding to the outlier data are the individual cells with abnormal changes. 5) Perform DRT analysis and ECM fitting on the EIS data of abnormal cells and cells with abnormal changes, and compare them with the EIS data of normal cells to evaluate the reasons for the abnormal performance of EIS data from multiple dimensions. 6) Combine the causes and mechanisms of abnormal EIS data to diagnose the types of faults in individual cells.

[0023] Specifically, step 1) includes: Step 11), continuously and periodically measure the EIS data of each individual cell in the energy storage battery module to be diagnosed; Step 12) Perform KK verification on the EIS data and preprocess the data to remove obvious abnormal data and remeasure the EIS data of the corresponding single cell. The processed data is used for subsequent clustering.

[0024] Specifically, step 2) includes: Step 21), obtain the 100-10Hz phase angle, 0.1-0.01Hz real part, and 100-0.01Hz imaginary part of the EIS data from the same measurement batch; Step 22) Perform adaptive DBSCAN-LoAD clustering on the EIS data features to obtain the center data in the dataset and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are abnormal individual cells.

[0025] Specifically, step 3) includes: Step 31) The EIS data of the latest measurement batch of the same single cell are compared with the EIS data of a certain historical measurement batch at the same frequency points to obtain the change of EIS data of each single cell in the module. Step 32) Perform interval mapping normalization on the EIS data changes obtained after subtraction to amplify the differences between EIS data.

[0026] Specifically, step 4) includes: Step 41), calculate and obtain the real and imaginary features of the EIS data changes in the low-to-medium frequency region; Step 42) Perform adaptive DBSCAN-LoAD clustering on the EIS data features obtained in Step 41) to obtain the central data that best represents the changes in EIS data of this measurement batch, and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are individual cells with abnormal changes.

[0027] Furthermore, the steps of the adaptive DBSCAN-LoAD clustering algorithm are as follows: Data preprocessing: The features of the input EIS data are standardized to eliminate the influence of different dimensions on subsequent density calculations and to construct a feature matrix suitable for density clustering. Parameter initialization: Set the basic algorithm parameters neighborhood radius ε and core point determination threshold minPts to provide initialization conditions for subsequent local density calculation and core point determination; Local density calculation: The local adaptive density (LoAD) index is used to quantify the local clustering degree of each sample point. The formula is as follows:

[0028] in: , Let i and j be the i-th and j-th sample points, respectively. for and The Euclidean distance between them; LoAD(x) is the kernel function bandwidth, used to control the range of influence of the neighborhood; i The higher the value, the greater the density of the area where that point is located; Core point determination: Based on DBSCAN density clustering rules, core points are determined by combining the LoAD density value with preset parameters; if If an ε-neighborhood contains no fewer than minPts samples, it is determined to be a core point, and all sample points within its neighborhood are assigned to the same density reachable region. Clustering formation and optimization: Initial clusters are formed by expanding the density reachability relationship with the core point as the center; low-density noise points are marked, and boundary points are assigned to a secondary location. Finally, the optimized clustering results are output to achieve accurate division of homogeneous working condition clusters.

[0029] This invention employs the adaptive DBSCAN-LoAD clustering algorithm for outlier data labeling, enabling adaptive search of shape and density parameters.

[0030] Specifically, step 5) includes: Step 51) Obtain EIS data of abnormal and normal single cells in one batch through Step 2), and obtain EIS data of abnormal and normal single cells in two batches through Step 4). Step 52), DRT analysis and ECM fitting are performed on the EIS data of abnormal and normal single cells; the parameters of each element in the ECM model are obtained through ECM fitting; the DRT analysis comprehensively considers the changes in peak area and position at different time scales, of which 10 -2 The three peaks in the ~10Hz range are the focus of consideration and analysis; Step 53) Determine the reasons for the discrepancies between the EIS data of normal cells and the EIS data of abnormal cells and cells with abnormal changes in the two dimensions of ECM and DRT.

[0031] Specifically, in step 6), ECM fitting is combined with DRT analysis and battery internal mechanism analysis to determine the mechanism cause of the discrepancy in EIS data. Based on the mechanism analysis, the abnormal causes of increased inconsistency in energy storage battery modules include abnormal temperature, abnormal SOC, abnormal aging, and abnormal internal short circuit.

[0032] The above method was applied as follows: a module consisting of six 280Ah CATL prismatic lithium iron phosphate batteries was used as the subject of a simulation experiment. Differences between the batteries within the module were considered, and EIS testing was conducted. Figure 2This is the battery EIS data obtained from the first test. Subsequently, an adaptive DBSCAN-LoAD clustering algorithm was used for EIS clustering analysis. The principle of the algorithm is as follows: Figure 3 As shown, DBSCAN is a density-based spatial clustering algorithm with noise. By inputting the x-coordinate of the intersection point of EIS data with the real axis, the x-coordinate of the second minimum point of EIS data, and the slope of the straight line after the minimum point as the clustering basis, the discrete data EIS1 with obvious anomalies and the discrete data EIS2 with partial anomalies are obtained, and the center data is EIS6.

[0033] The battery was then discharged to obtain a battery module in its second state, and EIS testing was performed. The obtained battery EIS data is as follows: Figure 4 As shown, the following three features were obtained: the x-coordinate of the intersection of the battery EIS data and the real axis, the x-coordinate of the second minimum point of the EIS data, and the slope of the straight line after the minimum point. Then, cluster analysis was performed to obtain discrete data EIS1 and EIS2 that showed obvious inconsistency with other batteries, and the obtained center data was EIS6.

[0034] The real and imaginary parts of the EIS from the second test and the first test were subtracted at the corresponding frequencies to obtain the difference between the real and imaginary parts at different frequencies in the two tests. Then, adaptive DBSCAN-LoAD clustering analysis was performed on this part of the data, focusing on the EIS data in the range of 0.01-100Hz, ignoring the inductive reactance caused by the battery and circuit structure in the high-frequency region. The difference data in the range of 0.01-100Hz was smoothed and filtered, and then the smoothed data was normalized by interval mapping.

[0035] The normalized data was input into the adaptive DBSCAN-LoAD clustering algorithm for cluster analysis. It was found that the trend of change between EIS1 and EIS2 measurements was significantly abnormal compared to the EIS of other single cells. The obtained EIS6 is the central and relatively representative EIS of a single cell.

[0036] The above steps determine that EIS1 and EIS2 are EIS data from abnormal individual cells, while EIS6 is the EIS data from a relatively normal individual cell. Subsequently, DRT analysis and ECM parameter fitting are performed on the EIS data obtained from the second test of the individual cells to obtain the causes of EIS anomalies and abnormal changes in both ECM and DRT dimensions. The DRT analysis uses the GAUSSIAN discretization method and second-order regularized derivatives to analyze all data. The resulting DRT corresponding to the EIS is shown below. Figure 6As shown, the P1 peak of EIS1 shows a slight increase, indicating an increase in the charge transfer impedance of ion migration in the single cell. The physical processes represented by P2 and P3 do not change significantly, indicating that the temperature of the single cell has not changed significantly. The P4 peak shows a slight decrease, indicating a decrease in the solid-phase transport impedance of the single cell. Therefore, it is preliminarily speculated that the cause of the anomaly is that the SOC of the single cell is relatively low. Figure 5 The decrease in real impedance in (a) is more pronounced, indicating a further increase in the SOC difference between individual cells during this process; a significant change in the time constant of peak P3 in EIS2 is observed, while the areas and peak center time constants of peaks P1 and P2 do not show significant differences from the corresponding peaks of normal individual cells, indicating a fault strongly correlated with low-frequency impedance, i.e., solid-phase transport changes. Figure 5 The change in the imaginary impedance in (b) and existing research suggest that an external short-circuit fault has occurred.

[0037] After performing DRT analysis on the EIS, ECM fitting of the first-order RC circuit was performed, and the constructed ECM model is as follows. Figure 7 As shown in (a), R1 and L1 in parallel represent the inductive reactance of the battery, R2 is the ohmic impedance of the battery, and R3 represents the charge transfer impedance of the battery; Q1 represents the constant-phase element, used to describe the non-ideal capacitive behavior of the electrode / electrolyte interface; Ws1 is connected in series at the end of the circuit, representing the semi-infinite diffusion impedance. More specifically, Q1_T is the admittance parameter Y0 (the amplitude coefficient of the constant-phase element CPE), Q1_P is the exponential factor n (reflecting the degree of deviation from the ideal capacitance), Ws1_R is the diffusion characteristic resistance, Ws1_T is the diffusion time constant, and Ws1_P is the diffusion exponential factor; the results of ECM parameter identification based on EIS data are as follows. Figure 7 As shown in (b), it is found that R3 of EIS1 is relatively large and accompanied by changes in diffusion impedance, which corroborates the above content about DRT analysis; Ws1_T of Ws element of EIS2 increases, showing obvious abnormality under the condition of SOC approximation, which corroborates the external short circuit fault inference in DRT analysis.

[0038] Example 2 This embodiment provides a multi-dimensional inconsistency diagnosis system for energy storage battery modules based on low-to-medium frequency EIS, used to implement the multi-dimensional inconsistency diagnosis method for energy storage battery modules described in Embodiment 1, such as... Figure 8 As shown, it consists of an EIS data measurement unit, a first clustering unit, an EIS data change acquisition unit, a second clustering unit, an anomaly cause assessment unit, and a fault type diagnosis unit.

[0039] The EIS data measurement unit is used to measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed.

[0040] The first clustering unit is used to cluster EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells, so as to identify the abnormal individual cells.

[0041] The EIS data change acquisition unit is used to calculate the difference between the EIS data of the same single cell in different batches of modules to obtain the change of EIS data between two consecutive measurements.

[0042] The second clustering unit: clusters the EIS data change data obtained by the EIS data change acquisition unit to obtain the central data and outlier data in the data change data. The individual cells corresponding to the outlier data are the individual cells with abnormal changes; in order to find the individual cells with abnormal changes.

[0043] The aforementioned anomaly cause assessment unit is used to perform DRT analysis and ECM fitting on the EIS data of abnormal single cells and abnormally changing single cells with the EIS data of normal single cells, and to evaluate the causes of abnormal EIS data performance from multiple dimensions.

[0044] The aforementioned fault type diagnosis unit is used to combine the causes and mechanisms of abnormal EIS data to diagnose the fault types of individual cells.

[0045] It should be noted that each unit in the aforementioned multi-dimensional inconsistency diagnosis system for energy storage battery modules based on low- and medium-frequency EIS can be implemented entirely or partially through software, hardware, or a combination thereof. These units can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each unit. For specific limitations regarding the multi-dimensional inconsistency diagnosis system for energy storage battery modules based on low- and medium-frequency EIS, please refer to the limitations of the multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low- and medium-frequency EIS (i.e., Example 1) above; both have the same function and role, and will not be repeated here.

[0046] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0047] This specification and accompanying drawings are merely illustrative examples of the present invention and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the present invention. Clearly, those skilled in the art can make various alterations and modifications to the present invention without departing from its scope. Therefore, if such modifications and variations fall within the scope of the present invention and its equivalents, the present invention intends to include these modifications and variations.

Claims

1. A multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low-to-medium frequency EIS, characterized in that, Including the following steps: 1) Measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed. 2) Cluster the EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells. 3) The difference between the EIS data of the same single cell in the modules of different measurement batches is calculated to obtain the change of EIS data between the two measurements. 4) Cluster the EIS data changes obtained in step 3) to obtain the central data and outlier data in the data changes. The individual cells corresponding to the outlier data are the individual cells with abnormal changes. 5) Perform DRT analysis and ECM fitting on the EIS data of abnormal cells and cells with abnormal changes, and compare them with the EIS data of normal cells to evaluate the reasons for the abnormal performance of EIS data from multiple dimensions. 6) Combine the causes and mechanisms of abnormal EIS data to diagnose the types of faults in individual cells.

2. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low- and medium-frequency EIS according to claim 1, characterized in that, Step 1) includes: Step 11), continuously and periodically measure the EIS data of each individual cell in the energy storage battery module to be diagnosed; Step 12) Perform KK verification on the EIS data and preprocess the data to remove obvious abnormal data and remeasure the EIS data of the corresponding single cell.

3. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on medium- and low-frequency EIS according to claim 1, characterized in that, Step 2) includes: Step 21), obtain the 100-10Hz phase angle, 0.1-0.01Hz real part, and 100-0.01Hz imaginary part of the EIS data from the same measurement batch; Step 22) Perform adaptive DBSCAN-LoAD clustering on the EIS data features to obtain the center data in the dataset and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are abnormal individual cells.

4. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on medium- and low-frequency EIS according to claim 1, characterized in that, Step 3) includes: Step 31) The EIS data of the latest measurement batch of the same single cell are compared with the EIS data of a certain historical measurement batch at the same frequency points to obtain the change of EIS data of each single cell in the module. Step 32) Perform interval mapping normalization on the EIS data changes obtained after subtraction to amplify the differences between EIS data.

5. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on medium- and low-frequency EIS according to claim 4, characterized in that, Step 4) includes: Step 41), calculate and obtain the real and imaginary features of the EIS data changes in the low-to-medium frequency region; Step 42) Perform adaptive DBSCAN-LoAD clustering on the EIS data features obtained in Step 41) to obtain the central data that best represents the changes in EIS data of this measurement batch, and identify outlier data with obvious outlier characteristics. The individual cells corresponding to the outlier data are individual cells with abnormal changes.

6. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on medium- and low-frequency EIS according to claim 3 or 5, characterized in that, The steps of the adaptive DBSCAN-LoAD clustering algorithm are as follows: Data preprocessing: Standardizing the features of the input EIS data; Parameter initialization: Set the basic algorithm parameters: neighborhood radius ε and core point determination threshold minPts; Local density calculation: The local adaptive density (LoAD) index is used to quantify the local clustering degree of each sample point. The formula is as follows: in: , Let i and j be the i-th and j-th sample points, respectively. for and The Euclidean distance between them; The kernel function bandwidth; Core point determination: Based on DBSCAN density clustering rules, core points are determined by combining the LoAD density value with preset parameters; if If an ε-neighborhood contains no fewer than minPts samples, it is determined to be a core point, and all sample points within its neighborhood are assigned to the same density reachable region. Clustering formation and optimization: Initial clusters are formed by expanding the density reachability relationship with the core point as the center; low-density noise points are marked, and boundary points are assigned to a secondary location. Finally, the optimized clustering results are output to achieve accurate division of homogeneous working condition clusters.

7. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on medium- and low-frequency EIS according to claim 1, characterized in that, Step 5) includes: Step 51) Obtain EIS data of abnormal and normal single cells in one batch through Step 2), and obtain EIS data of abnormal and normal single cells in two batches through Step 4). Step 52) Perform DRT analysis and ECM fitting on the EIS data of abnormal and normal single cells, and obtain the parameters of each component in the ECM model through ECM fitting. Step 53) Determine the reasons for the discrepancies between the EIS data of normal cells and the EIS data of abnormal cells and cells with abnormal changes in the two dimensions of ECM and DRT.

8. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low- and medium-frequency EIS according to claim 7, characterized in that, In step 52), the analysis using DRT comprehensively considers the changes in peak area and position at different time scales, where 10 -2 The three peaks in the ~10Hz range are the focus of consideration and analysis.

9. The multi-dimensional inconsistency diagnosis method for energy storage battery modules based on low- and medium-frequency EIS according to claim 7, characterized in that, In step 6), ECM fitting is combined with DRT analysis and battery internal mechanism analysis to determine the mechanism cause of the discrepancy in EIS data. Based on the mechanism analysis, the battery abnormality causes that lead to increased inconsistency in energy storage battery modules include temperature abnormality, SOC abnormality, aging abnormality and internal short circuit abnormality.

10. A multi-dimensional inconsistency diagnostic system for energy storage battery modules based on low-to-medium frequency EIS, used to implement the multi-dimensional inconsistency diagnostic method for energy storage battery modules according to any one of claims 1-9, characterized in that, include: EIS data measurement unit: used to measure and preprocess the EIS data of each individual cell in the energy storage battery module to be diagnosed. First clustering unit: used to cluster EIS data obtained from the same measurement batch to obtain the central data and outlier data in the dataset. The individual cells corresponding to the outlier data are abnormal individual cells. EIS data change acquisition unit: used to calculate the difference between the EIS data of the same single cell in different batches of modules to obtain the change of EIS data between two consecutive measurements. The second clustering unit: clusters the EIS data change obtained by the EIS data change acquisition unit to obtain the central data and outlier data in the data change. The individual cells corresponding to the outlier data are the individual cells with abnormal changes. Anomaly Cause Assessment Unit: Used to perform DRT analysis and ECM fitting on the EIS data of abnormal cells and cells with abnormal changes, and the EIS data of normal cells, to assess the causes of abnormal EIS data performance from multiple dimensions. Fault Type Diagnosis Unit: Used to combine the causes and mechanisms of abnormal EIS data to diagnose the fault types of individual cells.

Images