A power battery capacity abnormality early warning method

Abstract

The application discloses a kind of power battery capacity abnormal early warning method, belong to power battery health monitoring technical field.The method is first obtained battery pack charging condition data, and independent charging section is divided and the corresponding battery pack health state SOH of each charging section is calculated;Adopt the two-level architecture of " whole package coarse screening-battery cell fine positioning", only to the charging section of SOH anomaly is executed battery cell level positioning: extraction low SOC and high SOC feature interval monomer voltage data, combined with voltage ranking proportion and the joint determination rule of normalized permutation entropy, accurately identify capacity abnormal battery cell;Final output early warning result and abnormal battery cell identification.The application does not need to disassemble battery pack, only rely on dynamic charging operation data can realize whole package capacity evaluation and single battery abnormal positioning, high in calculation efficiency, positioning accuracy is high, can greatly reduce power battery operation and echelon utilization cost, applicable to electric vehicle, battery swap station and energy storage station Large-scale battery full life cycle monitoring.

Description

Technical Field

[0001] This invention relates to the field of power battery health monitoring and battery management technology, and in particular to a method for early warning of abnormal power battery capacity. Background Technology

[0002] During cyclic charging and discharging and long-term storage, the usable capacity of lithium-ion batteries will irreversibly decrease. A battery pack is composed of a large number of individual cells connected in series. Cells with abnormal numbers in the pack will create a "bottleneck effect", causing the usable capacity of the entire pack to drop prematurely. In addition, overcharging and over-discharging are prone to occur during charging and discharging, which may induce thermal runaway risks.

[0003] Existing power battery capacity assessment technologies have the following main shortcomings: Most solutions can only assess the overall State of Health (SOH) of the battery pack and cannot pinpoint specific cells with abnormal capacity. After-sales repairs usually involve replacing the entire pack, resulting in extremely high maintenance costs. Traditional capacity testing requires disassembling the battery pack and conducting long-term static testing, which is a destructive test and cannot rely on vehicle operation big data to achieve batch, non-destructive remote screening. Some deep learning-based cell positioning solutions have high computational complexity and are difficult to adapt to real-time monitoring scenarios involving massive numbers of batteries, such as vehicles, battery swapping stations, and large-scale energy storage power stations.

[0004] Therefore, there is an urgent need for a power battery capacity anomaly early warning method that does not require disassembling the battery pack, has high computational efficiency, and can accurately locate individual abnormal cells. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for early warning of abnormal power battery capacity. Based on dynamic charging condition data, it can realize the health assessment of the entire battery pack capacity and the non-destructive and accurate location of abnormal individual cells, taking into account both computational efficiency and location accuracy.

[0006] To achieve the above objectives, the present invention provides a method for early warning of abnormal power battery capacity, comprising the following steps: S1 acquires the charging condition timing data of the target battery pack, divides it into multiple independent charging segments, and calculates the battery pack health status (SOH) for each charging segment. S2 Single Charging Segment Anomaly Detection: If the SOH of the current charging segment is lower than the preset SOH anomaly threshold, it is marked as a capacity anomaly charging segment, and the capacity anomaly cell location step is executed; if the SOH is higher than the preset SOH anomaly threshold, cell location is skipped, and the next charging segment is processed. S3 Abnormal Capacity Cell Location: For abnormal capacity charging segments, extract the voltage data of each individual cell in the low SOC range and the high SOC range respectively; count the voltage ranking percentage of each individual cell in the corresponding range, and calculate the permutation entropy of the voltage ranking sequence of each individual cell; based on the joint judgment condition of voltage ranking percentage and permutation entropy in the low SOC range and the high SOC range, identify the abnormal capacity cells. S4 outputs a warning signal for abnormal battery pack capacity and identifies the abnormal battery cells.

[0007] Furthermore, in step S1, the specific method for dividing the charging segments into independent sections is as follows: The charging condition data is traversed in chronological order. If the time interval between two adjacent frames is greater than the preset time interval threshold, the next frame is determined to be the starting frame of a new charging segment. After dividing the charging segments, select the effective charging segments. The effective charging segments meet the following conditions: the SOC difference of a single segment is greater than the preset SOC difference threshold, and the starting SOC of the charging segment is less than the preset starting SOC threshold.

[0008] Furthermore, the specific steps for calculating the battery pack health status (SOH) in step S1 are as follows: S11 identifies current abrupt change points within the charging segment, i.e., frames where the absolute value of the current difference between two adjacent frames is greater than or equal to a preset current abrupt change threshold; calculates the instantaneous internal resistance corresponding to each current abrupt change point, and takes the average value of the internal resistance of all abrupt change points as the internal resistance characteristic value of the charging segment; if there are no current abrupt change points within the charging segment, a preset empirical internal resistance value is used as the internal resistance characteristic value. S12 is based on the Rint equivalent circuit model. It uses the lowest single cell voltage, the first frame current and the internal resistance characteristic value of the first frame to calculate the starting open circuit voltage, and uses the highest single cell voltage, the last frame current and the internal resistance characteristic value of the last frame to calculate the ending open circuit voltage. S13 Based on the pre-calibrated SOC-OCV curve, the corresponding initial SOC and termination SOC are obtained by mapping the initial open-circuit voltage and the termination open-circuit voltage; S14 uses the ampere-hour integration method to calculate the total charge input in the charging section, and combines the change in SOC to infer the current available capacity of the battery pack. Finally, the battery pack SOH is obtained from the ratio of available capacity to nominal capacity.

[0009] Furthermore, the formula for calculating the instantaneous internal resistance in step S11 is as follows: ， ; Where N is the total number of data frames in this charging segment. , The first i The highest single-cell voltage and charging current of the frame. , These represent the highest single-unit voltage difference and current difference between two adjacent frames, respectively. If there are q current abrupt change points within the charging segment, then the characteristic value of the internal resistance, i.e., the average internal resistance, is: ; in This represents the instantaneous internal resistance corresponding to each current abrupt change point.

[0010] Furthermore, the formula for calculating the open-circuit voltage in step S12 is as follows: ; ; in, , These are the lowest single-cell voltage and charging current in the first frame of charging, respectively. , These represent the highest single-cell voltage and charging current in the last frame of charging, respectively. This is the characteristic value of the internal resistance.

[0011] Furthermore, in step S3: The low SOC range is the range where the SOC is less than or equal to the preset low SOC threshold, and the individual cell voltages within this range are ranked in ascending order. The high SOC range is the range where the SOC is greater than or equal to the preset high SOC threshold, and the individual cell voltages within this range are ranked in descending order. The voltage ranking percentage is the ratio of the number of frames in which the corresponding single cell appears in the top ranking position to the total number of data frames in the corresponding SOC range.

[0012] Furthermore, the permutation entropy mentioned in step S3 is the normalized permutation entropy, and its calculation steps are as follows: S31 sets the embedding dimension and delay time, performs phase space reconstruction on the voltage sorting time series of a single cell, and obtains multiple reconstruction vectors with a length equal to the embedding dimension. S32 sorts the elements in each reconstructed vector by numerical value, maps them to the corresponding permutation patterns, and counts the probability of occurrence of each permutation pattern. S33 calculates the basic permutation entropy based on the Shannon entropy formula, and then normalizes it by using the natural logarithm of the number of permutations corresponding to the embedding dimension as the denominator to obtain the normalized permutation entropy.

[0013] Furthermore, the joint determination condition mentioned in step S3 is specifically as follows: A single battery cell is determined to be a cell with abnormal capacity when it simultaneously meets all of the following conditions: The percentage of top-ranked entries in the low SOC range is greater than or equal to the preset threshold for the percentage of low SOC range entries. The percentage of top-ranked entries in the high SOC range is greater than or equal to the preset high range percentage threshold. The normalized permutation entropy of the low SOC interval is less than or equal to the preset low interval entropy threshold. The normalized permutation entropy of the high SOC interval is less than or equal to the preset high interval entropy threshold.

[0014] Furthermore, step S4 specifically involves: calculating the percentage of abnormal charging segments among all valid charging segments within the calculation period; when this percentage exceeds a preset percentage threshold, outputting a battery pack capacity abnormality warning signal and the identification information of the identified abnormal capacity cells.

[0015] Furthermore, step S4, when outputting the warning signal, also includes: If the number of identified cells with abnormal capacity is within a preset limit, a targeted replacement prompt for the corresponding cells will be output. If no concentrated abnormal capacity cells are identified, the system will output a message indicating uniform capacity degradation across the entire pack and a recommendation to replace the entire pack.

[0016] The beneficial effects of this invention are as follows: (1) Non-destructive testing and dynamic operating conditions: Based on the Rint equivalent circuit model, the actual OCV and SOC are deduced from the dynamic charging data. There is no need to leave the data static for a long time or disassemble the battery pack. Capacity assessment can be achieved by relying only on the regular operating data. It has a wide range of applicable scenarios and is easy to implement in batches. (2) Precise positioning and reduced operation and maintenance costs: The combined judgment scheme of dual SOC characteristic intervals, voltage ranking ratio and arrangement entropy can accurately locate single cells with abnormal capacity, support targeted replacement, replace the traditional whole pack replacement scheme, and significantly reduce after-sales and cascade utilization costs. (3) High computational efficiency and adaptable to large-scale scenarios: The two-level architecture of "whole package coarse screening - cell precise positioning" is adopted. Cell-level calculations are not required during normal charging, resulting in low computational load. Furthermore, based on statistical algorithms, the complexity is much lower than that of deep learning solutions, which can support large-scale monitoring of the entire life cycle of electric vehicles, battery swapping stations, and energy storage stations. (4) Reliable results and low misjudgment rate: The abnormality of the battery cell is determined by the joint determination of multi-dimensional features, and the final warning can be made by combining the proportion of abnormal charging segments throughout the entire cycle, which effectively reduces the risk of misjudgment caused by single-segment data fluctuations and the warning results are highly stable.

[0017] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0018] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the invention.

[0019] Figure 1A flowchart illustrating an abnormal power battery capacity early warning method according to an embodiment of the present invention is shown.

[0020] Figure 2 A schematic diagram of the equivalent circuit model of a lithium-ion battery Rint according to an embodiment of the present invention is shown. Detailed Implementation

[0021] Preferred embodiments of the invention will now be described in more detail. While preferred embodiments of the invention are described below, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0022] See Figure 1 This disclosure provides a method for early warning of abnormal power battery capacity, comprising the following steps: Pre-parameter calibration: Before executing the early warning algorithm, the SOC-OCV curve of a single battery cell must be calibrated under normal temperature conditions to establish the mapping relationship between open-circuit voltage and state of charge. The specific process is as follows: Discharge the target battery cell to 0% SOC at a standard discharge rate, let it stand for 1-2 hours until the voltage stabilizes, and record the initial open-circuit voltage; charge with a constant current of 0.1C or less, stop charging after every 5% of SOC is charged, let it stand for 1-2 hours until the voltage stabilizes, and record the open-circuit voltage at the corresponding SOC, and complete the full range test from 0% to 100% in sequence; finally, use a polynomial fitting method to fit the discrete test points into a continuous SOC-OCV curve for subsequent open-circuit voltage to state of charge mapping calculation.

[0023] S1 Data Preprocessing and Battery Pack SOH Calculation: Charging condition time-series data of the target battery pack is extracted from the cloud big data platform according to a preset calculation cycle (e.g., once a month). Fields include battery pack code / vehicle VIN, timestamp, SOC, total charging current, and voltage of each individual cell. (m is the total number of cells connected in series), highest voltage of a single cell Minimum voltage of a single unit .

[0024] The raw data is cleaned to remove four types of invalid data frames: data frames with empty time, data frames with empty SOC or values ​​outside the range of [0,1], data frames with empty current, and data frames with empty or zero individual voltage or values ​​outside the normal operating voltage range.

[0025] The cleaned data is traversed chronologically to divide it into independent charging segments: if the time interval between two adjacent frames is greater than a preset time interval threshold (e.g., 600s), the next frame is determined to be the starting frame of a new charging segment, and each charging segment is calculated independently. Valid charging segments are filtered based on the following criteria: the SOC difference of a single segment is greater than a preset SOC difference threshold (e.g., 60%), and the starting SOC of the charging segment is less than a preset starting SOC threshold (e.g., 20%). This ensures that the charging process covers a sufficiently wide energy range and improves the reliability of the calculation results.

[0026] based on Figure 2 The Rint equivalent circuit model shown (the cell is equivalent to an ideal open-circuit voltage source OCV and an ohmic internal resistance) (in series), calculate the corresponding battery pack SOH for each effective charging segment.

[0027] S1.1 Calculate the characteristic value of internal resistance: Regardless of whether there are sudden current changes during this charging process, the instantaneous internal resistance corresponding to the highest single cell is first calculated frame by frame: ， ; Where N is the total number of data frames in this charging segment. , These represent the highest single-cell voltage and charging current in the i-th frame, respectively. , These represent the highest single-unit voltage difference and current difference between two adjacent frames, respectively.

[0028] Identify current abrupt changes within the charging segment: If the current difference between two adjacent frames satisfies If a preset current surge threshold (e.g., 10A) is set, then that point is determined to be a current surge point. If there are q current surge points within a charging segment, the average value of the internal resistance of all surge points is taken as the characteristic value of the internal resistance of this segment. ; If there is no sudden current change point in the charging segment and the internal resistance cannot be reliably calculated through dynamic data, then a preset empirical internal resistance value (such as 0.5mΩ) is used as the characteristic value of the internal resistance of this segment.

[0029] S1.2 Calculate the charging start and end open-circuit voltages and SOC: Based on Kirchhoff's voltage law, the true open-circuit voltage is obtained by subtracting the internal resistance voltage drop from the real-time terminal voltage: The initial OCV is calculated using the lowest single-cell voltage at the start of charging (frame 1): ; The charging termination time (frame N) is calculated using the highest single-cell voltage to determine the termination OCV: ; in , These are the minimum single-cell voltage and charging current for the first frame, respectively. , These represent the highest single-cell voltage and charging current in the Nth frame, respectively.

[0030] By looking up the pre-calibrated SOC-OCV curve in a table, the result is obtained... , Corresponding start With termination Calculate the change in SOC: ; S1.3 Calculate the available capacity and SOH: The total amount of electricity charged in this charging segment is calculated using the ampere-hour integration method: ; Time difference , Let N be the time of the i-th frame, and N be the total number of data frames in the charging segment.

[0031] By combining the change in SOC, we can deduce the current fully charged usable capacity of the battery pack: ; Final calculation of battery pack state of health (SOH): ; in This refers to the nominal capacity of the battery pack at the time of manufacture.

[0032] S2 Single Charging Segment Anomaly Detection: The calculated SOH of the current charging segment is compared with the preset SOH abnormal threshold (e.g., 70%): if SOH ≤ preset threshold, the charging segment is marked as a capacity abnormal charging segment and the cell-level positioning is performed in step S3; if SOH > preset threshold, the battery status of this segment is determined to be normal, the cell positioning step is skipped, and the next charging segment is processed directly.

[0033] S3 Abnormal Capacity Cell Location: For charging segments marked as having abnormal capacity, abnormal cells are identified based on voltage statistical characteristics of dual SOC intervals.

[0034] S3.1 Characteristic interval voltage sorting: Set a low SOC threshold (e.g., 30%) and a high SOC threshold (e.g., 90%), and extract the individual cell voltage data within the two intervals respectively: For data frames with SOC ≤ low SOC threshold, sort all individual cell voltages at the same time in ascending order, with lower voltages having smaller rank numbers; For data frames with SOC ≥ high SOC threshold, sort all individual cell voltages at the same time in descending order, with higher voltages having smaller rank numbers.

[0035] S3.2 Voltage Ranking Percentage Statistics

[0036] For the k-th individual cell (k=1,2,…,m), the proportion of data frames ranked j in the corresponding interval to the total number of data frames in that interval is calculated to obtain the ranking ratio rank_low_ratio(k,j) for the low SOC interval and the ranking ratio rank_high_ratio(k,j) for the high SOC interval; where j is the ranking number, and only the top rankings are counted (e.g., j≤5).

[0037] S3.3 Calculation of normalized permutation entropy: For each battery cell, the normalized permutation entropy is calculated based on the voltage sorting time series. This entropy is used to quantify the stability of the ranking sequence. The specific process involves setting the embedding dimension. With delay time (like =3, =1); Perform sliding window phase space reconstruction on the sorted sequence of length L to generate multiple sequences of length L. The reconstructed vectors have a total number of L-(p-1) vectors. For each reconstructed vector, elements are arranged in ascending numerical order, and the original position indices of the elements are recorded to form the arrangement pattern. The embedding dimension is [missing information]. Coexistence Count the possible permutations; count the frequency of each permutation and divide by the total number of vectors to obtain the corresponding probability of occurrence. ; Calculate the entropy of the basic permutation based on the Shannon entropy formula: ; With maximum entropy Normalizing the denominator yields the normalized permutation entropy with values ​​in the range [0,1]. ; Calculate the normalized permutation entropy of each cell in the low SOC range sequentially. Normalized permutation entropy with high SOC interval The smaller the value, the more stable the cell's voltage ranking and the more significant its capacity anomaly characteristics.

[0038] Numerical Example: The sorted sequence of a certain battery cell in the low SOC range is [1,1,2,1,1,1,2,1], with an embedding dimension of [1,1,2,1,1,1,2,1]. Delay time =1, generating a total of 6 reconstruction vectors; the mapping yields the permutation pattern (0,1,2) appearing 3 times, (0,2,1) appearing 2 times, and (1,2,0) appearing once, with corresponding probabilities of 0.5, 0.333, and 0.167 respectively; the entropy of the basic permutation is calculated. After normalization .

[0039] S3.4 Jointly determine abnormal battery cells: When the k-th cell simultaneously meets all of the following conditions, it is marked as a cell with abnormal capacity: the proportion of the top-ranked cells in the low SOC range is greater than or equal to the preset low range proportion threshold; the proportion of the top-ranked cells in the high SOC range is greater than or equal to the preset high range proportion threshold; the normalized permutation entropy of the low SOC range is less than or equal to the preset low range entropy threshold; and the normalized permutation entropy of the high SOC range is less than or equal to the preset high range entropy threshold.

[0040] S4 warning result output: The system outputs a battery pack capacity anomaly warning signal and the identification information of the identified abnormal cells. In a preferred embodiment, before outputting the warning result, a full-cycle statistical determination is performed: the total number of all valid charging segments within the cycle is calculated. And the number of charging segments with abnormal capacity. Calculate the proportion of abnormal charging segments When the percentage of abnormal charging segments exceeds the preset percentage threshold (e.g., 50%), a battery pack capacity abnormality warning signal and abnormal cell identification information will be output.

[0041] When outputting warning results, if the identified abnormal capacity cells are concentrated within a preset number (e.g., 1-2 cells), a targeted replacement prompt for the corresponding cells will be output simultaneously; if no concentrated abnormal cells are identified, it will be determined that the entire pack is uniformly degraded, and a replacement prompt for the entire pack will be output simultaneously.

[0042] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for early warning of abnormal capacity in power batteries, characterized in that, Includes the following steps: S1 acquires the charging condition timing data of the target battery pack, divides it into multiple independent charging segments, and calculates the battery pack health status (SOH) for each charging segment. S2 Single Charging Segment Anomaly Detection: If the SOH of the current charging segment is lower than the preset SOH anomaly threshold, it is marked as a capacity anomaly charging segment, and the capacity anomaly cell location step is executed; if the SOH is higher than the preset SOH anomaly threshold, cell location is skipped, and the next charging segment is processed. S3 Abnormal Capacity Cell Location: For abnormal capacity charging segments, extract the voltage data of each individual cell in the low SOC range and the high SOC range respectively; count the voltage ranking percentage of each individual cell in the corresponding range, and calculate the permutation entropy of the voltage ranking sequence of each individual cell; based on the joint judgment condition of voltage ranking percentage and permutation entropy in the low SOC range and the high SOC range, identify the abnormal capacity cells. S4 outputs a warning signal for abnormal battery pack capacity and identifies the abnormal battery cells.

2. The method for early warning of abnormal power battery capacity according to claim 1, characterized in that, In step S1, the specific method for dividing the charging segments into independent sections is as follows: The charging condition data is traversed in chronological order. If the time interval between two adjacent frames is greater than the preset time interval threshold, the next frame is determined to be the starting frame of a new charging segment. After dividing the charging segments, select the effective charging segments. The effective charging segments meet the following conditions: the SOC difference of a single segment is greater than the preset SOC difference threshold, and the starting SOC of the charging segment is less than the preset starting SOC threshold.

3. The method for early warning of abnormal power battery capacity according to claim 1, characterized in that, The specific steps for calculating the battery pack health status (SOH) in step S1 are as follows: S11 identifies current abrupt change points within the charging segment, i.e., frames where the absolute value of the current difference between two adjacent frames is greater than or equal to a preset current abrupt change threshold; calculates the instantaneous internal resistance corresponding to each current abrupt change point, and takes the average value of the internal resistance of all abrupt change points as the internal resistance characteristic value of the charging segment; if there are no current abrupt change points within the charging segment, a preset empirical internal resistance value is used as the internal resistance characteristic value. S12 is based on the Rint equivalent circuit model. It uses the lowest single cell voltage, the first frame current and the internal resistance characteristic value of the first frame to calculate the starting open circuit voltage, and uses the highest single cell voltage, the last frame current and the internal resistance characteristic value of the last frame to calculate the ending open circuit voltage. S13 Based on the pre-calibrated SOC-OCV curve, the corresponding initial SOC and termination SOC are obtained by mapping the initial open-circuit voltage and the termination open-circuit voltage; S14 uses the ampere-hour integration method to calculate the total charge input in the charging section, and combines the change in SOC to infer the current available capacity of the battery pack. Finally, the battery pack SOH is obtained from the ratio of available capacity to nominal capacity.

4. The power battery capacity anomaly early warning method according to claim 3, characterized in that, The formula for calculating the instantaneous internal resistance in step S11 is: ， ； Where N is the total number of data frames in this charging segment. , The first i The highest single-cell voltage and charging current of the frame. , These represent the highest single-unit voltage difference and current difference between two adjacent frames, respectively. If there are q current abrupt change points within the charging segment, then the characteristic value of the internal resistance, i.e., the average internal resistance, is: ； in This represents the instantaneous internal resistance corresponding to each current abrupt change point.

5. The power battery capacity anomaly early warning method according to claim 3, characterized in that, The formula for calculating the open-circuit voltage in step S12 is: ； ； in, , These are the lowest single-cell voltage and charging current in the first frame of charging, respectively. , These represent the highest single-cell voltage and charging current in the last frame of charging, respectively. This is the characteristic value of the internal resistance.

6. The method for early warning of abnormal power battery capacity according to claim 1, characterized in that, In step S3: The low SOC range is the range where the SOC is less than or equal to the preset low SOC threshold, and the individual cell voltages within this range are ranked in ascending order. The high SOC range is the range where the SOC is greater than or equal to the preset high SOC threshold, and the individual cell voltages within this range are ranked in descending order. The voltage ranking percentage is the ratio of the number of frames in which the corresponding single cell appears in the top ranking position to the total number of data frames in the corresponding SOC range.

7. The method for early warning of abnormal power battery capacity according to claim 1, characterized in that, The permutation entropy mentioned in step S3 is the normalized permutation entropy, and its calculation steps are as follows: S31 sets the embedding dimension and delay time, performs phase space reconstruction on the voltage sorting time series of a single cell, and obtains multiple reconstruction vectors with a length equal to the embedding dimension. S32 sorts the elements in each reconstructed vector by numerical value, maps them to the corresponding permutation patterns, and counts the probability of occurrence of each permutation pattern. S33 calculates the basic permutation entropy based on the Shannon entropy formula, and then normalizes it by using the natural logarithm of the number of permutations corresponding to the embedding dimension as the denominator to obtain the normalized permutation entropy.

8. The method for early warning of abnormal power battery capacity according to claim 7, characterized in that, The joint determination condition mentioned in step S3 is as follows: A single battery cell is determined to be a cell with abnormal capacity when it simultaneously meets all of the following conditions: The percentage of top-ranked entries in the low SOC range is greater than or equal to the preset threshold for the percentage of low SOC range entries. The percentage of top-ranked entries in the high SOC range is greater than or equal to the preset high range percentage threshold. The normalized permutation entropy of the low SOC interval is less than or equal to the preset low interval entropy threshold. The normalized permutation entropy of the high SOC interval is less than or equal to the preset high interval entropy threshold.

9. The method for early warning of abnormal power battery capacity according to claim 2, characterized in that, Step S4 specifically involves: calculating the percentage of abnormal charging segments among all valid charging segments within the calculation period; when this percentage exceeds a preset percentage threshold, outputting a battery pack capacity abnormality warning signal and the identification information of the identified abnormal capacity cells.

10. The method for early warning of abnormal power battery capacity according to claim 1, characterized in that, Step S4, when outputting the warning signal, also includes: If the number of identified cells with abnormal capacity is within a preset limit, a targeted replacement prompt for the corresponding cells will be output. If no concentrated abnormal capacity cells are identified, the system will output a message indicating uniform capacity degradation across the entire pack and a recommendation to replace the entire pack.

Images