Intelligent online capacity checking management system for storage battery pack

Abstract

The invention relates to the technical field of battery state monitoring, in particular to an intelligent online capacity checking management system for a storage battery pack, which comprises a jump sensing module, a lag correlation module, a starting point confirmation module, a state diagnosis module and a correction output module. According to the invention, the method achieves the precise recognition of a voltage hysteresis effect through the positioning of the difference value of a current sudden change point, the tracking of a fluctuation convergence interval, the construction of a dynamic sudden change buffer structure, the enhancement of the capturing capability of a key change behavior, the extraction of a response time point in combination with the change trend of a voltage derivative, and the combined binding of a charge state based on the voltage difference value and current sampling. The real-time performance and the pertinence of capacity evaluation are improved, the sudden change period is identified through the charge state difference value, a synchronous mark is established through voltage stability analysis, correction mapping is carried out according to the state difference, a capacity correction path with continuity, traceability and dynamic adaptability is formed, and the precision and the stability of capacity management of the storage battery pack are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of battery status monitoring, and in particular to an intelligent online capacity management system for a battery pack. Background Art

[0002] The field of battery status monitoring technology includes the real-time acquisition, analysis and management of battery operating status, electrical performance parameters and health indicators. The core content includes continuous measurement and recording of key parameters such as voltage, current, temperature, capacity, internal resistance, etc. of batteries or battery packs during use, and the judgment of battery charge status, remaining capacity and health status through data processing. It covers the entire process from battery data collection, status assessment, life prediction to fault warning, and is widely used in power, communication, transportation, energy storage and other industries. Its purpose is to improve the safety, reliability and maintenance efficiency of battery use, and provide basic support for energy management.

[0003] Among them, the intelligent online nuclear capacity management system for battery packs refers to a system device used in battery pack operation scenarios to implement online capacity accounting and status management. It mainly aims at the problems of capacity attenuation, inaccurate power evaluation and maintenance dependence on manual labor in the long-term operation of battery packs. It proposes a method for capacity accounting through voltage, current and time data, combines periodic voltage collection and discharge process records, calculates the actual available capacity of the battery pack, and makes trend judgments based on multi-cycle operation data to assist management. Its means include capacity calculation based on time series current integration, comparison basis for judging capacity changes based on static voltage attenuation, and capacity calibration method based on data collected under constant discharge conditions. The capacity evaluation task is completed through a clear electrical parameter data collection path and electrochemical performance judgment model.

[0004] In the existing battery capacity monitoring of storage batteries, periodic integration and fixed feature point acquisition are mostly the core, which cannot cope with the complex disturbances generated by the current signal under the mutation state, resulting in insufficient coverage of capacity identification for nonlinear responses. The processing of voltage changes relies on interval averaging or static analysis, lacks refined identification of the time point of the mutation response, resulting in an increase in the matching error between state evolution and parameter response. During the operation of the battery, in the event of complex scenarios such as sudden increase in load and sudden drop in voltage, conventional methods are difficult to locate the real impact node on the time axis, thus affecting the real-time and accuracy of capacity trend judgment. The horizontal linkage and trend pairing between data have not formed a closed-loop mechanism, and the state judgment and capacity correction links are disconnected, and the capacity change path cannot be tracked based on the operation trajectory. The accumulation of capacity errors formed during long-term operation often cannot be effectively alleviated by single-point correction, which in turn affects the formulation of capacity estimation and maintenance strategies and increases the uncertainty of battery management. Summary of the invention

[0005] The object of the present invention is to solve the disadvantages existing in the prior art, and to propose an intelligent online capacity verification management system for a battery pack.

[0006] To achieve the above object, the present invention adopts the following technical solutions: An intelligent online capacity verification management system for a battery pack includes: The jump perception module obtains the current sampling sequence during the operation cycle of the battery pack, selects the starting point of current mutation, tracks the convergence interval of the difference value of each cycle after the starting point of mutation towards zero, and generates a data set of mutation buffer segments; The hysteresis correlation module, based on the data set of the mutation buffer segments, extracts the continuous voltage data after the jump end time point, selects the starting point of response to count the duration of the negative trend, and obtains the jump response time deviation data by comparing the response time limit; The starting point confirmation module, based on the jump response time deviation data, extracts the voltage value corresponding to the starting point of response, determines whether the difference from the voltage value of the previous cycle exceeds the capacity change trigger threshold, counts the current and voltage values corresponding to the response point, and obtains the capacity sampling data packet; The state diagnosis module, based on the capacity sampling data packet, calculates the difference in the state of charge of each cycle, locates the key cycle of state mutation, analyzes whether the subsequent voltage difference set is less than the stable threshold value, and if satisfied, generates a state voltage synchronization mark cycle; The correction output module, according to the state voltage synchronization mark cycle, incorporates the corresponding state of charge value into the capacity verification correction record sequence, calculates the capacity correction amplitude, and obtains the capacity adjustment record of the battery pack.

[0007] As a further solution of the present invention, the data set of the mutation buffer segments includes the current upper limit, the current lower limit, and the cycle length; the jump response time deviation data includes the starting point of response, the voltage change trend, and the duration of the negative trend; the capacity sampling data packet includes the current value, the voltage value, and the state of charge; the state voltage synchronization mark cycle includes the key cycle of state mutation, the voltage difference set, and the stability judgment flag; the capacity adjustment record includes the state of charge value, the correction source of the capacity trend node, and the capacity correction amplitude.

[0008] As a further solution of the present invention, the jump perception module includes: The difference extraction sub-module, based on the current sampling sequence during the operation cycle of the battery pack, sequentially selects the current values of each sampling point and its adjacent points before and after, calculates the difference between the current sampling point and the previous sampling point and the difference between the current sampling point and the next sampling point, obtains the difference set before and after each sampling point, and obtains the current difference sequence; The mutation starting point recognition sub-module, according to the current difference sequence, determines whether each difference exceeds the jump current threshold, screens the sequence segments where the continuous differences all exceed the jump current threshold, locates the starting position of the sequence segment as the mutation starting point, and obtains the current mutation starting point index sequence; Based on the current mutation starting point index sequence, the mutation buffer segment construction sub-module starts from each mutation starting point, combines the difference sequence in subsequent cycles, determines whether the difference converges to zero step by step, records the end point of the convergence interval as the end point of the buffer segment, records the maximum current value, the minimum current value and the cycle length within the interval, and obtains the mutation buffer segment data set.

[0009] As a further solution of the present invention, the hysteresis correlation module includes: The voltage sequence extraction sub-module extracts the continuous voltage data segment after the jump end time point from the original time sequence based on the jump end time point marked in the mutation buffer segment data set, and obtains the voltage change sequence; Based on the voltage change sequence, the response starting point recognition sub-module performs differential processing on the voltage values of every two adjacent time points in the sequence, calculates the first-order derivative of the voltage and marks the derivative sign, screens the time point segments where the derivative values are continuously negative, and uses the formula: ; Determine the response start time point , identify the interval length from the start time point to before the trend reversal, and obtain the negative duration sequence, where is the voltage value at the kth moment, is the sampling time interval, is the sign function of the voltage derivative, n is the minimum number of consecutive negative cycles of the voltage derivative, is the kth sampling time point, is the negative trend threshold of the derivative; the deviation data acquisition sub-module calculates the time interval between the response start time point sequence and the jump end time point data in the mutation buffer segment data set, and statistically calculates the difference between the time interval and the preset response time limit as the offset, and obtains the jump response time deviation data.

[0010] As a further solution of the present invention, the start point confirmation module includes: Based on the jump response time deviation data, the voltage deviation judgment sub-module extracts the voltage value corresponding to each response start point, obtains the voltage value of the previous cycle, calculates the difference, and compares the difference with the capacity change trigger threshold to determine whether it exceeds the capacity change trigger threshold, and obtains the trigger valid index sequence; The response point binding sub-module combines the trigger valid index sequence, extracts the voltage value, current value and state of charge information at the corresponding time point, performs synchronous recording and joint marking, and obtains the state of charge mapping data group; The sampling data generation sub-module binds and arranges the combined voltage value, current value and state of charge information according to the state of charge mapping data group, constructs a standard data structure and unifies the recording format, and obtains the capacity sampling data packet.

[0011] As a further solution of the present invention, the state diagnosis module includes: The rate judgment sub-module calculates the difference in the state of charge between consecutive sampling time points for each cycle based on the state of charge sequence in the recorded time period of the capacity sampling data packet, determines whether the difference exceeds the charge rate mutation identification threshold, screens the cycle index positions that meet the conditions, and obtains the charge rate mutation cycle sequence; The key cycle extraction sub-module obtains the voltage value sequence corresponding to the cycle positions according to the charge rate mutation cycle sequence, performs sliding window processing with a fixed cycle width, constructs a set of differences between adjacent cycle voltage values within each window, and uses the formula: ; Calculate the voltage fluctuation stability of the th cycle, screen the stability, extract the cycle positions less than the stability threshold value, and obtain the key cycle sequence of state mutation, where is the voltage value of the th cycle, N is the number of cycles in the sliding window, is the absolute difference between adjacent voltage values; the stable window marking sub-module performs a marking operation at each cycle position according to the key cycle sequence of state mutation, synchronously binds the state mutation cycle with the stable voltage cycle within the subsequent sliding window, establishes a cycle synchronization information structure, and obtains the state voltage synchronization marked cycle.

[0012] As a further solution of the present invention, the correction output module includes: The correction record inclusion sub-module extracts the state of charge values corresponding to the cycles based on the state voltage synchronization marked cycle, incorporates them into the correction sequence structure in sequence, and attaches the current capacity trend node position to each correction value to establish a mapping index between the correction and the trend points, and obtains the capacity correction reference sequence; The capacity difference calculation sub-module calculates the numerical difference based on the correction node in the capacity correction reference sequence and its corresponding initial capacity starting value, combines the voltage fluctuation stability and the voltage response fluctuations of the previous and subsequent cycles, and uses the formula: ; Calculate the capacity correction amplitude of the battery pack for the jth cycle, establish the relationship between the correction amplitude and the difference and change rate between nodes, and obtain the capacity correction amplitude sequence, where is the state of charge value of the jth cycle, is the starting value of the capacity, and are the voltage values of the jth cycle and the j - 1th cycle respectively, is the voltage fluctuation stability of the j-th cycle, and m is the number of periods for trend continuation; the adjustment record generation sub-module combines each capacity correction amplitude with the corresponding state of charge and trend identifier according to the node positioning information in the capacity correction amplitude sequence and the capacity correction reference sequence to obtain the battery pack capacity adjustment record.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by performing difference analysis to identify the starting point of current mutation and tracking the boundary of the fluctuation period converging to zero, a mutation buffer structure with dynamic representativeness can be constructed, expanding the recognition range of the current change process of the battery pack. Combining the trend change of the first-order derivative of the voltage after the jump ends, the continuous segment with a negative derivative is extracted as the response time point, avoiding the misjudgment risk brought by fixed time delay judgment, and achieving accurate tracking of the voltage hysteresis characteristic. The voltage difference at the response point is combined with the current sampling value to bind the state of charge, enabling the capacity recognition to transition from static measurement to dynamic calibration, improving the real-time performance and context relevance of capacity recognition. By performing continuous state of charge difference analysis to identify the key periods of state mutation and forming synchronization marks in combination with the stable trend of voltage change, the voltage behavior corresponding to capacity mutation is effectively isolated, enhancing the logical consistency of the capacity trend evolution path. Pairing the state data with the capacity trend nodes, extracting the correction amplitude and incorporating it into the correction record, enabling the capacity adjustment to have the ability of continuous tracking, trend correction and multi-period feedback, and supporting the battery pack capacity assessment to maintain accuracy, continuity and traceability under changing working conditions. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is the system flow chart of the present invention; Figure 2 is the flow chart of the jump perception module of the present invention; Figure 3 is the flow chart of the hysteresis correlation module of the present invention; Figure 4 is the flow chart of the starting point confirmation module of the present invention; Figure 5 is the flow chart of the state diagnosis module of the present invention; Figure 6 is the flow chart of the correction output module of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0014] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0015] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more, unless otherwise specifically defined.

[0016] Please refer to Figure 1 , an intelligent online capacity verification management system for a battery pack includes: The jump perception module obtains the current sampling sequence during the operation cycle of the battery pack, calculates the difference between any sampling point and the front and rear points, selects the starting point of the continuous difference exceeding the jump current threshold as the starting point of the current mutation, tracks the convergence interval of the difference in each cycle after the starting point of the mutation to zero, records the current upper limit, lower limit, and cycle length, and generates a set of mutation buffer segment data; The hysteresis correlation module extracts the continuous voltage data in the time series after the jump end time point marked in the set of mutation buffer segment data, calculates the change trend of the first derivative, selects the initial time point with continuously negative derivative values as the starting point of the response, and statistically calculates the duration of the negative trend, compares it with the response time limit, and obtains the jump response time deviation data; The starting point confirmation module extracts the voltage value corresponding to the starting point of the response based on the jump response time deviation data, determines whether the difference from the voltage value of the previous cycle exceeds the capacity change trigger threshold. If it is satisfied, it statistically calculates the current and voltage values corresponding to the response point, records and binds the current state of charge, and obtains the capacity sampling data packet; The state diagnosis module calculates the difference in the state of charge between consecutive sampling periods based on the state of charge sequence recorded in the capacity sampling data packet, determines whether it exceeds the charge rate mutation recognition threshold. If it is satisfied, it locates the key cycle of the state mutation, obtains the corresponding voltage sampling data, and uses the sliding window method to analyze whether the subsequent voltage difference set is less than the stable threshold value. If it is satisfied, it generates a state voltage synchronization marker cycle; The correction output module incorporates the corresponding state of charge value into the capacity verification correction record sequence according to the state of charge value corresponding to the state voltage synchronization marker cycle, marks the correction source of the current capacity trend node, calculates the capacity correction amplitude by combining the difference between the current capacity starting point and the corresponding node, and obtains the capacity adjustment record of the battery pack.

[0017] The mutation buffer segment data set includes the current upper limit, current lower limit, and cycle length. The jump response time deviation data includes the response starting point, voltage change trend, and duration of the negative trend. The capacity sampling data packet includes the current value, voltage value, and charge state. The state voltage synchronization mark cycle includes the state mutation key cycle, voltage difference set, and stability judgment mark. The capacity adjustment record includes the charge state value, capacity trend node correction source, and capacity correction amplitude.

[0018] See also Figure 2 , the jump perception module includes: The difference extraction submodule selects the current value of each sampling point and its adjacent points in turn based on the current sampling sequence in the battery pack operation cycle, calculates the difference between the current sampling point and the previous sampling point and the difference between the current sampling point and the next sampling point, obtains the difference set before and after each sampling point, and obtains the current difference sequence; Obtain the current sampling sequence during the battery pack operation cycle. The sequence is composed of a current sensor deployed at the output end of the battery pack with a sampling interval of 1 second. Continuous acquisition, for example, the current values ​​(unit: ampere A) collected at 10 consecutive sampling points constitute a sequence: ; This sequence is recorded in the system's storage unit, and the difference extraction submodule first retrieves this current sampling sequence from the storage unit , and then process each sampling point in the sequence in turn, but skip the first and last two points to ensure that each processed point has adjacent points before and after. For the index k in the sequence (where ) sampling point , the module will simultaneously read the current value of the previous sampling point and the current value of the next sampling point , and then perform two subtraction operations, the first to calculate the current sampling point With the previous sampling point The current difference , the second calculation of the next sampling point With the current sampling point The current difference , for example, when processing the sampling point with index k=2 in the sequence When , the previous point is , the latter point is , we can calculate: ,as well as , these two differences The elements in the difference set before and after the sampling point with index 2 are formed. Then, the module continues to process the sampling point with index k=3 , the previous point , a little later , calculated , , form the elements of the set of the differences before and after the sampling point with index 3 , and continue this process until all sampling points with index are processed , its previous point , and its next point , calculate to get , and form , after processing all sampling points within the allowed range, the module combines all calculated values in chronological order. For example, after processing the sequence , the forward difference sequence obtained is as follows: = = = , This sequence is the required current difference sequence, and this sequence is then passed to the mutation starting point recognition sub-module to obtain the current difference sequence.

[0019] Based on the current difference sequence, the mutation starting point recognition sub-module determines whether each difference exceeds the jump current threshold, filters out the sequence segments where consecutive differences all exceed the jump current threshold, locates the starting position of the sequence segment as the mutation starting point, and obtains the current mutation starting point index sequence; Based on the current difference sequence, for example , the mutation starting point recognition sub-module first needs to set a jump current threshold . The setting of this threshold is based on the statistical analysis of the current fluctuation range caused by internal chemical reaction fluctuations, sensor noise, and minor load changes of this type of battery pack under normal operating conditions. Usually, the upper limit of the fluctuation in the 99% confidence interval of historical data is selected as a reference. For example, if the analysis shows that the normal fluctuation is usually within , then can be set. After setting this threshold, the module checks each difference in the sequence one by one, and performs a judgment operation: compare with . For the sequence and the threshold , the judgment result sequence is [False, False, True, False, False, True, False, False] (corresponding to the values ). Next, the module filters out the segments in the above judgment result sequence that are continuously True. In this example, there are only two individual True values, at index 2 (corresponding to the difference 4.5) and index 5 (corresponding to the difference -3.3), and there is no continuously True sequence segment. Suppose we have another difference sequence , using the same threshold , the judgment result is [False, False, True, True, True, False, False]. At this time, there is a continuous sequence segment of True, and the indexes of the original difference sequence covered by it are 2, 3, 4 (corresponding to the differences 2.5, 3.1, 2.8). The module locates the starting position of this continuous True sequence segment, that is, index 2, and records this starting index. If there are multiple such continuous segments, record the starting index of each continuous segment. For example, if the judgment result is [F, T, T, F, T, T, T, F], two starting indexes, 1 and 4, will be recorded. Returning to our first example and the judgment result [F, F, T, F, F, T, F, F]. Since there is no paragraph with a continuous difference exceeding the threshold, but the original description requires screening "sequence segments where the continuous differences all exceed the threshold", this may imply that a single point exceeding the threshold is also counted as a "continuous segment" of length 1, or the threshold or processing logic needs to be adjusted. Assuming the logic is to find any point exceeding the threshold as a potential starting point, then in this example, indexes 2 and 5 will be recorded, forming a list of mutation starting point indexes [2, 5]. (Please note: If "continuous" is strictly required, the result of this example is an empty list. To enable the subsequent steps to proceed, we assume that a single point exceeding the threshold is also marked, or the threshold is adjusted to , then the judgment result is [F, F, T, F, F, T, F, F], the absolute values of the differences are [0.1, 0.3, 4.5, 0.1, 0.2, 3.3, 0.2, 0.1], and the indexes exceeding the threshold are 2 and 5, and there is still no continuous segment. The following will start from indexes 2 and 5), and finally output these recorded indexes to obtain the current mutation starting point index sequence.

[0020] The mutation buffer segment construction sub-module, according to the current mutation starting point index sequence, starts from each mutation starting point, combines the difference sequence in the subsequent cycles, judges whether the difference gradually converges to zero, records the end point of the convergence interval as the end point of the buffer segment, and records the maximum current, minimum current and cycle length within the interval to obtain the mutation buffer segment data set; According to the current mutation starting point index sequence, such as [2, 5], and the original current difference sequence and the original current sequence: ; Process each starting index one by one. Process the starting index 2. Starting from this point (that is, the difference ), combine the subsequent difference sequence , judge whether the difference gradually converges to zero. The judgment criterion for convergence is whether the absolute value of the difference is less than a preset convergence threshold , and this threshold is usually set as the jump current threshold A smaller proportion, such as If , then: Now check the difference after index 2: Since the first subsequent difference is already less than the convergence threshold, indicating that the current change has quickly stabilized, record the end index of the convergence interval as 3. The original difference sequence indices covered by the buffer segment of this mutation event are from 2 to 3 (i.e., differences 4.5 and 0.1), and the corresponding original current sequence has indices 2, 3, 4 (because the difference ), so the current values involved are . The module then looks for the maximum current and the minimum current within this interval and records the period length corresponding to this buffer segment, that is, the number of sampling points involved, as 3 periods (or time length seconds, depending on the definition). Then process the next mutation start index 5, with the corresponding difference being . The subsequent difference sequence is . Check the subsequent differences: , which is already less than the convergence threshold . Record the end index of the convergence interval as 6. The original difference indices covered by this buffer segment are from 5 to 6 (differences -3.3 and -0.2), and the corresponding original current sequence has indices 5, 6, 7, and the current values involved are . Look for the maximum current and the minimum current within this interval and record the period length as 3 periods (or 2 seconds). Store the data obtained from each processing, including the start index, end index, maximum current within the interval, minimum current, and period length, as a data structure. For example, form records: (start_idx = 2, end_idx = 3, I_max = 15.1, I_min = 10.5, length = 3); (start_idx = 5, end_idx = 6, I_max = 15.3, I_min = 11.8, length = 3); All these records together constitute the mutation buffer segment data set for use by subsequent modules to obtain the mutation buffer segment data set.

[0021] Table 1 Mutation Buffer Segment Data Table Starting index Ending index Maximum current (A) Minimum current (A) Cycle length 2 3 15.1 10.5 3 5 6 15.3 11.8 3 As shown in Table 1, this table lists the specific data of two buffer segments identified by the mutation buffer segment construction sub-module, including the start and end indices in the original difference sequence, the maximum and minimum current values within the corresponding original current sequence interval, and the number of sampling points included in the buffer segment.

[0022] Please refer to Figure 3 , the hysteresis correlation module includes: The voltage sequence extraction sub-module extracts a continuous voltage data segment after the jump end time point from the original time sequence based on the marked jump end time points in the mutation buffer segment data set, and obtains the voltage change sequence; Based on the aforementioned mutation buffer segment data set, such as the data shown in Table 1, this sub-module needs to access the original voltage time sequence V collected and stored synchronously with the current sequence , assuming the corresponding voltage sequence is Volts (V), and this sequence corresponds exactly to the time stamps of the current sequence . Taking the first record (start_idx = 2, end_idx = 3,...) in the mutation buffer segment data set as an example, the end time point of the jump (the current change tends to be stable) marked by this record corresponds to the index 3 of the original difference sequence, which corresponds to the index 4 of the original current / voltage sequence (because is associated with , and the buffer segment ends when the difference is stable, that is relative to the change is small), so the corresponding voltage time point is index 4, that is . The module starts from after this end time point (index 4) and extracts a continuous voltage data segment. The length of the extracted data segment is a preset parameter, for example, set to sampling points, then extracts voltage values starting from index 4 + 1 = 5, that is , and the corresponding voltage values are . This sequence constitutes the voltage change sequence associated with the first mutation event, and then processes the second record (start_idx = 5, end_idx = 6,...). Its end time point corresponds to the original difference sequence index 6, that is, the original voltage sequence index 7, corresponding to the voltage . Starting from the next point (index 8), it extracts points, but since the original voltage sequence only goes to index 9, it can only extract , that is . If a fixed length is required, this event may be discarded or specially processed due to insufficient data. Assuming we allow extraction to the end, the extracted sequence is , store the voltage data segments successfully extracted each time to form a set containing multiple voltage change sequences for use by the next sub-module to obtain the voltage change sequences.

[0023] Based on the voltage change sequence, the response start point identification sub-module performs a difference operation on the voltage values at every two adjacent time points in the sequence, calculates the first-order derivative of the voltage and marks the derivative sign, and filters out the time point segments where the derivative values are continuously negative. The formula is used: ; Determine the response start time point , identify the interval length from the start time point to before the trend reversal to obtain the negative duration sequence. Among them, is the voltage value at the k-th moment, is the sampling time interval, is the sign function of the voltage derivative, n is the minimum number of periods for which the voltage derivative is continuously negative, is the k-th sampling time point, is the derivative negative trend threshold; based on the extracted voltage change sequence, the first sequence and the second sequence , the response start point identification sub-module processes each sequence. Taking as an example, the original index corresponding to this sequence is . First, perform a difference operation on the voltage values at every two adjacent time points in the sequence. Assuming the sampling time interval seconds, calculate the approximate value of the first-order derivative of the voltage . For , the calculated derivative sequence is . Then the module marks the sign of each derivative to obtain the sign sequence [1, 1, -1, -1]. Then filter out the time point segments where the derivative values are continuously negative. In this example, the continuously negative segment appears in the last two derivatives, corresponding to the derivatives at the original indices k = 7 and k = 8 (i.e., and ). Next, apply the formula to determine the response start time point . Explanation of the formula parameters: : The voltage response start time point to be found, : The k-th sampling time point in the voltage change sequence, The voltage value at the time point , : Sampling time interval (unit: second s). For example, : The voltage change rate at the time point (approximate value of the first-order derivative, unit: volt per second V / s), : Sign function of the voltage change rate, which is 1 when the voltage increases, -1 when it decreases, and 0 when it remains unchanged. : This term is actually equal to , that is, the voltage change rate itself. n: The window size for calculating the average voltage change rate, which means how many consecutive cycles' average change rates need to meet the conditions. Set n = 2. This value is set based on the understanding of the battery voltage response characteristics, believing that at least two consecutive sampling points' negative trend is considered an effective response starting point to avoid single-point noise interference. Derivative negative trend threshold (unit: V / s), set This threshold is based on the evaluation of the voltage measurement noise level, and a value slightly larger than the normal fluctuation caused by noise is selected. For example, if the change rate caused by voltage noise is usually within , then choosing 0.005 V / s can effectively filter the noise. The summation symbol means starting from the current time point k, continuously accumulating the voltage change rates of n time points. means calculating the arithmetic mean of these n change rates. means requiring this average voltage change rate to be less than the negative threshold, that is, having a significant negative (descending) trend. means among all that meet the conditions, selecting the one with the earliest time (the smallest index) as the response starting point. , Formula logic: This formula finds the starting point where the voltage begins to continuously decline by calculating the sliding average of the voltage change rate. The sliding average can smooth the noise, n ensures the persistence of the trend, ensures the significance of the decline, and min ensures that the found is the very beginning position of the descending trend. The benefit of the formula is that by combining the sliding average and the negative threshold to judge the voltage response starting point, compared with the method that only relies on single-point derivative or simple threshold, it can more robustly identify the true voltage decline starting point caused by current jump, effectively avoiding misjudgment caused by noise spikes or short-term fluctuations; Numerical example: Using the derivative sequence , the derivative corresponding to the original voltage sequence index , set , when k = 5 (considering i = 5, 6), average derivative: , does not meet , when k = 6 (considering i = 6, 7), average derivative: , does not meet , when k = 7 (considering i = 7, 8), average derivative: , meets , so the smallest k that meets the conditions is 7, and the corresponding response starting time point is the time point at index 7 in the original voltage sequence , which indicates that at moment, the voltage begins to show a continuous and significant downward trend. This time point is considered the starting point of the voltage response after the current jump. Then, the module identifies the length of the interval where the voltage derivative is continuously negative starting from . The derivative sequence starts from index 7 and is . The length of being continuously negative is 2 cycles. The module records this negative duration, for example, stores it as [2]. Repeat this process for all the extracted voltage sequences to obtain a series of response start time points and the corresponding negative durations, and obtain the negative duration sequence.

[0024] The deviation data acquisition sub-module calculates the time interval between the response start time point sequence and the jump end time point data in the mutation buffer segment data set, and statistically calculates the difference between the time interval and the preset response time limit as the offset to obtain the jump response time deviation data; According to the response start time point sequence, for example, for the identified for the first mutation event, and the mutation buffer segment data associated with this event recorded in the "mutation buffer segment construction sub-module" (see the first row of Table 1), especially the jump end time point therein (corresponding to the original sequence index 4, denoted as ), the deviation data acquisition sub-module calculates the time interval between the two. In this example, , since the time point index represents the sampling sequence number and the sampling interval is , the time interval is seconds. Next, the module compares the calculated time interval with a preset response time limit and calculates the difference as the offset . The setting of the preset response time limit is based on the understanding of the electrochemical response speed of this type of battery and the requirements of the system application scenario. For example, for a system that requires a quick response, it may be required that the voltage starts to change significantly within 1 second after the current step stabilizes. Therefore, seconds can be set. This value can be obtained by analyzing the data distribution of a large number of normal operating conditions , for example, taking its 80% quantile, such as seconds, and performing the calculation: seconds. This offset represents the delay of the actually observed voltage response start time relative to the expected response time limit. For each identified and the corresponding All perform this calculation, count all the calculated offsets, form a list of offset data, such as [2.0,...] (if subsequent events are processed), and obtain the jump response time deviation data.

[0025] Please refer to Figure 4 , and the starting point confirmation module includes: Based on the jump response time deviation data, the voltage deviation judgment sub-module extracts the voltage value corresponding to each response starting point, obtains the voltage value of the previous cycle, calculates the difference, and compares the difference with the capacity change trigger threshold to determine whether it exceeds the capacity change trigger threshold, and obtains the trigger valid index sequence; Based on the jump response time deviation data and the associated response start time point (for example ) and the original voltage sequence , the voltage deviation judgment sub-module first extracts each corresponding voltage value , for , extract , then, the module also needs to obtain a voltage value of the "previous cycle" as a comparison reference. Here, the "previous cycle" refers to the stable state voltage before the current jump that triggers this response. One way is to take the voltage at the point before the starting point of the mutation buffer section data (the first row of Table 1, start_idx = 2) corresponding to this current jump, that is , take this voltage as , calculate the voltage difference , then, take the absolute value of this difference and compare it with a preset capacity change trigger threshold , The setting of [the threshold] takes into account the voltage platform characteristics of the battery in a specific SOC range and the influence of capacity attenuation on the voltage response amplitude. For example, when the capacity drops significantly, the same current change may cause a greater voltage drop. This threshold can be determined through experiments or simulations to determine how much voltage response deviation usually indicates a capacity change that needs attention. Assume that for the current working condition, set , make a judgment , compare whether 0.03 is greater than The result is no , therefore, this response event is not considered a valid trigger. If the calculated is greater than , for example, assume , then , , at this time 0.06 > 0.05, the judgment is true, then the time index of this response starting point Recorded into the trigger valid index sequence, the module repeats this process for all response events, screening out all indexes that meet the conditions to obtain the trigger valid index sequence.

[0026] The response point binding sub-module combines the trigger valid index sequence, extracts the voltage value, current value, and state of charge information at the corresponding time points, performs synchronous recording and joint marking to obtain the state of charge mapping data group; Combined with the trigger valid index sequence, assuming that an effective index list has been obtained after screening, such as [t_{15}, t_{28}] (using a new index example to avoid confusion with the previous example), and the original synchronously acquired voltage sequence V, current sequence and the state of charge (SOC) sequence S estimated by the battery management system (BMS), the response point binding sub-module processes each time point index in the effective index list in turn. For the first effective index , the module extracts the data corresponding to this time point from the three original sequences: the voltage value , the current value , and the state of charge value , and queries to obtain: , synchronously record and jointly mark these three values with the timestamp to form a data unit, such as (timestamp=t_{15}, voltage=3.70, current=18.5, soc=55.8), and process the next effective index , to obtain (timestamp=t_{28}, voltage=3.68, current=19.0, soc=52.3), and assemble the data units extracted from all effective time points to form a data set, where each element contains the key state parameters at the same moment, to obtain the state of charge mapping data group.

[0027] The sampling data generation sub-module binds and arranges the combined voltage value, current value, and state of charge information according to the state of charge mapping data group, constructs a standard data structure and unifies the recording format to obtain the capacity sampling data packet; According to the state of charge mapping data group, for example, it contains: (timestamp=t_{15}, voltage=3.70, current=18.5, soc=55.8); (timestamp=t_{28}, voltage=3.68, current=19.0, soc=52.3); Based on the recorded data groups, the sampling data generation sub-module binds and organizes these combined voltage values, current values, and state of charge information. This is mainly to ensure the unity and standardization of the data format for subsequent processing by analysis modules or storage systems. This may involve organizing the data into specific data structures, such as an array of JSON objects, a Pandas DataFrame, or database records. For example, a standard data structure can be defined, where each structure contains fields: timestamp (such as Unix timestamp or a specific format string), voltage (floating point number, unit V), current (floating point number, unit A, with positive and negative values possibly distinguished to represent charging and discharging), and state of charge (floating point number, unit %). The module will map each record in the data group into this standard structure. For example, the first record is converted to: {'timestamp':1678886415,'voltage':3.70,'current':-18.5,'soc':55.8} (assuming the corresponding timestamp, and a negative current indicates discharging); The second record is converted to: {'timestamp':1678886428,'voltage':3.68,'current':-19.0,'soc':52.3}; Organize all the converted standard data structures together to form an ordered (sorted by timestamp) data packet. This data packet contains all the state sampling points that are considered valid and reflect the battery's state under specific dynamic responses, obtaining the capacity sampling data packet.

[0028] Please refer to Figure 5 , the state diagnosis module includes: The rate judgment sub-module calculates the difference in the state of charge between consecutive sampling time points for each cycle based on the state of charge sequence during the recorded time period in the capacity sampling data packet. It determines whether the difference exceeds the state of charge rate mutation identification threshold, filters the cycle index positions that meet the conditions, and obtains the state of charge rate mutation cycle sequence; Based on the aforementioned capacity sampling data packet, which contains records arranged in chronological order, each record includes a timestamp, voltage, current, and state of charge (SOC). For example, the SOC sequence part extracted from the data packet is: ;

[0029] The rate judgment sub-module calculates the difference in the state of charge between consecutive sampling time points for each cycle (i.e., for each sampling time point) , for example, calculate , , …, , assuming , then , , Next, the module takes the absolute value of each difference and compares it with a preset recognition threshold for sudden changes in the charging rate . The setting of this threshold needs to consider the normal SOC estimation update rate and the possible SOC jump amplitude caused by model calibration, Coulomb efficiency compensation adjustment, or specific events (such as OCV calibration after a long-term static state). The goal is to identify abnormal SOC changes that far exceed the normal Coulomb integration change. For example, if the SOC changes by no more than 0.5% per second during normal operation, but occasional internal state refreshes of the BMS may cause jumps of up to 2%, then is set for judgment: , and all cycle index positions k that satisfy conditions are screened out. In this example, the satisfied index is k = 27. The module records these indexes and obtains the charging rate mutation cycle sequence

[27] .

[0030] According to the charging rate mutation cycle sequence, the key cycle extraction sub-module obtains the voltage value sequence at the corresponding cycle positions, performs sliding window processing with a fixed cycle width, constructs the difference set of adjacent cycle voltage values within each window, and uses the formula: ; Calculate the voltage fluctuation stability of the th cycle, screen the stability, extract the cycle positions less than the stability threshold value, and obtain the key cycle sequence of state mutation. Among them, is the voltage value of the th cycle, N is the number of cycles in the sliding window, is the absolute difference between adjacent voltage values; According to the charging rate mutation cycle sequence output by the previous sub-module, such as

[27] , and the voltage sequence V in the capacity sampling data packet, the key cycle extraction sub-module obtains the voltage value sequence near the corresponding position of the mutation cycle k = 27. For example, it extracts the voltage value sequence that expands several points before and after with k = 27 as the center (for example, a total of 10 points are extracted, from k = 22 to k = 31) , and then performs sliding window processing on this voltage segment with a fixed cycle width N. The sliding window width N = 5 cycles is set. The selection of this width N aims to balance noise smoothing and sensitivity to voltage dynamic changes, and usually selects the time length corresponding to several seconds, such as 5 seconds (if the sampling interval is 1 second). For each window, the difference set of adjacent cycle voltage values within the window is constructed and , where is the starting index of the window, and then the formula is applied to calculate the Voltage fluctuation stability over one cycle , Explanation of formula parameters: : Represents the voltage fluctuation stability index within a window of length N starting from the -th cycle. The smaller the value, the more stable the voltage. : The voltage value of the -th cycle (here U is used to be consistent with V in the previous text, representing voltage), N: The number of cycles of the sliding window (window size), for example, N = 5. : The absolute difference between the voltage value of the -th cycle within the window and its previous cycle. : Represents the sum of all terms from index to within the window. : Calculates the average value of the absolute differences of the voltages within the window, measuring the average fluctuation amplitude. : The square of the voltage difference within the window. : Calculates the square root of the sum of the squares of the root mean squares of the voltage differences within the window (similar to the standard deviation, but not divided by N - 1), measuring the overall energy or dispersion of the fluctuations, and is more sensitive to large jumps. Formula logic: This formula evaluates the voltage stability by combining the average absolute deviation of the voltage changes within the window and the L2 norm (not normalized) of the voltage changes. The first term focuses on continuous small fluctuations, and the second term focuses on large instantaneous fluctuations. The sum of the two gives a comprehensive stability index. The advantage of the formula is that it simultaneously considers the average amplitude and peak amplitude of the voltage fluctuations (through squaring and taking the square root), and can describe the stable state of the voltage more comprehensively than a single index (such as only using the average absolute difference or standard deviation), especially being able to distinguish the different effects of continuous small noise and short-term large disturbances on stability; Numerical example: Assume that the voltage sequence segment extracted near k = 27 is: ; (The index corresponds to ), calculate the stability starting from the window (the window contains , and the values from to to are required). Assume , calculate the difference within the window to : : i = 27: , i = 28: , i = 29: , i = 30: , i = 31: ; Sum of absolute differences: ; Average absolute difference: ; Sum of squared differences: ; Square root of the sum of squares: ; Calculate The calculation module filters all values calculated for the sliding window and extracts those cycle positions that are less than a preset stability threshold , and the setting of the stability threshold is based on the statistical analysis of the voltage fluctuation stability of the battery under stable operating conditions (such as standing still or constant current charge and discharge at a low rate). For example, if is usually less than 0.02 under stable conditions, it can be set with a certain margin for judgment: , which meets the condition, so the index is recorded. If the calculated for other windows is also less than , its starting index is also recorded. This result indicates that within the window starting from cycle k = 27 where the charging rate changes suddenly, the voltage shows high stability ( is less than the threshold), which may mean that the jump in SOC is not caused by a drastic change in the external operating conditions but by an internal state adjustment. This cycle is considered a critical cycle, and a sequence of critical cycles of state mutation is obtained.

[0031] The stable window marking sub-module performs a marking operation at each cycle position according to the sequence of critical cycles of state mutation, synchronously binds the state mutation cycle with the stable voltage cycles within the subsequent sliding window, establishes a cycle synchronization information structure, and obtains the state voltage synchronization marking cycle; According to the sequence of critical cycles of state mutation, such as including indices [27, 45], etc., the stable window marking sub-module performs a marking operation on each critical cycle index. Taking the critical cycle index as an example, the module marks it as a critical moment of state mutation. At the same time, it synchronously binds it with all cycles within the sliding window that was used in the calculation and was judged to be stable. In the example of the previous module, when calculating , the voltage values from to were used (window width N = 5), and the calculated is less than the stability threshold , indicating that this window is a set of stable voltage cycles. The module establishes an association between the critical cycle and this stable window , and an information structure can be created, for example: (key_period=t_{27}, stable_window=[t_{27},t_{28},t_{29},t_{30},t_{31}]); Do the same for all other indices in the key period sequence, such as ), and if is also less than , then create the corresponding record: (key_period=t_{45}, stable_window=[t_{45},...,t_{49}]); These information structures together constitute the cycle synchronization information, which links the key time points of the SOC rate mutation with a period of voltage stability immediately following (or including) it, and obtains the state voltage synchronization marker period.

[0032] Please refer to Figure 6 , the correction output module includes: The correction record incorporation sub-module extracts the state of charge values of the corresponding period based on the state voltage synchronization marker period, incorporates them into the correction sequence structure in turn, and attaches the current capacity trend node position to each correction value to establish a mapping index between the correction and the trend points, and obtains the capacity correction reference sequence; Based on the state voltage synchronization marker period information, such as including: (key_period=t_{27}, stable_window=[...]); (key_period=t_{45}, stable_window=[...]); For a set of records such as these, the correction record incorporation sub-module processes each record in turn. For the record (key_period=t_{27},...), extract the key period corresponding state of charge (SOC) value, which is obtained from the original capacity sampling data packet, such as (Note: Here, the value after the SOC jump should be used, that is, the value in the previous module example or itself, depending on the specific definition. Here, it is assumed to be the value after the jump , or just use the value of the record, assuming it is the value recorded in the record, assuming it is just the record in The jump value (52.3%) is used as a candidate correction value and incorporated into a correction sequence structure. Meanwhile, the module needs to attach an identifier of the current capacity trend node position to this correction value. This identifier indicates the stage at which this correction point is located in the overall battery capacity attenuation or change trend (e.g., "initial acceleration attenuation", "mid-term stability", "late-stage dive", etc.). The determination of this identifier may depend on longer-term historical capacity data and attenuation models. Assuming that according to historical analysis, the current battery is in the mid-term stable stage of capacity attenuation, the identifier "Mid-Stable" is attached, thus forming a correction reference record: (correction_soc=52.3,trend_pos='Mid-Stable',time=t_{27}); Perform the same operation on all other records in the set (such as the corresponding records), extract (such as 48.1%), determine its trend node position (such as still 'Mid-Stable'), and form a record: (correction_soc=48.1,trend_pos='Mid-Stable',time=t_{45}); All these records are combined to establish a mapping index between the candidate correction SOC values and their positions in the capacity attenuation trend, and a capacity correction reference sequence is obtained.

[0033] Based on the correction nodes in the capacity correction reference sequence and their corresponding initial capacity starting values, the capacity difference calculation sub-module calculates the numerical difference. Combining the voltage fluctuation stability and the voltage response fluctuations in the previous and subsequent cycles, the formula is used: ; Calculate the capacity correction amplitude of the battery pack in the j-th cycle , establish the relationship between the correction amplitude and the difference and change rate between nodes, and obtain the capacity correction amplitude sequence, where is the state of charge value in the j-th cycle, is the starting value of the capacity, and are the voltage values in the j-th cycle and the (j - 1)-th cycle respectively, is the voltage fluctuation stability in the j-th cycle, and m is the number of cycles for the trend continuation; Based on the obtained capacity correction reference sequence, for example, it contains: (correction_soc=52.3,trend_pos='Mid-Stable',time=t_{27}); (correction_soc = 48.1, trend_pos = 'Mid - Stable', time = t_{45}); sequences such as those recorded, and an initial capacity starting value is required This represents the reference SOC value for the current capacity evaluation period, or a certain reference point SOC in theory when there is no attenuation, depending on the specific correction strategy. Assume is the SOC estimation value at the start of this round of evaluation. For example , the capacity difference calculation sub - module processes each record in the correction reference sequence. Taking the record with j = 27 (time ) as an example, its corrected node SOC value is , and the module applies the formula to calculate the capacity correction amplitude of the battery pack for the j - th cycle Parameter description of the formula: : The capacity correction amplitude calculated for the j - th cycle, dimensionless, is a comprehensive indicator : The state - of - charge value for the j - th cycle (obtained from the correction reference sequence), expressed as a decimal. For example , : The starting capacity value (reference SOC value), expressed as a decimal. For example , : The voltage value for the j - th cycle, obtained from the original data. For example , : The voltage value for the (j - 1) - th cycle. For example , : The voltage fluctuation stability for the j - th cycle (calculated by the key cycle extraction sub - module). For example , and and subsequent values, m: The number of cycles for trend continuation, used to examine the change trend of voltage stability for a period of time after the correction point. Set m = 3. The setting of this value is based on how long after the correction point the stability change is hoped to be observed to assist in judging the reliability of the correction. For example, observe for 3 seconds : The absolute difference between the correction point SOC and the reference SOC, reflecting the deviation degree of SOC : The absolute change amount of voltage at the correction point, reflecting the voltage stability at that time : Calculate the sum of the absolute change amounts of voltage fluctuation stability for m cycles starting from the correction point j : Calculate the average rate of change of the computational stability S within m cycles after the correction point, which reflects whether the voltage stability is tending to be stable or still changing. The formula logic: This formula calculates the correction amplitude by summing three parts: the absolute deviation of the SOC, the voltage change at the correction moment, and the change trend of the voltage stability within a short period of time after the correction moment. A larger value means a large SOC deviation, or a voltage jump at the correction moment, or the voltage stability is still changing after the correction (which may indicate that the system is not fully stable). Considering these factors comprehensively to judge the necessity and amplitude of the correction. The benefit of the formula is that it not only considers the static deviation of the SOC ( ), but also incorporates dynamic information (voltage change ) and quasi-static information (stability change trend ), enabling the calculation of the correction amplitude to take into account the system state stability of the measurement point itself and for a period of time after it, improving the robustness of the correction decision; Numerical example: Calculate , given , , , , it is necessary to , , (because m = 3, k = 27, 28, 29 are required, involving , , ). Assume that through calculation, is obtained. The first part: , the second part: , the third part: Calculate , k = 27: , k = 28: , k = 29: ; Summation: ; Average: ; Calculate module calculates the corresponding value for all records in the correction reference sequence, forming a correction amplitude sequence, such as [0.06050,...]. This result is a quantization index, with a relatively small value, mainly contributed by the SOC deviation, indicating that although there is a deviation in the SOC, the voltage at the correction point is stable and the subsequent stability change is small. It is possible to determine the actual correction weight or whether to perform the correction based on the size of the value. For example, set a threshold . Because , it is considered necessary to perform the correction, obtaining a capacity correction amplitude sequence.

[0034] The adjustment record generation sub-module combines each capacity correction magnitude with the corresponding state of charge and trend identifier according to the node positioning information in the capacity correction magnitude sequence and the capacity correction reference sequence, and obtains the capacity adjustment record of the battery pack; According to the calculated capacity correction magnitude sequence, such as including , etc., and the capacity correction reference sequence established in the "correction record inclusion sub-module", which includes the state of charge (SOC) value and trend identifier information corresponding to each correction point, such as: (correction_soc=52.3,trend_pos='Mid-Stable',time=t_{27}); The adjustment record generation sub-module combines these two parts of information. For the time point , the module combines the calculated correction magnitude with the corresponding state of charge and the trend identifier 'Mid-Stable' to form a complete adjustment record, which clarifies at which time point, based on what correction driving force (magnitude), what the current SOC reading is, and which stage of capacity change it is in. For example, the record can be generated as: (time=t_{27},adjustment_magnitude=0.06050,soc_at_point=52.3,trend_phase='Mid-Stable'); This operation is performed on each value in the capacity correction magnitude sequence and its corresponding reference sequence information, and all the generated records are collected. This set is the final output, which can be used as input by the capacity estimation algorithm in the BMS to decide how to adjust the current battery capacity (SOH) estimation or SOC estimation model parameters, and obtain the capacity adjustment record of the battery pack.

[0035] The above is only the preferred embodiment of the present invention, and does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still belong to the protection scope of the technical solution of the present invention.

Claims

1. An intelligent online capacity management system for battery packs, characterized in that: The system comprises: The jump sensing module obtains the current sampling sequence within the operation cycle of the battery pack, selects the starting point of the current mutation, tracks the interval where the difference of each cycle converges to zero after the mutation starting point, and generates a mutation buffer segment data set; The hysteresis association module extracts the continuous voltage data after the jump end time point based on the mutation buffer segment data set, selects the response starting point to count the duration of the negative trend, and compares the response time limit to obtain the jump response time deviation data; The starting point confirmation module extracts the voltage value corresponding to the response starting point based on the jump response time deviation data, determines whether the difference with the voltage value of the previous cycle exceeds the capacity change trigger threshold, counts the current and voltage values ​​corresponding to the response point, and obtains the capacity sampling data packet; The state diagnosis module calculates the periodic state of charge difference based on the capacity sampling data packet, locates the key period of state mutation, analyzes whether the subsequent voltage difference sets are all less than the stable threshold value, and generates a state voltage synchronization mark period if it is satisfied; The correction output module incorporates the corresponding state of charge value into the core capacity correction record sequence according to the state voltage synchronization mark period, calculates the capacity correction amplitude, and obtains the battery pack capacity adjustment record.

2. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The mutation buffer segment data set includes the current upper limit, the current lower limit, and the cycle length; the jump response time deviation data includes the response starting point, the voltage change trend, and the duration of the negative trend; the capacity sampling data packet includes the current value, the voltage value, and the state of charge; the state voltage synchronization mark period includes the state mutation key period, the voltage difference set, and the stability judgment mark; the capacity adjustment record includes the state of charge value, the capacity trend node correction source, and the capacity correction amplitude.

3. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The transition sensing module comprises: The difference extraction submodule selects the current value of each sampling point and its adjacent points in turn based on the current sampling sequence in the battery pack operation cycle, calculates the difference between the current sampling point and the previous sampling point and the difference between the current sampling point and the next sampling point, obtains the difference set before and after each sampling point, and obtains the current difference sequence; The mutation starting point identification submodule determines whether each difference exceeds the jump current threshold according to the current difference sequence, selects the sequence segment whose continuous difference exceeds the jump current threshold, locates the starting position of the sequence segment as the mutation starting point, and obtains the current mutation starting point index sequence; The mutation buffer segment construction submodule starts from each mutation starting point according to the current mutation starting point index sequence, combines the difference sequence in the subsequent cycle, determines whether the difference gradually converges to zero, records the end point of the convergence interval as the buffer segment termination point, records the maximum current value, the minimum current value and the cycle length in the interval, and obtains the mutation buffer segment data set.

4. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The hysteresis association module comprises: The voltage sequence extraction submodule extracts the continuous voltage data segments after the transition end time point from the original time series based on the transition end time point marked in the mutation buffer segment data set, and obtains the voltage change sequence; The response starting point identification submodule performs differential processing on the voltage values ​​of each two adjacent time points in the sequence based on the voltage change sequence, calculates the first-order derivative of the voltage and marks the derivative sign, and selects the time point segments where the derivative value is continuously negative, using the formula: ; Determine the response start time , identify the interval length from the starting time point to the trend reversal, and obtain the negative duration sequence, where, is the voltage value at the kth moment, is the sampling time interval, is the sign function of the voltage derivative, n is the minimum number of cycles in which the voltage derivative is continuously negative, is the kth sampling time point, is the negative trend threshold of the derivative; The deviation data acquisition submodule calculates the time interval between the response start time point sequence and the jump end time point data in the mutation buffer segment data set, and calculates the difference between the time interval and the preset response time limit as an offset to obtain the jump response time deviation data.

5. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The starting point confirmation module includes: The voltage deviation judgment submodule extracts the voltage value corresponding to each response starting point based on the jump response time deviation data, obtains the voltage value of the previous cycle, calculates the difference, and compares the difference with the capacity change trigger threshold to determine whether it exceeds the capacity change trigger threshold and obtain the trigger valid index sequence; The response point binding submodule combines the trigger valid index sequence to extract the voltage value, current value and charge state information at the corresponding time point, performs synchronous recording and joint marking, and obtains the charge state mapping data group; The sampling data generation submodule binds and organizes the combined voltage value, current value and charge state information according to the charge state mapping data group, constructs a standard data structure and unifies the recording format, and obtains a capacity sampling data packet.

6. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The status diagnosis module comprises: The rate judgment submodule calculates the charge state difference between consecutive sampling time points period by period based on the charge state sequence of the time period recorded in the capacity sampling data packet, determines whether the difference exceeds the charge rate mutation recognition threshold, selects the period index position that meets the conditions, and obtains the charge rate mutation period sequence; The key cycle extraction submodule obtains the voltage value sequence of the corresponding cycle position according to the charging rate mutation cycle sequence, performs sliding window processing according to the fixed cycle width, and constructs the difference set of adjacent cycle voltage values ​​in each window, using the formula: ; Calculate the Voltage fluctuation stability per cycle , the stability is screened, the period position less than the stability threshold is extracted, and the key period sequence of state mutation is obtained, where: For the The voltage value of a cycle, N is the number of cycles of the sliding window, is the absolute difference between adjacent voltage values; The stable window marking submodule performs a marking operation at each cycle position according to the state mutation key cycle sequence, synchronously binds the state mutation cycle with the stable voltage cycle in the rear sliding window, establishes a cycle synchronization information structure, and obtains the state voltage synchronization marking cycle.

7. The intelligent online capacity management system for storage battery packs according to claim 1 is characterized in that: The correction output module comprises: The correction record incorporation submodule extracts the state of charge value of the corresponding period based on the state voltage synchronization mark period, and sequentially incorporates it into the correction sequence structure, and adds the current capacity trend node position to each correction value, establishes a mapping index between the correction and trend points, and obtains a capacity correction reference sequence; The capacity difference calculation submodule calculates the value difference based on the correction node in the capacity correction reference sequence and its corresponding initial capacity starting point value, and uses the formula: ; Calculate the battery capacity correction amplitude for the jth cycle , establish the relationship between the correction amplitude and the difference between nodes and the change rate, and obtain the capacity correction amplitude sequence, where, is the state of charge value of the jth cycle, is the capacity starting value, and are the voltage values ​​of the jth cycle and j-1th cycle respectively, is the voltage fluctuation stability in the jth cycle, and m is the number of trend continuation cycles; The adjustment record generation submodule combines each capacity correction amplitude with the corresponding state of charge and trend identifier according to the node positioning information in the capacity correction amplitude sequence and the capacity correction reference sequence to obtain the battery pack capacity adjustment record.

Images