Fault diagnosis method and system for power energy storage system

Abstract

The present application relates to the technical field of fault diagnosis, in particular to a power energy storage system fault diagnosis method and system, comprising the following steps: collecting single voltage and current to generate discharge voltage time sequence record, calculating adjacent voltage difference to generate voltage back-off sequence, comparing same position difference to identify difference position and generate slope offset record, dividing time length section to compare offset position and generate mapping record, judging offset position to establish corresponding relationship and generate energy storage system diagnosis, in the present application, multi-dimensional data is collected to establish time sequence record and generate change sequence, time sequence coupling relationship between parameters is deeply mined, difference position is identified to generate offset record, early weak performance degradation characteristics are effectively captured, time length section is divided to establish position mapping, information processing delay caused by massive data is eliminated, position is directly judged to establish deep corresponding relationship, breaking the conventional limitations of fixed threshold setting, greatly enhancing the accuracy of equipment state determination, improving overall reliability.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis technology, and in particular to fault diagnosis methods and systems for power storage systems. Background Technology

[0002] The field of fault diagnosis technology involves a technical system for detecting, analyzing, and determining the operating status of electrical equipment, energy storage devices, and related power systems. Its core aspects include operating parameter acquisition, status feature extraction, abnormal state identification, and fault type discrimination. It is widely used in power systems, new energy systems, and electrochemical energy storage systems. This technical field typically involves continuous monitoring of voltage, current, temperature, and operating status information, combined with analysis of changes in electrical characteristics to determine whether equipment exhibits abnormal operating conditions or potential faults, thus forming a systematic diagnostic technology system for the operating status of power equipment. Traditional fault diagnosis methods for power energy storage systems refer to methods for identifying and judging abnormal states that may occur during the operation of battery energy storage systems or energy storage units. Typically, voltage acquisition circuits, current sampling resistors, and temperature sensors are arranged in the energy storage battery pack and its connection circuits to collect the voltage, charging and discharging current, and surface temperature of each battery cell in real time. The collected voltage difference, current change, and temperature change are compared with preset thresholds one by one. When the voltage of a single cell deviates from the set voltage range or the temperature exceeds the set temperature value, it is judged as an abnormal state. At the same time, the location of abnormal batteries is determined by detecting the voltage of each battery cell and measuring the current in the charging and discharging circuit. Based on the degree of voltage imbalance, current fluctuation, and temperature change trend, the overcharge, over-discharge, loose connection, or battery degradation states of the energy storage system are classified and judged.

[0003] Existing conventional solutions collect individual operating parameters in real time and compare the changes with preset thresholds to determine abnormal states. However, this comparison method often ignores the deep temporal coupling between multi-dimensional parameters, making it difficult to accurately capture early and subtle performance degradation phenomena. The method of detecting each individual parameter is prone to heavy data processing delays when dealing with large-scale battery pack data. Fixed threshold settings make the classification and determination of equipment status lack flexibility and accuracy when facing variable dynamic conditions, which seriously weakens the efficiency and reliability of the overall equipment operating status systematic diagnosis. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a fault diagnosis method for power storage systems, comprising the following steps: S1: Collect the individual cell voltage, circuit current, and sampling time of the series circuit of the energy storage battery module, filter the individual cell voltages in the same current range, arrange the individual cell voltages corresponding to the sampling time, establish the correspondence between the sampling time and the individual cell voltage, and generate a discharge voltage timing record. S2: Calculate the voltage difference of individual cells at adjacent sampling times in the discharge voltage time sequence record, arrange the differences and establish the correspondence between sampling times and differences to generate a voltage drop change sequence; S3: Compare the voltage drop change sequence with the difference at the same position to obtain the difference sequence, sort the difference sequence and identify continuous difference positions, establish the correspondence between continuous difference positions and sampling time, and generate slope offset segment records. S4: Divide the discharge voltage timing record into discharge duration segments and establish a front segment interval, a middle segment interval, and a rear segment interval. Compare the slope offset segment record positions and establish a correspondence between the front segment interval, the middle segment interval, and the rear segment interval and the slope offset segment record. Generate segment position mapping records. S5: Determine the offset section position of the section position mapping record and establish the correspondence between internal resistance anomaly, electrode activity decay, conduction path anomaly and energy storage unit number, and generate power energy storage system fault diagnosis results.

[0005] As a further aspect of the present invention, the discharge voltage timing record includes a sampling time sequence, a set of corresponding individual voltage values, and a current interval identifier; the voltage drop change sequence includes a set of voltage differences between adjacent times, an index of the time corresponding to the difference, and an identifier of the difference change amplitude; the slope offset segment record includes a set of continuous difference positions, a corresponding sampling time segment, and a difference sorting identifier; the segment position mapping record includes an identifier for the first segment, a identifier for the middle segment, and an identifier for the last segment; and the fault diagnosis result of the power storage system includes an identifier for abnormal internal resistance, an identifier for electrode activity decay, and an identifier for abnormal conduction path.

[0006] As a further aspect of the present invention, the continuous difference position refers to the set of sampling positions in the difference sequence where the difference value of adjacent sampling points continuously exceeds a preset threshold and appears continuously in the time series.

[0007] As a further aspect of the present invention, the slope offset segment refers to the voltage change slope being modified relative to the historical average voltage change trend, formed by the sampling time range corresponding to the continuous difference positions.

[0008] As a further aspect of the present invention, the specific steps of S1 are as follows: S101: Acquire the individual cell voltage sequence, loop current sequence, and sampling time sequence of the series circuit of the energy storage battery module. Perform current interval comparison on the loop current corresponding to the individual cell voltage sequence. Keep the individual cell voltage records that fall into the same current interval and remove the records outside the interval. Organize the records according to the current interval label to obtain the current interval voltage record set. S102: According to the current interval voltage record set, call the corresponding sampling time sequence, sort the sampling times in ascending order of time and synchronously adjust the corresponding individual voltage position, perform index binding on the sorted sampling times and the corresponding individual voltage, and establish a time-series voltage lookup table. S103: Based on the time-series voltage lookup table, call the sampling time value and the individual cell voltage value to perform key-value pairing, associate the sampling time key-value item with the individual cell voltage key-value item one by one and connect them in time order to generate a discharge voltage time-series record.

[0009] As a further aspect of the present invention, the specific steps of S2 are as follows: S201: Obtain the discharge voltage time sequence record, retrieve the individual voltage value corresponding to the sampling time, extract the voltage value pairs of adjacent sampling times in time sequence, calculate and record the difference value for each voltage value pair, organize the record order according to the sampling time index, and obtain the adjacent voltage difference sequence. S202: According to the adjacent voltage difference sequence, call the corresponding sampling time index, sort and verify the difference records according to the increasing relationship of sampling time, establish an adjacency relationship between the sampling time numbers of the previous and next samples and bind the corresponding difference records, and establish a sampling adjacency index table; S203: Based on the sampling adjacency index table, call the adjacent sampling time number and the difference record to perform key-value association, connect the adjacent sampling time number and the difference record in time order to form a sequence structure, and generate a voltage drop change sequence.

[0010] As a further aspect of the present invention, the specific steps of S3 are as follows: S301: Obtain the voltage drop change sequence, retrieve the sequence position difference value, perform item-by-item comparison operation on the same sequence position difference value, calculate the difference amplitude of adjacent position difference and record the difference sign, and arrange all comparison results according to the sequence position number to obtain the voltage difference sequence. S302: According to the voltage difference sequence, call the position difference value, sort the difference value and rearrange it according to the sequence position index, perform continuity judgment on the sorted position number, extract the continuous number segment and record the start and end position number of the segment, and generate a continuous difference segment table. S303: Based on the continuous difference segment table, call the segment start and end position numbers, retrieve the corresponding sampling time number and perform key-value association, establish a mapping structure between the continuous difference segment number and the corresponding sampling time number, organize and connect them in chronological order, and generate slope offset segment records.

[0011] As a further aspect of the present invention, the specific steps of S4 are as follows: S401: Obtain the slope offset segment record, retrieve the corresponding sampling time number, extract the start sampling time and end sampling time from the discharge voltage time sequence record and calculate the discharge duration value, organize the position numbers in time order, and obtain the discharge duration position sequence. S402: Based on the discharge duration position sequence, call the discharge duration value, perform interval division on the position number, split the discharge duration sequence into a front interval number set, a middle interval number set, and a rear interval number set, and generate a discharge segment number table. S403: Based on the discharge segment number table, call the segment number set, retrieve the corresponding position number of the slope offset segment record, associate the first segment interval, the middle segment interval, and the last segment interval with the slope offset segment position using key-value pairs, and organize them in chronological order to generate a segment position mapping record.

[0012] As a further aspect of the present invention, the specific steps of S5 are as follows: S501: Obtain the segment location mapping record, retrieve the correspondence between the slope offset segment location number and the set of front segment interval numbers, the set of middle segment interval numbers, and the set of rear segment interval numbers, perform segment ownership judgment on the location number and organize it according to the energy storage unit number to obtain the offset segment location table. S502: According to the offset section location table, call the energy storage cell number and section assignment result, perform fault type discrimination on the section assignment result, mark the front section as internal resistance abnormality, mark the middle section as electrode activity decay, mark the rear section as conduction path abnormality, and organize them according to the energy storage cell number to generate a fault type identification table. S503: Based on the fault type identifier table, call the energy storage cell number and fault type identifier, perform key-value association between internal resistance abnormality, electrode activity degradation, conduction path abnormality and energy storage cell number, and organize them in numerical order to generate power energy storage system fault diagnosis results.

[0013] A fault diagnosis system for power storage systems includes: The discharge timing construction module is used to implement S1: collect the individual cell voltage, loop current and sampling time of the series circuit of the energy storage battery module, filter the individual cell voltages in the same current range, arrange the individual cell voltages corresponding to the sampling time and establish the correspondence between the sampling time and the individual cell voltage, and generate the discharge voltage timing record. The voltage drop analysis module is used to implement S2: calculate the voltage difference of individual cells at adjacent sampling times of the discharge voltage time sequence record, arrange the differences and establish the correspondence between sampling times and differences, and generate a voltage drop change sequence; The slope offset identification module is used to implement S3: compare the difference between the same position in the voltage drop change sequence to obtain the difference sequence, sort the difference sequence and identify continuous difference positions, establish the correspondence between continuous difference positions and sampling time, and generate slope offset segment records. The segment mapping and positioning module is used to implement S4: dividing the discharge voltage timing record into discharge duration segments and establishing a front segment interval, a middle segment interval, and a rear segment interval; comparing the position of the slope offset segment record and establishing a correspondence between the front segment interval, the middle segment interval, and the rear segment interval and the slope offset segment record; and generating a segment position mapping record. The fault diagnosis output module is used to implement S5: determine the offset section position of the section position mapping record and establish the correspondence between internal resistance abnormality, electrode activity decay, conduction path abnormality and energy storage unit number, and generate the fault diagnosis result of the power energy storage system.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by collecting individual unit voltage and current and sampling time to establish time-series records and generate voltage drop change sequences, the deep temporal coupling relationship between multi-dimensional parameters is deeply explored. By comparing the differences at the same position to obtain the difference sequence and identifying continuous difference positions to generate slope offset segment records, the invention effectively captures early and weak performance degradation characteristics, divides discharge duration segments and compares the positions of mapped segments, eliminates the information processing delay caused by detecting large-scale data one by one, directly judges the position of offset segments to establish the correspondence between various anomalies, breaks the limitations of fixed threshold settings, greatly enhances the flexibility and accuracy of equipment status classification and judgment, and comprehensively improves the overall efficiency and operational reliability of fault diagnosis in power energy storage systems. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0018] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0019] Please see Figure 1 This invention provides a fault diagnosis method for a power energy storage system, comprising the following steps: S1: Collect the individual cell voltage, circuit current, and sampling time through the series circuit of the energy storage battery module. Filter the individual cell voltages in the same current range, arrange the filtered individual cell voltages in the order of sampling time, establish the correspondence between sampling time and individual cell voltage, and generate a discharge voltage timing record. S2: Based on the discharge voltage time sequence record, perform difference calculation on the individual cell voltage at adjacent sampling times, arrange the difference results according to the sampling time order and establish the correspondence between adjacent sampling times to generate a voltage drop change sequence; S3: Based on the voltage drop change sequence, perform a difference comparison at the same location to obtain a difference sequence, sort the difference sequence and identify continuous difference positions, establish a mapping relationship between continuous difference positions and sampling time, and generate slope offset segment records. S4: Based on the slope offset segment record, the discharge duration corresponding to the discharge voltage time sequence record is divided into segments, and the mapping relationship between the first segment interval, the middle segment interval and the last segment interval and the slope offset segment position is established to generate the segment position mapping record. S5: Based on the segment location mapping record, the location of the offset segment is judged, the correspondence between internal resistance anomaly, electrode activity decay, conduction path anomaly and energy storage cell number is established, and the fault diagnosis result of the power energy storage system is generated.

[0020] The discharge voltage time sequence record includes the sampling time sequence, the corresponding single-cell voltage value set, and the current interval identifier. The voltage drop change sequence includes the voltage difference set between adjacent times, the time index corresponding to the difference, and the difference change amplitude identifier. The slope offset segment record includes the continuous difference position set, the corresponding sampling time segment, and the difference sorting identifier. The segment position mapping record includes the first segment interval identifier, the middle segment interval identifier, and the last segment interval identifier. The power storage system fault diagnosis results include internal resistance abnormality identifier, electrode activity decay identifier, and conduction path abnormality identifier.

[0021] Please see Figure 2 The specific steps of S1 are as follows: S101: Acquire the individual cell voltage sequence, loop current sequence, and sampling time sequence of the series circuit of the energy storage battery module. Perform current interval comparison on the loop current corresponding to the individual cell voltage sequence. Keep the individual cell voltage records that fall into the same current interval and remove the records outside the interval. Organize the records according to the current interval label to obtain the current interval voltage record set. The system uses a high-precision Hall current sensor and multiple voltage acquisition channels to read the individual cell voltage sequence, loop current sequence, and sampling time sequence in the series circuit of the energy storage battery module. The individual cell voltage sequence is defined as a data set containing several consecutive floating-point numbers, the loop current sequence as a corresponding set of floating-point numbers of the same length, and the sampling time sequence as a set of long integers with absolute timestamps. These three sets of sequences are imported into a temporary data buffer. For the individual cell voltage sequence, a missing value imputation operation is performed. Elements in the sequence are scanned one by one. If a voltage value at a certain position is found to be null, the preceding and following valid voltage values ​​at that position are extracted, summed, and the result is divided by 2. The average value is then directly filled into the null position. The same missing value imputation operation is performed for the loop current sequence. Preset target stable discharge current lower and upper threshold values ​​are retrieved. The target stable discharge current lower threshold is set to 45 amps, and the target stable discharge current upper threshold is set to 55 amps. Construct an empty temporary record list, setting the initial index number to 1, and extract the first current value from the loop current sequence. Compare this current value with the lower threshold of 45 amps. If the extracted current value is greater than or equal to 45 amps, continue comparing it with the upper threshold of 55 amps. If the current value is simultaneously less than or equal to 55 amps, the data corresponding to this index number is considered to meet the interval condition. If the condition is met, extract the corresponding individual voltage value, loop current value, and sampling time value from the individual voltage sequence, loop current sequence, and sampling time sequence, respectively, based on the current index number. Combine these three values ​​into a complete data record entry and append this entry to the end of the temporary record list. If the extracted current value is less than 45 amps or greater than 55 amps, discard the data corresponding to this index number without performing any append operation. Increment the index number by 1 and repeat the above extraction and comparison operations until all values ​​in the loop current sequence have been traversed. All entries in the temporary record list are extracted and rearranged according to the classification requirements of the current range labels, integrating all records into a current range voltage record set. For example, if the current value read at the 150th sampling time is 48 amps and the single-cell voltage value is 3.2 volts, since 48 amps falls between 45 amps and 55 amps, this set of values ​​is fully retained and recorded. If the current value read at the 151st sampling time is 42 amps, which does not meet the lower limit requirement, the data corresponding to this time is directly discarded. The advantage of this operation logic is that by comparing the upper and lower threshold boundaries, interference from current values ​​in unstable discharge ranges is eliminated.

[0022] S102: Based on the current interval voltage record set, call the corresponding sampling time sequence, sort the sampling times in ascending order of time and synchronously adjust the corresponding individual voltage position, perform index binding on the sorted sampling times and the corresponding individual voltage, and establish a time-series voltage lookup table. The pre-processed current-range voltage record set is retrieved, and all independently existing sampling time values ​​are extracted from this record set. These extracted values ​​are stored independently in a temporary time array. Simultaneously, the individual unit voltage values ​​that strictly correspond to each sampling time value are extracted and stored in parallel in a temporary voltage array. A time-increasing sort operation is performed on the sampling time values ​​in the temporary time array. The sorting process uses quicksort logic. A base timestamp is randomly assigned in the temporary time array, and left and right scan pointers are established. The left scan pointer moves from the beginning of the array to the right, searching for values ​​greater than the base timestamp, while the right scan pointer moves from the end of the array to the left, searching for values ​​less than the base timestamp. When both pointers find values ​​that meet the criteria, their positions are swapped. This scanning and swapping process continues until the left and right scan pointers meet. Then, the base timestamp is placed at the position where the two pointers meet, dividing the entire array into a left half (values ​​less than the base timestamp) and a right half (values ​​greater than the base timestamp). Repeat the above-described reference specification and pointer scan swapping operations for both the left and right parts until all sampled time values ​​in the temporary time array exhibit a strict ascending order. Simultaneously with each swapping of temporary time array elements, the corresponding individual voltage values ​​at the same index in the temporary voltage array are required to undergo a fully synchronized swapping operation. After completing all sorting and adjustment operations, perform cross-array index binding operations on the sorted sampled time array and the synchronized adjusted individual voltage array. Following a sequential order, extract the first sorted sampled time value and its corresponding first individual voltage value, assigning them a shared sequence identification code. Then, extract the second set of values ​​and assign the next sequence identification code, and so on. Integrate all data pairs bound by identification codes into a two-dimensional data table structure to establish a time-series voltage lookup table. For example, consider three sets of unordered original records with sampling times of 1620000005, 1620000001, and 1620000003, corresponding to single-unit voltages of 3.25 volts, 3.28 volts, and 3.26 volts, respectively. In the first round of sorting, 1620000003 is selected as the reference, and 1620000001 is moved to its left, while 1620000005 is moved to its right. During this swap, the voltage of 3.28 volts corresponding to 1620000001 moves to the front, and the voltage of 3.25 volts corresponding to 1620000005 moves to the back. The final sorted time series is (1620000001, 1620000003, 1620000005), and the corresponding voltage series is (3.28 volts, 3.26 volts, 3.25 volts).The advantage of this operational logic is that by performing a dual synchronization operation of timestamp values ​​and voltage values, it avoids the risk of misalignment in the timing reassembly process.

[0023] S103: Based on the time-series voltage lookup table, call the sampling time value and the individual cell voltage value to perform key-value pairing, associate the sampling time key-value item with the individual cell voltage key-value item one by one and connect them in time order to generate a discharge voltage time-series record; The timing voltage lookup table constructed in the preceding process is directly read. For each row in the table, key-value pairing operations at the data encapsulation level are performed between the sampled time value and the individual voltage value. An empty hash map structure space is dynamically allocated in memory. The traversal starting point is set to the first row of the lookup table. The sampled time value in that row is extracted as the unique identifier key data, and the individual voltage value in that row is extracted as the corresponding core value data. The extracted identifier key data and core value data are combined to construct an independent key-value pair unit, which is then written into the dynamically allocated hash map structure space. Subsequently, the data extraction pointer is moved down one row, and the sampled time value and individual voltage value of the second row are extracted to construct the second key-value pair unit and write it into the hash map structure space. This process of pointer movement, value extraction, key-value construction, and data writing is repeated until the data extraction pointer reaches the last row of the timing voltage lookup table and the final pairing write is completed. After all key-value pairings are completed, all key-value pair units are retrieved from the hash map structure space. Based on the sampling time values ​​contained within each key-value pair, a strict time monotonically increasing determination operation is performed. Starting with the key-value pair with the smallest timestamp value, the next key-value pair with a slightly larger timestamp value is connected to the end of the current key-value pair via the tail pointer of the virtual data chain. Then, the next key-value pair with an even larger timestamp value is connected to the end of the previous key-value pair. This end-to-end connection chain concatenates the originally discrete key-value pairs into a data sequence structure with a clear sequential direction. This concatenated data sequence structure is output and saved. For example, three rows of data are extracted from the lookup table: time 1620000001 with voltage 3.28 volts, time 1620000003 with voltage 3.26 volts, and time 1620000005 with voltage 3.25 volts. After performing the pairing operation, key-value pair unit 1 is generated with a key of 1620000001 and a value of 3.28 volts; key-value pair unit 2 is generated with a key of 1620000003 and a value of 3.26 volts; and key-value pair unit 3 is generated with a key of 1620000005 and a value of 3.25 volts. The key values ​​of these units are then extracted. Since 1620000001 is the smallest, key-value pair unit 1 is placed first. The remaining units are compared. Since 1620000003 is greater than 1620000001, key-value pair unit 2 is connected to the end of the tail pointer of unit 1. Since 1620000005 is greater than 1620000003, key-value pair unit 3 is connected to the end of the tail pointer of unit 2. The final sequence structure is the discharge voltage timing record. The advantage of this operational logic is that it optimizes the speed of continuous memory access for large monitoring data sequences by chaining pointers within the hash key-value structure.

[0024] Please see Figure 3The specific steps of S2 are as follows: S201: Obtain the discharge voltage time sequence record, retrieve the individual cell voltage value corresponding to the sampling time, extract the voltage value pairs of adjacent sampling times in time sequence, calculate and record the difference value for each voltage value pair, organize the record order according to the sampling time index, and obtain the adjacent voltage difference sequence. The structured discharge voltage timing record set contains sampling times arranged strictly in chronological order and their unique corresponding individual cell voltage values. An initial read pointer is set, pointing to the first data entry in the record set. Using this read pointer, the first sampling time and its corresponding first individual cell voltage value are extracted. Then, the read pointer is moved one position forward, pointing to the second data entry, and the second sampling time and its corresponding second individual cell voltage value are extracted. At this point, the first and second individual cell voltage values ​​together form a pair of voltage values ​​with adjacent sampling times in chronological order. A difference calculation operation is performed on this voltage value pair. Specifically, the second individual cell voltage value is used as the minuend, and the first individual cell voltage value is used as the subtrahend to perform a subtraction operation, thereby obtaining the voltage change between these two adjacent times, and this change is defined as the adjacent voltage difference value. After obtaining the first difference value, the pointer continues to move forward, extracting the third individual voltage value, and forming a new pair of adjacent voltage values ​​with the second individual voltage value, performing the same subtraction operation. This process is repeated until the pointer reaches the last entry in the discharge voltage time sequence record set. During this process, for each adjacent voltage difference value calculated, the sampling time index number corresponding to the current minuend is extracted, and a mandatory binding relationship is established between the difference value and the sampling time index number. All adjacent voltage difference values ​​with sampling time index numbers are sorted and arranged in chronological order, ultimately generating a complete adjacent voltage difference sequence. For example, in actual monitoring, the individual voltage value read at the first sampling time is 3.250 volts, and the individual voltage value at the second sampling time is 3.245 volts. Following the aforementioned operational logic, 3.245 volts is used as the minuend, and 3.250 volts is used as the subtrahend for subtraction. The actual operation is 3.245 minus 3.250, resulting in a voltage difference of -0.005 volts. This -0.005 volt difference is then recorded and bound to index number 2 at the second sampling time. The single-unit voltage value at the third sampling time is 3.238 volts. This is subtracted from the 3.245 volts at the second sampling time, resulting in a voltage difference of -0.007 volts, which is then bound to index number 3. The advantage of this operational logic is that by directly subtracting the preceding and subsequent single-unit voltage values, it enhances the ability to extract features of local voltage fluctuations.

[0025] S202: Based on the adjacent voltage difference sequence, call the corresponding sampling time index, sort and verify the difference records according to the ascending relationship of sampling time, establish the adjacency relationship between the sampling time numbers of the previous and next samples and bind the corresponding difference records, and establish a sampling adjacency index table; All data records within the adjacent voltage difference sequence are extracted. The difference value and the pre-bound corresponding sampling time index number are read from each record. A sorting and verification operation is performed on these difference records with index numbers to ensure that the sampling time order is consistent with the actual valid data. Specifically, the records are first arranged sequentially according to the sampling time index number, and the verification benchmark is set as a strictly monotonically increasing sequence of natural numbers. Then, adjacent sampling time index numbers are extracted sequentially from the sorted sequence. The difference between the latter and former index numbers is calculated, i.e., subtracting the former index number as the subtrahend and the latter as the minuend. When the calculation result equals 1, the two index numbers are deemed to meet the increasing continuity requirement, indicating that the sampled data remains normally continuous within this interval. When the calculation result is greater than 1, it is determined that there are missing sampling points in this interval. Without deleting the original valid data, a supplementary difference is generated by performing mean interpolation on adjacent difference records, and the missing sampling time index numbers are simultaneously filled in, thereby restoring the continuity of the index sequence. If an abnormal index order is detected, the correct time order is restored by re-sorting. After completing all index continuity checks and necessary padding, the adjacency relationships in the graph structure are constructed using the validated adjacent sampling time numbers: the previous sampling time number is defined as the source node, the immediately following sampling time number is defined as the target node, and a one-way directed edge is established between them. Simultaneously, the corresponding voltage difference record is bound as the weight attribute of this directed edge. Subsequently, the above node definition and directed edge construction operations are performed sequentially on all sampling indices in the sequence. Finally, all source nodes, target nodes, and corresponding edge weight difference records are uniformly integrated and structurally arranged to generate a data mapping table, which is defined as the sampling adjacency index table to represent the difference connection relationship between consecutive sampling times. Table 1 shows the construction results of some sampling adjacency indices.

[0026] Table 1. Sampling Adjacency Index Mapping Relationship Table Source node index number Target node index number Edge weight difference value Incremental check status 1 2 -0.005 pass 2 3 -0.007 pass 3 4 -0.006 pass As shown in Table 1, the difference propagation path between adjacent time points is clearly demonstrated through a rigorous analysis of the correspondence between source and target nodes. The advantage of this operational logic is that it solidifies the sequential logical topology of the voltage time series by constructing directed edge adjacency relationships and assigning weights using graph theory.

[0027] S203: Based on the sampling adjacency index table, call the adjacent sampling time number and the difference record to perform key-value association, connect the adjacent sampling time number and the difference record in time order to form a sequence structure, and generate a voltage drop change sequence; Based on the sampling adjacency index table constructed in the previous steps, the source node index number, target node index number, and difference record used as weight are called row by row in the table. A contiguous address space is dynamically allocated in memory to store the new sequence. The source node index number and target node index number are extracted from the first row of data and combined into a composite search key. This composite search key is associated with the extracted difference record at the memory address level. After the association is completed, the difference record is pushed to the first position of the newly allocated contiguous address space. Subsequently, the source node index number and target node index number of the second row of data are read and used as a composite search key to associate with the corresponding difference record. After confirming that the source node index number of the second row of data is completely consistent with the target node index number of the first row of data, the difference record corresponding to the second row of data is stored immediately after the physical memory address of the first difference record, i.e., at the second position. According to the connection relationship between the beginning and end of the adjacent sampling time numbers, the independent difference records are physically connected into a seamless chain sequence structure according to the chronological flow order. After traversing all rows of data in the sampling adjacency index table, all difference records within the contiguous address space are encapsulated and output, ultimately generating a complete voltage drop change sequence. For example, retrieving data from Table 1, the first row is extracted from source node 1 and target node 2, and their corresponding edge weight difference value of -0.005 is associated and placed in the first position of the address space. The second row is extracted from source node 2 and target node 3. Since source node 2 is connected to the preceding target node 2, its corresponding edge weight difference value of -0.007 is placed in the second position. The third row is extracted from source node 3 and target node 4, and their corresponding difference value of -0.006 is placed in the third position. Finally, a chain sequence structure containing -0.005, -0.007, and -0.006 is generated. The advantage of this operation logic is that, through the consistency comparison of the composite search key and the continuous append operation of the memory address, the physical storage coherence of the difference data is strengthened.

[0028] Please see Figure 4 The specific steps of S3 are as follows: S301: Obtain the voltage drop change sequence, retrieve the sequence position difference value, perform item-by-item comparison operation on the same sequence position difference value, calculate the difference amplitude of adjacent position difference and record the difference sign, and arrange all comparison results according to the sequence position number to obtain the voltage difference sequence. Obtain the voltage drop change sequence generated in the previous steps, and read the position difference values ​​of each sequence arranged in memory address order. Set up a sliding window for data comparison, specifying its coverage width as two adjacent sequence positions. Initialize this window and place it at the very beginning of the sequence, completely selecting the first and second position difference values. Perform a step-by-step comparison operation on these two adjacent position difference values ​​within the same sequence. Specifically, use the second position difference value as the subtrahend and the first position difference value as the subtraction value, perform a standard subtraction operation, and obtain the absolute difference value after subtraction. This absolute difference value is then recorded as the magnitude of the adjacent position difference. After calculating the magnitude, perform a sign determination operation on the original result value obtained from this subtraction operation. If the original result of the subtraction operation is greater than 0, a positive sign is entered; if the original result is less than 0, a negative sign is entered; and if the original result is equal to 0, a zero sign is entered. The calculated difference magnitude between adjacent positions and the entered difference sign are used as a composite comparison result unit. Then, the sliding operation window is shifted one position towards the end of the sequence to select the difference values ​​at the second and third positions, and the standard subtraction operation and difference sign determination are repeated. This sliding and comparison process continues until the edge of the operation window touches and processes the last difference value in the voltage drop sequence. All generated composite comparison result units are extracted and arranged vertically in ascending order of their corresponding operation window starting sequence position numbers, thus constructing a structured voltage difference sequence. For example, in actual data processing, the difference value at the first position in the sequence is -0.005 volts, and the difference value at the second position is -0.007 volts. Following the comparison logic, -0.007 volts is used as the minuend, and -0.005 volts is used as the subtraction term. The actual calculation process is -0.007 volts plus 0.005 volts, resulting in an original value of -0.002 volts. Therefore, its absolute value is extracted, recording the difference between adjacent positions as 0.002 volts. Since the original result is less than 0, the difference is recorded as negative, and this unit is assigned position number 2. The difference value at the third position is -0.006 volts. Subtracting this from the difference value at the second position (-0.007 volts) yields an original value of +0.001 volts. The difference is recorded as 0.001 volts, with a positive sign, and is assigned position number 3. The advantage of this operational logic is that it quantifies the dynamic fluctuation trend of battery discharge degradation rate by using the difference operation between adjacent elements and the sign discrimination logic within the sliding window.

[0029] S302: Call the position difference values ​​according to the voltage difference sequence, sort the difference values ​​and rearrange them according to the sequence position index, perform continuity judgment on the sorted position numbers, extract continuous number segments and record the start and end position numbers of the segments, and generate a continuous difference segment table. Based on the structured voltage difference sequence, the difference amplitude values ​​and corresponding difference signs of all adjacent positions stored in the sequence are retrieved, maintaining the original sampling time order of the data. Global sorting of the difference amplitude values ​​is not performed to avoid disrupting the continuity of the time series. While maintaining the original time index order, continuous anomaly identification processing is performed on the difference amplitude values ​​in the sequence. Specifically, a preset difference amplitude discrimination threshold or anomaly discrimination rule is retrieved, and the difference amplitude values ​​are read item by item in time order, determining whether they meet the abnormal fluctuation judgment conditions. When a difference amplitude value meets the anomaly condition, that position is recorded as the anomaly start point, and the continuity check continues along the time order for subsequent data. If the difference amplitude values ​​of subsequent adjacent positions continuously meet the same anomaly judgment conditions, this phenomenon is accumulated and recorded as a continuous anomaly sequence, while maintaining its corresponding original sequence position index number. When the number of data entries that continuously meet the anomaly judgment conditions reaches the preset continuity discrimination benchmark length, the time segment composed of the original sequence position index numbers is extracted and confirmed as a continuous abnormal fluctuation segment, where the continuity discrimination benchmark length is set to 5. Subsequently, the first original sequence position index number in the abnormal segment is extracted as the segment start position number, and the last original sequence position index number in the abnormal segment is extracted as the segment end position number. These two are combined and recorded to form an abnormal segment identifier. By performing the above time-series traversal detection operation on the entire voltage difference sequence, all continuous abnormal fluctuation segments that meet the conditions can be identified and uniformly organized into a continuous difference segment table. For example, during the sequence traversal, when data entries with original sequence position index numbers 15, 16, 17, 18, 19, and 20 all meet the abnormal difference amplitude judgment condition, the length of this continuous segment is 6, which is greater than or equal to the continuity judgment benchmark length of 5. Therefore, this time segment is extracted as a continuous abnormal fluctuation segment, and its start position number is recorded as 15 and its end position number as 20. Through the above processing method, accurate identification of concentrated time regions of continuous abnormal fluctuations in voltage change rate is achieved without destroying the original time series structure.

[0030] S303: Based on the continuous difference segment table, call the segment start and end position numbers, retrieve the corresponding sampling time number and perform key-value association, establish a mapping structure between the continuous difference segment number and the corresponding sampling time number, organize and connect them in time order, and generate slope offset segment records. The process iteratively retrieves the start and end position numbers of each segment from the table. It then returns to the index corresponding to the original discharge voltage timing record, retrieving and extracting the physically corresponding original start sampling time number based on the retrieved segment start position number. Similarly, it retrieves and extracts the physically corresponding original end sampling time number based on the retrieved segment end position number. For each consecutive difference segment, it performs a low-level key-value association operation between the segment's auto-incrementing number and the extracted original start and end sampling time numbers. Specifically, it creates a new data object entity in the database table, setting the auto-incrementing number of the consecutive difference segments as the primary key field. Within this data object entity, it sets the original start and end sampling time numbers as foreign key fields, thus constructing a stable mapping structure between the numbers of the consecutive difference segments and the specific sampling time numbers. All constructed data object entities are then extracted to the memory cache, and the original start sampling time numbers within each entity are extracted. A timestamp-level comparison operation is then performed on these original start sampling time numbers. Based on the original starting sampling time numbers in ascending chronological order, these data object entities are rearranged by joining their first and last physical addresses in the memory cache. After rearranging and sequential joining, the rearranged set of data object entities is exported and persistently stored, generating the final slope offset segment records. For example, the starting position number 15 and ending position number 20 of a segment recorded in the continuous difference segment table are retrieved. The index is compared and retrieved, confirming that the original starting sampling time number corresponding to number 15 is 1620000030, and the original ending sampling time number corresponding to number 20 is 1620000040. A data object entity is constructed, assigned an auto-incrementing number of 1, and 1620000030 and 1620000040 are written into the associated attributes of this entity. For another segment starting position number 50 and ending position number 58, the corresponding original starting sampling time number is retrieved as 1620000100, and the original ending sampling time number is 1620000116. A second data object entity is constructed, assigned an auto-incrementing ID of 2, and associated with its corresponding time stamp. Finally, comparing 1620000030 and 1620000100, it is confirmed that the former is earlier; therefore, the entity with auto-incrementing ID 1 is linked before the entity with auto-incrementing ID 2. The advantage of this operation logic is that by mapping primary and foreign key entities and comparing and sorting based on timestamp size, it breaks down the information barrier between the abstract data index and the actual time coordinates.

[0031] Please see Figure 5 The specific steps of S4 are as follows: S401: Obtain the slope offset segment record, retrieve the corresponding sampling time number, extract the start sampling time and end sampling time from the discharge voltage time sequence record and calculate the discharge duration value, organize the position numbers in time order, and obtain the discharge duration position sequence. The system retrieves the start and end sampling timestamps of each segment within the record set. It then retrieves the complete discharge voltage timing record master table from the global database. Using the retrieved start sampling time number as the search keyword, it performs a full table scan to extract the corresponding absolute timestamp data, identifying it as the start sampling time of the current discharge cycle. Similarly, it uses the end sampling time number as the search keyword to perform a matching scan in the master table, extracting the corresponding absolute timestamp data, and identifying it as the end sampling time of the current discharge cycle. After obtaining these two key timestamp values, it calculates the discharge duration. The calculation logic is as follows: the absolute timestamp corresponding to the end sampling time is used as the minuend, and the absolute timestamp corresponding to the start sampling time is used as the subtrahend; a direct subtraction operation is performed between the two. The difference obtained from this subtraction operation clearly represents the actual physical time length experienced during this battery discharge process, and this result is recorded as the discharge duration value. After completing the calculation, it extracts all consecutive position numbers from the original discharge voltage timing record corresponding to this discharge duration value. These consecutive position numbers are extracted and arranged strictly in chronological order using a one-dimensional linear arrangement to construct a continuous position number array. For example, in the actual extraction process, the true absolute timestamp data corresponding to the starting sampling time of a certain discharge cycle is obtained as 1620000000. The true absolute timestamp data corresponding to the ending sampling time of the same discharge cycle is extracted as 1620003600. According to the aforementioned subtraction operation logic, 1620003600 is used as the minuend, and 1620000000 is used as the subtrahend for subtraction. The calculated difference is 3600 seconds. Therefore, the total discharge duration of this cycle is recorded as 3600 seconds. Subsequently, all relative position numbers starting from 1 within this 3600-second span are extracted and arranged in ascending order from 1, 2, 3 up to 3600, finally obtaining the discharge duration position sequence. The advantage of this operational logic is that it accurately extracts the absolute time span parameter of the complete discharge cycle through direct subtraction of timestamp values ​​and one-dimensional array permutation operations.

[0032] S402: Based on the discharge duration position sequence, call the discharge duration value, perform interval division on the position number, split the discharge duration sequence into a set of front interval numbers, a set of middle interval numbers, and a set of rear interval numbers, and generate a discharge segment number table. Retrieve the discharge duration position sequence containing all position numbers, and simultaneously extract the total discharge duration value calculated from the preceding sequence. Perform a proportional physical interval division operation on the position numbers in this discharge duration position sequence. Retrieve the preset proportional splitting coefficients set in the configuration library, setting the splitting coefficient for the first segment to 0.3, the splitting coefficient for the middle segment to 0.4, and the splitting coefficient for the last segment to 0.3. Extract the total sequence length value corresponding to the total discharge duration, and perform multiplication operations on the aforementioned three splitting coefficients for each of these values. Specifically, multiply the total sequence length value by 0.3 to obtain the upper limit value of the number of elements that the first segment should contain. Multiply the total sequence length value by 0.4 to obtain the number of elements that the middle segment should contain. Multiply the total sequence length value by 0.3 again to obtain the number of elements that the last segment should contain. Based on the calculated quantity boundaries, perform a slicing and separation operation on the position numbers starting from the beginning of the discharge duration position sequence. Extract all position numbers starting from the first position number until the cumulative number of position numbers equals the upper limit of the number of elements contained in the preceding interval. Combine these extracted position numbers to form the preceding interval number set. Then, continue extracting position numbers from the breakpoint of the slice until the cumulative number of extracted position numbers equals the number of elements contained in the middle interval. Combine these slice number sets to form the middle interval number set. Finally, extract all remaining position numbers and combine them to form the following interval number set. Integrate and correlate these three separated number sets to generate a discharge segment numbering table with clearly defined stage boundaries. Table 2 shows the actual interval division boundary data for this discharge segment numbering.

[0033] Table 2 Data Table for Delineating the Boundary of Discharge Section Numbers Interval phase division Proportional split coefficient Starting position number End position number First section 0.3 1 1080 Middle section 0.4 1081 2520 The latter section 0.3 2521 3600 As shown in Table 2, the sequence with a total length of 3600 was physically truncated using a specific splitting coefficient. The advantage of this operational logic is that it constructs a three-stage analysis framework that fits the physical characteristics of battery discharge by using the multiplication of the total length value and the splitting coefficient for truncation.

[0034] S403: Based on the discharge segment number table, call the segment number set, retrieve the corresponding position number of the slope offset segment record, associate the first segment interval, middle segment interval and last segment interval with the slope offset segment position using key value and organize them in time order to generate segment position mapping record; The completed discharge segment number table is extracted, and the isolated front, middle, and rear segment number sets are retrieved from it. Simultaneously, slope offset segment records containing specific fault location information are searched and retrieved from memory. For each specific position number in this slope offset segment record, an inclusion-based search is performed in the three independent segment number sets. A loop search pointer is set to extract the center position number value of the first slope offset segment record. This center position number value is first pushed to the front segment number set for traversal comparison. If an element with an exact match is found in the front segment number set, the underlying key-value association mechanism is immediately triggered. The slope offset segment record identifier to which the center position number belongs is forcibly bound to the front segment identifier as a key-value pair. If no matching element is found in the front segment, the center position number value is pushed to the middle segment number set for continued traversal comparison. If a match is found, it is bound to the middle segment identifier as a key-value pair. If no match is found, the record is pushed into the subsequent interval number set for comparison and binding. After processing all slope offset segment records according to this logic, all generated key-value pairs are extracted centrally. Based on the magnitude of the center position number contained within the key-value pair, a strict ascending monotonically increasing sorting is performed. After sorting, these pair data are serialized and output to generate segment position mapping records arranged in chronological order. For example, the center position number values ​​of the retrieved slope offset segment records are 800 and 3000. The first value 800 is extracted according to the boundary set in Table 2. It is then compared with the preceding interval number set, i.e., the range from 1 to 1080, confirming that 800 is included between 1 and 1080. Therefore, the unique identifier of this offset record and the preceding interval label are combined into a key-value pair and stored in the cache. Subsequently, the second value 3000 is extracted. After the comparisons in the first and middle intervals failed, the comparison was performed on the last interval number set, i.e., the range from 2521 to 3600, confirming that 3000 was included in this range. Its identifier was then used to construct a key-value pair with the last interval label. Since 800 is less than 3000, the key-value pair corresponding to 800 was placed before the key-value pair corresponding to 3000. The advantage of this operational logic is that, through equal-value matching comparisons across set elements and label binding operations, the specific discharge stage of the abnormal voltage fluctuation phenomenon is clearly defined.

[0035] Please see Figure 6 The specific steps of S5 are as follows: S501: Obtain the segment location mapping record, retrieve the correspondence between the slope offset segment location number and the set of the preceding segment number, the set of the middle segment number, and the set of the following segment number, perform segment ownership judgment on the location number and organize it according to the energy storage unit number to obtain the offset segment location table. Extract the segment location mapping record data table encapsulated in the above operations. Scan the record table row by row, extracting the slope offset segment location number recorded in each row, and the classification label identifier that strictly corresponds to it in the front segment number set, middle segment number set, or back segment number set. Perform a segment final assignment determination operation based on the classification label for each extracted location number. Construct three independent data collection storage buckets in memory, named the front segment collection bucket, middle segment collection bucket, and back segment collection bucket, respectively. Read the first mapping record. If the classification label of the record is identified as a front segment identifier, write the slope offset segment location number directly into the front segment collection bucket. If the classification label of the record is identified as a middle segment identifier, write it into the middle segment collection bucket. If it is identified as a back segment identifier, write it into the back segment collection bucket. After completing the collection and allocation of all location numbers, call the hardware topology management database to extract the factory serial number of the energy storage unit hardware that caused the above slope offset. A horizontal data concatenation operation is performed between these hardware serial numbers and the location numbers within each collection bucket. Specifically, within each collection bucket, the energy storage unit serial number is used as the first data field, and the allocation result to that bucket's segment is used as the second data field, and these are combined in parallel. The recombined and concatenated data records from the three collection buckets are extracted, and globally unified and vertically arranged according to the ascending order of the energy storage unit serial numbers to generate a complete structured data table, which is defined as an offset segment location table. For example, if energy storage unit number 101 is scanned, it will be offset at location number 800 with a preceding segment identifier. Simultaneously, if energy storage unit number 105 is scanned, it will be offset at location number 3000 with a following segment identifier. The allocation determination and writing operation is initiated, writing 800 to the preceding collection bucket and 3000 to the following collection bucket. Then, the hardware serial numbers 101 and 105 are extracted. Within the first collection bucket, a data row is generated where the first column is 101, representing the first interval. Within the second collection bucket, a data row is generated where the first column is 105, representing the second interval. Finally, these two rows are extracted. Since 101 is less than 105, the row containing 101 is placed first, and the row containing 105 is placed below it. The advantage of this operational logic is that, through the allocation of independent storage buckets based on tags and the merging of hardware numbers, the corresponding topological structure of abnormal behavior of individual batteries and the time periods in which they occur is identified.

[0036] S502: Based on the offset section location table, call the energy storage cell number and section assignment result, perform fault type discrimination on the section assignment result, mark the front section as internal resistance abnormality, mark the middle section as electrode activity decay, mark the rear section as conduction path abnormality, and organize them according to the energy storage cell number to generate a fault type identification table. The system retrieves the structured offset segment location table and extracts the specific energy storage cell number and pre-assigned segment attribution result field from each data row. Logical judgment is then performed on the extracted segment attribution result field. The built-in fault type knowledge graph matching rule library is then retrieved. The first matching rule is executed: if the retrieved segment attribution result field string is identical to the preceding segment, the corresponding judgment logic is triggered, directly changing the label of the preceding segment to the fault type string of internal resistance abnormality through string replacement. The second matching rule is executed: if the retrieved segment attribution result field string is identical to the middle segment, the corresponding judgment logic is triggered, forcibly changing the label of the middle segment to the fault type string of electrode activity degradation through string replacement. The third matching rule is executed: if the retrieved segment attribution result field string is identical to the following segment, the corresponding judgment logic is triggered, changing the label of the following segment to the fault type string of conduction path abnormality through string replacement. After completing all string matching and replacement judgment operations, the energy storage cell number in the first column of the table is extracted again. The generated specific fault type string is combined with the energy storage unit number to form a new data row record. All newly generated data row records are then rearranged and physically placed according to the ascending order of the energy storage unit number values, ultimately generating a fault type identifier table that intuitively reflects the hardware health status. Table 3 details the generation results of the correspondence between hardware number and fault type identifier.

[0037] Table 3. Single Cell Fault Type Identification Mapping Table Energy storage unit hardware serial number Segment Attribution Results String replacement matching rules Fault type final identification 101 First section Matching rule 1 Internal resistance abnormality 102 Middle section Matching rule 2 Electrode activity decline 105 The latter section Matching rule 3 abnormal conduction path As shown in Table 3, abnormal phenomena at different discharge stages are directly transformed into specific physical component defect qualitative results through preset rules. The advantage of this operational logic is that it establishes a diagnostic link between abstract timing stages and concrete failure mechanisms by using equal-value matching and forced substitution assignment operations of string feature values.

[0038] S503: Based on the fault type identifier table, call the energy storage cell number and fault type identifier, perform key-value association between internal resistance abnormality, electrode activity degradation, conduction path abnormality and energy storage cell number, and organize them in numerical order to generate power energy storage system fault diagnosis results. Extract a fault type identifier table containing clear physical diagnostic results. Retrieve the energy storage unit number data and its associated specific fault type identifier string, including strings for abnormal internal resistance, electrode activity degradation, or conduction path abnormality, row by row. Within the buffer area of ​​the diagnostic result output module, allocate an independent result encapsulation data package for each energy storage unit number. For the fault type identifier string (such as abnormal internal resistance) extracted from the first row, extract the complete content of the string and push it, along with the corresponding energy storage unit number, into the memory address of the first target result encapsulation data package. Perform a key-value entity association operation within this memory address. Specifically, set the energy storage unit number as the primary key for this data package and set the corresponding fault type identifier string as the detailed description load of the data package. Perform the above extraction, pushing, and load setting entity association operations sequentially on all data rows in the table. After all data has been processed and each has generated its own result encapsulation data package, read the serial numbers from the primary keys of all data packages. Perform a final bubble sort comparison operation on these numbers. The algorithm compares the primary key numbers of two adjacent data packets. If the former number is greater than the latter, the two data packets are swapped in the buffer area until all result-encapsulated data packets are neatly arranged in ascending order of primary key numbers. After final sorting, a data push command is triggered on the output port. The sorted result-encapsulated data packets are serialized into a complete report document file and broadcast to external terminals, thus officially generating and issuing the power storage system fault diagnosis result report. For example, if two unsorted data packets are extracted from the buffer area, the first with primary key 105 describing a conduction path anomaly, and the second with primary key 101 describing an internal resistance anomaly, a bubble comparison operation is performed. The numbers 105 and 101 are extracted and compared by subtraction. The result of subtracting 101 from 105 is greater than 0. Since the former number is greater than the latter number, a swap command is triggered, moving the data packet with primary key 101 before the data packet with primary key 105. After this position swapping and reorganization, the internal resistance anomaly diagnostic report for battery #101 will be output before the report for battery #105. The advantage of this operational logic is that, through the construction of the primary key load entity and the bubbling position swapping comparison operation, the structural clarity of the large-scale cluster battery diagnostic reports at the final output level is ensured.

[0039] Please see Figure 7 A fault diagnosis system for power storage systems, including: The discharge timing construction module is used to implement S1: collect the individual cell voltage, loop current and sampling time of the series circuit of the energy storage battery module, filter the individual cell voltages in the same current range, arrange the individual cell voltages corresponding to the sampling time and establish the correspondence between the sampling time and the individual cell voltage, and generate the discharge voltage timing record. The voltage drop analysis module is used to implement S2: calculate the voltage difference of individual cells at adjacent sampling times of the discharge voltage time sequence record, arrange the differences and establish the correspondence between sampling time and difference, and generate a voltage drop change sequence; The slope offset identification module is used to implement S3: compare the difference between the same position in the voltage drop change sequence to obtain the difference sequence, sort the difference sequence and identify the continuous difference position, establish the correspondence between the continuous difference position and the sampling time, and generate the slope offset segment record. The segment mapping and positioning module is used to implement S4: dividing the discharge voltage timing record into discharge duration segments and establishing the first segment interval, the middle segment interval, and the last segment interval; comparing the slope offset segment record positions and establishing the correspondence between the first segment interval, the middle segment interval, and the last segment interval and the slope offset segment record; and generating segment position mapping records. The fault diagnosis output module is used to implement S5: determine the offset section position of the section location mapping record and establish the correspondence between internal resistance abnormality, electrode activity decay, conduction path abnormality and energy storage cell number, and generate the fault diagnosis result of the power energy storage system.

[0040] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the technical solution.

Claims

1. A fault diagnosis method for power storage systems, characterized in that, Includes the following steps: S1: Collect the individual cell voltage, circuit current, and sampling time of the series circuit of the energy storage battery module, filter the individual cell voltages in the same current range, arrange the individual cell voltages corresponding to the sampling time, establish the correspondence between the sampling time and the individual cell voltage, and generate a discharge voltage timing record. S2: Calculate the voltage difference of individual cells at adjacent sampling times in the discharge voltage time sequence record, arrange the differences and establish the correspondence between sampling times and differences to generate a voltage drop change sequence; S3: Compare the voltage drop change sequence with the difference at the same position to obtain the difference sequence, sort the difference sequence and identify continuous difference positions, establish the correspondence between continuous difference positions and sampling time, and generate slope offset segment records. S4: Divide the discharge voltage timing record into discharge duration segments and establish a front segment interval, a middle segment interval, and a rear segment interval. Compare the slope offset segment record positions and establish a correspondence between the front segment interval, the middle segment interval, and the rear segment interval and the slope offset segment record. Generate segment position mapping records. S5: Determine the offset section position of the section position mapping record and establish the correspondence between internal resistance anomaly, electrode activity decay, conduction path anomaly and energy storage unit number, and generate the fault diagnosis result of the power energy storage system. The specific steps of S3 are as follows: S301: Obtain the voltage drop change sequence, retrieve the sequence position difference value, perform item-by-item comparison operation on the same sequence position difference value, calculate the difference amplitude of adjacent position difference and record the difference sign, and arrange all comparison results according to the sequence position number to obtain the voltage difference sequence. S302: According to the voltage difference sequence, call the position difference value, sort the difference value and rearrange it according to the sequence position index, perform continuity judgment on the sorted position number, extract the continuous number segment and record the start and end position number of the segment, and generate a continuous difference segment table. S303: Based on the continuous difference segment table, call the segment start and end position number, retrieve the corresponding sampling time number and perform key-value association, establish a mapping structure between the continuous difference segment number and the corresponding sampling time number, organize and connect them in time order, and generate slope offset segment record; The specific steps of S5 are as follows: S501: Obtain the segment location mapping record, retrieve the correspondence between the slope offset segment location number and the set of front segment interval numbers, the set of middle segment interval numbers, and the set of rear segment interval numbers, perform segment ownership judgment on the location number and organize it according to the energy storage unit number to obtain the offset segment location table. S502: According to the offset section location table, call the energy storage cell number and section assignment result, perform fault type discrimination on the section assignment result, mark the front section as internal resistance abnormality, mark the middle section as electrode activity decay, mark the rear section as conduction path abnormality, and organize them according to the energy storage cell number to generate a fault type identification table. S503: Based on the fault type identifier table, call the energy storage cell number and fault type identifier, perform key-value association between internal resistance abnormality, electrode activity degradation, conduction path abnormality and energy storage cell number, and organize them in numerical order to generate power energy storage system fault diagnosis results.

2. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The discharge voltage timing record includes a sampling time sequence, a set of corresponding individual voltage values, and a current interval identifier. The voltage drop change sequence includes a set of voltage differences between adjacent times, an index of the time corresponding to the difference, and an identifier of the difference change amplitude. The slope offset segment record includes a set of continuous difference positions, a corresponding sampling time segment, and a difference sorting identifier. The segment position mapping record includes an identifier for the first segment, a identifier for the middle segment, and an identifier for the last segment. The fault diagnosis results of the power storage system include an identifier for abnormal internal resistance, an identifier for electrode activity decay, and an identifier for abnormal conduction path.

3. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The continuous difference position refers to the set of sampling positions in the difference sequence where the difference value of adjacent sampling points continuously exceeds a preset threshold and appears continuously in the time series.

4. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The slope offset section refers to the voltage change slope modified relative to the historical average voltage change trend, formed by the sampling time range corresponding to continuous difference positions.

5. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Acquire the individual cell voltage sequence, loop current sequence, and sampling time sequence of the series circuit of the energy storage battery module. Perform current interval comparison on the loop current corresponding to the individual cell voltage sequence. Keep the individual cell voltage records that fall into the same current interval and remove the records outside the interval. Organize the records according to the current interval label to obtain the current interval voltage record set. S102: According to the current interval voltage record set, call the corresponding sampling time sequence, sort the sampling times in ascending order of time and synchronously adjust the corresponding individual voltage position, perform index binding on the sorted sampling times and the corresponding individual voltage, and establish a time-series voltage lookup table. S103: Based on the time-series voltage lookup table, call the sampling time value and the individual cell voltage value to perform key-value pairing, associate the sampling time key-value item with the individual cell voltage key-value item one by one and connect them in time order to generate a discharge voltage time-series record.

6. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Obtain the discharge voltage time sequence record, retrieve the individual voltage value corresponding to the sampling time, extract the voltage value pairs of adjacent sampling times in time sequence, calculate and record the difference value for each voltage value pair, organize the record order according to the sampling time index, and obtain the adjacent voltage difference sequence. S202: According to the adjacent voltage difference sequence, call the corresponding sampling time index, sort and verify the difference records according to the increasing relationship of sampling time, establish the adjacency relationship between the sampling time numbers of the previous and next samples and bind the corresponding difference records, and establish a sampling adjacency index table; S203: Based on the sampling adjacency index table, call the adjacent sampling time number and the difference record to perform key-value association, connect the adjacent sampling time number and the difference record in time order to form a sequence structure, and generate a voltage drop change sequence.

7. The fault diagnosis method for a power energy storage system according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Obtain the slope offset segment record, retrieve the corresponding sampling time number, extract the start sampling time and end sampling time from the discharge voltage time sequence record and calculate the discharge duration value, organize the position numbers in time order, and obtain the discharge duration position sequence. S402: Based on the discharge duration position sequence, call the discharge duration value, perform interval division on the position number, split the discharge duration sequence into a front interval number set, a middle interval number set, and a rear interval number set, and generate a discharge segment number table; S403: Based on the discharge segment number table, call the segment number set, retrieve the corresponding position number of the slope offset segment record, associate the first segment interval, the middle segment interval, and the last segment interval with the slope offset segment position using key-value pairs, and organize them in chronological order to generate a segment position mapping record.

8. A fault diagnosis system for power storage systems, characterized in that, The system is used to implement the fault diagnosis method for the power storage system according to any one of claims 1-7, including: The discharge timing construction module is used to implement S1: collect the individual cell voltage, loop current and sampling time of the series circuit of the energy storage battery module, filter the individual cell voltages in the same current range, arrange the individual cell voltages corresponding to the sampling time and establish the correspondence between the sampling time and the individual cell voltage, and generate the discharge voltage timing record. The voltage drop analysis module is used to implement S2: calculate the voltage difference of individual cells at adjacent sampling times of the discharge voltage time sequence record, arrange the differences and establish the correspondence between sampling times and differences, and generate a voltage drop change sequence; The slope offset identification module is used to implement S3: compare the difference between the same position in the voltage drop change sequence to obtain the difference sequence, sort the difference sequence and identify continuous difference positions, establish the correspondence between continuous difference positions and sampling time, and generate slope offset segment records. The specific steps of S3 are as follows: S301: Obtain the voltage drop change sequence, retrieve the sequence position difference value, perform item-by-item comparison operation on the same sequence position difference value, calculate the difference amplitude of adjacent position difference and record the difference sign, and arrange all comparison results according to the sequence position number to obtain the voltage difference sequence. S302: According to the voltage difference sequence, call the position difference value, sort the difference value and rearrange it according to the sequence position index, perform continuity judgment on the sorted position number, extract the continuous number segment and record the start and end position number of the segment, and generate a continuous difference segment table. S303: Based on the continuous difference segment table, call the segment start and end position number, retrieve the corresponding sampling time number and perform key-value association, establish a mapping structure between the continuous difference segment number and the corresponding sampling time number, organize and connect them in time order, and generate slope offset segment record; The segment mapping and positioning module is used to implement S4: dividing the discharge voltage timing record into discharge duration segments and establishing a front segment interval, a middle segment interval, and a rear segment interval; comparing the position of the slope offset segment record and establishing a correspondence between the front segment interval, the middle segment interval, and the rear segment interval and the slope offset segment record; and generating a segment position mapping record. The fault diagnosis output module is used to implement S5: determine the offset section position of the section position mapping record and establish the correspondence between internal resistance abnormality, electrode activity decay, conduction path abnormality and energy storage cell number, and generate the fault diagnosis result of the power energy storage system. The specific steps of S5 are as follows: S501: Obtain the segment location mapping record, retrieve the correspondence between the slope offset segment location number and the set of front segment interval numbers, the set of middle segment interval numbers, and the set of rear segment interval numbers, perform segment ownership judgment on the location number and organize it according to the energy storage unit number to obtain the offset segment location table. S502: According to the offset section location table, call the energy storage cell number and section assignment result, perform fault type discrimination on the section assignment result, mark the front section as internal resistance abnormality, mark the middle section as electrode activity decay, mark the rear section as conduction path abnormality, and organize them according to the energy storage cell number to generate a fault type identification table. S503: Based on the fault type identifier table, call the energy storage cell number and fault type identifier, perform key-value association between internal resistance abnormality, electrode activity degradation, conduction path abnormality and energy storage cell number, and organize them in numerical order to generate power energy storage system fault diagnosis results.

Images