An edge-computing-based data preprocessing method for lithium battery energy storage station

Abstract

The application discloses a lithium battery energy storage station data preprocessing method based on edge calculation and relates to the technical field of energy storage data preprocessing.The method comprises the following steps: selecting voltage values and temperature values in topological code sequence based on a current page and a current available interface; generating object codes according to time codes, topological codes and object types, and positioning data positions of the current page; checking whether the data positions form effective collection values; judging missing, out-of-bound and same-side deviation of voltage values or temperature values of the data positions; forming missing processing records, evidence processing records, correction processing records or normal processing records respectively; and obtaining a cleaning set by arranging the object codes.The application achieves the effects of guaranteeing data traceability accuracy and cleaning logic strictness, and improves the reliability of lithium battery energy storage station data preprocessing results.

Description

Technical Field

[0001] This invention relates to the field of energy storage data preprocessing technology, and in particular to a data preprocessing method for lithium battery energy storage stations based on edge computing. Background Technology

[0002] In the field of lithium battery energy storage station operation monitoring, to ensure system safety and the accuracy of energy efficiency assessment, it is usually necessary to collect and process massive amounts of high-frequency battery operation data in real time. The conventional approach often employs a centralized cloud computing architecture, directly uploading time-series data such as voltage, temperature, and current from all measuring points within the station to a remote cloud platform for unified cleaning and analysis. This method relies on the powerful computing resources of the cloud, enabling the execution of complex statistical algorithms. By establishing a global data model to identify outliers and perform filtering, it is one of the mainstream models for industrial big data processing. Furthermore, edge computing technology is gradually being introduced into this field. By deploying computing nodes near the data source, it aims to alleviate network bandwidth pressure and achieve low-latency responses. Lightweight scripts are typically used to perform preliminary format conversion and simple threshold filtering on the raw data stream to adapt to the limited computing resources at the edge.

[0003] However, conventional edge processing methods have certain limitations. On the one hand, existing edge preprocessing often lacks a dynamic judgment mechanism for data stability, and mostly directly uses fixed thresholds or static models, making it difficult to adapt to the dynamically changing electrical characteristics of batteries under different charging and discharging conditions, resulting in a disconnect between cleaning rules and actual physical states. On the other hand, when dealing with missing or abrupt data, existing methods usually only perform simple linear interpolation or discarding, lacking in-depth mining of the data topological hierarchy and failing to utilize the spatial distribution characteristics of adjacent measurement points to assist in correction, thus affecting the physical reliability of data reconstruction. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a lithium battery energy storage station data preprocessing method based on edge computing to solve the problems of insufficient cleaning accuracy caused by the reliance on static thresholds and lack of topological correlation in data repair in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a data preprocessing method for lithium battery energy storage stations based on edge computing. The method includes: activating edge nodes, forming continuous sampling time slices and time codes; collecting measurement point operation data and topology hierarchy information within the sampling time slices; forming topology codes based on device addresses and installation locations; grouping measurement point operation data into pages according to time codes and topology codes; using pages as processing objects, filtering stable pages based on the source of measurement points within a page, operating status, and page difference results, and summarizing them into a stable page set; calculating voltage and temperature boundaries based on voltage and temperature page differences in the stable page set, generating boundary boards, establishing a mapping between boundary boards and the stable page set, and obtaining the currently available boundary board after confirming the consistency of page difference counts; and selecting voltage and temperature values ​​according to the topology code order based on the current page and the currently available boundary board. The system generates object codes based on time codes, topology codes, and object types, locates the data position on the current page, verifies whether the data position forms a valid acquisition value, and judges the voltage or temperature values ​​at the data position for missing, out-of-bounds, and same-side deviation, respectively forming missing processing records, evidence processing records, correction processing records, or normal processing records, and arranges them according to object codes to obtain a clean set; extracts pages to form a page table, converts the clean set into a clean record table, merges continuous evidence processing records in the clean record table into evidence segments according to topology codes and time codes, establishes a local retrieval structure based on the page table, clean record table, and evidence segments, and replays and verifies the net value and difference return value according to the pointing relationship between clean records and page tables and boundary plates, generates a replay result set, and summarizes it into the data preprocessing results of the lithium battery energy storage station.

[0008] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method described in this invention, the steps for forming continuous sampling time slices and time codes, collecting measurement point operation data and topology hierarchy information within the sampling time slices, and forming topology codes based on device addresses and installation locations are as follows: Edge nodes read their local clocks, determine the start time as the first sampling starting point, and divide the sampling time slices. Each sampling time slice records its start and end times, with the start time of the next sampling time slice following the end time of the previous sampling time slice; edge nodes extract the year, month, and day from the local clock. As the sampling date, the start time is extracted from the sampling time slice as the sampling time, and the slice number is incremented according to the generation order of the sampling time slice. The sampling date, sampling time, and slice number are combined into a time code. The edge node collects the measurement point operation data and topology hierarchy information within the sampling time slice. The number of each level is determined according to the equipment address and installation location of the measurement point, and the topology code is combined according to the physical hierarchy relationship of power station, battery compartment, battery cluster, battery pack, module, and individual unit. The measurement point operation data includes voltage value, temperature value, current value, state of charge, equipment status bit, communication status bit, and operating condition status bit.

[0009] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the step of grouping the measurement point operation data into pages according to time code and topology code is as follows: associating the time code and topology code with the measurement point operation data in the current sampling time slice to generate a page; when the voltage value or temperature value is not collected, a missing mark is formed at the corresponding data position; when the voltage value and temperature value have been collected, an original value pointer is generated according to the time code, topology code and data type; determining the module to which the measurement point belongs according to the topology code, and aggregating the same type of voltage value or temperature value under the same module that is not missing in the current page, and generating the voltage value neighbor base value and temperature value neighbor base value through the median value; taking the voltage value or temperature value that is not missing in the current page as the object, combining the voltage value neighbor base value or temperature value neighbor base value, and the valid original value under the same topology code in the previous page to calculate the voltage page difference or temperature page difference.

[0010] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the steps for calculating voltage and temperature boundaries based on voltage and temperature page differences of the stable page set and generating boundary plates are as follows: taking a page as the judgment object, generating a source verification status based on the source of the measurement point within the page, generating an operation verification status based on the operation status within the page, and generating a page difference verification status based on the page difference result; when the source verification status, operation verification status, and page difference verification status all meet the stable page conditions, generating a passed page screening mark, and setting the page as the target... Pages that should pass the page screening are identified as stable pages. Stable pages that pass the page screening are aggregated into a stable page set. Voltage page difference and temperature page difference are extracted from the stable page set, and arranged in numerical order to generate voltage page difference sequences and temperature page difference sequences. Identical page differences are merged to obtain candidate values ​​for voltage page difference and temperature page difference. The number of page differences not greater than each candidate value for voltage page difference and temperature page difference is counted sequentially. Based on the number of page differences, voltage boundaries and temperature boundaries are determined. Boundary plates are generated by combining the boundary plate number, boundary plate generation time, number of page differences, and stable page time range.

[0011] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method described in this invention, the steps for establishing the mapping between boundary plates and stable page sets, and generating currently available boundary plates after confirming the consistency of page difference counts, are as follows: Based on the boundary plate number and stable page time range in the boundary plate, stable pages whose time codes fall within the stable page time range are extracted from the stable page set, and sorted in order of time codes from first to last. The time code, topology code, voltage page difference, and temperature page difference of each stable page are extracted, and a data processing method is generated according to the boundary plate number, time code, and topology code. Stable page correspondence code; associate the stable page correspondence code with the voltage page difference, temperature page difference, stable page start time code, and stable page end time code in the corresponding stable page, and summarize them in the order of stable page time codes to generate a boundary plate correspondence table; compare the actual number of voltage page differences and temperature page differences in the boundary plate correspondence table with the number of voltage page differences and temperature page differences in the boundary plate, respectively. When the numbers are consistent, the boundary plate is used as a usable boundary plate, and the current usable boundary plate is determined from the usable boundary plates according to the boundary plate generation time, so as to obtain the current voltage boundary and the current temperature boundary.

[0012] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the step of verifying whether the data location has formed a valid acquisition value is as follows: Based on the current page and the currently available interface, the measurement points are arranged in ascending order of the topology code within the current page, and the voltage value and temperature value of the same measurement point are respectively used as different processing objects to generate a processing object sequence; the current processing object is extracted from the processing object sequence, the time code of the current page is used as the object time code, the topology code of the measurement point corresponding to the current processing object is used as the object topology code, and the object type is determined according to the data category of the current processing object to generate an object code; the data location in the current page is located according to the object code, and it is verified whether the data location has formed a valid acquisition value. If no valid acquisition value has been formed, a state of missing pending object is generated; if a valid acquisition value has been formed, a state of no missing pending object is generated.

[0013] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the steps for forming missing processing records and evidence processing records are as follows: Locate the data position of the current page along the object code, and obtain the corresponding original value pointer, page difference, and neighboring base value according to the object type. Determine the obtained neighboring base value as the object's neighboring base value. Obtain the valid original value under the same topology code from the previous page, and extract the current voltage boundary or current temperature boundary from the currently available boundary board. Determine the boundary board number of the currently available boundary board as the object boundary board number, and generate the object to be processed; When an object is missing, a net value is formed using the already determined object neighboring values, and the original value pointers associated with the object neighboring values ​​are arranged in topological code order to form a set of reference pointers. A missing processing record is generated by combining the object boundary version number. When the object to be processed is not missing, the object page difference is compared with the current voltage boundary or the current temperature boundary. When the object page difference exceeds the corresponding boundary, an evidence reference pointer is generated based on the source of the object page difference, and the voltage value or temperature value corresponding to the object code in the current page is used as the evidence net value. An evidence processing record is generated by combining the evidence reference pointer and the object boundary version number.

[0014] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the steps for obtaining the cleaning set by arranging the objects by object code are as follows: For objects to be processed that have not generated missing processing records and evidence processing records, check whether the original value pointer, object neighbor base value, previous page valid original value, object page difference, current voltage boundary or current temperature boundary corresponding to the object type and object boundary version number are complete, and generate a corrected pending state or a corrected undecidable state; when generating a corrected pending state, compare the object page difference with the corresponding boundary, and determine when... If the current original value deviates from the object's neighboring value and the valid original value on the previous page, and if so, the median value among the current original value, the object's neighboring value, and the valid original value on the previous page is taken as the corrected net value, and a corrected processing record is generated. For objects to be processed that have not generated a corrected processing record, the voltage or temperature value corresponding to the object code on the current page is taken as the normal net value, and a normal processing record is generated. The missing processing record, evidence processing record, corrected processing record, and normal processing record are arranged according to the object code to obtain the cleansing set.

[0015] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the steps of merging continuous evidence processing records in the cleaning record table into evidence segments according to topology and time code, and establishing a local retrieval structure from the page table, cleaning record table, and evidence segments are as follows: Read the already formed pages and arrange them in order of time code from first to last to generate a page table. In the page table, retain the time code, topology code, acquisition status, voltage value, temperature value, original value pointer, neighboring base value, and page difference corresponding to each measurement point, and establish the correspondence between the acquisition position within the page and the pointer in the cleaning record; read the cleaning records in the cleaning set in order of object code from smallest to largest, and group the time code, topology code, object type, net value, reference pointer, difference return value, and boundary plate in the cleaning record into the same record item to generate a cleaning record table, and extract the evidence processing records to form an evidence processing record sequence; generate evidence segments based on record groups in the evidence processing record sequence that have the same topology code, the same object type, continuous time codes, and no other processing records in between, and form retrieval keys and corresponding items for the page table, cleaning record table, and evidence segments respectively, and establish a local retrieval structure according to the retrieval key order.

[0016] As a preferred embodiment of the edge computing-based lithium battery energy storage station data preprocessing method of the present invention, the following steps are taken to summarize the lithium battery energy storage station data preprocessing results: The missing processing records and evidence processing records in the cleaning record table are replayed and verified. For missing processing records, the corresponding data is obtained according to the page table position and reference pointer pointed to by the cleaning record, a new net value is regenerated, and the new net value is compared with the net value in the cleaning record. A missing and inapplicable mark is added to the difference return value in the missing processing record. For evidence processing records, the corresponding data is obtained according to the page table position, reference pointer, and boundary of the cleaning record, a new net value is regenerated, and the new net value is compared with the net value in the cleaning record. The new net asset value, difference return value, and page difference out-of-bounds situations are compared with the cleaning records; the correction processing records and normal processing records in the cleaning record table are replayed for verification. The correction processing records are compared with the median value of the current original value, the object's neighboring value, and the valid original value of the previous page for net asset value and difference return value. The normal processing records are compared with the current original value for net asset value and difference return value. A replay result item is generated based on the verification result of each cleaning record. The replay result item corresponding to the evidence processing record is associated with the evidence segment covering the time code. All replay result items are summarized into a replay result set, which is then combined with the page table, cleaning record table, evidence segment, and local retrieval structure to form the data preprocessing result of the lithium battery energy storage station.

[0017] The beneficial effects of this invention are as follows: by calculating the voltage and temperature boundaries based on the voltage and temperature page differences of the stable page set and generating boundary plates, an adaptive benchmark reference is provided for data cleaning, thereby improving the confidence of edge-side data processing; by combining time codes, topology codes, and object types to generate object codes and using object codes to locate data positions, the accuracy of data traceability and the rigor of cleaning logic are ensured, thereby improving the reliability of data preprocessing results for lithium battery energy storage stations. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0019] Figure 1 This is a flowchart of a data preprocessing method for lithium battery energy storage stations based on edge computing.

[0020] Figure 2 Generate a flowchart for page differences.

[0021] Figure 3 Generate a flowchart for the currently available interface.

[0022] Figure 4 Generate a flowchart for cleaning the collection.

[0023] Figure 5 The graph shows the changes in voltage and temperature limits as a function of the boundary plate.

[0024] Figure 6 A distribution chart showing the number of cleaning records for different object types. Detailed Implementation

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0027] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0028] Reference Figures 1-6 This is one embodiment of the present invention, which provides a data preprocessing method for lithium battery energy storage stations based on edge computing, including the following steps:

[0029] S1. Start the edge node to form a continuous sampling time slice and time code. Collect the measurement point operation data and topology level information within the sampling time slice. Form the topology code according to the device address and installation location. Group the measurement point operation data into pages according to the time code and topology code.

[0030] Start the edge node and read the local clock. The example uses a sampling frequency of 1 Hz for data acquisition. The start time of the edge node is used as the first sampling starting point. Sampling time slices are obtained by dividing the time into 1-second intervals. The first sampling time slice corresponds to the time range from the start time to 1 second after start, the second sampling time slice corresponds to the time range from 1 second after start to 2 seconds after start, and so on. Sampling time slices are generated sequentially at the same time intervals. Each sampling time slice includes a start time and an end time.

[0031] The local clock's year, month, and day are extracted as the sampling date, and the start time of the sampling time slice is extracted as the sampling time slice. Starting from the first sampling time slice, the slice numbers are incremented by natural numbers to obtain the slice number. The time code is obtained by concatenating the sampling date, sampling time, and slice number in that order.

[0032] Edge nodes directly collect voltage, temperature, current, state of charge, device status bits, communication status bits, and operating condition status bits within the sampling time slice. Simultaneously, they collect topology-level information in the order of power station, battery compartment, battery cluster, battery pack, module, and individual unit. Edge nodes assign numbers to each level based on the device address and installation location identifier of each measurement point. The station number within the power station to which the measurement point belongs is used as the power station number; the compartment number within the battery compartment to which the measurement point belongs is used as the battery compartment number; the cluster number within the battery cluster to which the measurement point belongs is used as the battery cluster number; the pack number within the battery pack to which the measurement point belongs is used as the battery pack number; the module number within the module to which the measurement point belongs is used as the module number; and the individual unit number to which the measurement point belongs is used as the individual unit number. The topology code is formed by combining the power station number, battery compartment number, battery cluster number, battery pack number, module number, and individual unit number in the order of power station number, battery compartment number, battery cluster number, battery pack number, module number, and individual unit number.

[0033] It should be noted that the device status bit indicates whether the device is operating normally within the current sampling time slice, including normal, alarm, fault, or shutdown; the communication status bit indicates whether the data transmission link is operating normally within the current sampling time slice, including normal, delayed, interrupted, or packet loss; the operating condition status bit indicates the operating condition of the energy storage object within the current sampling time slice, including charging, discharging, or stationary. Topology hierarchy information refers to hierarchical location data representing the physical hierarchy of equipment within the lithium battery energy storage station and the installation location of measurement points, used to clarify the specific physical location from which each collected value originates.

[0034] For each measurement point in the current sampling time slice, an associated time code and corresponding topology code are established. The voltage, temperature, current, state of charge, device status bits, communication status bits, and operating condition status bits corresponding to the measurement point in the same sampling time slice are then grouped into pages. If the voltage or temperature value of a certain measurement point is not collected within the current sampling time slice, a missing marker is generated at the corresponding location of the measurement point. The missing marker corresponds to the time code and topology code already associated with the measurement point.

[0035] For the acquired voltage values, the corresponding time code and topology code are extracted, and the data type is determined to be voltage. The original value pointer of the voltage value is obtained by arranging the data according to the time code, topology code, and data type. For the acquired temperature values, the corresponding time code and topology code are extracted, and the data type is determined to be temperature. The original value pointer of the temperature value is obtained by arranging the data according to the time code, topology code, and data type. For voltage or temperature values ​​with missing tags, no valid original value pointer is generated; only the missing tag, time code, and topology code are retained.

[0036] It should be noted that the original value pointer is used to identify the location information of the original acquisition position of the voltage or temperature value within the current page.

[0037] Extract the operating condition status bit, current value, and state of charge (SOC) of the current page, as well as the SOC of the same energy storage object from the previous page. If the previous page does not exist, generate a failed operating condition verification flag.

[0038] If the operating status bit is charging, check whether the current direction of the current page is consistent with the charging direction, and check whether the current page's state of charge is not less than the previous page's state of charge. If both conditions are met, a passing operating status verification mark is generated; if either condition is not met, a failing operating status verification mark is generated. If the operating status bit is discharging, check whether the current direction of the current page is consistent with the discharging direction, and check whether the current page's state of charge is not greater than the previous page's state of charge. If both conditions are met, a passing operating status verification mark is generated; if either condition is not met, a failing operating status verification mark is generated. If the operating status bit is stationary, a passing operating status verification mark is generated.

[0039] Starting from the beginning of the topology code, the following are sequentially retained: power station number, battery compartment number, battery cluster number, battery pack number, and module number, resulting in a module topology code prefix. Within the current page, voltage values ​​of the same module topology code prefix that are of the same voltage type and have no missing markers are extracted and arranged in ascending order of value, resulting in a voltage value sequence. The median of this voltage value sequence is then calculated to obtain the neighboring base value. Similarly, within the current page, temperature values ​​of the same module topology code prefix that are of the same temperature type and have no missing markers are extracted and arranged in ascending order of value, resulting in a temperature value sequence. The median of this temperature value sequence is then calculated to obtain the neighboring base value.

[0040] It should be noted that the module topology code prefix refers to the topology positioning prefix used to indicate that the measurement point belongs to the same module level, which facilitates the search for reference data within the same module; the voltage value neighbor base value refers to the spatial reference base used to represent the voltage value within the same module, serving as a reference value to determine whether the current voltage value deviates from other measurement points within the same module; the temperature value neighbor base value refers to the spatial reference base used to represent the temperature value within the same module, serving as a reference value to determine whether the current temperature value deviates from other measurement points within the same module.

[0041] If there are no identical data types and no missing voltage or temperature values ​​under the same module topology code prefix, then no neighboring base value is generated, and no neighboring base value is generated without a label.

[0042] For any voltage value on the current page that does not have a missing tag, extract the corresponding time code, topological code, original value pointer, and voltage value neighboring base value. If a voltage value neighboring base value has already been generated, then the voltage value neighboring base value is used as the neighboring base reference for the current voltage value; if the voltage value corresponding to the neighboring base value has not been tagged, then it is confirmed that the current voltage value does not have a neighboring base reference.

[0043] Based on the current voltage value's topology and data type, search the previous page for a voltage value with the same topology, voltage type, and no missing marker. If the voltage value exists on the previous page and a valid original value pointer has been generated for that voltage value, then it is determined as the valid original value for the previous page. If the voltage value does not exist on the previous page, the voltage value with the same topology on the previous page has a missing marker, or the voltage value with the same topology on the previous page has not generated a valid original value pointer, then a valid original value for the previous page is not generated.

[0044] If the voltage value of the current page has already generated a neighboring base reference, then the voltage page difference is calculated using the following expression:

[0045] ;

[0046] in, For voltage page difference, This is the voltage value for the current page. The voltage value of the current page is the neighboring base value. This is the voltage value from the previous page.

[0047] It should be noted that when neighboring base values ​​for voltage values ​​have been generated, and the voltage values ​​corresponding to the same topological code on the previous page have been extracted, the deviation of the current voltage value relative to the neighboring base values ​​is included in the voltage page difference. This reflects the spatial deviation of the current measurement point within the same module. Simultaneously, the change in the current voltage value relative to the voltage value on the previous page is also included in the voltage page difference, reflecting the continuous change of the current measurement point along the time direction. This allows for the simultaneous expression of spatial anomalies and temporal abrupt changes within the same page difference result. The voltage page difference is used to represent the degree of deviation of the current page voltage value from the voltage reference value of the same module or the voltage value of the same measurement point on the previous page.

[0048] If the current voltage value has a neighboring base value that has not been marked, and the voltage value corresponding to the same topological code in the previous page has been extracted, then the voltage page difference for the unmarked neighboring base value is calculated as follows:

[0049] ;

[0050] in, This represents the voltage page difference for which neighboring base values ​​were not generated.

[0051] It should be noted that when neighboring base values ​​for voltage values ​​are not generated, spatial references within the same module cannot be used. If a voltage value corresponding to the same topology code exists on the previous page and has an original value pointer, then the voltage page difference can only be generated based on the change between the current voltage value and the voltage value of the same topology code on the previous page. This setting is because the voltage value on the previous page has already been generated by the preceding page and belongs to the same measurement point as the current voltage value, providing a reference for continuous time direction comparison when neighboring base references are lacking.

[0052] If the current voltage value has already generated a neighboring base value, and the voltage value corresponding to the same topological code in the previous page has not been extracted, then calculate the voltage page difference for the voltage value corresponding to the same topological code in the previous page that has not been extracted, expressed as:

[0053] ;

[0054] in, This represents the voltage page difference where the voltage value corresponding to the same topology code on the previous page was not extracted.

[0055] It should be noted that if the neighboring base value corresponding to the current voltage value has not been marked, and the voltage value corresponding to the same topology code in the previous page has been extracted, the voltage page difference is generated only based on the deviation of the current voltage value from the voltage value neighboring base value. This avoids fabricating the voltage value of the previous page or using the default value, and ensures that the page difference still comes from the already generated current page, voltage value neighboring base value, time code and topology code.

[0056] If the neighboring base value corresponding to the current voltage value is not marked, and the voltage value corresponding to the same topology code in the previous page is not extracted, then no voltage page difference is generated, and a page difference not marked is generated.

[0057] Calculate the temperature page difference using the same method as calculating the voltage page difference.

[0058] It should be noted that the temperature page difference is used to indicate the degree of deviation of a certain temperature value on the current page from the temperature reference value of the same module or the temperature value of the same measuring point on the previous page.

[0059] S2. Using pages as the processing object, stable pages are selected based on the source of measurement points within the page, operating status, and page difference results, and summarized into a stable page set. Voltage and temperature limits are calculated based on the voltage and temperature page differences of the stable page set, and a limit plate is generated. A mapping between the limit plate and the stable page set is established. The currently available limit plate is obtained after the consistency of the number of page differences is confirmed.

[0060] Edge nodes extract the current page's timecode, the topology code associated with each measurement point, device status bits, communication status bits, operating condition verification flags, missing flags at corresponding voltage and temperature values, and the voltage page difference and temperature page difference for each voltage and temperature value, respectively. If the current page has a timecode, a valid timecode state is generated; otherwise, an invalid timecode state is generated. If every measurement point within the current page is associated with a topology code, a valid topology code state is generated; otherwise, an invalid topology code state is generated. If the device status bit is normal, a valid device state is generated; otherwise, an invalid device state is generated. If the communication status bit is normal, a valid communication state is generated; otherwise, an invalid communication state is generated. If the communication status bit is delayed, interrupted, or lost, an invalid communication state is generated. If the operating condition verification flag passes, a valid operating condition state is generated; otherwise, an invalid operating condition state is generated. If there are no missing markers in the current page, a complete acquisition status is generated; if a missing marker exists at the corresponding position of any voltage or temperature value in the current page, an incomplete acquisition status is generated. If both voltage and temperature page differences have been generated in the current page, a valid page difference status is generated; if any voltage or temperature value has a page difference but no marker has been generated, an invalid page difference status is generated.

[0061] If the current page has a valid timecode, valid topology code, valid device, valid communication, valid operating condition, complete acquisition, or valid page difference status, the edge node generates a passed page filtering mark. If the current page has any of the following invalid states: invalid timecode, invalid topology code, invalid device, invalid communication, invalid operating condition, missing acquisition, or invalid page difference status, the edge node generates a failed page filtering mark.

[0062] Pages marked as passed by the page filter are designated as stable pages, and all stable pages are aggregated into a stable page set.

[0063] It should be noted that a stable page refers to a usable sampling page that simultaneously meets the requirements of time code, topology code, device status, communication status, operating condition verification, data acquisition integrity, and page difference validity.

[0064] The edge nodes extract all voltage page differences from the stable page set as voltage page difference samples, and sort them in ascending order to obtain the voltage page difference sequence.

[0065] Edge nodes sequentially read the voltage page difference sequence, using the first read voltage page difference as the first voltage page difference candidate value. They then read the second voltage page difference; if the second voltage page difference differs from the first candidate value, a new candidate value is generated; otherwise, they continue reading the next voltage page difference. This process continues, comparing each read voltage page difference with the last generated candidate value. If the new candidate value differs, a new candidate value is generated; otherwise, they continue reading the next voltage page difference, until the entire voltage page difference sequence has been traversed.

[0066] Candidate voltage page difference values ​​are selected sequentially and compared with the voltage page difference sequence. The number of voltage page differences in the voltage page difference sequence that are not greater than the candidate voltage page difference values ​​is counted. The total number of voltage page differences in the voltage page difference sequence is taken as the voltage page difference count, and the voltage boundary is calculated using the following expression:

[0067] ;

[0068] in, Voltage boundary, It is a minimum value function. Candidate values ​​for voltage page difference. This represents the number of voltage page differences in the voltage page difference sequence that are not greater than the candidate voltage page difference value. This represents the number of voltage page differences.

[0069] It should be noted that the voltage boundary refers to the upper boundary of voltage deviation formed by the voltage page difference in the stable page. It is used to represent the upper limit of the normal deviation range of the voltage value relative to the neighboring base value and the original effective voltage value of the previous page when the current lithium battery energy storage station is in a state where the equipment status bit is normal, the communication status bit is normal, the operating condition verification mark has passed, the data acquisition is complete, and the page difference has been generated. If the maximum voltage page difference in the voltage page difference sequence is directly used as the voltage boundary, a single large voltage page difference will make the voltage boundary too high; if the average value or linear weighted value is used, it will introduce a dependence on the distribution pattern of the page difference. Therefore, the edge node uses the coverage boundary of the voltage page difference sequence to generate the voltage boundary, and selects the smallest candidate value that meets the coverage ratio as the voltage boundary. The coverage ratio is set to 0.95. The coverage ratio is used to determine the coverage degree of the normal deviation range in the stable page difference sample. If the value is too low, the voltage boundary or temperature boundary will be too small, increasing the probability of normal data being misjudged as abnormal. If the value is too high, the voltage boundary or temperature boundary will be too large, reducing the ability to identify real abnormal jumps. The edge node coverage ratio is limited to between 0.90 and 0.98, and in this embodiment, it is set to 0.95, in order to achieve a balance between covering most stable page difference samples and excluding a small number of large page differences at the highest end, thereby improving the stability of the boundary generation results and the reliability of anomaly identification.

[0070] Using the same method as for calculating the voltage limit, a temperature page difference sequence is constructed, and candidate values ​​of temperature page differences are extracted. The number of temperature page differences in the temperature page difference sequence that are not greater than the candidate values ​​of temperature page differences is counted. The total number of temperature page differences in the temperature page difference sequence is counted as the number of temperature page differences, and then the temperature limit is calculated.

[0071] It should be noted that the temperature boundary is the upper limit of temperature deviation determined by the stable page set. It is used to represent the upper limit of the allowable deviation range of the temperature value relative to the temperature reference value of the same module or the temperature value of the same measurement point on the previous page under normal conditions where the equipment, communication, operating conditions, acquisition integrity and page difference all meet the requirements.

[0072] The edge node reads the local clock and uses the current time as the boundary plate generation time. From the boundary plate generation time, the edge node extracts the year, month, and day as the boundary plate date, and the hour, minute, and second as the boundary plate time. The edge node obtains the number of boundary plates already generated locally and increments it by one bit to get the current boundary plate number. The edge node combines the boundary plate date, boundary plate time, and boundary plate number in the order they appear to obtain the boundary plate number. If no boundary plate exists locally, the boundary plate number is set to one; if a boundary plate already exists locally, the incremented result of the previous boundary plate number is used as the current boundary plate number.

[0073] Edge nodes extract the timecodes of each stable page in the stable page set and sort them in ascending order. The first sorted timecode is designated as the stable page start timecode, and the last sorted timecode as the stable page end timecode. If there is only one stable page in the stable page set, its timecode serves as both the stable page start and end timecode. The time range between the stable page start and end timecodes is defined as the stable page time range.

[0074] The boundary plates are compiled in the following order: boundary plate number, boundary plate generation time, number of voltage page differences, number of temperature page differences, voltage boundary, temperature boundary, and stable page time range.

[0075] It should be noted that the boundary version refers to the boundary version data used to record the currently available voltage boundary, temperature boundary and their corresponding stable range.

[0076] After obtaining the boundary map, the edge node establishes a correspondence between the current boundary map and the corresponding stable page set. The edge node extracts the boundary map number, stable page time range, number of voltage page differences, number of temperature page differences, voltage boundary, and temperature boundary from the current boundary map. Simultaneously, it extracts the timecode, topology code, voltage page difference, and temperature page difference for each stable page in the stable page set. The edge node uses the boundary map number as the unique identifier of the current boundary map, the stable page time range as the time boundary of the stable page set corresponding to the current boundary map, and identifies stable pages in the stable page set whose timecodes fall between the start and end timecodes of the stable page as the stable pages corresponding to the current boundary map.

[0077] Edge nodes sort the stable pages corresponding to the current boundary board according to their timecodes in ascending order. For each sorted stable page, the edge node extracts its timecode and the topology code, voltage page difference, and temperature page difference for each measurement point within that stable page. These are then arranged in the order of boundary board number, stable page timecode, and stable page topology code to obtain the stable page corresponding code. If multiple topology codes exist within the same stable page, a stable page corresponding code is generated for each topology code.

[0078] The boundary plate number, stabilization time code, topology code, voltage page difference, temperature page difference, stabilization start time code, and stabilization end time code are summarized into a boundary plate correspondence table according to the stabilization time code in the order from first to last.

[0079] The edge node checks whether the number of voltage page differences in the boundary table matches the number of voltage page differences in the current boundary, and also checks whether the number of temperature page differences in the boundary table matches the number of temperature page differences in the current boundary. If the actual number of voltage page differences in the boundary table equals the number of voltage page differences in the current boundary, and the actual number of temperature page differences in the boundary table equals the number of temperature page differences in the current boundary, the edge node generates a boundary correspondence confirmation flag and designates the current boundary as a usable boundary. If the actual number of voltage page differences in the boundary table does not equal the number of voltage page differences in the current boundary, or the actual number of temperature page differences in the boundary table does not equal the number of temperature page differences in the current boundary, the edge node generates a boundary correspondence anomaly flag. After generating the boundary correspondence anomaly flag, the edge node does not designate the current boundary as a usable boundary, and continues to extract subsequent pages, regenerates the stable page set, voltage page difference sequence, temperature page difference sequence, voltage boundary, and temperature boundary, and regenerates a new boundary, establishing a correspondence between the new boundary and the new stable page set.

[0080] If there is only one available boundary plate for an edge node, then that boundary plate is used as the current available boundary plate. If there are multiple available boundary plates, then the available boundary plate with the latest generation time is used as the current available boundary plate. The voltage and temperature boundaries from the current available boundary plate are extracted as the current voltage and temperature boundaries.

[0081] If a usable boundary version has not yet been generated, the edge node retains the corresponding time code, topology code, original value pointer, missing tag, neighbor base value, and page difference from the already formed page, and continues to extract pages, and continues to determine the stable page condition and generate a stable page set until a usable boundary version is generated and the current usable boundary version is determined. Then, the edge node uses the already retained page as the current page.

[0082] To illustrate the correspondence between the boundary plate generation process and the voltage and temperature boundaries in this invention. Figure 5 This diagram illustrates the changes in voltage and temperature limits under different limit designations. The horizontal axis represents the limit designation, the left vertical axis represents the voltage limit, and the right vertical axis represents the temperature limit; the solid blue line represents the voltage limit, and the dashed red line represents the temperature limit. Figure 5 This demonstrates the voltage and temperature bounds generated by edge nodes under different stable page sets. Figure 5 It can be seen that the voltage and temperature limits corresponding to different limit plate numbers are different, indicating that the present invention can generate corresponding limit plates based on the voltage and temperature page differences in the stable page set, and use the limit plate as the benchmark reference when cleaning the current page data, thus reflecting the correspondence between the voltage limit, temperature limit and limit plate.

[0083] S3. Based on the current page and the currently available boundary, select voltage and temperature values ​​in order of topology code. Generate object code according to time code, topology code and object type. Locate the data position on the current page. Check whether the data position forms a valid acquisition value. Judge the voltage or temperature value at the data position for missing, out-of-bounds and same-side deviation. Generate missing processing record, evidence processing record, correction processing record or normal processing record respectively. Arrange them according to object code to obtain the clean set.

[0084] Edge nodes process the voltage and temperature values ​​of the current page. First, they extract the topology codes associated with all measurement points within the current page. Then, they select measurement points sequentially in ascending order of topology codes. If both voltage and temperature values ​​exist for the same measurement point, the edge node selects the voltage value first, followed by the temperature value. For each selected voltage or temperature value, the edge node extracts the timecode of the current page and uses it as the object's timecode. Next, it extracts the topology codes associated with the corresponding measurement point and uses them as the object's topology code. The object type is determined based on the selected data: if the selected data is a voltage value, the object type is voltage; if the selected data is a temperature value, the object type is temperature. The object codes are then sorted according to the order of object timecode, object topology code, and object type to generate object codes.

[0085] Edge nodes locate the corresponding data position on the current page based on the object code. If the object type is voltage, the edge node locates the voltage value position on the current page corresponding to the object's time code and topology code; if the object type is temperature, the edge node locates the temperature value position on the current page corresponding to the object's time code and topology code. Edge nodes extract missing markers from the data positions. If a missing marker exists at the data position, a missing state is generated indicating the object to be processed is missing; if no missing marker exists at the data position, a missing state is generated indicating the object to be processed is not missing.

[0086] Edge nodes extract the original value pointer based on the object code. If the object type is voltage, and a valid original value pointer has already been formed for the corresponding voltage value in the current page, a valid voltage original value pointer state is generated, and the original value pointer of the voltage value is used as the original value pointer of the voltage object of the object to be processed; if the corresponding voltage value has a missing tag or has not formed a valid original value pointer, an empty voltage original value pointer state is generated. If the object type is temperature, and a valid original value pointer has already been formed for the corresponding temperature value in the current page, a valid temperature original value pointer state is generated, and the original value pointer of the temperature value is used as the original value pointer of the temperature object of the object to be processed; if the corresponding temperature value has a missing tag or has not formed a valid original value pointer, an empty temperature original value pointer state is generated.

[0087] Edge nodes extract the neighboring base value status based on the object code. If the object type is voltage, the edge node extracts the voltage neighboring base value under the corresponding topology code in the current page. If the voltage neighboring base value has already been generated, the neighboring base value status is determined to be valid, and this voltage neighboring base value is used as the voltage object neighboring base value for the object to be processed. If the voltage neighboring base value has not been generated, the neighboring base value status is determined to be empty. If the object type is temperature, the edge node extracts the temperature neighboring base value under the corresponding topology code in the current page. If the temperature neighboring base value has already been generated, the neighboring base value status is determined to be valid, and this temperature neighboring base value is used as the temperature object neighboring base value for the object to be processed. If the temperature neighboring base value has not been generated, the neighboring base value status is determined to be empty.

[0088] It should be noted that the voltage object neighbor base value refers to the value extracted from the voltage value neighbor base values ​​already generated by the module to which the current object belongs in the current page when the current object being processed is a voltage value.

[0089] Edge nodes extract page difference states based on object codes. If the object type is voltage, the edge node extracts the voltage page difference under the corresponding topology code in the current page. If the voltage page difference has already been generated, a valid voltage page difference state is generated, and the voltage page difference is used as the voltage object page difference for the object to be processed; if the voltage page difference has not been generated, an empty voltage page difference state is generated. If the object type is temperature, the edge node extracts the temperature page difference under the corresponding topology code in the current page. If the temperature page difference has already been generated, a valid temperature page difference state is generated, and the temperature page difference is used as the temperature object page difference for the object to be processed; if the temperature page difference has not been generated, an empty temperature page difference state is generated.

[0090] Edge nodes search for valid original values ​​on the previous page based on the object code. If the object type is voltage, the edge node searches the previous page for a voltage value that has the same topology code as the object, is also of voltage type, has no missing markers, and has already formed a valid original value pointer. If a voltage value with the same topology code as the object, is also of voltage type, has no missing markers, and has already formed a valid original value pointer is found, a valid original value state for the previous page is generated and used as the valid original value for the previous page of the object to be processed. At the same time, the original value pointer of the voltage value is used as the original value pointer for the previous page. If no voltage value with the same topology code as the object, is also of voltage type, has no missing markers, and has already formed a valid original value pointer is found, an empty original value state for the previous page is generated. If the object type is temperature, the edge node searches for a temperature value in the previous page that has the same topology code as the object, is of the same type as temperature, has no missing markers, and has already formed a valid original value pointer. If a temperature value with the same topology code as the object, is of the same type as temperature, has no missing markers, and has already formed a valid original value pointer is found, a valid original value state for the previous page temperature is generated and used as the valid original value state for the previous page temperature of the object to be processed. At the same time, the original value pointer of the temperature value is used as the original value pointer for the previous page. If a temperature value with the same topology code as the object, is of the same type as temperature, has no missing markers, and has already formed a valid original value pointer is not found, an empty original value state for the previous page temperature is generated.

[0091] Edge nodes extract the corresponding boundary from the currently available boundary templates based on the object type. If the object type is voltage, the edge node extracts the current voltage boundary from the currently available boundary templates; if the current voltage boundary exists, is in volts, and its value is not less than zero, a valid voltage boundary state is generated. If the object type is temperature, the edge node extracts the current temperature boundary from the currently available boundary templates; if the current temperature boundary exists, is in degrees Celsius, and its value is not less than zero, a valid temperature boundary state is generated. The edge node also extracts the currently available boundary template number and uses this number as the object boundary template number for the object to be processed.

[0092] The edge node summarizes the object code, object time code, object topology code, object type, whether the object to be processed is missing or not, voltage object original value pointer or temperature object original value pointer, voltage object neighbor base value or temperature object neighbor base value, voltage object page difference or temperature object page difference, previous page valid original value of voltage or temperature, previous page original value pointer, current voltage boundary or current temperature boundary, and object boundary version number to generate the object to be processed.

[0093] Edge nodes determine the missing status of the object to be processed. If the object is missing, they check if the object's neighboring value exists. If the neighboring value exists, it is used as the net value of the current object. The edge nodes extract the original value pointers that form a relationship with the object's neighboring value. The edge nodes arrange these original value pointers in ascending order of the topological code corresponding to each original value pointer, generating a set of reference pointers. If the object's neighboring value does not exist, the edge nodes check if a valid original value exists on the previous page. If a valid original value exists on the previous page, it is used as the net value of the current object, and the original value pointer on the previous page is used as a reference pointer. If neither the object's neighboring value nor a valid original value from the previous page exists, the missing marker at the corresponding data position in the current page is retained, and page extraction continues until either the object's neighboring value or a valid original value from the previous page is obtained, at which point the net value of the current object is formed.

[0094] It should be noted that net value refers to the data value used to represent the numerical result obtained after the object to be processed has been processed by the edge nodes. The reference pointer set refers to the set of pointers used to record one or more original data locations referenced by the net value, which facilitates tracing the source of the net value.

[0095] Edge nodes extract the object timecode of the object to be processed as the missing timecode, the object topology code of the object to be processed as the missing topology code, the object type of the object to be processed as the missing object type, the missing state of the object to be processed as the missing processing state, the net value as the missing net value, the set of reference pointers as the missing reference pointers, and the object boundary version number of the object to be processed as the missing boundary version number. Since the missing state is the state where the object to be processed is already missing, there is no valid original value pointer corresponding to the object code in the current page, so the missing original value pointer is empty, and a missing difference return value with a missing inapplicable flag is added. The missing processing records are generated by sorting the missing timecode, missing topology code, missing object type, missing processing state, missing original value pointer, missing net value, missing reference pointer, missing difference return value, and missing boundary version number.

[0096] It should be noted that the missing processing record refers to the processing result record generated when there is a missing object to be processed. It is used to record the location information, processing status, net value source and corresponding boundary information of the missing object.

[0097] If the missing status of the object to be processed is the non-missing status, then check if the object boundary number exists; if the object boundary number exists, then generate a valid boundary number status; if the object boundary number does not exist, then generate an empty boundary number status, the edge node determines the object boundary number as an empty value, determines the boundary number verification result as a missing item status, and associates the missing item status with the object code.

[0098] When generating a valid boundary version number, if the object type is voltage, the edge node extracts the voltage object's original value pointer, voltage object page difference, and current voltage boundary.

[0099] Verify if the voltage object's original value pointer is in a valid state. If the voltage object's original value pointer is valid, the edge node locates the voltage value position in the current page corresponding to the object's time code and topology code based on the voltage object's original value pointer, and uses the voltage value at that position as the evidence voltage's current original value. If the voltage object's original value pointer is not valid, the edge node relocates the voltage value position in the current page based on the object's time code, topology code, and object type, re-extracts the missing marker and original value pointer at the voltage value position, and re-forms the object to be processed based on the re-extracted results.

[0100] Verify if the voltage object page difference is in a valid state. If it is, extract its value and unit. When the unit is volts and the value is not less than zero, the edge node generates three states: voltage page difference existence pass, voltage page difference unit pass, and voltage page difference value pass. These states are then associated with the object code to generate a voltage page difference comparable state, and the voltage object page difference is used as the page difference to be compared. When the unit is not volts or the value is less than zero, an incomparable state is generated. If the voltage object page difference is not in a valid state, an incomparable state is generated.

[0101] Verify if the current voltage boundary is in a valid state. If it is, extract its value and unit. If the unit is volts and the value is not less than zero, the edge node generates a voltage boundary existence pass state, a voltage boundary unit pass state, and a voltage boundary value pass state. These states are then associated according to the object code to generate comparable voltage boundary states, with the current voltage boundary used as the comparison boundary. If the unit is not volts or the value is less than zero, an incomparable voltage boundary state is generated. If the current voltage boundary is not in a valid state, an incomparable voltage boundary state is generated.

[0102] When the object type is voltage, the missing state is the state where the object to be processed is not missing, the voltage original value pointer is valid, the voltage page difference is comparable, the voltage boundary is comparable, and the boundary version number is valid, the edge node compares the page difference to be compared with the boundary to be compared. If the page difference to be compared is greater than the boundary to be compared, a voltage out-of-bounds state is generated; if the page difference to be compared is not greater than the boundary to be compared, a voltage not out-of-bounds state is generated.

[0103] If the object type is temperature, the edge node uses the same method as when the object type is voltage to check the original value pointer of the temperature object, the page difference of the temperature object, and the current temperature boundary, and generates a temperature out-of-bounds state or a temperature not out-of-bounds state.

[0104] When the object type is voltage, the edge node extracts the voltage value corresponding to the object's time code and topology code in the current page as the net voltage evidence value and generates a voltage evidence difference return value with a value of 0; when the object type is temperature, the edge node extracts the temperature value corresponding to the object's time code and topology code in the current page as the net temperature evidence value and generates a temperature evidence difference return value with a value of 0.

[0105] It should be noted that the difference return value refers to the difference between the current original value and the net value after processing, which is used to represent the change in value before and after data cleaning; the evidence difference return value is the voltage evidence difference return value when the object type is voltage, and the temperature evidence difference return value when the object type is temperature.

[0106] Edge nodes generate evidence reference pointers. If the object page difference is formed by both the object's neighboring base values ​​and the valid original values ​​from the previous page, the edge node first extracts the original value pointers that are associated with the object's neighboring base values, and then extracts the original value pointers from the previous page. The original value pointers that are associated with the object's neighboring base values ​​are arranged in ascending order of the topological code of the corresponding measurement point, and are continued from the original value pointers of the previous page to generate a set of evidence reference sources. If the object page difference is formed only by the object's neighboring base values, the edge node extracts the original value pointers that are associated with the object's neighboring base values, and arranges them in ascending order of the topological code of the corresponding measurement point to generate a set of evidence reference sources. If the object page difference is formed only by the valid original values ​​from the previous page, the edge node extracts the original value pointers from the previous page and uses the original value pointers from the previous page as the set of evidence reference sources.

[0107] Edge nodes extract the object timecode of the object to be processed as the evidence timecode, the object topology code of the object to be processed as the evidence topology code, and the voltage or temperature out-of-bounds status as the evidence processing status. If the object type is voltage, the voltage object original value pointer is extracted as the voltage evidence original value pointer; if the object type is temperature, the temperature object original value pointer is extracted as the temperature evidence original value pointer. Edge nodes extract the evidence net value as the evidence net value, the evidence citation source set as the evidence citation pointer, the difference return value as the evidence difference return value, and the object boundary version number as the evidence boundary version number. The evidence processing record is generated by sorting the evidence timecode, evidence topology code, evidence processing status, voltage or temperature evidence original value pointer, evidence net value, evidence citation pointer, evidence difference return value, and evidence boundary version number.

[0108] It should be noted that the evidence processing record refers to the processing result record generated when the object to be processed is determined to be in an out-of-bounds state. It is used to record the location information, original value pointer, net value, reference pointer, difference return value and boundary information of the out-of-bounds object.

[0109] If no missing processing record is generated for the object to be processed, and no evidence processing record is generated, then extract the object type of the object to be processed.

[0110] If the object type of the object to be processed is voltage, the edge node checks the voltage object's original value pointer. If the voltage object's original value pointer is in a valid state, the edge node locates the voltage value position in the current page corresponding to the object's time code and topology code based on the voltage object's original value pointer, and uses the voltage value at that position as the corrected voltage's current original value. If the voltage object's original value pointer is in a null state, the edge node extracts the object's time code, object topology code, and object type from the object code, relocates the voltage value position in the current page, and extracts the missing marker state and original value pointer state of the voltage value position. If a missing marker corresponding to the voltage value position exists in the missing marker field, it is arranged in the order of object code, voltage value position identifier, and missing marker field value to generate a voltage missing marker confirmation status. If no missing marker corresponding to the voltage value position exists in the missing marker field, it is arranged in the order of extracted object code, voltage value position identifier, and missing marker null value to generate a voltage missing marker unconfirmed status. If the original value pointer field is empty, or the original value pointer field cannot locate the voltage value position corresponding to the object time code and object topology code in the current page, it is arranged in the order of object code, voltage value position identifier, and original value pointer null value or original value pointer mismatch value to generate a voltage pointer missing confirmation status.

[0111] If there is no missing marker at the voltage value location, and the original value pointer field can locate the voltage value location in the current page corresponding to the object's time code and topology code, then the edge node will re-determine the original value pointer field as the voltage object's original value pointer and extract the current original voltage value based on the voltage object's original value pointer. If a voltage missing marker confirmation state or a voltage pointer missing confirmation state is generated, a voltage original value missing state is generated in the order of object code, voltage original value pointer empty state, and voltage missing marker confirmation state or voltage pointer missing confirmation state, and the voltage original value missing state is associated with the object code.

[0112] Edge nodes verify the voltage object's neighboring base values. If the voltage object's neighboring base value is valid, the value and unit of the voltage object's neighboring base value are extracted; if the unit is volts, the voltage object's neighboring base value is used as the voltage correction neighboring base value. If the voltage object's neighboring base value is empty, it is arranged in the order of object code and empty neighboring base value state to generate a voltage neighboring base value missing state. If the voltage object's neighboring base value already exists but the unit is not volts, it is arranged in the order of object code, voltage object's neighboring base value value, and voltage object's neighboring base value unit to generate a voltage neighboring base value missing state, and the voltage neighboring base value missing state is associated with the object code.

[0113] It should be noted that the voltage correction neighbor base value refers to the voltage object neighbor base value called during the correction determination phase.

[0114] The edge node checks the previous page's valid original voltage value. If the previous page's valid original voltage value is in the previous page's valid original voltage state, then the value and unit of the previous page's valid original voltage value are extracted; when the unit is volts, the previous page's valid original voltage value is used as the voltage correction for the previous page's original value. If the previous page's valid original voltage value is in the previous page's empty state, then it is arranged according to the object code and the previous page's empty state, generating a previous page's missing original voltage value state. If the previous page's valid original voltage value already exists but the unit is not volts, then it is arranged according to the object code, the previous page's valid original voltage value's value, and the previous page's valid original voltage value's unit, generating a previous page's missing original voltage value state, and associating the previous page's missing original voltage value state with the object code.

[0115] Edge nodes verify voltage object page differences. If the voltage object page difference is valid, its value and unit are extracted; if the unit is volts and the value is not less than zero, the voltage object page difference is used as the voltage correction page difference. If the voltage object page difference is empty, it is arranged in the order of object code and empty page difference state to generate a voltage page difference missing state. If the voltage object page difference already exists but the unit is not volts, or the value of the voltage object page difference is less than zero, it is arranged in the order of object code, voltage object page difference value, and voltage object page difference unit to generate a voltage page difference missing state. The voltage page difference missing state is then associated with the object code.

[0116] Edge nodes check the current voltage boundary. If the current voltage boundary is valid, its value and unit are extracted; if the unit is volts and the value is not less than zero, the current voltage boundary is used as the voltage correction boundary. If the current voltage boundary is not valid, it is arranged in the order of object code and invalid voltage boundary states to generate a voltage boundary missing state. If the current voltage boundary already exists but its unit is not volts, or its value is less than zero, it is combined in the order of object code, current voltage boundary value, and current voltage boundary unit to generate a voltage boundary missing state, and then associated with the object code.

[0117] Edge nodes verify the object boundary number. If the object boundary number exists, it is used as the voltage correction boundary number. If the object boundary number is empty, it is arranged in the order of object code and boundary number null value, generating a boundary number missing status, and associating the boundary number missing status with the object code.

[0118] After verifying the voltage object's original value pointer, voltage object's neighboring base value, previous page's valid original voltage value, voltage object's page difference, current voltage boundary, and object boundary version number item by item, the edge node extracts the object code and queries whether the object code has been associated with the missing voltage original value status, missing voltage neighboring base value status, missing previous page's original voltage value status, missing voltage page difference status, missing voltage boundary status, or missing boundary version number status.

[0119] If the object code is not associated with any of the above missing states, and the current original voltage value, voltage correction neighboring base value, voltage correction previous page original value, voltage correction page difference, voltage correction boundary, and voltage correction boundary version number have all been formed, then the object code, current original voltage value, voltage correction neighboring base value, voltage correction previous page original value, voltage correction page difference, voltage correction boundary, and voltage correction boundary version number are arranged in the order of generating a voltage correction pending state. If the object code is already associated with any of the following missing states: voltage original value missing state, voltage neighboring base value missing state, previous page voltage original value missing state, voltage page difference missing state, voltage boundary missing state, or boundary version number missing state, then the object code and the associated missing states are arranged in the order of generating a voltage correction undeterminable state.

[0120] When a voltage correction pending state is generated, the voltage correction page difference is compared with the voltage correction boundary. If the voltage correction page difference is not greater than the voltage correction boundary, a voltage page difference correction passed state is generated; if the voltage correction page difference is greater than the voltage correction boundary, a voltage correction undecidable state is generated.

[0121] Compare the current original voltage value, the neighboring base value of the voltage object, and the valid original voltage value of the previous page. If the current original voltage value is simultaneously greater than both the neighboring base value of the voltage object and the valid original voltage value of the previous page, or simultaneously less than both the neighboring base value of the voltage object and the valid original voltage value of the previous page, it indicates that the current original voltage value is in the same deviation direction relative to the neighboring base value of the voltage object and the valid original voltage value of the previous page, and is considered as the same-side voltage state. If the current original voltage value is not simultaneously greater than both the neighboring base value of the voltage object and the valid original voltage value of the previous page, or not simultaneously less than both the neighboring base value of the voltage object and the valid original voltage value of the previous page, it indicates that the current original voltage value is not in the same deviation direction relative to the neighboring base value of the voltage object and the valid original voltage value of the previous page.

[0122] It should be noted that the voltage on the same side state refers to a state in which the current original voltage value is in the same direction of being higher or lower than the neighboring base value of the voltage object and the effective original voltage value of the previous page.

[0123] If the current original voltage value is in the same deviation direction relative to the voltage object's neighboring base value and the previous page's valid original voltage value, then the median value among the current original voltage value, the voltage correction neighboring base value, and the voltage correction previous page's original value is taken as the net voltage correction value, expressed as:

[0124] ;

[0125] in, Net voltage correction value, This is the current original value of the voltage. For the neighboring base values ​​of the voltage object, This is the original effective voltage value from the previous page. It is a function for maximizing the value.

[0126] It should be noted that the current original voltage value, the voltage object neighbor base value, and the previous page's valid original voltage value reflect the current acquisition result of the current measurement point, the spatial reference result of the same type of measurement point within the same module, and the time reference result of the previous sampling time slice of the same measurement point, respectively. When the current original voltage value deviates from the latter two on the same side, it indicates that the current original value has deviated from the consistent range of the spatial reference and the time reference. However, since the evidence processing conditions have not been met, the deviation value is not directly retained as the net value. Instead, the value in the middle position is selected from the three generated values ​​with clear sources as the correction result, so that the net value is simultaneously constrained by the current acquisition, spatial reference, and time reference.

[0127] The difference between the current original voltage value and the net voltage correction value is calculated to obtain the voltage correction difference return value.

[0128] Edge nodes extract original value pointers that form a relationship with the neighboring base values ​​of the voltage object, and arrange them in ascending order according to the topology code of the corresponding measurement point to obtain a set of voltage neighboring base reference pointers; extract the original value pointers of the previous page, and arrange them in the order of the set of voltage neighboring base reference pointers and the original value pointers of the previous page to obtain voltage correction reference pointers.

[0129] The voltage correction processing record is obtained by arranging the following information in the order of object time code, object topology code, voltage same-side status, voltage object original value pointer, voltage correction net value, voltage correction reference pointer, voltage correction difference return value, and object boundary version number.

[0130] It should be noted that the voltage correction processing record refers to the processing result record generated when the voltage object to be processed meets the correction conditions. It is used to record the positioning information, original value pointer, correction net value, correction reference pointer, difference return value and boundary information of the voltage object.

[0131] If the object type is temperature, the same method as when the object type is voltage is used to sequentially check the temperature object original value pointer, temperature object neighbor base value, previous page valid original temperature value, temperature object page difference, current temperature boundary and object boundary version number, forming a temperature neighbor base value missing status, previous page temperature original value missing status, temperature page difference missing status, temperature boundary missing status, boundary version number missing status, temperature correction pending status or temperature correction undecidable status, and calculate the temperature correction net value and temperature correction difference return value, generate a temperature correction reference pointer, and obtain the temperature correction processing record.

[0132] It should be noted that the temperature correction processing record refers to the processing result record generated when the temperature object to be processed meets the correction conditions. It is used to record the location information, original value pointer, correction net value, correction reference pointer, difference return value and boundary information of the temperature object.

[0133] If no voltage correction or temperature correction record is generated, the edge node extracts the object type. If the object type is voltage, the voltage value corresponding to the object's time code and topology code in the current page is extracted as the voltage normal net value, and a voltage normal difference return value of 0 is generated. The edge node associates the non-missing state, the state without generated evidence processing record, and the state without generated voltage correction processing record according to the object code to generate a voltage normal state. The voltage normal processing record is generated in the following order: object time code, object topology code, normal processing state, voltage object original value pointer, voltage normal net value, voltage normal difference return value, and object boundary version number.

[0134] If the object type is temperature, generate a normal temperature processing record using the same method as if the object type is voltage.

[0135] It should be noted that the voltage normal processing record refers to the processing result record generated when the voltage object to be processed is not missing, not out of bounds, and does not require correction. It is used to record the location information, original value pointer, normal net value, difference return value, and boundary information of the voltage object. The temperature normal processing record refers to the processing result record generated when the temperature object to be processed is not missing, not out of bounds, and does not require correction. It is used to record the location information, original value pointer, normal net value, difference return value, and boundary information of the temperature object.

[0136] Each missing record, evidence processing record, correction processing record, and normal processing record is treated as a cleansing record and arranged in ascending order of object code to obtain a cleansing set.

[0137] It should be noted that a cleaning record refers to a single processing result record formed after the object to be processed has undergone missing processing, evidence processing, correction processing, or normal processing.

[0138] To illustrate the correspondence between object types and cleaning records in this invention, Figure 6 This displays the number of missing records, evidence processing records, correction processing records, and normal processing records generated for voltage and temperature objects, respectively. The horizontal axis represents the object type, and the vertical axis represents the number of records. Green represents missing records, yellow represents evidence processing records, blue represents correction processing records, and red represents the number of normal processing records. Figure 6 It can be seen that both voltage and temperature objects can form missing processing records, evidence processing records, correction processing records and normal processing records according to the processing logic of the present invention. This shows that the present invention can classify and process data of different object types and record the processing results through object codes, cleaning sets and cleaning record tables.

[0139] S4. Extract pages to form a page table, convert the cleaned set into a cleaned record table, merge continuous evidence processing records in the cleaned record table into evidence segments according to topological codes and time codes, establish a local retrieval structure based on the page table, cleaned record table and evidence segments, replay and verify the net value and difference return value according to the pointing relationship between the cleaned records and the page table and boundary, generate a replay result set, and summarize it into the data preprocessing results of the lithium battery energy storage station.

[0140] Edge nodes read the already formed pages and arrange them in order of timecode from earliest to latest to obtain the page table.

[0141] If a voltage or temperature value has a missing marker, the page table retains the corresponding time code, topology code, and missing marker for the measurement point; if a voltage or temperature value has already formed a valid original value pointer, the page table retains the original value pointer; if a voltage or temperature value has already formed a neighboring base value, the page table retains the neighboring base value; if a voltage or temperature value has already formed a page difference, the page difference retains the page difference.

[0142] After the page table is formed, the edge nodes extract the cleansing records in ascending order of object code, and summarize the time code, topology code, object type, original value pointer, net value, reference pointer, difference return value and boundary of each cleansing record to obtain the cleansing record table.

[0143] The edge node extracts all evidence processing records from the cleaned record table and arranges them in ascending order of topology code and chronological order of time code to obtain an evidence processing record sequence. If the evidence processing record sequence is not empty, the first evidence processing record in the sequence is read, its topology code is extracted as the topology code to be merged, its object type is extracted as the object type to be merged, its time code is extracted as the starting time code to be merged, and the first evidence processing record is set as the first record in the record group to be merged.

[0144] The edge node continues reading the next evidence processing record in the evidence processing record sequence and extracts its topology code and timecode. The edge node compares the topology code of the next evidence processing record with the topology code to be merged, performs a continuity comparison of the timecode of the next evidence processing record with the timecode of the last record in the record group to be merged, and checks the cleaning record table to see if there are any missing, corrected, or normally processed records between the timecode of the last record in the record group to be merged and the timecode of the next evidence processing record.

[0145] If the topology code of the next evidence processing record is the same as the topology code to be merged, the object type of the next evidence processing record is the same as the object type to be merged, and the timecode of the next evidence processing record is consecutive to the timecode of the last record in the record group to be merged, and there are no missing processing records, corrected processing records, or normal processing records between them, then the edge node will append the next evidence processing record to the end of the record group to be merged. Continue reading the next evidence processing record in the evidence processing record sequence, and repeat the topology code comparison, object type comparison, timecode continuity comparison, and intermediate record verification.

[0146] If the topology code of the next evidence processing record is different from the topology code to be merged, or the object type of the next evidence processing record is different from the object type to be merged, or the time code of the next evidence processing record is not continuous with the time code of the last record in the record group to be merged, or there are missing processing records, corrected processing records, or normal processing records between them, then the edge node stops expanding the record group to be merged and treats the record group to be merged as an evidence segment.

[0147] It should be noted that an evidence segment refers to an abnormal evidence segment formed by merging evidence processing records that appear consecutively under the same topological code and the same object type. It is used to centrally characterize the out-of-bounds situation of the same object type at the same measurement point within a continuous time range.

[0148] Add sequentially increasing segment numbers to the evidence segments; use the timecode of the first evidence processing record in the record group to be merged as the start timecode of the evidence segment, and the timecode of the last evidence processing record in the record group to be merged as the end timecode of the evidence segment; use the topology code to be merged as the topology code of the evidence segment; count the number of evidence processing records in the record group to be merged as the evidence quantity of the evidence segment; extract the object codes corresponding to each evidence processing record in the record group to be merged and summarize them into the object code set of the evidence segment; use the boundary number corresponding to the evidence processing record in the record group to be merged as the boundary of the evidence segment.

[0149] After generating an evidence segment, the edge node will take the next evidence processing record that caused the interruption as the first record in the new record group to be merged, and generate a new evidence segment, until the last evidence processing record is read and the last evidence segment is generated.

[0150] When generating each evidence segment, the edge node first searches the page difference corresponding to each evidence processing record in the page table one by one, based on the object code set in the evidence segment. If the object type is voltage, the corresponding voltage page difference is extracted, and the voltage page differences corresponding to each evidence processing record in the evidence segment are compared with each other, with the largest voltage page difference being taken as the maximum voltage page difference of the evidence segment; if the object type is temperature, the corresponding temperature page difference is extracted, and the temperature page differences corresponding to each evidence processing record in the evidence segment are compared with each other, with the largest temperature page difference being taken as the maximum temperature page difference of the evidence segment.

[0151] For a page in the page table, extract the topology code and time code of the measurement points within the page, and set the boundary plate to empty. Arrange the topology code and time code of the measurement points within the page in order as the page table retrieval key. Extract the time code, topology code, object type, voltage or temperature value, missing marker, original value pointer, neighbor base value, and page difference of the measurement points within the page in the page table as the corresponding content of the page table. Summarize the page table retrieval key and the corresponding content of the page table into a page table corresponding item. For a cleaning record in the cleaning record table, extract the topology code, time code, and boundary plate of the cleaning record itself, and arrange them in order as the cleaning record retrieval key. At the same time, extract the original value pointer, net value, reference pointer, and difference return value of the cleaning record as the corresponding content of the cleaning record. Summarize the cleaning record retrieval key and the corresponding content of the cleaning record into a cleaning record corresponding item. For each evidence segment, its topology code, start time code, end time code, and boundary plate are extracted. These are then arranged in the order of the evidence segment's topology code, start time code, end time code, and boundary plate to form the evidence segment retrieval key. Simultaneously, the segment number, number of pieces of evidence, maximum voltage page difference or maximum temperature page difference, and object code set are extracted from the evidence segment as its corresponding content. The evidence segment retrieval key and the corresponding content are then summarized into evidence segment corresponding items. The page table corresponding items, cleaning record corresponding items, and evidence segment corresponding items are arranged in ascending order of the retrieval key; if the retrieval keys are the same, they are arranged in ascending order of the time code; if the time codes are also the same, they are arranged in the order of voltage preceding temperature within the object type, forming a local retrieval structure.

[0152] It should be noted that the maximum voltage page difference or maximum temperature page difference in the evidence segment is used to characterize the degree of page difference deviation of the corresponding object type within the evidence segment, the number of evidence is used to characterize the number of consecutive evidence processing records within the evidence segment, and the object code set is used to trace the evidence processing records covered by the evidence segment. The segment number, the number of evidence, the maximum voltage page difference or maximum temperature page difference, and the object code set are stored in the local retrieval structure as the corresponding content of the evidence segment along with the evidence segment retrieval key for the auxiliary display and playback tracing of the evidence segment.

[0153] If the evidence processing record sequence is empty, the edge nodes will not generate evidence segments, and the page table and the cleaning record table will be used to form search keys and corresponding items, respectively. The local search structure will be established according to the search key order.

[0154] It should be noted that the page table retrieval key refers to the index identifier used to locate the corresponding item in the page table, characterized by the topology code and time code of the test point within the page. The cleaning record retrieval key refers to the index identifier used to locate the corresponding item in the cleaning record table, characterized by the topology code, time code, and boundary of the cleaning record itself. The evidence segment retrieval key refers to the index identifier used to locate the corresponding item in the evidence segment, characterized by the topology code, start time code, end time code, and boundary of the evidence segment itself. The local retrieval structure is used to represent the local association index relationship between the page table, the cleaning record table, and the evidence segment, and to support edge nodes or external systems in quickly locating the corresponding page data, cleaning record, and evidence segment according to the time code, topology code, and object type, and to perform traceability queries and playback verification of the cleaning results.

[0155] Verify the cleaning record. If the current cleaning record is a missing record, the edge node locates the page with the same timecode and topology code in the page table based on the timecode and topology code in the cleaning record, and checks whether there is a missing marker in the page. If the reference pointer in the cleaning record points to the original value pointer that forms a relationship with the neighboring base value, the edge node retrieves the voltage or temperature value that participated in the formation of the neighboring base value from the page table based on the reference pointer, and re-obtains the neighboring base value according to the neighboring base value formation method. The re-obtained neighboring base value is then used as the new net value for verification. If the reference pointer in the cleaning record points to a valid original value on the previous page, the edge node retrieves the valid original value on the previous page from the page table based on the reference pointer, and uses the valid original value on the previous page as the new net value for verification. If the reference pointer in the cleaning record is null, a null value is used as the new net value for verification. If the new net value matches the net value in the cleaning record, a playback result with a pass status is generated; if the new net value does not match the net value in the cleaning record, an playback result with an abnormal status is generated.

[0156] It should be noted that the playback result is used to indicate whether the cleaning record is consistent with the page table, reference pointer, difference return value and boundary information after playback verification.

[0157] If the current cleansing record belongs to the evidence processing record, the edge node retrieves the voltage or temperature value of the current page from the page table based on the original value pointer in the cleansing record, and uses the voltage or temperature value of the current page as the new net value for verification. The edge node retrieves the original value pointer that forms a relationship with the neighboring base value and the valid original value of the previous page based on the reference pointer in the cleansing record, and retrieves the voltage or temperature boundary based on the boundary plate in the cleansing record. If the object type is voltage, the edge node retrieves the voltage page difference from the page table and checks whether the voltage page difference is greater than the voltage boundary. If the object type is temperature, the edge node retrieves the temperature page difference from the page table and checks whether the temperature page difference is greater than the temperature boundary. If the new net value matches the net value in the cleansing record, the difference return value is zero, and the page difference is greater than the boundary, a playback result with a "passed" status is generated; if any item is inconsistent, an playback result with an "abnormal" status is generated.

[0158] If the current cleaning record is a correction record, the edge node retrieves the voltage or temperature value of the current page from the page table based on the original value pointer in the cleaning record, and retrieves the neighboring base value and the valid original value of the previous page from the page table based on the reference pointer in the cleaning record. The edge node uses the median value among the current voltage or temperature value, the neighboring base value, and the valid original value of the previous page as the new net value for verification. The difference between the voltage or temperature value of the current page and the corresponding net value of the same type is calculated to obtain a new difference return value. If the new net value is consistent with the net value in the cleaning record, and the new difference return value is consistent with the difference return value in the cleaning record, a playback result with a passing status is generated; if either the net value or the difference return value is inconsistent, an playback result with an abnormal status is generated.

[0159] If the current cleaning record is a normal processing record, the edge node retrieves the voltage or temperature value of the current page from the page table based on the original value pointer in the cleaning record, and uses the voltage or temperature value of the current page as the new net value for verification. The edge node checks whether the new net value matches the net value in the cleaning record and whether the difference return value is zero. If the new net value matches the net value in the cleaning record and the difference return value is zero, a playback result with a "passed" status is generated; if either is inconsistent, a playback result with an "abnormal" status is generated.

[0160] After verifying the current cleansing record, the edge node generates a playback result item for that record in the following order: timecode, topology code, object type, original value, net value, reference pointer, difference return value, boundary plate, and playback result. If the current cleansing record is an evidence processing record, the edge node searches for evidence segments with the same topology code, boundary plate, and time range covering the timecode, based on the timecode, topology code, and boundary plate in the cleansing record. This evidence segment is then added to the playback result item, establishing a connection between the evidence segment and the playback result item. All playback results are then aggregated to obtain the playback result set.

[0161] The edge nodes summarize the page table, cleaning record table, evidence segment, local retrieval structure and playback result set to obtain the data preprocessing results of the lithium battery energy storage station.

[0162] In summary, this invention improves the confidence of edge-side data processing by calculating voltage and temperature boundaries based on voltage and temperature page differences in the stable page set and generating boundary plates, thus providing an adaptive benchmark for data cleaning. Furthermore, by combining time codes, topology codes, and object types to generate object codes and using these object codes to locate data positions, the invention ensures accurate data traceability and rigorous cleaning logic, thereby enhancing the reliability of data preprocessing results for lithium battery energy storage stations.

[0163] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data preprocessing method for lithium battery energy storage stations based on edge computing, characterized in that, include: Start the edge node to form continuous sampling time slices and time codes. Collect the measurement point operation data and topology level information within the sampling time slice. Form the topology code according to the device address and installation location. Group the measurement point operation data into pages according to the time code and topology code. Using pages as the processing object, stable pages are selected based on the source of measurement points within the page, operating status, and page difference results, and then summarized into a stable page set. Voltage and temperature limits are calculated based on the voltage and temperature page differences of the stable page set, and a limit board is generated. A mapping between the limit board and the stable page set is established, and the currently available limit board is obtained after the consistency of the number of page differences is confirmed. Based on the current page and the currently available boundary, select voltage and temperature values ​​in order of topology code, generate object codes according to time code, topology code and object type, locate the data position on the current page, check whether the data position forms a valid acquisition value, judge the voltage or temperature value of the data position for missing, out-of-bounds and same-side deviation, and form missing processing record, evidence processing record, correction processing record or normal processing record respectively, and arrange them according to object code to obtain the clean set; Extract pages to form a page table, convert the clean set into a clean record table, merge continuous evidence processing records in the clean record table into evidence segments according to topological code and time code, establish a local retrieval structure based on the page table, clean record table and evidence segments, replay and verify net value and difference return value according to the pointing relationship between clean records and page table and boundary, generate a replay result set, and summarize it into the data preprocessing results of lithium battery energy storage station; The steps for establishing the mapping between boundary pages and the stable page set, and generating the currently available boundary page after confirming the consistency of page difference count, are as follows: Based on the boundary number and stabilization time range in the boundary, extract the stabilization pages whose time codes fall within the stabilization time range from the stabilization page set, sort them in order of time codes from first to last, extract the time code, topology code, voltage page difference, and temperature page difference for each stabilization page, and generate the corresponding stabilization page code according to the boundary number, time code, and topology code. Associate the page stabilization code with the voltage page difference, temperature page difference, page stabilization start time code, and page stabilization end time code in the corresponding page stabilization, and summarize them in the order of page stabilization time codes to generate a boundary page correspondence table; The actual number of voltage page differences and temperature page differences in the boundary plate correspondence table are compared with the number of voltage page differences and temperature page differences in the boundary plate, respectively. When the numbers are consistent, the boundary plate is used as a usable boundary plate. Based on the boundary plate generation time, the current usable boundary plate is determined from the usable boundary plates, and the current voltage boundary and the current temperature boundary are obtained.

2. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 1, characterized in that, The process of forming continuous sampling time slices and time codes, collecting measurement point operation data and topology hierarchy information within the sampling time slices, and generating topology codes based on device addresses and installation locations is as follows: The edge node reads the local clock, determines the start time as the first sampling start point, and divides the sampling time slices. Each sampling time slice records the start time and end time, and the start time of the next sampling time slice is connected to the end time of the previous sampling time slice. Edge nodes extract the year, month, and day from the local clock as the sampling date, extract the start time from the sampling time slice as the sampling time, and increment the slice number according to the generation order of the sampling time slices. The sampling date, sampling time, and slice number are then combined into a time code. Edge nodes collect measurement point operation data and topology hierarchy information within the sampling time slice, determine the numbering of each level according to the equipment address and installation location of the measurement point, and combine them into topology codes according to the physical hierarchy of power station, battery compartment, battery cluster, battery pack, module and individual unit; The measurement point operation data includes voltage value, temperature value, current value, state of charge, equipment status bit, communication status bit, and operating condition status bit.

3. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 2, characterized in that, The steps for grouping the measurement point operation data into pages according to time code and topology code are as follows: Associate the time code and topology code with the measurement point running data within the current sampling time slice to generate a page. When the voltage or temperature value is not collected, a missing mark is formed at the corresponding data position. When the voltage and temperature values ​​have been collected, the original value pointer is generated according to the time code, topology code and data type. The module to which the measurement point belongs is determined based on the topology code, and the same type of voltage or temperature values ​​that are not missing under the same module are aggregated in the current page. The neighboring base values ​​of voltage and temperature are generated by calculating the median value. The voltage page difference or temperature page difference is calculated by taking the voltage or temperature values ​​that are not missing in the current page as objects, combining the voltage or temperature neighboring base values, and the valid original values ​​under the same topology code in the previous page.

4. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 1, characterized in that, The steps for calculating voltage and temperature limits based on the voltage and temperature page differences of the stable page set, and generating the limit board, are as follows: Using the page as the judgment object, a source verification status is generated based on the source of the measurement points within the page, an operation verification status is generated based on the operation status within the page, and a page difference verification status is generated based on the page difference result. When the source verification status, operation verification status, and page difference verification status all meet the page stability condition, a page screening mark that has passed is generated, and the corresponding page is determined as a stable page. The stable pages marked as passed by the page filter are summarized into a stable page set. The voltage page difference and temperature page difference are extracted from the stable page set, and arranged in numerical order to generate voltage page difference sequence and temperature page difference sequence respectively. The same page difference is merged to obtain the voltage page difference candidate value and temperature page difference candidate value. The number of page differences not exceeding each voltage page difference candidate value and temperature page difference candidate value is counted sequentially. The voltage boundary and temperature boundary are determined according to the number of page differences. The boundary plate is generated by combining the boundary plate number, boundary plate generation time, number of page differences and page stabilization time range.

5. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 1, characterized in that, The steps for verifying whether a valid data location has been collected are as follows: Based on the current page and the currently available boundary, the measurement points are arranged in ascending order of the topology code within the current page, and the voltage and temperature values ​​of the same measurement point are treated as different processing objects to generate a sequence of processing objects. Extract the current processing object from the processing object sequence, use the current page's time code as the object's time code, use the topology code of the measurement point corresponding to the current processing object as the object's topology code, and determine the object type based on the data category of the current processing object to generate the object code; The system locates the data position on the current page based on the object code and checks whether the data position has formed a valid collection value. If no valid collection value has been formed, a state indicating that the object to be processed is missing is generated. If a valid collection value has been formed, a state indicating that the object to be processed is not missing is generated.

6. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 5, characterized in that, The steps for creating missing item processing records and evidence processing records are as follows: Locate the data position of the current page along the object code, and obtain the corresponding original value pointer, page difference and neighbor base value according to the object type. Determine the obtained neighbor base value as the object neighbor base value. Obtain the valid original value under the same topology code from the previous page, and extract the current voltage boundary or current temperature boundary from the currently available boundary board. Determine the boundary board number of the currently available boundary board as the object boundary board number, and generate the object to be processed. When the object to be processed is in a missing state, the net value is formed with the already determined object neighbor base value, and the original value pointers associated with the object neighbor base value are arranged into a set of reference pointers in topological code order, and a missing processing record is generated by combining the object boundary version number; When the object to be processed is in a non-missing state, the object page difference is compared with the current voltage limit or the current temperature limit. When the object page difference exceeds the corresponding limit, an evidence reference pointer is generated based on the source of the object page difference, and the voltage value or temperature value corresponding to the object code in the current page is used as the evidence net value. An evidence processing record is generated by combining the evidence reference pointer and the object limit version number.

7. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 6, characterized in that, The steps for obtaining the cleaned set by arranging it according to object code are as follows: For objects to be processed that have not generated missing processing records and evidence processing records, check whether the original value pointer, object neighbor base value, previous page valid original value, object page difference, current voltage boundary or current temperature boundary corresponding to the object type and object boundary version number are complete, and generate a corrected pending state or corrected undecidable state. When generating the correction pending state, the object page difference is compared with the corresponding boundary, and it is determined whether the current original value deviates from the object neighbor base value and the previous page valid original value on the same side. If the current original value deviates from the object neighbor base value and the previous page valid original value on the same side, the median value of the current original value, the object neighbor base value and the previous page valid original value is taken as the correction net value, and the correction processing record is generated. For objects awaiting processing that have not yet generated a correction processing record, the voltage or temperature value corresponding to the object code on the current page will be used as the normal net value to generate a normal processing record. Arrange the missing processing records, evidence processing records, correction processing records, and normal processing records according to their object codes to obtain the cleaned set.

8. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 1 or 7, characterized in that, The steps for merging continuous evidence processing records from the cleaning record table into evidence segments based on topological and time codes, and establishing a local retrieval structure using the page table, cleaning record table, and evidence segments, are as follows: Read the already formed pages and arrange them in order of time code from first to last to generate a page table. In the page table, retain the time code, topology code, acquisition status, voltage value, temperature value, original value pointer, neighboring base value and page difference corresponding to each measurement point, and establish the correspondence between the acquisition position in the page and the pointer in the cleaning record. Read the cleansing records in the cleansing set in ascending order of object code, and group the time code, topology code, object type, net value, reference pointer, difference return value and boundary plate in the cleansing record into the same record item to generate a cleansing record table, and extract the evidence processing records to form an evidence processing record sequence; Evidence segments are generated from record groups in the evidence processing record sequence that have the same topological code, the same object type, consecutive time codes, and no other processing records in between. The page table, the cleaning record table, and the evidence segments are respectively used to form search keys and corresponding items, and a local search structure is established according to the search key order.

9. The data preprocessing method for lithium battery energy storage stations based on edge computing as described in claim 8, characterized in that, The data preprocessing results of the lithium battery energy storage station are summarized as follows: The missing handling records and evidence handling records in the cleaning record table are replayed and verified. For missing handling records, the corresponding data is obtained according to the page table position and reference pointer pointed to by the cleaning record, a new net value is regenerated, and the new net value is compared with the net value in the cleaning record. A missing and inapplicable mark is added to the difference return value in the missing handling record. For evidence handling records, the corresponding data is obtained according to the page table position, reference pointer and boundary of the cleaning record, a new net value is regenerated, and the new net value, difference return value and page difference out-of-bounds situation are compared with the cleaning record. The correction and normal processing records in the cleaning record table are replayed and verified. The correction and processing records are compared with the net value and the difference return value by the median value of the current original value, the object's neighboring base value and the previous page's valid original value. The normal processing records are compared with the net value and the difference return value by the current original value. Based on the verification results of each cleaning record, a playback result item is generated. The playback result item corresponding to the evidence processing record is associated with the evidence segment covering the time code. All playback result items are summarized into a playback result set, which is then combined with the page table, cleaning record table, evidence segment, and local retrieval structure to form the data preprocessing result of the lithium battery energy storage station.

Images