A method, equipment, and medium for constructing a data middle platform based on the Industrial Internet

By acquiring and processing multi-source data in the industrial internet data platform, identifying and supplementing data changes, and generating data processing range records, the problem of incomplete data range determination after changes in multi-level data processing relationships in the data platform is solved, thus achieving accuracy and completeness in data processing.

CN123068487APending Publication Date: 2026-11-10国家能源集团永州发电有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611190894.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-06
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing industrial internet data platforms struggle to fully determine the scope of affected data after changes in multi-level data processing relationships, leading to omissions in the data processing scope, newly added related records, and records outside the scope referenced by query results, making further corrections difficult.

Method used

By acquiring multi-source industrial data and data processing tasks from the industrial internet platform, recording input-output relationships and data versions, generating basic records, identifying changes in fields, objects, and processing rules, cyclically supplementing records to be processed along time windows and data association grouping and aggregation relationships, tracing back to actual reference records, generating data processing input-output reference range confirmation records, establishing query indexes, and supplementing query differences to the data processing range.

Benefits of technology

The data platform construction results have been improved, ensuring that newly added related records and affected grouped records are included in the processing scope, correcting records outside the scope of sorting, pagination, and grouping queries, and improving the accuracy and completeness of data processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123068487A_ABST
    Figure CN123068487A_ABST
Patent Text Reader

Abstract

This invention discloses a method, device, and medium for constructing a data middle platform based on the Industrial Internet, relating to the field of Industrial Internet data processing technology. The method includes: acquiring multi-source industrial data, industrial objects, and processing tasks; recording the input-output relationships and data versions of source records; identifying changes in fields, objects, processing rules, and source records; cyclically supplementing records to be processed along time windows, data associations, and grouping aggregation relationships; executing candidate processing, corresponding to current outputs and candidate outputs, tracing back to the actual referenced records from the first difference task, and supplementing missing inputs and outputs to be completed; generating candidate data asset versions and query indexes; performing corresponding version queries under the same user input and data cutoff boundary on current and candidate versions; supplementing query differences not corresponding to output changes into the data processing scope; and generating the data middle platform construction result. This invention improves the completeness of determining the scope of impact of data changes and correcting the scope of query results.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of industrial internet data processing technology, and in particular to a method, device and medium for constructing a data middle platform based on the industrial internet. Background Technology

[0002] Industrial Internet platforms typically aggregate production process data and business data from distributed control systems, plant-level monitoring information systems, equipment management systems, edge gateways, relational databases, and message queues. Through field standardization, object association, time window processing, data aggregation, partitioned storage, and query services, they form a data platform for equipment operation analysis, production reports, business management, and digital twin applications.

[0003] However, existing methods still have two limitations: First, after changes in field mapping, processing rules, or supplementary data, the impact will continue to propagate along the time window, data association, and group aggregation. When determining the processing scope based on a fixed time range or the original data lineage, newly added associated records after recalculation may be missed. Second, after data assets are updated, query sorting, pagination, and grouping results may still reference records outside the scope, making it difficult to continue to correct the data processing scope. Summary of the Invention

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

[0005] Therefore, this invention provides a method for constructing a data platform based on the Industrial Internet, which solves the problem that the Industrial Internet data platform is unable to fully determine the scope of affected data and continue to supplement the processing scope based on query differences after changes in multi-level data processing relationships.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for constructing a data platform based on the Industrial Internet, comprising: acquiring multi-source industrial data, industrial object data, and data processing tasks from an Industrial Internet platform; recording the input-output relationships and data versions of source records in each task; generating a basic record for constructing the Industrial Internet data platform; identifying fields, objects, processing rules, and changes in source records based on the basic record for constructing the Industrial Internet data platform; using the field values ​​before and after the changes to iteratively supplement the records to be processed along time windows, data associations, and grouping aggregation relationships; generating a data change processing range record; executing candidate processing through the data change processing range record; corresponding to the current output and candidate output; tracing back the actual reference record from the processing task that first produced a difference; supplementing missing inputs and outputs to be completed; generating a data processing input-output reference range confirmation record; generating candidate data asset versions and establishing a query index based on the data processing input-output reference range confirmation record; performing a corresponding version query under the same user input and data cutoff boundary on the current version and candidate version; and back-supplementing the query differences that could not be matched to output changes to the data processing range; generating a data platform construction result.

[0007] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the step of acquiring multi-source industrial data, industrial object data, and data processing tasks from the Industrial Internet platform, and recording the input-output relationships and data versions of source records in each task includes: registering multi-source industrial data according to data source number, field structure version, and industrial object number, and generating source record numbers; during field conversion, time window division, data association, group aggregation, and data asset writing, recording task execution batches, input records, output records, field values ​​before and after task execution, record reference relationships, and data versions, and generating basic records for the construction of the Industrial Internet data platform according to the task sequence.

[0008] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the following steps are taken: the basic record identification fields, objects, processing rules, and source record changes of the data platform based on the Industrial Internet are identified; the records to be processed are supplemented cyclically by using the field values ​​before and after the change along the time window, data association, and grouping aggregation relationship; and the data change processing range record is generated by: comparing the data source field structure, industrial object correspondence, data processing rules, and source record content of the current version and the candidate version respectively; determining the change effective time according to the field structure effective time, industrial object version effective time, processing rule start time, and source record business time; and extracting the initial records to be processed based on the change effective time and data processing path.

[0009] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the following steps are included: The construction of the basic record identification field, object, processing rule, and source record changes based on the Industrial Internet data platform; the generation of a data change processing range record by cyclically supplementing the records to be processed using the field values ​​before and after the change along the time window, data association, and grouping aggregation relationships; performing data processing on the initial records to be processed according to candidate versions; comparing the time values, associated field values, and grouping field values ​​before and after the change; reading records within the current time window and candidate time windows when the time value changes; reading current association results and candidate association results when the associated field value changes; reading records within the current group and candidate group when the grouping field value changes; and reprocessing newly added records from the data processing task they first participated in until no new source records, intermediate records, grouping field values, data partitions, or output row keys are generated in each round, thus generating a data change processing range record.

[0010] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the step of executing candidate processing through data change processing range records, corresponding to the current output and candidate output, includes: matching the current output and candidate output according to the corresponding fields of the output record configured by the business primary key, combined unique field, or data processing task; marking the current output as in use, updated, or ended; marking the candidate output as in use, updated, or newly added; and determining the data processing task with the first occurrence of output record quantity, field value, source record, data version, or output row key difference according to the task execution order.

[0011] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the step of tracing back the actual reference records from the processing task that first generates a difference, supplementing missing inputs and outputs to be completed, and generating a data processing input / output reference range confirmation record includes: reading the source records and intermediate records actually referenced by the candidate output from the data processing task that first generates a difference, adding the reference records that are not within the data change processing range to the pending processing records; reading the current output that is no longer generated in the candidate processing and the data processing task that references the current output, adding the corresponding input records and tasks to the re-execution range, repeating the candidate processing and output status determination until all candidate output reference records are within the pending processing record range and all current affected outputs have completed status marking, and generating a data processing input / output reference range confirmation record.

[0012] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the step of generating candidate data asset versions and establishing a query index based on the data processing input / output reference range confirmation record includes: writing candidate outputs in the updated and newly added states into candidate data asset records according to the candidate output status; establishing a version reference relationship between candidate outputs in the continued state and the corresponding current data asset records; writing the current outputs in the ended state into the data asset version record to be ended; establishing a candidate version query index according to the data asset number, data partition, data asset row key and candidate version number, and establishing a joint query index with the industrial object number, data project number and data time.

[0013] As a preferred embodiment of the data platform construction method based on the Industrial Internet described in this invention, the step of performing corresponding version queries on the current version and candidate versions under the same user input and data cutoff boundary, and back-complementing the query differences that fail to correspond to output changes to the data processing scope to generate the data platform construction result includes: within the same data construction snapshot range, performing data service queries using the current data asset version and candidate data asset version respectively, and corresponding the query differences with the candidate output status and data service configuration changes; for return record differences, grouping differences, sorting differences, or pagination boundary differences that cannot form a corresponding relationship, tracing the source record according to the reference relationship between data asset records, intermediate records, and source records and supplementing the data processing scope, repeatedly generating candidate data asset versions and performing data service queries until no new processing records or uncorresponding query differences are generated, and generating the data platform construction result.

[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the data platform construction method based on the Industrial Internet as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the data platform construction method based on the Industrial Internet as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By cyclically supplementing the records to be processed along the time window, data association, and group aggregation relationship, a data change processing range record is generated, so that newly added associated records and affected group records are included in the processing range. This is used to determine the data recalculation boundary of the industrial internet data platform and improve the completeness of the identification of the scope of change impact. By establishing candidate data asset versions and query indexes, corresponding version queries are performed on the current version and candidate versions under the same user input and data cutoff boundary. Query differences that fail to correspond to output changes are supplemented into the data processing range. This is used to correct records outside the scope involved in sorting, pagination, and grouping queries and improve the completeness of the data platform construction results. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the method for constructing a data middle platform based on the Industrial Internet.

[0019] Figure 2 Define the flowchart for the scope of data change processing.

[0020] Figure 3 Flowchart for confirming the input / output reference range for data processing.

[0021] Figure 4 Flowchart for constructing candidate data asset versions and confirming query scope.

[0022] Figure 5 A comparison chart showing the completeness rate of the range of records to be processed under different changing scenarios.

[0023] Figure 6 A graph showing the change in the number of unexplained query discrepancies under different data service query scenarios. Detailed Implementation

[0024] 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.

[0025] 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.

[0026] 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.

[0027] Reference Figures 1-6 This is one embodiment of the present invention, which provides a method for constructing a data middle platform based on the Industrial Internet, including the following steps: S1. Acquire multi-source industrial data, industrial object data, and data processing tasks from the industrial internet platform, record the input-output relationships and data versions of the source records in each task, and generate basic records for building the industrial internet data platform.

[0028] Acquire the data source configuration, equipment ledger, field structure, data processing task configuration, data asset structure, and data service configuration of the industrial internet platform; register the data source, field structure, industrial object, and field correspondence versions; generate source record numbers for received data; perform format parsing, field conversion, time window division, data association, group aggregation, and data asset writing according to the initially constructed record set; record the actual input of the processing task, the field values ​​involved in association and aggregation processing before and after task execution, and the reference relationship between output records and input records; associate the data asset row key, data partition, version, and data service call results; establish a query index; and generate the basic records for building the industrial internet data platform.

[0029] The Industrial Internet platform receives industrial equipment data, production process data, and production and operation data through industrial communication protocols, database interfaces, message queues, and application programming interfaces (APIs). Industrial communication protocols include the Open Platform Unified Communication Architecture Protocol, Modbus communication protocol, and power system remote control communication protocol. Database interfaces read relational database tables from distributed control systems, plant-level monitoring information systems, and equipment management, defect management, maintenance management, fuel management, and environmental management systems. The Industrial Internet data platform server registers each access source as a data source object, recording the data source number, data source type, data providing system number, access method, interface version, acquisition task number, field structure version, and validity period.

[0030] The Industrial Internet data platform server includes at least one data processor, memory, database storage device, and network communication interface. The memory stores program instructions for performing data access, data processing tasks, record reference relationship tracing, candidate data asset version construction, query playback, and version release. The Industrial Internet data platform server executes program instructions to read, compare, match, classify, update, and store source records, data processing task configurations, intermediate records, data asset records, query indexes, and data service call records. The Industrial Internet data platform server reads data change processing range records, executes candidate data processing according to the data processing task configuration, matches current output records with candidate output records, determines the processing task where the result first differs, and traces the source record along the intermediate record source relationship.

[0031] Industrial measurement point data records include the edge gateway number, acquisition channel, measurement point address, sampling period, and source time fields; relational database table data records the database instance number, table name, primary key field, and incremental identification field; message queue data records the message topic, message partition, message offset, and message generation time; and application programming interface (API) data records the interface number, returned data path, and call batch. If the primary key field, message offset, or time and sequence fields that can distinguish consecutive records from the same measurement point are missing, the alternative unique field, database log sequence number, or sequence number within the acquisition batch is read sequentially. If a stable identifier still cannot be obtained, a record identifier missing condition flag is generated, and the corresponding data is only written to the original data storage area. No source record number is generated, and it is not involved in modification, deletion, or version matching. The record field value sequence hash value is only used to check for duplicate records within the same acquisition batch.

[0032] The industrial internet data platform server reads field names, field order, field data types, field lengths, allowed null values, data units, and time formats. It converts the field attributes into a unified character encoding according to the field order and concatenates them using attribute separators, field separators, and null value markers to form a field structure string. A 256-bit secure hash algorithm (SHA-256) is used to calculate the hash value of the field structure string, and the hash value, field list, interface version, and expiration time are written into the data source structure record. Upon subsequent reads, the hash values ​​are compared, and each field attribute is checked. If they match, the current version is used; if any attribute differs, a new field structure version is generated, and the expiration time is written to the original version.

[0033] The industrial internet data platform server acquires equipment ledgers, equipment hierarchy tables, component lists, measurement point lists, and object code tables, and generates equipment object records, production area records, and business object records accordingly. Equipment object records are registered according to the hierarchical relationship between power plants, generating units, production systems, equipment, and components, recording the industrial object number, object type, parent object number, equipment ledger code, system number, commissioning time, decommissioning time, and ledger version. Production area object records are derived from the production area code table, and business object records are derived from the business system primary key or business document number. Industrial objects are database records in the industrial internet platform used to identify the associated entities of equipment, components, production areas, or business records.

[0034] When equipment is replaced, components are disassembled, measurement points are relocated, or ledger codes are adjusted, an expiration time is written to the original industrial object record, and a new industrial object version record is generated. If the source record has a source system generation time, a valid industrial object version is queried based on this; if this time is empty, a time source marker is written based on the platform's receiving time. When the equipment ledger code, object type, and validity time all correspond to one industrial object record, a corresponding industrial field record is generated; if the object type, system, or validity time does not meet the requirements, a field conflict record is generated; if the same code corresponds to multiple valid records, a pending confirmation field record is written; if no equipment ledger code is provided, the measurement point list is queried according to the combination of the data-providing system number, system code, equipment name, installation location, and measurement point name. If a unique match is found, a correspondence is established; if multiple records correspond or no corresponding record is found, a pending confirmation field record or a non-corresponding field record is written respectively. Data with the aforementioned abnormal records is not entered into the standard data storage area.

[0035] The industrial internet data platform server reads the standard field number, standard data item name, standard data type, standard unit, object type, and standard field version from the standard field table. The source field number consists of the data source number, field structure version, and field order. The corresponding record for the industrial field includes the source field number, industrial object number, standard field number, source data type, target data type, source data unit, target data unit, data type conversion rule, unit conversion rule, time conversion rule, effective time of the corresponding relationship, expiration time of the corresponding relationship, and version of the corresponding relationship. When the source unit and target unit belong to the same dimension and the unit conversion table has corresponding rules, unit conversion is performed; when there are no conversion rules, the dimensions are different, or the conversion result exceeds the allowable range of the target data type, a conversion rule missing flag, a dimension inconsistency flag, or a value conversion exceeding the limit flag is written respectively, and the relevant fields are not entered into the standard data storage area.

[0036] The Industrial Internet data platform server reads the initial construction time range and data source range from the data migration task configuration. If no time range is configured, the intersection of the historical time range of the data source and the access time of the Industrial Internet platform is used. The Industrial Internet data platform server extracts records within this range from the original data storage area. If the data generation time is empty, the platform reception time is used, and a time source marker is written. The extraction results are sorted according to the data source number, data time, and source record number to generate an initial construction record set.

[0037] The Industrial Internet data platform server generates a source record number for each piece of data in the initially constructed record set. Relational database records are generated using the data source number, table number, primary key value, and field structure version. If a primary key is missing, the database log sequence number, incremental sequence number, or alternative unique field can be used; if none can be obtained, a hash value of the record field value sequence is used, and the same hash value and field value are retrieved across different acquisition batches. Existing source record numbers are used when field values ​​are consistent; a new source record version is generated when field values ​​change but the corresponding fields in the record are consistent. Independent source record numbers are generated when a cross-batch correspondence cannot be established. Message queue records, industrial measurement point records, and application programming interface records generate source record numbers according to their corresponding message location, acquisition location, and invocation location, respectively.

[0038] The source record's basic record settings include the source system generation time, edge gateway acquisition time, platform reception time, and source record business time fields. The source record's business time prioritizes the source system generation time after conversion according to the data source's time zone; if the source system generation time is missing, the edge gateway acquisition time and platform reception time are used in sequence. The data time, source time, and the business time corresponding to the data content are stored as the original source time field. Whenever used to determine the effective time of changes, time windows, the range of records to be processed, and query time boundaries, the source record's business time is uniformly read.

[0039] When source record numbers are the same and field value sequences are identical, one source record is retained and the number of repeated receptions is accumulated. When source record numbers are the same but field value sequences are different, separate source record versions are generated, recording the field value hash, reception time, and version order. Each source record saves the source system generation time, edge gateway acquisition time, platform reception time, and storage time; missing time fields are written with null values. The source time is converted to Coordinated Universal Time (UTC) according to the time zone and time format configured in the data source. The original time text, time zone, and conversion result are written together into the source record's base record. If the time format cannot be parsed, a time conversion failure flag is written, and the corresponding record is not included in the time window task.

[0040] The Industrial Internet data platform server reads data processing task configuration records. These tasks include format parsing, field conversion, time window division, data association, grouping and aggregation, sorting within partitions, and data asset writing. Each task record includes a task number, task type, input and output data objects, input and output field sets, processing rule version, preceding task number, execution order, and storage location. The Industrial Internet data platform server arranges the execution order according to the preceding task number and the input and output data objects. A task circular reference flag is generated when the task output is used as input again by one or more tasks; a task configuration missing flag is generated when the preceding task does not exist, the input field is not output by the preceding task, or the processing rule version does not exist. If any flag exists, no corresponding task execution record is generated.

[0041] During task execution, the Industrial Internet data platform server generates a task execution batch number and records the source record number or intermediate record number actually read in this batch. The format parsing task reads the data payload according to the field structure version and outputs parsed fields, unrecognized fields, missing fields, and fields with inconsistent types. If the data payload cannot be parsed, the original data payload is saved and a parsing failure flag is written. The field conversion task performs field name conversion, data type conversion, unit conversion, and writes the industrial object number based on the corresponding records for the industrial fields. It also records the field numbers, field types, and field values ​​involved in time window segmentation, data association, grouping aggregation, or sorting before and after task execution, generating records corresponding to the field values ​​before and after the task execution.

[0042] The time window task uses an interval that includes the window start time but excludes the window end time. Fixed time windows divide records based on the window start time and window duration; sliding time windows determine all window numbers corresponding to the same record based on the window duration and window movement interval; session time windows divide consecutive records for the same industrial object based on the session interval, generating a new window number when the time difference between adjacent records is greater than the session interval. Each window output record saves the window number, window start time, window end time, industrial object number, and input record number.

[0043] Data association tasks perform equivalence association, time-validated association, or range association based on industrial object number, equipment ledger code, business document number, data project number, batch number, or time partition number. Task configuration registers the association type and whether one-to-many associations are allowed. One association result is generated when matching a single record; separate association results are generated when matching multiple records and one-to-many associations are allowed; a discrepancy flag is written when matching multiple records but one-to-many associations are not allowed; and an overlap flag is written when multiple time-validated associations exist simultaneously. Each association result saves the input record number, associated field value, associated record version, and time overlap range for each side.

[0044] Grouping and aggregation tasks are performed by industry object number, data item number, time window number, business type number, or equipment status code, and then perform counting, summation, minimum value, maximum value, average value, first record extraction, or last record extraction. The first and last records are determined by the time field and the output row key; the handling of null values ​​is configured in the task settings to either ignore null values ​​or treat them as zero. Each aggregation result saves the grouping field value, aggregation type, input record number, and aggregation output row key.

[0045] The Industrial Internet data platform server stores intermediate record source relationship records using a one-to-one correspondence method. Each relationship record includes the task execution batch number, output record number, input record number, reference order, reference type, and relationship version. Data asset writing tasks write processing results to the data asset storage area according to the data asset table structure. When a business primary key exists, the data asset row key is generated using the data asset number, business primary key value, and data version. When no business primary key exists but a combined unique field is configured, the row key is generated using the combined unique field value, data asset number, and data version. When neither exists, the row key is generated using the data asset number, task execution batch number, source record number, and data version. When using hash values ​​to construct row keys, the original key field string before hash calculation is also saved.

[0046] If the output row key does not exist, a new data asset record is written. If the output row key already exists and the output field value, source record number set, processing rule version, and industrial field correspondence version are all the same, the current version is retained. If any of these are different, a new data asset version record is generated, and the expiration time is written to the original version. The industrial internet data platform server establishes a source record query index in the raw data storage area using the data source number and source record number, establishes a source relationship index in the intermediate result storage area using the task execution batch number, output record number, and input record number, establishes a data asset query index in the data asset storage area using the data asset number, data partition number, data asset row key, and data version, and establishes a joint query index using the industrial object number, data project number, and data time.

[0047] After the initial write of data assets, the Industrial Internet Data Platform server reads the data service configuration, recording the service number, filter fields, grouping fields, sorting fields, pagination parameters, return fields, referenced data asset numbers, and allowed data versions. The Industrial Internet Data Platform server executes a basic query using default query conditions and the current data version, recording the actual data asset partitions read, the returned row key order, and pagination boundaries, generating a basic data service call record; this record is maintained throughout platform operation. If the sorting field is missing, the referenced data asset version is missing, or the pagination parameters do not meet the configuration, an exception flag is added to the data service configuration, and no corresponding query results are output.

[0048] The industrial internet data platform server associates various records according to data source number, source record number, task sequence, data asset number, and data service number to generate data processing path records. When a task execution record has output records but no input records, an input source missing flag is written; when a data asset record is not associated with a task execution batch, an asset source missing flag is written; when a data service call record references a non-existent data asset row key, a service reference missing flag is written. If any missing flag exists, the corresponding data processing path is not written with its current valid status.

[0049] After completing the record association and integrity verification, the Industrial Internet Data Platform server generates the basic records for building the Industrial Internet Data Platform. These basic records include data source structure records, industrial object records, industrial field correspondence records, source record basic records, data processing task configuration records, data processing task execution records, field value correspondence records before and after processing task execution, intermediate record source relationship records, data processing path records, data asset version records, data service call records, and query index records. These records serve as the data foundation for the Industrial Internet platform to read changes in source records, field structure, processing rules, and the scope of data service references.

[0050] S2. Based on the industrial internet data platform, construct basic record identification fields, objects, processing rules and source record changes, and use the field values ​​before and after the change to cyclically supplement the records to be processed along the time window, data association and grouping aggregation relationship to generate a record of the data change processing range.

[0051] The process involves acquiring basic records for building the industrial internet data platform, the data source structure read in this instance, candidate industrial object records, candidate industrial field corresponding records, candidate data processing task configurations, and source records received within the inspection period. The current version is compared with candidate versions to generate various change records. Initial records to be processed are extracted based on the change's effective time, data processing path records, and task execution records. Current processing results are read and candidate processing results are generated according to the candidate configuration. The range of records to be processed is expanded based on time values, associated field values, grouping field values, data partition values, output row keys, and data service reference relationships until all sets of newly added records are empty, generating a data change processing range record. Candidate versions refer to the data source field structure, industrial object correspondence, data processing rules, and versions of source records that have been added or modified but have not yet taken effect; these are used to identify changes and extract initial records to be processed.

[0052] The Industrial Internet Data Platform server reads data source structure records, industrial object records, industrial field correspondence records, source record basic records, data processing task configuration records, data processing task execution records, field value correspondence records before and after processing task execution, intermediate record source relationship records, data processing path records, data asset version records, data service call records, and query index records from the Industrial Internet Data Platform's basic construction records. It also reads data source field structures, equipment ledgers, measurement point lists, industrial field correspondence configurations, data processing task configurations, and source records obtained by the Industrial Internet platform during the inspection cycle. The inspection cycle records the inspection start time, inspection end time, data source number range, and task number range. Message queue data is read according to message offset, relational database data is read according to incremental identification fields or database log sequence numbers, and industrial measurement point and application interface data are read according to collection batches and call batches. If no new source records are added during the inspection cycle but the configuration version changes, a version comparison is performed.

[0053] The Industrial Internet Data Platform server records all valid versions in the basic records of the Industrial Internet Data Platform construction as the current version, and records data read this time but not yet replaced by the current version as candidate versions. The Industrial Internet Data Platform server compares the structures of two types of data sources according to the data source number. If any of the following differs: number of fields, field order, field name, field data type, field length, nullable state, data unit, time format, or time zone, a data source structure change record is generated. For newly added fields, the first reception time is used as the candidate effective time; for deleted fields, the sum of the last valid record time and the sampling period is used, or, if the sampling period is not configured, the first reception time of the candidate field structure is used; if the interface version changes but the field list remains the same, only the interface version change is recorded; if the field structure hash values ​​are the same but the field lists are different, a hash value conflict flag is saved.

[0054] The industrial internet data platform server compares current industrial object records with candidate industrial object records based on the industrial object number, equipment ledger code, and validity period. When the object type, system number, parent object number, equipment ledger code, commissioning time, decommissioning time, or measurement point affiliation changes, an industrial object change record is generated. If a switchover time exists for an equipment change record, that time is used as the candidate validity period; if not, the new equipment commissioning time is used; if neither exists, the candidate ledger version validity period is used. If multiple times exist simultaneously with different values, the source of each time is saved and a pending confirmation flag is added to the validity period. When one current industrial object corresponds to multiple candidate industrial objects, multiple current industrial objects correspond to one candidate industrial object, or the correspondence between old and new codes is missing, a pending confirmation record for the corresponding industrial object is generated, and the relevant candidate industrial object record is not written to the validity status.

[0055] The industrial internet data platform server compares the current industrial field corresponding record with the candidate industrial field corresponding record according to the source field number. If any of the following is different: industrial object number, standard field number, target data type, target data unit, unit conversion rule, time conversion rule, or valid time, a corresponding change record for the industrial field is generated. When splitting fields, the candidate source fields and data sources are recorded; when merging fields, the candidate standard fields and value combination rules are recorded. If there are conflicting field records, unconfirmed field records, uncorresponding field records, missing unit conversion rule markers, inconsistent unit markers, or numerical conversion exceeding limits markers, the candidate industrial field corresponding record is not written into the valid status.

[0056] The industrial internet data platform server compares the current data processing task configuration with the candidate data processing task configuration according to the task number. If any of the following differs: task type, input / output data objects, input / output field sets, field transformation rules, time window type and parameters, associated fields, association type, association validity time conditions, grouping fields, aggregation operations, null value handling rules, partitioning fields, data asset row key components, or data service filtering, grouping, sorting, and pagination configurations, a data processing rule change record is generated. If a candidate task has a missing predecessor task, an input field not output by a predecessor task, a missing rule version, or a task circular reference, a task configuration exception flag is generated, and the corresponding candidate task is not executed.

[0057] The industrial internet data platform server checks the source records received within the current inspection period. A new source record is generated if the source record number does not exist; a modified source record is generated if the source record number exists but the field hash values ​​are different; a deleted source record is generated when the data source returns a deletion marker or a database log record is deleted; a delayed arrival record is generated when the source system's generation time is earlier than the inspection start time and the platform's reception time is within the inspection period; a record gap marker is generated when the message offset, collection sequence number, or database incremental sequence number is discontinuous, and the gap record is registered as a new source record upon arrival; duplicate records with the same source record number and field value sequence are only counted for the number of times they are received.

[0058] The aforementioned change records constitute the change source record, storing the change type, data source number, field number or source record number, current version, candidate version, field values ​​before and after the change, candidate effective time, and time source. When a source record is modified or deleted, the business time corresponding to the data content is used to determine the scope to be processed, while the database modification time, deletion operation time, or platform reception time is used to determine the inspection cycle. When the source system generation time is empty, the platform reception time is used, and a time source marker is written in it.

[0059] The Industrial Internet Data Platform server extracts initial pending records based on the change source records. New source records and delayed arrival records are directly written to the initial pending record set; the pre-change processing result is recorded as null. Modifying a source record involves writing both the current source record version and the candidate source record version. Deleting a source record retrieves the current source record version, original intermediate record, original window number, original association result, original grouping field value, original data partition, and original output row key stored in the Industrial Internet Data Platform's basic construction records; the post-change processing result is recorded as null. When an industrial field changes, the source record query index retrieves the same source field within the time frame from the candidate's effective time to the end of the check; when an industrial object changes, the current source record corresponding to the candidate industrial object number and falling within that time range is retrieved.

[0060] When data processing rules change, the industrial internet data platform server reads the rule's effective start time from the candidate rule configuration; if not configured, it uses the rule version's effective time; if still not registered, it uses the rule's publication completion time. When time window rules change, it reads all input records entering the corresponding time window task within the effective time range; when association rules change, it reads records entering each side of the association task; when grouping fields, aggregation operations, or null value handling rules change, it reads records entering the corresponding group aggregation task; when partition fields or data asset row key components change, it reads the source record corresponding to the affected data asset partition; when data service configuration changes, it reads the data asset record referenced by the corresponding data service and its source record. Each initial pending record saves the source record number, version, current intermediate record number, first participating task number, change source record number, and data time range.

[0061] The Industrial Internet Data Platform server establishes candidate processing partitions in the intermediate result storage area, using independent candidate version numbers and task execution batch numbers, and storing input / output record numbers and configuration versions. For source records with existing processing results, the server reads the results before the change from the basic records constructed by the Industrial Internet Data Platform; for newly added source records and delayed arrival records, the results before the change are empty; for modified source records, the server reads the processing results corresponding to the current source record version; for deleted source records, the server reads the current processing results without generating candidate outputs; when processing rules change but source records remain unchanged, the server reads the current results and re-executes using the candidate processing rules. Records in the candidate processing partitions do not overwrite the current data asset version.

[0062] The industrial internet data platform server executes format parsing, field conversion, time window division, data association, grouping and aggregation, data partitioning, and data asset row key generation on the initial records to be processed, according to the task order recorded in the data processing path record. It records the input records, output records, field values ​​before and after execution, and output row keys for each task. If the input record number and version, field structure version, industrial field correspondence version, processing rule version, and previous task output version are all the same as existing candidate task execution records, the corresponding task will not be executed repeatedly.

[0063] After field transformation, the industrial internet data platform server compares the current processing result with the candidate processing result for the industrial object number, standard field number, time value, associated field value, grouping field value, aggregated field value, partitioning field value, sorting field value, and output row key, generating a field value change record. The current field value of a newly added source record is empty, and the candidate field value of a deleted source record is empty. If the field value remains unchanged but the rule version, field correspondence version, or source record set changes, the candidate task execution record is still retained.

[0064] When time values ​​or time window rules change, the industrial internet data platform server determines the window range according to the current rules and candidate rules, and reads the input records within both ranges. A fixed time window reads all records corresponding to the current window number and candidate window number; a sliding time window reads all windows and records within each window where the changed record enters under both rules; a session time window reads records of the same industrial object sequentially from the changed position forward and backward, calculating the time difference between adjacent records. When the time difference between adjacent records in a certain direction is simultaneously greater than both the current session interval and the candidate session interval, reading in that direction stops. When a newly added or modified record is located between two current session windows, and its time difference with both the previous and next records is not greater than the candidate session interval, both session windows and the intermediate record are added to the processing range; when a deleted or modified record causes the time difference between adjacent records within the original session window to be greater than the candidate session interval, the records before and after the changed position are added to the processing range respectively.

[0065] When the values ​​of associated fields or the association rules change, the industrial internet data platform server queries records on each side of the association task according to the current association rules and candidate association rules, and adds the union of the two query results to the processing range. Equivalent associations read records with the same associated field values; time-valid associations require the data time to be within the valid time of the associated record; interval associations read records with overlapping time ranges. Intra-associations, left-side retained associations, right-side retained associations, or all-retained associations process unmatched records according to the task configuration. If a candidate rule does not allow one-to-many associations but multiple records are found, or multiple association versions are valid at the same time, a flag indicating inconsistent number of associated records or overlapping association versions is generated respectively.

[0066] When the grouping field value, aggregation field value, or grouping aggregation rule changes, the industrial internet data platform server queries the grouped records according to the current grouping field and the candidate grouping field, respectively. Adding a new record adds the existing record from the candidate group; deleting a record adds the remaining records from the current group; when the grouping field value changes, all records in both groups are added. When the grouping field changes from a single field to a combined field, the aggregation operation changes, or the null value handling rule changes, the union of the grouped records formed by the two types of rules is added. When there are multi-level aggregation tasks, the current aggregation output is used as the input for the next level aggregation, and the affected next-level grouped records are read.

[0067] When the partition field value, data asset row key component fields, or industrial object number changes, the Industrial Internet Data Platform server reads the data asset records corresponding to the current data partition and candidate data partition, the current output row key and candidate output row key, and the current industrial object number and candidate industrial object number, respectively, and obtains the source record through the intermediate record source relationship index. If the current output row key exists but the candidate output row key does not exist, the current output row key is written to the pending version record; if the candidate output row key already exists, the source record number set, processing rule version, and industrial field correspondence version of the corresponding data asset record are read. When the sorting field or data service configuration changes, the Industrial Internet Data Platform server reads the corresponding data service number, filter field, grouping field, sorting field, number of paginated records, current return row key range, and pagination boundary to generate the affected data service record.

[0068] After each round of candidate processing, the Industrial Internet Data Platform server generates the following sets: source records, intermediate records, field value change records, windows, associated records, grouping results, data partitions, output records, and data service sets. Each set uses the source record number and version, intermediate record number and version, field value change record number and effective time, window number and window rule version, associated record number and association rule version, grouping field value and statistical time range, data partition number and data asset version, output row key and output version, and data service number and data service configuration version as unique keys, and compares them with the corresponding cumulative set. Only unique keys that do not exist in the cumulative set are written to the newly added set in this round and then merged into the cumulative set. When any newly added set contains records, the corresponding records are sent to the next round of candidate processing starting from the first data processing task added; the range expansion stops when any type of newly added set is empty.

[0069] The Industrial Internet Data Platform Server reads the data platform construction task configuration records of the Industrial Internet Platform, obtains the maximum number of repeated executions for a single data platform construction, and, based on the data processing path records, time window task configurations, data association task configurations, group aggregation task configurations, and data service configurations, calculates the number of tasks included in the longest data processing path, the maximum number of overlapping time windows, the maximum number of data association levels, the maximum number of group aggregation levels, and the maximum number of data service reference levels. The Industrial Internet Data Platform Server determines the smaller value between the sum of the aforementioned statistical values ​​and the maximum number of repeated executions for a single data platform construction as the data platform construction protection count, which must be no less than one. The data platform construction protection count, task configuration version, and each statistical value are written into the Industrial Internet Data Platform construction basic record.

[0070] The data change processing scope expansion, data processing input / output reference scope verification, and data service query replay are each configured with a scope expansion count, a reference verification count, and a query replay count, respectively. These counts are incremented once for each round of corresponding processing. A data change scope expansion protection stop flag is generated when the scope expansion count reaches the data platform construction protection count and new pending records are still being generated. Similarly, a data processing reference verification protection stop flag is generated when the reference verification count reaches the data platform construction protection count and there are still missing inputs or unclassified outputs. Finally, a data service query replay protection stop flag is generated when the query replay count reaches the data platform construction protection count and there are still newly added processing records or unexplained query differences.

[0071] The industrial internet data platform server aggregates change source records, initial pending records, new records from each round, candidate task execution records, field value change records, windows, associations, groupings and data asset partition ranges, current output row keys, candidate output row keys, pending version records, and affected data service records to generate a data change processing range record. If there are pending confirmation records for industrial objects, pending confirmation flags for effective times, abnormal task configuration flags, parsing failure flags, inconsistent number of associated records flags, overlapping associated versions flags, or task execution protection stop flags, the corresponding data change processing range record is written to the pending confirmation status; if the aforementioned records or flags do not exist and all types of new sets are empty, the range is written to the determined status.

[0072] S3. By recording the data change processing range, execute candidate processing, corresponding to the current output and candidate output, trace back the actual reference record from the processing task that first produced the difference, supplement the missing inputs and outputs to be completed, and generate a data processing input and output reference range confirmation record.

[0073] The system reads data change processing scope records, current data processing task execution records, intermediate record source relationship records, data asset version records, and data processing task configuration records from the industrial internet data platform construction foundation records. It then performs candidate data processing on data change processing scopes with defined scope states. Based on the corresponding fields of the output records configured in the data processing task configuration, it matches current output records and candidate output records, determining the current output record's reuse, update, or termination status, as well as the candidate output record's reuse, update, or addition status. From the processing task where the result first differs, it reads the source and intermediate records actually referenced by the candidate output, and reads current output records that are no longer generated in the candidate processing. It adds tasks with missing input records and those referencing output records to be terminated to the re-execution task and input record scope, repeating candidate processing and status classification until all candidate output reference records are included in the cumulative pending record set, both current and candidate outputs have completed unique status classification, and there are no conflicts in the candidate output row keys. Finally, it generates a data processing input / output reference scope confirmation record.

[0074] The industrial internet data platform server reads the record status, change source record, initial pending record, candidate task execution record, window number range, associated record range, grouping field value range, data asset partition range, current output row key, and candidate output row key of the data change processing range record. When the record status is "pending confirmation," it reads the pending confirmation record corresponding to the industrial object, the pending confirmation flag for the effective time, the flag for abnormal task configuration, the flag for parsing failure, the flag for inconsistent number of associated records, the flag for overlapping associated versions, and the flag for task execution protection stop. It saves the stop position and the generated candidate task execution record, without entering the output record correspondence and status classification. When the record status is "range determined," it extracts candidate processing inputs according to the source record number and version, the first processing task number, and the processing task range.

[0075] The Industrial Internet data platform server establishes a result verification partition in the intermediate result storage area. This result verification partition is a database partition divided according to the verification batch number, recording the verification batch number, data change processing scope record number, current version number, candidate version number, candidate task execution record, candidate output record, candidate intermediate record source relationship record, and output record processing status. Before candidate processing is completed, records in the result verification partition are not written to the current data asset partition.

[0076] The Industrial Internet Data Platform server reads source records within its range from the raw data storage area, current intermediate records and their source relationships from the intermediate result storage area, and current data asset records from the data asset storage area. New source records and delayed arrival records do not have current processing results, and the current output record set is recorded as empty. Modified source records read both the current source record version and the candidate source record version. Deleted source records are not considered valid input for candidate data processing tasks. The Industrial Internet Data Platform server reads the current intermediate records, window outputs, associated outputs, grouped outputs, and data asset records generated by the source record according to the current data processing path, and writes them into the current version's affected output record set. When field structures, industrial field correspondences, or processing rules change, the same source record is used to obtain the current task execution result, and the corresponding task is re-executed according to the candidate configuration.

[0077] The industrial internet data platform server arranges task paths based on the preceding task numbers and task succession relationships in the data processing task configuration record. The task path is determined by the source record's initial participation in the task, the succession numbers of each task, and the target data asset number. When the same source record enters multiple time windows, related results, or grouped results, separate task path numbers are generated. Candidate processing begins with the task first involved in the change, reads the unchanged preceding output records as input, and executes until the candidate data asset record is generated. The candidate data asset record includes the candidate data asset row key, candidate data partition number, output field values, a set of source record numbers, and corresponding versions, but is not written to the current data asset partition.

[0078] Each candidate task execution generates a candidate task execution batch number. The industrial internet data platform server stores the source relationship records of candidate intermediate records using a one-line-one-reference method. Each record includes the candidate task execution batch number, candidate output record number, input record number, input record version, reference order, reference type, and candidate processing rule version. When a candidate task is re-executed and the set of input records changes, the original candidate task execution record is written to an invalidation status, a new candidate task execution batch is generated, and the status classification only reads the currently valid candidate task execution records.

[0079] The industrial internet data platform server reads the corresponding fields from the output records in the data processing task configuration record. Time window tasks use the task number, industrial object number, data project number, and window number; data association tasks use the task number, associated master record number, and association result sequence number; grouping and aggregation tasks use the task number, grouping field value, and statistical time range; data asset records use the data asset number and business primary key or combined unique field. If the task configuration does not register the corresponding fields for the output records, and no business primary key or combined unique field exists, the current output is written to the pending version output record, and the candidate output is written to the new status.

[0080] The industrial internet data platform server matches current and candidate output records according to the business primary key, combined unique field, and corresponding fields of the output record. When a field involved in the matching is null, database null values ​​are only matched against database null values; empty strings, zero values, and empty sets are not treated as database null values. When the same corresponding field matches only one current output and one candidate output, a one-to-one correspondence is established; when one-to-many, many-to-one, or many-to-many relationships occur, the output quantity relationship in the task configuration is read. If the task configuration allows it, the matching continues according to the association result sequence number, window number, grouping field value, or candidate output row key; if the task configuration does not allow it, a mismatch flag is generated for the corresponding number of output records.

[0081] After unique matching is completed, the industrial internet data platform server compares the current output with the candidate output in terms of output field set, field data type, field value, source record number set, field structure version, industrial field correspondence version, and processing rule version. Numeric fields are compared after conversion according to the data asset field structure registration data type, decimal places, and rounding method; time fields are compared after conversion to Coordinated Universal Time; character fields are compared according to unified character encoding; ordered arrays are compared item by item according to storage order; unordered sets are compared after being sorted according to the configured element sorting field; and record sets are compared after being sorted according to the record primary key or combined unique field. If no set sorting field is configured and the element order is not fixed, a missing set comparison rule flag is generated.

[0082] When the current output and the candidate output have the same output row key, output field value, source record number set, and versions, both are written to the "Continue Using" status. When the corresponding fields of the output records are the same, but there are differences in the output row key, field value, source record number set, or any version, both are written to the "Update" status. When there is no corresponding current record for the candidate output, it is written to the "Add" status. When there is no corresponding candidate record for the current output, it is written to the "End" status, and a record for the version to be ended is generated. When a candidate output adds a field, deletes a field, or its field type changes, it is written to the "Update" status. When a field conversion fails, an output field is missing, or a set comparison rule is missing, a candidate output processing exception record is generated, and no status classification is performed.

[0083] The industrial internet data platform server checks the uniqueness of the status of both the current version's affected output records and the candidate output records. Each current version's affected output record should be written to one of the following states: continued, updated, or terminated; each candidate output record should be written to one of the following states: continued, updated, or added. When the same record corresponds to both states, the corresponding fields of the output record, the candidate task execution batch number, the source record number set, and the processing rule version are read and rematched; if a unique state still cannot be formed, an output state conflict flag is generated.

[0084] The industrial internet data platform server searches for the processing task where the result first differs according to the execution order of the task path. If both the current output and candidate output of the preceding task are in use, it continues reading the next task; if there are changes in the number of output records, field values, source records, versions, or output row keys, the task is written to the record of the processing task where the result first differs. Corresponding records are generated for multiple task paths.

[0085] For processing tasks where the results differ for the first time, the Industrial Internet data platform server reads the input records directly referenced by the candidate output from the candidate intermediate record source relationship record, and then reads the source records layer by layer along the intermediate record source relationship, generating a set of candidate output actual referenced source records and a set of candidate output actual referenced intermediate records. These two sets are compared with the cumulative set of pending source records and the cumulative set of pending intermediate records, respectively. Records not included in the cumulative set are generated as supplementary input record records to be processed, recording the missing record number and version, the first time the record participated in the task, the candidate output record number, and the reason for supplementation, and are added to the scope of re-execution tasks and input records.

[0086] When candidate output references time window results, the industrial internet data platform server queries the original data storage area according to the start and end times and boundary inclusion rules of the candidate window, verifying the records within the window against the source record set of the candidate window. If a queried but unreferenced record has a source record deletion flag, a duplicate record exclusion flag, a parsing failure flag, a time conversion failure flag, an industrial field corresponding to a record awaiting confirmation flag, an object version expiration flag, a field conversion failure flag, or a task filtering condition not met flag, the exclusion reason is saved; if the aforementioned flags do not exist, a supplementary record for pending input records is generated. When a session time window is merged or split, the output of the current window that will no longer be generated is written to the end state, and the output of the newly generated candidate window is written to the new state.

[0087] When a candidate output references a related data result, the industrial internet data platform server queries the input data tables of each side in the related task configuration according to the candidate related field value, related type, and valid time condition, and compares the query record set with the source record set of the candidate related result. If a query record exists but is not referenced by the candidate result, a supplementary record of the input record to be processed is generated; if the record referenced by the candidate result does not meet the candidate related conditions, the related task number, the related side input record, and the related conditions are written into the re-execution task and the range of input records, the candidate related result for this round is written into an invalid state, and then re-execution is carried out according to the candidate related rules.

[0088] When a candidate output references a grouped aggregation result, the industrial internet data platform server queries the grouped input records according to the candidate grouping field values ​​and statistical time range, and compares them with the source record set of the candidate aggregation result. If there are newly added, deleted, or source version changed records, all input records within the group are added to the re-execution task and input record range. In multi-level aggregation, when a lower-level output is in an add, update, or end state, the upper-level grouping task and input records referencing that lower-level output are added to the re-execution range.

[0089] For output records of versions pending completion, the Industrial Internet Data Platform server reads the source relationship records and data processing path records of the current intermediate records to determine the source record that generated the record, the preceding intermediate record, and the time window task, data association task, grouping and aggregation task, data asset writing task, and data service that reference the record. If the output record of the version pending completion has been used as input for other tasks, the corresponding task number and other input records are added to the scope of re-execution tasks and input records; if it has already been written to data assets, the data asset number, data partition number, data asset row key, and current version number are saved, and the current data asset record is not deleted from the result verification partition.

[0090] The Industrial Internet data platform server checks the uniqueness of the row keys of candidate data assets. Within the same candidate version, if the row keys of candidate data assets are the same and the output field values, source record number sets, and versions are all identical, only one candidate output record is retained, and the duplicate generation task batch is registered. If the row keys are the same but the field values, source record number sets, or any version are different, a candidate output row key conflict flag is generated. If the same hash value corresponds to different original key field strings, a row key hash conflict flag is generated.

[0091] After each round of verification, the industrial internet data platform server maintains the following sets of records: cumulative pending source records, cumulative pending intermediate records, cumulative candidate output records, cumulative pending final version output records, cumulative re-executed task set, and cumulative result difference task set. The records for the current round are compared with their corresponding cumulative sets. Only records not present in the cumulative sets are added to the newly added set for this round, and the newly added set is then merged into the cumulative sets. When supplementary pending input records are generated or tasks are re-executed, candidate processing and status classification are re-executed starting from the first time the corresponding record participates in the task.

[0092] The result verification stops when the supplementary set of pending input records, the set of unclassified current output records, the set of unclassified candidate output records, the set of re-executed tasks and input record ranges, and the set of candidate output row key conflicts are all empty, and the source records obtained by tracing the valid candidate outputs through intermediate records are all included in the cumulative set of pending source records. A result verification protection stop flag is generated when the task execution protection round number is reached, and the current record is not written to the reference range confirmation status.

[0093] The Industrial Internet data platform server aggregates candidate task execution records, candidate intermediate record source relationships, current version affected output records, candidate output records, output record processing status, processing task records where the results first differ, the set of candidate output actual reference source records, the set of candidate output actual reference intermediate records, supplementary records of pending input records, output records of versions to be completed, re-executed tasks and input record ranges, and candidate output row key check results, generating a data processing input / output reference range confirmation record. A pending confirmation status is written when there are inconsistencies in the number of output records, missing set comparison rules, abnormal candidate output processing records, output status conflict markers, candidate output row key conflict markers, row key hash conflict markers, or result verification protection stop markers. A reference range confirmation status is written when none of the aforementioned records or markers exist, and both the current and candidate outputs have completed unique status classification, and all candidate output actual reference records have been included in the cumulative pending record set.

[0094] S4. Based on the data processing input and output reference range, confirm the record to generate candidate data asset versions and establish a query index. Perform corresponding version queries on the current version and candidate versions under the same user input and data cutoff boundary. Back-fill the query differences that fail to correspond to the output changes into the data processing range to generate the data platform construction results.

[0095] Read the data processing input / output reference range confirmation record, candidate output record, pending version output record, data asset structure record, and affected data service record. Generate candidate data asset records, candidate data asset version reference records, and pending data asset version records based on the processing status of candidate output records, and establish a candidate version query index covering candidate records and the referenced current records. Within the same data construction snapshot range, execute query replays using the current data asset version and current data service configuration, and candidate data asset versions and candidate data service configurations respectively. Match the query differences with the changes in the processing status of candidate output records and data service configurations. For differences that cannot be matched, trace the source record according to the data asset source relationship and supplement the processing range. Repeat the generation of candidate data asset versions and execution of query replays until there are no new processing records, candidate version record status conflicts, missing candidate query indexes, and unexplained query differences. Generate candidate data asset version construction records and data service query range confirmation records.

[0096] The Industrial Internet data platform server reads the record status of the data processing input / output reference range confirmation record. When the record status is "Pending Confirmation," it reads the output record corresponding quantity inconsistency flag, candidate output processing abnormal record, output status conflict flag, candidate output row key conflict flag, and result verification protection stop flag. It saves the candidate task execution batch, stop task number, candidate output record, and pending version output record, but does not execute candidate data asset version generation. When the record status is "Reference Range Confirmation," it reads the current version affected output record, candidate output record, candidate output record processing status, candidate intermediate record source relationship, pending version output record, candidate data asset row key, candidate data partition number, and affected data service number.

[0097] The Industrial Internet Data Platform server reads data asset structure records from the basic records built on the Industrial Internet Data Platform. These records include data asset number, data asset type, field name and order, field data type, allowed null value status, business primary key field, composite unique field, partition field, partition time granularity, sorting field, data version field, and source record field. If a candidate output lacks a field that is not allowed to be null, the field type cannot be converted according to the conversion rules, the partition field is unavailable, or the data asset structure version does not exist, a candidate data asset structure anomaly record is generated, and the corresponding candidate output is not included in the candidate data asset version.

[0098] The industrial internet data platform server reads the receiving boundaries of each data source based on the end time of the inspection cycle, generates candidate data assets, and constructs snapshot records. Message queue data records message topics, message partitions, and end offsets; relational database data records the end sequence number of the database log or the end value of the incremental field; industrial measurement point data records the acquisition task number and the end acquisition batch; and application programming interface data records the interface number and the end call batch. Current version queries, candidate version queries, and query difference tracing must not read data exceeding the snapshot boundary. Newly received data during query playback is written to the source record set of the next inspection cycle.

[0099] The Industrial Internet data platform server generates candidate construction batch numbers and candidate data asset version numbers, and establishes candidate data asset partitions in the data asset storage area. These partitions are divided according to the data asset number, candidate version number, and candidate data partition number, and store the candidate data asset row key, output field values, source record number set, candidate task execution batch number, field structure version, industrial field correspondence version, processing rule version, and candidate output record processing status. The validity period of existing data asset records is not modified before the query scope is confirmed.

[0100] When a candidate output record is in a "continuing" state, the Industrial Internet Data Platform server generates a candidate data asset version reference record, recording the candidate data asset version number, current data asset number, current data asset row key, current record version, candidate data partition number, and reference source. The current record's field values ​​are not copied. When a candidate output record is in an "updating" state, the candidate field values, the candidate source record number set, and the candidate configuration version are written to a new candidate data asset record, and the correspondence between the current record version and the candidate record version is saved. When a candidate output record is in a "creating" state, a candidate data asset record is directly generated. When a current output record is in a "ending" state, a data asset version record to be ended is generated, recording the current data asset row key, current record version, candidate version number, ending reason, and change source record number.

[0101] When the partition field of a candidate output record changes, the industrial internet data platform server registers the current data partition and the candidate data partition. The current record is written to the data asset version record to be completed, and the candidate output is written to the candidate data partition. When the row key of the candidate output changes but the corresponding field of the output record does not change, a row key substitution relationship record is generated, saving the current row key, the candidate row key, the substitution relationship type, and the record number of the change source.

[0102] Candidate data asset versions consist of a base version reference record, candidate data asset records, and records of data asset versions to be terminated. The base version reference record points to the current data asset version that was valid at the start of this construction and is used to inherit currently valid records outside the scope of processing. When executing a candidate version query, the candidate data asset records are read first, followed by the currently valid records in the base version, excluding the current row keys corresponding to records of data asset versions to be terminated and those replaced by candidate data asset records; if a candidate record has the same row key as a base version record, only the candidate record is retained. Before release, it is verified that each currently valid record in the base version is in an inheritance, replacement, or termination state, and that each data asset row key in the candidate version corresponds to only one valid record. If a base version reference does not exist, the base version has expired, or the same current record is simultaneously in an inheritance and termination state, a missing base version reference flag, an invalid base version reference flag, or a candidate version record state conflict flag is generated, respectively.

[0103] The Industrial Internet data platform server verifies the integrity of candidate data asset versions. Each candidate output record in the "continuing," "updating," or "adding" state should correspond to a candidate data asset version reference record or a candidate data asset record; each output record for a version to be terminated should correspond to one current data asset record and one data asset version record for a version to be terminated; the candidate version numbers used for the three types of records should be the same; the partition number of the candidate data asset record should correspond to the partition field value; the source record should be traceable to the source record through the source relationship of the candidate intermediate record. Within the same candidate version, the same candidate data asset row key can only correspond to one valid logical record.

[0104] The Industrial Internet Data Platform server establishes a candidate version query index for candidate data asset versions. The index entries include those corresponding to candidate data asset records and those referencing candidate data asset versions, but exclude those corresponding to records of data asset versions awaiting termination. The Industrial Internet Data Platform server establishes a primary query index using the data asset number, candidate data partition number, candidate data asset row key, and candidate version number; a joint query index using the industrial object number, data project number, and data time; and a source query index using the source record number and candidate task execution batch number. When the data service configuration includes grouping fields, sorting fields, or cursor fields, a joint index is established between the corresponding fields and the data asset row key. Each candidate record and referencing record should have a candidate version index entry; records awaiting termination should not retain valid candidate version index entries.

[0105] The Industrial Internet Data Platform server reads affected data service records, current data service configurations, candidate data service configurations, and data service call records. The data service configuration records include the service number, referenced data asset number, filter fields and operators, grouping fields, aggregation fields, sorting fields and sorting direction, pagination method, number of page records, return fields, query time field, and allowed data versions. The Industrial Internet Data Platform server writes historical call parameters within the data cutoff boundary into the query replay input set and generates a standard query set based on the affected data partition, affected fields, candidate data asset row keys, data asset row keys to be terminated, grouping field values ​​before and after changes, and sorting / pagination boundaries. The standard query set sets up affected partition queries, row key-oriented queries, changed field filtering queries, changed grouping queries, and sorting / pagination boundary queries. Query parameters, number of page records, and return fields are read from the corresponding data service configuration; if the corresponding configuration item does not exist, this type of query is not generated. The Industrial Internet Data Platform server merges the historical call parameters and the standard query set and deletes duplicate queries to form a query replay input record. Both current data asset version queries and candidate data asset version queries are executed according to this record.

[0106] When the data service configuration remains unchanged, the current data asset version query and the candidate data asset version query use the same user input parameters, filtering conditions, grouping conditions, sorting conditions, pagination parameters, and returned fields. When the data service configuration changes, under the same user input parameters and candidate data asset snapshot construction boundaries, the current data asset version query uses the current data service configuration, and the candidate data asset version query uses the candidate data service configuration. Allowed configuration items to change include the referenced data asset number, filtering fields and operators, grouping fields, aggregation fields, sorting fields and sorting direction, pagination method, number of records per page, returned fields, and allowed data versions. When field names change, they are converted to the current field name and candidate field name respectively, based on the corresponding records in the data service field. Each query replay generates a query replay number and records the user input parameters, data cutoff boundaries, current data service configuration version, candidate data service configuration version, and candidate data asset snapshot construction version.

[0107] Regular queries compare the set of returned fields, field data types, field values, number of returned records, and the set of data asset row keys. When a sorting field is configured but pagination is not implemented, a sorting field value sequence is formed according to field order, sorting direction, and null value arrangement rules. If field values ​​are the same, they are sorted by the data asset row key, and the order of the returned row keys is compared. Page number pagination queries compare the total number of records, the number of records per page, the first row key of the page, the last row key of the page, and adjacent page boundaries. Cursor pagination queries compare the start and end cursors, the cursor field value sequence, the query direction, and boundary inclusion rules. Grouped queries compare group field values, the set of data asset row keys referenced by the group, the number of records participating in the aggregation, and the aggregation field value. Data subscription reads compare the subscription start and end sequence numbers, duplicate row keys, missing row keys, record time order, and data asset version. File exchange tasks compare the file field structure, the number of file rows, the file row order, and the data asset row key corresponding to each row.

[0108] The Industrial Internet data platform server generates query difference units based on service number, query playback number, difference type, current data asset row key, candidate data asset row key, difference field, grouping field value, sorting field value sequence, and pagination boundary. When a candidate query reads records outside the candidate version range, a candidate query data version reference anomaly flag is generated; when a candidate data asset record meets the query conditions but the candidate version index is not returned, a candidate query index missing flag is generated. If none of the aforementioned anomalies exist, the candidate output record processing status is used as the corresponding record: when the candidate data asset row key of a newly added status record is the same as the newly returned row key, the current data asset row key of an ended status record is the same as the reduced returned row key, and the corresponding field of the output record of an updated status record is the same and the difference field belongs to the update field, a candidate output change corresponding difference record is generated; grouped differences should also satisfy that the changed record belongs to the corresponding group, and sorting or pagination differences should also satisfy that the sorting field value sequence of the changed record is between the current position and the candidate position, or between the current pagination boundary and the candidate pagination boundary.

[0109] The industrial internet data platform server maintains the current data asset version, only replacing the current data service configuration with the candidate data service configuration and performing a configuration verification query once. When the verification query produces the same return row key, difference field, grouping result, sorting position, or pagination boundary as the query difference unit, a difference record corresponding to the data service configuration change is generated. When the query difference simultaneously meets the conditions corresponding to both the candidate output change and the data service configuration change, a difference record corresponding to both the output change and the data service configuration change is generated; if only one condition is met, the corresponding difference record is written; if neither condition is met, an unexplained query difference record is written. The query difference classification record saves the corresponding candidate output record number, data asset row key, difference field, source record number, data service configuration version, grouping field value, sorting field value sequence, and pagination boundary as the basis for determining whether to supplement the data processing scope.

[0110] When unexplained query discrepancies involve increases, decreases, or changes in field values ​​of the returned row keys, the Industrial Internet Data Platform server reads candidate task execution batches based on the data asset row keys and traces the source records through the source relationships of the candidate intermediate records to generate supplementary query input records. When grouped queries generate unexplained new groups, group cancellations, or changes in aggregation values, all data asset row keys within the query time range are read according to the current group field values ​​and candidate group field values, and then the source records not included in the processing scope are traced. When sorting or pagination generates unexplained discrepancies, records between the current boundary and the candidate boundary are read according to the combined order of the sort field value sequence and the data asset row keys, and the source records not within the cumulative processing scope are traced.

[0111] When the current row key corresponding to the pending data asset version record still appears in the candidate query results, the Industrial Internet Data Platform server reads the actual data asset version and candidate version index item referenced by the candidate query. A candidate query data version reference anomaly flag is generated when the candidate query references the current version record; a candidate query index not removed flag is generated when the candidate version index still contains the pending row key. A candidate query returned record omission flag is generated when a newly added or updated candidate record meets the filtering conditions but does not appear in the candidate query results.

[0112] The industrial internet data platform server sends the supplementary records of the query input records to the data change processing scope determination and data processing input / output reference scope confirmation processes, and regenerates candidate output records, candidate output record processing status, and candidate data asset records. When reprocessing causes changes in candidate data partitions, candidate data asset row keys, grouping field values, sorting field value sequences, or pagination boundaries, it writes an invalidation status to the relevant candidate records, reference records, index items, and query replay records in the original candidate construction batch, generates a new candidate construction batch, and re-executes the affected data service query.

[0113] The Industrial Internet data platform server establishes a set of anomalies that are prohibited from being confirmed and released. This set includes markers for gaps, anomalies in candidate data asset structure, missing or invalid base version references, conflicting candidate version record states, missing or unremoved candidate query indexes, anomalies in candidate query data version references, missing candidate query returned records, unexplained query differences, and data service query replay protection stops. When any of these markers or records are generated, the corresponding candidate data asset version cannot be written into the query scope confirmation status, nor can it enter the version release process.

[0114] After each round of query replay, the Industrial Internet Data Platform server maintains the following sets: cumulative query input record supplementary set, cumulative candidate data partition set, cumulative grouping field value set, cumulative sorting boundary set, cumulative pagination boundary set, cumulative unexplained query difference set, and cumulative affected data service set. Each set is uniquely keyed based on record number and version, grouping field value and query time range, sorting boundary and data service configuration version, and pagination boundary and query replay number. The current round's results are compared with the corresponding cumulative set, and only records not present in the cumulative set are added to the new set for this round. If a new record exists for this round, data processing and query replay are re-executed. When all new sets for this round are empty, the prohibited confirmation and release exception set is empty, and the pending data asset version record has been excluded from the candidate query results and candidate version query index, query range confirmation is stopped and a query range confirmation status is written. When the query replay count reaches the number of times the data platform has been built for protection and there are still new processing records or unexplained query differences, a data service query replay protection stop flag is generated and written to the prohibited confirmation and release exception set.

[0115] The Industrial Internet Data Platform server aggregates candidate data asset construction snapshot records, candidate data asset records, candidate data asset version reference records, pending data asset version records, candidate version query indexes, query replay records, and query difference classification records, generating candidate data asset version construction records and data service query scope confirmation records. Only when the data service query scope confirmation record is in a query scope confirmation state and the prohibited confirmation and release exception set is empty, the Industrial Internet Data Platform server generates a version release batch number, marks the candidate data asset version as a valid version within the same database transaction, writes the current record corresponding to the pending data asset version record to the valid end time, enables the candidate version query index, disables the corresponding current version query index, and updates the data service reference version and current data asset version number to the candidate data asset version number. When the prohibited confirmation and release exception set is not empty, the candidate data asset version remains in a pending confirmation state, and the current data asset version, current query index, and data service reference version are not updated.

[0116] To verify the coverage of the data change processing range record and the data processing input / output reference range confirmation record on the change propagation record, four scenarios were set up: changes in field correspondence, changes in data processing rules, delayed arrival of source records, and deletion of source records. Each scenario was configured with 1200 source records, and 60 were randomly selected as change records. Each data processing path sequentially passed through field transformation, time window, data association, grouping and aggregation, and data asset writing tasks. The longest data processing path contained eight tasks, the maximum overlap of time windows was three layers, and the number of data association and grouping / aggregation levels were both two levels. Each scenario was independently repeated twenty times. Using the actual affected record set determined by the full candidate processing as a benchmark, the proportion of the number of actually affected records included in the current processing stage to the total number of actually affected records was recorded as the completeness rate of the record range to be processed. The average and standard deviation of the repeated experiment results for each scenario were also recorded.

[0117] like Figure 5 As shown, after cyclically supplementing records along the time window, data association, and grouping aggregation relationships, the completeness rate of the range of records to be processed under each change scenario is higher than the initial set of records to be processed. Continuing to trace back from the processing task where the result first differs to the actual referenced candidate output records, and supplementing missing inputs and outputs to be completed, the completeness rates of the range of records to be processed under the scenarios of changed field correspondence, changed data processing rules, delayed arrival of source records, and source record deletion reach 100%, 99.8%, 100%, and 99.9%, respectively. This result indicates that the data change processing range records and the data processing input / output reference range confirmation records can cover the source records, intermediate records, and output records generated by the propagation of changes in the simulation scenario.

[0118] In the data service query simulation, scenarios involving changes in grouping field values, sorting field values, page number / pagination boundaries, and cursor field values ​​were set up. The Industrial Internet data platform server, based on the data processing input / output reference range, confirmed records, generated candidate data asset versions, and established a candidate version query index. Within the same candidate data asset construction snapshot range, it executed both current version queries and candidate version queries, mapping query differences to the candidate output record processing status and data service configuration changes. Differences that could not be mapped were written to the unexplained query difference record. By tracing the source records of unexplained query difference records, generating supplementary query input records, and re-executing candidate data asset version construction and query replay, the number of unexplained query difference records in each round was counted, forming... Figure 6 .

[0119] like Figure 6As shown, the number of unexplained query difference records for grouped queries, sorted queries, page-number paginated queries, and cursor paginated queries all decrease with each query replay round, and drops to zero when no new source records, grouping field values, sort boundaries, or pagination boundaries are generated within the processing scope. This result indicates that the data service query scope confirmation record can retroactively supplement records outside the scope discovered by the query layer into the data processing scope, and ensure that the grouping results, sorting order, and pagination boundaries of the candidate data asset versions are all confirmed accordingly.

[0120] This embodiment also provides a computer device applicable to the data platform construction method based on the Industrial Internet, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the data platform construction method based on the Industrial Internet proposed in the above embodiment.

[0121] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0122] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for constructing a data platform based on the Industrial Internet as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0123] In summary, this invention generates a data change processing range record by iteratively supplementing the records to be processed along time windows, data associations, and group aggregation relationships. This includes newly added associated records and affected group records within the processing range, which is used to determine the data recalculation boundary of the industrial internet data platform and improve the completeness of the identification of the scope of change impact. By establishing candidate data asset versions and query indexes, corresponding version queries are performed on the current version and candidate versions under the same user input and data cutoff boundary. Query differences that fail to correspond to output changes are supplemented into the data processing range, which is used to correct records outside the scope involved in sorting, pagination, and grouping queries, thereby improving the completeness of the data platform construction results.

[0124] 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 method for constructing a data middle platform based on the Industrial Internet, characterized in that, include: Acquire multi-source industrial data, industrial object data and data processing tasks from the industrial internet platform, record the input-output relationship and data version of the source records in each task, and generate basic records for building the industrial internet data platform. Based on the industrial internet data platform, the basic record identification fields, objects, processing rules and source record changes are constructed. The records to be processed are supplemented cyclically by using the field values ​​before and after the change along the time window, data association and grouping aggregation relationship to generate a record of the data change processing range. By recording the data change processing range, candidate processing is executed, corresponding to the current output and candidate output. The actual reference record is traced back from the processing task that first produced the difference, missing inputs and outputs to be completed are supplemented, and a data processing input and output reference range confirmation record is generated. Based on the data processing input and output reference range, confirm the record to generate candidate data asset versions and establish a query index. Perform corresponding version queries on the current version and candidate versions under the same user input and data cutoff boundary. Inversely supplement the data processing range with the query differences that fail to correspond to the output changes, and generate the data platform construction results.

2. The method for constructing a data middle platform based on the Industrial Internet as described in claim 1, characterized in that, The acquisition of multi-source industrial data, industrial object data, and data processing tasks from the industrial internet platform, including recording the input-output relationships and data versions of the source data in each task, includes: Register multi-source industrial data according to data source number, field structure version, and industrial object number, and generate source record number; During field conversion, time window division, data association, grouping aggregation, and data asset writing, the task execution batch, input records, output records, field values ​​before and after task execution, record reference relationships, and data versions are recorded, and basic records for building the industrial internet data platform are generated according to the task sequence.

3. The method for constructing a data middle platform based on the Industrial Internet as described in claim 2, characterized in that, The process involves building a basic record identification system based on an industrial internet data platform, identifying fields, objects, processing rules, and changes in source records. It then uses the field values ​​before and after the changes to iteratively supplement the records to be processed along a time window, data association, and grouping aggregation relationships, generating a record covering the data change processing range. The data source field structure, industrial object correspondence, data processing rules, and source record content of the current version and the candidate version are compared respectively. The change effective time is determined according to the effective time of the field structure, the effective time of the industrial object version, the start time of the processing rule, and the business time of the source record. The initial records to be processed are extracted based on the change effective time and the data processing path.

4. The method for constructing a data middle platform based on the Industrial Internet as described in claim 3, characterized in that, The process involves building a basic record identification system based on an industrial internet data platform, identifying fields, objects, processing rules, and changes in source records. It then uses the field values ​​before and after the changes to iteratively supplement the records to be processed along a time window, data association, and grouping aggregation relationships, generating a record covering the data change processing range. Data processing is performed on the initial records to be processed according to the candidate versions, comparing the time values, related field values, and grouping field values ​​before and after the change; When the time value changes, the records in the current time window and the candidate time window are read. When the value of the associated field changes, the records in the current association result and the candidate association result are read. When the value of the grouping field changes, the records in the current group and the candidate group are read. The newly added record is processed again from the data processing task in which it first participated until no new source record, intermediate record, grouping field value, data partition or output row key is generated in each round, and a record of the data change processing range is generated.

5. The method for constructing a data middle platform based on the Industrial Internet as described in claim 4, characterized in that, The process of performing candidate processing by recording the data change processing range, corresponding to the current output and candidate output, includes: Match the current output and candidate outputs according to the corresponding fields of the output records configured by the business primary key, combined unique field, or data processing task. Mark the current output as in use, updated, or ended, and mark the candidate output as in use, updated, or newly added, respectively. Determine the data processing task with the first occurrence of the output record quantity, field value, source record, data version, or output row key difference according to the task execution order.

6. The method for constructing a data middle platform based on the Industrial Internet as described in claim 5, characterized in that, The process of tracing back from the processing task that first generated a difference to the actual reference record, supplementing missing inputs and outputs to be completed, and generating a data processing input / output reference range confirmation record includes: Read the source and intermediate records actually referenced by the candidate output from the data processing task that first produces the difference, and add the referenced records that are not within the scope of data change processing to the record to be processed; Read the current outputs that are no longer generated in the candidate processing and the data processing tasks that reference the current outputs. Add the corresponding input records and tasks to the re-execution scope. Repeat the candidate processing and output status determination until all candidate output reference records are within the range of records to be processed and all current affected outputs have completed status marking. Generate a data processing input and output reference scope confirmation record.

7. The method for constructing a data middle platform based on the Industrial Internet as described in claim 6, characterized in that, The step of confirming records, generating candidate data asset versions, and establishing query indexes based on the data processing input / output reference range includes: Based on the candidate output status, write the candidate outputs of the updated status and the newly added status into the candidate data asset record, establish a version reference relationship between the candidate outputs of the continued status and the corresponding current data asset record, and write the current outputs of the ended status into the data asset version record to be ended. A candidate version query index is created based on the data asset number, data partition, data asset row key, and candidate version number. A joint query index is created based on the industrial object number, data project number, and data time.

8. The method for constructing a data middle platform based on the Industrial Internet as described in claim 7, characterized in that, The step of performing a corresponding version query on the current version and candidate versions under the same user input and data cutoff boundary, and then back-complementing the query differences that could not be mapped to the output changes into the data processing scope, generates the data platform construction results including: Within the same data construction snapshot scope, data service queries are executed using the current data asset version and candidate data asset versions respectively, and the query differences are correlated with the candidate output status and data service configuration changes; For differences in returned records, grouping differences, sorting differences, or pagination boundary differences that cannot be matched, trace the source record according to the reference relationship between data asset records, intermediate records, and source records, supplement the data processing scope, repeatedly generate candidate data asset versions and execute data service queries until no new processing records or unmatched query differences are generated, and generate the data platform construction results.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the data platform construction method based on the Industrial Internet as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the data platform construction method based on the Industrial Internet as described in any one of claims 1 to 8.