A method and system for master data cleaning and quality verification in manufacturing enterprises

Abstract

This application relates to the field of computer data processing, and discloses a method and system for cleaning and quality verification of master data in manufacturing enterprises. The method involves acquiring manufacturing master data records and manufacturing association records from a product lifecycle management system, an enterprise resource planning system, and a manufacturing execution system; determining candidate sets for merging or splitting the data; creating local relationship snapshots in isolated verification areas; creating virtual standard nodes and generating incremental cleaning packages; dividing manufacturing execution reference relationships into migrateable reference relationships and frozen reference relationships based on whether business objects have formed execution records; and constructing a shadow cleaning relationship graph. The method performs relationship constraint verification on the shadow cleaning relationship graph; if the verification passes, it performs master data merging or splitting and association relationship redirection; if the verification fails, the corresponding cleaning set is marked as isolated data.

Description

Technical Field

[0001] This invention relates to the field of computer data processing, and specifically to a method and system for cleaning and quality verification of master data in manufacturing enterprises. Background Technology

[0002] With the development of digital R&D and intelligent manufacturing systems in manufacturing enterprises, product lifecycle management systems, enterprise resource planning systems, and manufacturing execution systems are used to manage product design data, material and inventory data, and production execution data, respectively. Due to differences in the construction time, data models, coding rules, field structures, and lifecycle status definitions of each system, the same material, component, or process object may have different object identifiers, attribute names, units of measurement, and version statuses in different systems. Furthermore, the same object identifier may correspond to different technical states or different manufacturing uses. As historical data accumulates, these differences can easily lead to data quality issues such as duplicate master data, duplicate codes for different items, duplicate codes for the same item, and overlapping version validity periods.

[0003] Existing master data cleaning methods typically identify duplicate records based on object name, specifications, field similarity, or source system priority, and then form standard master data through field overwriting, encoding mapping, or record merging. Some solutions also verify field integrity, primary and foreign key relationships, or data consistency between systems after cleaning. However, manufacturing master data includes not only object attributes but also links to bills of materials, process routes, production orders, quality inspections, inventory batches, and assembly histories. Two records with similar object attributes may correspond to different bills of materials locations, process versions, or technical statuses. Merging based solely on attribute similarity may result in misaligned bills of materials relationships, confusion regarding process applicability, or multiple effective versions existing within the same valid range.

[0004] Furthermore, existing data cleansing methods typically migrate all relationships from the original manufacturing master data records to the cleaned standard records, without distinguishing between unexecuted business references and historical references that have already resulted in production performance, inspection results, inventory turnover records, or assembly histories. For business objects with existing execution records, changing their references during master data cleansing may lead to inconsistencies between historical production data and the actual material versions used, affecting product traceability and quality responsibility determination. Therefore, how to verify the impact of merging or splitting operations on bill of materials relationships, process applicability relationships, and manufacturing execution reference relationships before formally modifying master data, and how to maintain existing manufacturing execution facts while unifying subsequent business references, are problems that manufacturing enterprises need to solve during master data cleansing and quality verification. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for cleaning and quality verification of master data in manufacturing enterprises, thereby solving the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A method for cleaning and quality verification of master data in manufacturing enterprises includes the following steps: S1: Obtain manufacturing master data records from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, as well as manufacturing association records associated with the manufacturing master data records. Manufacturing association records include bill of materials relationships, process applicability relationships, and manufacturing execution reference relationships. S2: Determine the candidate cleaning set based on the object attributes, technical status, valid range, and manufacturing-related records of the manufacturing master data records, and determine whether the candidate cleaning set is a candidate merging type or a candidate splitting type; S3: Extract the manufacturing master data records and their directly related nodes and relationships from the candidate cleaning set to form a local relationship snapshot isolated from the formal master data storage area; When the candidate cleansing set is of the candidate merge type, create a virtual standard node in the local relation snapshot; when the candidate cleansing set is of the candidate split type, create at least two virtual standard nodes in the local relation snapshot; establish the node mapping relationship between the manufacturing master data record and the virtual standard node, and generate the cleansing incremental package; S4: Based on whether the business object corresponding to the manufacturing execution reference relationship has already formed an execution record, the manufacturing execution reference relationship is divided into a migrateable reference relationship and a frozen reference relationship; S5: Load the cleaning incremental package into the local relation snapshot, temporarily redirect the migration-to-be-migrated association identified by the cleaning incremental package to the corresponding virtual standard node according to the node mapping relationship, keep the frozen reference relationship pointing to the original manufacturing master data record, and set the original manufacturing master data record as a historical alias node associated with the corresponding virtual standard node to form a shadow cleaning relation graph; S6: Perform relation constraint verification on the shadow cleaning relation graph; When the relation constraint verification passes, the merging or splitting of the master data records and the redirection of the association relationships are performed in the formal master data storage area according to the cleaning incremental package, while keeping the frozen reference relationships pointing to the corresponding historical alias nodes; when the relation constraint verification fails, the candidate cleaning set is marked as isolated data.

[0007] Preferably, after obtaining the manufacturing master data record, the method further includes: Based on the source system identifier and manufacturing object type of the manufacturing master data record, the corresponding field mapping relationship is invoked to convert the source fields in different source systems into a unified field structure; Convert object attributes according to the data type corresponding to the unified field structure, convert the unit of measurement used in the object attributes to the benchmark unit of measurement of the corresponding manufacturing object type, and convert the technical status of different source systems to a unified technical status. The effective time and expiration time of the manufacturing master data record are converted into a valid interval defined by the start time and the end time, and the source system identifier and source record identifier are retained in the converted manufacturing master data record.

[0008] Preferably, the candidate cleaning set is determined, including: The identity fields used to distinguish the identity of manufacturing objects are determined according to the data model corresponding to the type of manufacturing object. The standardized field values ​​of each identity field are combined in a fixed field order to form identity attribute keys. The node identifiers, relationship types, relationship directions, technical status, and valid ranges associated with the manufacturing master data records are combined in a fixed data order to form manufacturing application characteristics; Among manufacturing master data records with the same type of manufacturing objects, those manufacturing master data records with different object identifiers, the same identity attribute key, and no conflicting bill of materials relationships or process applicability relationships within the intersecting valid interval are identified as candidate merge records; Manufacturing master data records with the same object identifier but different identity attribute keys, or manufacturing master data records with the same object identifier and identity attribute keys but associated with different structural positions or process versions within the intersecting valid interval, are identified as candidate split records; Corresponding candidate merge records are grouped into a candidate cleansing set of the candidate merge type, and corresponding candidate split records are grouped into a candidate cleansing set of the candidate split type.

[0009] Preferably, forming a local relation snapshot and generating a cleaning incremental package includes: The manufacturing master data records in the candidate cleaning set are identified as the central nodes. The associated nodes that are directly connected to the central nodes through the manufacturing association records are extracted, and the relationship identifier, relationship direction, technical status and effective range between the central nodes and the associated nodes are extracted. The central node, associated nodes, and their relationships are copied to a verification area isolated from the formal main data storage area, and the corresponding source node identifier and source relationship identifier are retained for the copied nodes and relationships to form a local relationship snapshot; For candidate merging types, each candidate merging record is mapped to the same virtual standard node; for candidate splitting types, each candidate splitting record is divided into at least two record groups according to the correspondence between identity attribute keys, technical status, valid range and manufacturing-related records, and each record group is mapped to the corresponding virtual standard node. Based on the node mapping relationship, determine the virtual standard node to be created, the manufacturing master data record to be converted into a historical alias node, the association relationship to be migrated, the frozen reference relationship to be kept, and the valid range to be adjusted, and write the corresponding node processing data, relationship processing data and valid range processing data into the cleaning incremental package.

[0010] Preferably, the object attributes of the virtual standard node are determined, including: Extract the release records, cross-system reception records, and manufacturing business reference records corresponding to the object attributes with different field values ​​in the candidate cleaning set; Based on the record formation time, release records, cross-system reception records, and manufacturing business reference records with the same object identifier, the same technical status, and corresponding valid intervals are sequentially connected to form a manufacturing business event chain; A complete event chain is defined as a chain of manufacturing business events that sequentially includes release records, cross-system reception records, and manufacturing business reference records, with the formation time of each record increasing sequentially. When the same object attribute corresponds to only one complete event chain, the field value corresponding to that complete event chain is written to the virtual standard node; when the same object attribute corresponds to at least two complete event chains, the field value corresponding to the complete event chain with the latest release time within the target valid interval is written to the virtual standard node; when the same object attribute does not have a complete event chain, the corresponding candidate cleaning set is marked as isolated data, and the generation of the corresponding cleaning incremental package is stopped.

[0011] Preferably, the manufacturing execution reference relationship is divided into migrateable reference relationships and frozen reference relationships, including: Read the business status of the business object corresponding to the manufacturing execution reference relationship, as well as the execution records associated with the business object; When the business status is draft, not issued, or withdrawn, and there is no execution record associated with the business object, the corresponding manufacturing execution reference relationship is determined as a transferable reference relationship; When the business status is issued, in progress, or completed, or when there are production performance records, quality inspection result records, inventory turnover records, or assembly history records associated with the business object, the corresponding manufacturing execution reference relationship will be determined as a frozen reference relationship. Write the source relation identifier, original pointing object identifier, and partitioning result of the migrateable and frozen reference relations into the relation processing data.

[0012] Preferably, the relation constraint verification includes: Traverse the shadow cleaning relationship graph from the parent node to the child node in the bill of materials relationship. When the traversal path returns to a node that already exists in the current traversal path, it is determined that the bill of materials loop check has failed. When two technical states pointing to the same virtual standard node within the same application scope have intersecting valid intervals and there is no version inheritance relationship between the two technical states, the conflict check for the valid intervals of the technical states fails. When the same virtual standard node is located under two different parent nodes of the same bill of materials version within the same valid range, and the corresponding bill of materials relationship is not marked as a replacement relationship, or is simultaneously associated with two process versions that do not have a version inheritance relationship, the manufacturing use compatibility check is determined to have failed. When the same manufacturing execution reference node points to at least two virtual standard nodes that are in effect within the same valid range, the uniqueness check of the migrated reference is determined to have failed. If the source relation identifier, start node identifier, or end node identifier of any frozen reference relationship is inconsistent in the local relation snapshot and the shadow cleansing relation graph, it is determined that the frozen reference pointing invariance check has failed. When all checks pass, the creation of the standard manufacturing master data record, the conversion of the original manufacturing master data record to the historical alias node, the redirection of the migration association and the adjustment of the effective range are encapsulated into the same cleaning transaction, and the cleaning transaction is written into the formal master data storage area.

[0013] A master data cleaning and quality verification system for manufacturing enterprises includes an interface component, a data processing server, a formal master data storage area, and a verification area isolated from the formal master data storage area. The data processing server is connected to the interface component, the formal master data storage area, and the verification area, respectively. The interface component obtains manufacturing master data records and manufacturing association records associated with the manufacturing master data records from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, and sends the obtained data to the data processing server; The data processing server determines the candidate cleaning set and candidate cleaning type based on the object attributes, technical status, valid range and manufacturing related records of the manufacturing master data records, and copies the manufacturing master data records and their directly related nodes and relationships in the candidate cleaning set to the verification area to form a local relationship snapshot. The data processing server creates virtual standard nodes in the local relation snapshot according to the candidate cleaning type, establishes a node mapping relationship between the manufacturing master data record and the virtual standard node, and generates a cleaning incremental package; according to whether the business object corresponding to the manufacturing execution reference relationship has formed an execution record, the manufacturing execution reference relationship is divided into a migrated reference relationship and a frozen reference relationship. The data processing server temporarily redirects the migration-to-be-migrated associations identified by the cleaning incremental package to the corresponding virtual standard nodes according to the node mapping relationship, keeps the frozen reference relationship pointing to the original manufacturing master data record, and sets the original manufacturing master data record as a historical alias node, forming a shadow cleaning relationship graph; The data processing server performs relation constraint verification on the shadow cleaning relation graph. When the relation constraint verification passes, it updates the formal master data storage area according to the cleaning increment package. When the relation constraint verification fails, it marks the corresponding candidate cleaning set as isolated data.

[0014] Preferably, the data processing server includes a standardization processing unit, a candidate cleaning and determination unit, an incremental package generation unit, and a shadow relationship construction unit; The standardization processing unit calls the corresponding field mapping relationship according to the source system identifier and the manufacturing object type to convert the manufacturing master data record into a unified field structure, and unifies the data type, unit of measurement, technical status and valid range of the manufacturing master data record; The candidate cleaning determination unit generates identity attribute keys based on the data model corresponding to the manufacturing object type, generates manufacturing use features based on the manufacturing association records, technical status, and effective range of the manufacturing master data record, and determines the candidate cleaning set and candidate cleaning type based on the identity attribute keys and manufacturing use features. The incremental package generation unit establishes a node mapping relationship between manufacturing master data records and virtual standard nodes based on the candidate cleaning type, and generates a cleaning incremental package containing node processing data, relationship processing data, and effective interval processing data based on the node mapping relationship. The shadow relationship construction unit loads the cleaning incremental package into the local relationship snapshot, temporarily redirects the relationships to be migrated according to the relationship processing data, and retains the original pointers of the frozen reference relationships to form a shadow cleaning relationship graph.

[0015] Preferably, the data processing server further includes a constraint verification unit, a transaction processing unit, and a historical mapping unit; The constraint verification unit detects the bill of materials cycle, technical status valid interval conflict, manufacturing purpose conflict, non-unique pointer of transferable reference, and pointer change of frozen reference relationship in the shadow cleaning relationship diagram, and generates relationship constraint verification results. When the relation constraint verification passes, the transaction processing unit encapsulates the creation of the standard manufacturing master data record, the conversion of the original manufacturing master data record to the historical alias node, the redirection of the association to be migrated, and the adjustment of the effective range into the same cleaning transaction, and writes the cleaning transaction into the formal master data storage area. The historical mapping unit establishes a correspondence between the source record identifier, the standard manufacturing master data record identifier, the historical alias node identifier, and the cleaning transaction identifier, and stores the correspondence in association with the frozen reference relationship that retains its original pointer.

[0016] In summary, the present invention has the following main beneficial effects: By acquiring manufacturing master data records and their bill of materials (BOM) relationships, process applicability relationships, and manufacturing execution reference relationships from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, candidate merge records and candidate split records are determined based on object attributes, technical status, effective range, and manufacturing application characteristics. Local relationship snapshots and virtual standard nodes are constructed in a verification area isolated from the formal master data storage area. This achieves the goal of simulating master data merging or splitting results in advance without changing the formal master data. It can identify problems such as BOM cycles, overlapping technical statuses, conflicting process applicability, and non-unique migration of association relationships before formal cleaning, reducing the risk of product structure errors or master data association disorder caused by direct cleaning.

[0017] By classifying manufacturing execution reference relationships into migrateable and frozen reference relationships based on whether the business objects corresponding to the manufacturing execution reference relationships have already formed execution records, the manufacturing execution reference relationships are divided into migrateable reference relationships and frozen reference relationships. Migratable reference relationships are temporarily redirected to virtual standard nodes, while frozen reference relationships continue to point to historical alias nodes formed by the conversion of the original manufacturing master data records. This achieves the goal of unifying subsequent business master data references and keeping existing manufacturing execution facts unchanged, avoiding changes in the pointing of existing production performance, quality inspection results, inventory flow records, or assembly history due to master data cleaning, and maintaining the consistency between the traceability relationship of historical manufacturing data and actual business facts.

[0018] By forming a cleaning incremental package from node-processed data, relationship-processed data, and valid interval-processed data, after the shadow cleaning relationship graph passes the relationship constraint verification and the source record version information has not changed, the creation of standard manufacturing master data records, the conversion of historical alias nodes, the redirection of association relationships, and the adjustment of valid intervals are encapsulated into the same cleaning transaction. This aims to prevent partial writes, relationship omissions, or snapshot data invalidation during the master data cleaning process. When any write operation fails, all write operations in the same cleaning transaction are rolled back, and the candidate cleaning set whose processing results cannot be uniquely determined is set as isolated data, thereby maintaining data consistency in the formal master data storage area. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1 refer to Figure 1 A method for cleaning and quality verification of master data in manufacturing enterprises includes the following steps: S1: Obtain manufacturing master data records from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, as well as manufacturing association records associated with the manufacturing master data records. Manufacturing association records include bill of materials relationships, process applicability relationships, and manufacturing execution reference relationships. S2: Determine the candidate cleaning set based on the object attributes, technical status, valid range, and manufacturing-related records of the manufacturing master data records, and determine whether the candidate cleaning set is a candidate merging type or a candidate splitting type; S3: Extract the manufacturing master data records and their directly related nodes and relationships from the candidate cleaning set to form a local relationship snapshot isolated from the formal master data storage area; When the candidate cleansing set is of the candidate merge type, create a virtual standard node in the local relation snapshot; when the candidate cleansing set is of the candidate split type, create at least two virtual standard nodes in the local relation snapshot; establish the node mapping relationship between the manufacturing master data record and the virtual standard node, and generate the cleansing incremental package; S4: Based on whether the business object corresponding to the manufacturing execution reference relationship has already formed an execution record, the manufacturing execution reference relationship is divided into a migrateable reference relationship and a frozen reference relationship; S5: Load the cleaning incremental package into the local relation snapshot, temporarily redirect the migration-to-be-migrated association identified by the cleaning incremental package to the corresponding virtual standard node according to the node mapping relationship, keep the frozen reference relationship pointing to the original manufacturing master data record, and set the original manufacturing master data record as a historical alias node associated with the corresponding virtual standard node to form a shadow cleaning relation graph; S6: Perform relation constraint verification on the shadow cleaning relation graph; When the relation constraint verification passes, the merging or splitting of the master data records and the redirection of the association relationships are performed in the formal master data storage area according to the cleaning incremental package, while keeping the frozen reference relationships pointing to the corresponding historical alias nodes; when the relation constraint verification fails, the candidate cleaning set is marked as isolated data.

[0022] A method and system for cleaning and quality verification of master data in manufacturing enterprises is provided. This system is used to uniformly process, identify candidates, perform relationship deduction before cleaning, verify relationship constraints, and perform formal cleaning after verification of manufacturing master data distributed in product lifecycle management systems, enterprise resource planning systems, and manufacturing execution systems.

[0023] In this embodiment, the manufacturing master data cleaning does not involve directly deleting, overwriting, or merging manufacturing master data records based on object name similarity, field frequency, or source system priority. Instead, a local relationship snapshot is first created in a verification area isolated from the formal master data storage area. Virtual standard nodes are then created based on the intended merge or split operation. Relationships that have not yet formed business execution facts are temporarily migrated to the virtual standard nodes, while manufacturing execution references that have already formed business execution facts continue to point to the original manufacturing master data records, thus forming a shadow cleaning relationship graph. Only after the shadow cleaning relationship graph passes the relationship constraint verification is the corresponding cleaning incremental package written to the formal master data storage area.

[0024] The manufacturing enterprise master data cleaning and quality verification system includes interface components, a data processing server, a formal master data storage area, and a verification area.

[0025] The interface components communicate with the product lifecycle management system, enterprise resource planning system, and manufacturing execution system respectively, and are used to obtain manufacturing master data records, manufacturing association records, engineering change records, version release records, cross-system reception records, and manufacturing business execution records.

[0026] Product lifecycle management systems provide data on materials, components, product structure, bill of materials versions, drawing versions, process versions, and engineering changes. Enterprise resource planning systems provide data on material codes, units of measurement, procurement, inventory, production planning, and material substitution. Manufacturing execution systems provide data on production orders, production performance, material consumption, batch records, and assembly history.

[0027] When a manufacturing enterprise configures a quality management system or laboratory management system, the interface component also obtains inspection tasks and results from the corresponding system and processes these tasks and results as manufacturing business execution records. If the corresponding system is not configured, no corresponding quality inspection data is generated.

[0028] The data processing server connects to the interface component, the formal master data storage area, and the verification area. The formal master data storage area stores source manufacturing master data records, standard manufacturing master data records, historical alias nodes, manufacturing association records, and historical mapping relationships. The verification area stores local relationship snapshots, virtual standard nodes, node mapping relationships, cleaning incremental packages, and shadow cleaning relationship graphs.

[0029] The official master data storage area and the validation area are isolated using different databases, different database schemas, or different tablespaces. The database account used in the validation area does not have write permissions to the official master data storage area. Writes to the official master data storage area are performed only by the transaction processing unit using an account with the corresponding write permissions.

[0030] The data processing server includes a standardization processing unit, a candidate cleaning and determination unit, an incremental package generation unit, a shadow relationship construction unit, a constraint verification unit, a transaction processing unit, and a historical mapping unit.

[0031] The standardization processing unit outputs standardized manufacturing master data records; the candidate cleaning determination unit outputs a candidate cleaning set and candidate cleaning types; the incremental package generation unit outputs node processing data, relationship processing data, and effective interval processing data; the shadow relationship construction unit outputs a shadow cleaning relationship graph; the constraint verification unit outputs relationship constraint verification results; the transaction processing unit executes cleaning transactions or forms isolation records based on the relationship constraint verification results; and the historical mapping unit establishes the correspondence between source records, standard manufacturing master data records, historical alias nodes, and cleaning transactions after a cleaning transaction is successfully submitted.

[0032] Each manufacturing master data record includes at least the source system identifier, source record identifier, source record version information, object identifier, manufacturing object type, object attributes, technical status, valid range, and last update time.

[0033] The source system identifier is used to distinguish whether a manufacturing master data record originates from the product lifecycle management system, enterprise resource planning system, or manufacturing execution system. The source record identifier is a unique identifier assigned to the record by the source system.

[0034] The source record version information preferentially uses the record version number provided by the source system. When the source system does not provide a record version number, the interface component records the last update time of the source record and generates a content summary based on the field content involved in this cleaning process. The last update time and the content summary are used together as the source record version information.

[0035] Manufacturing object types are used to distinguish between materials, parts, semi-finished products, finished products, process objects, and other manufacturing objects defined by the unified master data model.

[0036] Object attributes are a set of fields describing the characteristics of a manufactured object, including identity fields and non-identity fields. Identity fields are used to distinguish manufactured objects, while non-identity fields are used to supplement the description of the manufactured object. Different manufactured object types correspond to different sets of identity fields.

[0037] Technical status represents the version status or lifecycle status of the master data record. Standardized technical status includes at least three states: unreleased, active, and inactive.

[0038] The effective interval consists of the effective start time and the invalidation termination time. If the source system does not set an invalidation termination time, an open termination identifier is used. When comparing times, the open termination identifier is treated as occurring later than all known service times.

[0039] Let the first The valid range for a master data record or a related record is: , No. The valid range for a master data record or a related record is: Whether the two intersect is determined by the following formula: in, This indicates the result of the valid interval intersection judgment; and These represent the start times of the two valid intervals, respectively. and These represent the failure termination times of the two valid intervals, respectively. When When, determine the intersection of two valid intervals; when When two valid intervals do not intersect, it is determined that they are not intersecting.

[0040] The effective interval intersection judgment is used to determine the candidate cleaning set, verify the conflict of effective intervals of technical status, verify the compatibility of manufacturing use, and determine the correlation to be migrated. This embodiment does not set an empirical time threshold that allows a difference of several hours or days, but directly uses the effective start time and failure termination time recorded by the source system for judgment.

[0041] Each manufacturing association record includes at least the source relationship identifier, relationship type, start node identifier, end node identifier, relationship direction, technical status, valid range, and source system identifier.

[0042] Relationship types include bill of materials relationships, process applicable relationships, and manufacturing execution reference relationships.

[0043] The bill of materials relationship also includes the bill of materials version identifier, parent node identifier, child node identifier, structural location identifier, configuration condition identifier, usage, unit of measurement, and substitute group identifier.

[0044] The structural location identifier originates from the bill of materials line item in the product lifecycle management system or enterprise resource planning system. If the source system does not have a configured condition identifier or alternative group identifier, a fixed empty identifier is used for storage. This fixed empty identifier is distinct from any valid field value.

[0045] The application of processes also includes application scope identifiers, process version identifiers, process location identifiers, and mandatory application identifiers.

[0046] Manufacturing execution references represent references between production work orders, quality inspection tasks, inventory records, or assembly histories to manufacturing master data records.

[0047] The data processing server pre-stores a unified master data model table, a field mapping table, a unit of measurement conversion table, a technical status mapping table, a business status mapping table, an application scope configuration table, and a version inheritance table.

[0048] The unified master data model table records unified field names, field data types, field order, whether a field belongs to an identity field, whether a field can be null, and the corresponding benchmark unit of measurement for each field according to the manufacturing object type.

[0049] The identity field originates from the material coding rules, product data models, or master data management specifications actually adopted by the manufacturing enterprise, and is not temporarily determined by the data processing server during operation based on field name, field frequency, or field similarity.

[0050] The field mapping table is established based on the source system data dictionary and the unified master data model, and records at least the source system identifier, manufacturing object type, source field name, unified field name, and data model version.

[0051] The unit conversion table is established based on the manufacturing enterprise's unit of measurement directory or the conversion relationship of legal units of measurement, and at least records the source unit of measurement, the reference unit of measurement, and the conversion factor. The conversion factor is not generated by fitting or extrapolating from historical data.

[0052] The technical status mapping table is established based on the lifecycle state machine in the product lifecycle management system and the master data management system, and is used to convert the technical status of different source systems into a unified technical status.

[0053] The business status mapping table is established based on the business object state machine of the enterprise resource planning system and the manufacturing execution system. It is used to convert the source business status into draft status, unissued status, cancelled status, issued status, in-process status, or completed status.

[0054] The application scope configuration table records scope fields used to define the boundaries of the manufacturing master data, categorized by manufacturing object type. Scope fields are selected from the applicable product identifier, production organization identifier, bill of materials version identifier, and configuration condition identifier.

[0055] If two manufacturing association records use the same version of the application scope configuration table, and the values ​​of each scope field are identical, then the two manufacturing association records are determined to be in the same application scope.

[0056] The version succession table is established based on engineering change records or version release records in the product lifecycle management system. It must record at least the previous version identifier, the next version identifier, the effective date of succession, and the engineering change identifier. If only the version numbers are adjacent, but no engineering change record or version release record exists, it is uncertain whether a version succession relationship exists between the two versions.

[0057] After the interface component obtains the manufacturing master data records from various source systems, the standardization processing unit calls the corresponding field mapping table according to the source system identifier, manufacturing object type, and data model version to convert the source fields in different source systems into a unified field structure.

[0058] For numeric object attributes, the standardized processing unit performs conversion according to the data types specified in the unified master data model table. When data type conversion fails, the source field value and conversion failure flag are retained, the corresponding manufacturing master data record is marked as isolated data, and the reason for isolation is recorded as field type conversion failure.

[0059] For object attributes that include units of measurement, the standardization processing unit converts both the attribute value and the unit of measurement according to a unit conversion table. For example, the source value is multiplied by the corresponding conversion factor in the unit conversion table to obtain the attribute value expressed in the reference unit of measurement.

[0060] The standardization processing unit converts the technical status of the source system into a unified technical status based on the technical status mapping table, and converts the business status of the source system into a unified business status based on the business status mapping table.

[0061] When the source technical status cannot be mapped to a unified technical status, the corresponding manufacturing master data record is marked as isolated data. When the source business status cannot be mapped to a unified business status, the corresponding manufacturing execution reference relationship is treated as a frozen reference relationship to avoid changing the pointer of the manufacturing execution reference relationship when the business execution status is unclear.

[0062] The standardized processing unit converts the effective time and expiration time in the source record into a unified time format to form a valid range, and retains the source system identifier, source record identifier, source record version information and last update time.

[0063] When the field marked as the identity field in the unified master data model table is empty, the source field does not exist, or the field value conversion fails, the identity attribute key of the manufacturing master data record is not generated, the manufacturing master data record is marked as isolated data, and the isolation reason is recorded as incomplete identity field.

[0064] For non-identity fields in the unified master data model table that are allowed to be null, a fixed null identifier is used for storage. The fixed null identifier is not used in the determination of whether two valid field values ​​are the same.

[0065] After standardization, the candidate cleaning and determination unit reads the identity field and its order based on the manufacturing object type and data model version.

[0066] The candidate cleaning and determination unit combines the standardized field values ​​of each identity field according to the field order specified in the unified master data model table to form an identity attribute key, and records the corresponding data model version in the identity attribute key.

[0067] When storing identity attribute keys in summary form, the candidate cleaning and determination unit simultaneously retains all standardized field values ​​involved in the summary calculation. If the summary values ​​of two identity attribute keys are the same, the standardized field values ​​involved in the summary calculation are compared item by item; only when all standardized field values ​​are the same are the two identity attribute keys determined to be the same.

[0068] Manufacturing application characteristics consist of structural related data, process related data, execution reference data, technical status, and effective range.

[0069] The structural association data is arranged in the following order: bill of materials version identifier, parent node identifier, child node identifier, structural location identifier, configuration condition identifier, relationship direction, technical status, and effective range.

[0070] The process-related data are arranged in the following order: application scope identifier, process version identifier, process location identifier, mandatory application identifier, technical status, and effective range.

[0071] The execution reference data is arranged in the following order: business object identifier, manufacturing execution reference relationship direction, business status, execution record identifier, and valid range.

[0072] The above arrangement is stored in the unified master data model table. When the data model version changes, the newly generated identity attribute keys and manufacturing purpose characteristics record the new data model version. Identity attribute keys generated using different data model versions are not directly used for the same judgment; they need to be regenerated according to the same data model version first.

[0073] The candidate cleaning determination unit first groups the manufacturing master data records according to the manufacturing object type, without merging or splitting different manufacturing object types.

[0074] Under the same manufacturing object type, manufacturing master data records are identified as candidate merge records when they simultaneously meet the following conditions: Each manufacturing master data record has a different object identifier; The identity attribute keys of all manufacturing master data records are the same; The valid ranges of each manufacturing master data record intersect, or are connected continuously through version inheritance; Within the intersecting valid intervals and the same application scope, there are no conflicting bill of materials relationships or process applicability relationships.

[0075] Bill of Materials (BOM) relationship conflicts are determined based on the structural relationship key. The structural relationship key consists of the BOM version identifier, parent node identifier, structural location identifier, and configuration condition identifier.

[0076] If two bill of materials relationships that are in effect within the same valid interval have the same structural relationship key, but their usage, unit of measurement or substitute group identifiers are different, and there is no version succession relationship or substitution relationship between the two bill of materials relationships, then the two bill of materials relationships are determined to be conflicting.

[0077] If the same general component is referenced repeatedly in different parent nodes, different bill of materials versions, or different structural locations, it is uncertain whether this is a bill of materials relationship conflict.

[0078] Conflicts in process applicability are determined based on the process relationship key. The process relationship key consists of an application scope identifier and a process location identifier.

[0079] Within the intersecting valid interval, if two process application relationships have the same process relationship key, point to different effective process versions, there is no version inheritance relationship between the two process versions, and both process application relationships are marked as mandatory application, then the two process application relationships are determined to be in conflict.

[0080] Manufacturing master data records that meet the candidate merging criteria are grouped into the candidate cleaning set of the same candidate merging type.

[0081] For manufacturing master data records with the same object identifier, a record is identified as a candidate for splitting if it meets any of the following conditions: The same object identifier may correspond to two or more different identity attribute keys; The same object identifier and the same identity attribute key correspond to conflicting bill of materials relationships within the intersecting valid interval; The same object identifier and the same identity attribute key correspond to conflicting process application relationships within the intersecting valid intervals; The same object identifier corresponds to two effective technical states that do not have a version succession relationship within the same application scope and intersecting valid intervals.

[0082] Corresponding candidate split records are grouped into a candidate cleaning set of the same candidate split type.

[0083] The candidate cleaning set is only used as a verification object before cleaning and does not directly modify the formal main data storage area.

[0084] The candidate cleaning determination unit identifies the manufacturing master data record in the candidate cleaning set as the central node.

[0085] The data processing server extracts the associated nodes and corresponding manufacturing-related records that are directly connected to the central node through bill of materials relationships, process application relationships, or manufacturing execution reference relationships.

[0086] To complete the bill of materials cycle verification, the data processing server continues to extract all downstream structural nodes and bill of materials relationships that can be reached from the central node or directly associated structural nodes in the same bill of materials version, from the parent node to the child node, until it reaches a node that has no downstream bill of materials relationship or a node that has already been extracted.

[0087] For upstream bill of materials relationships pointing to the central node, the data processing server also extracts the corresponding parent node to preserve the structural position of the central node in the bill of materials.

[0088] The data processing server copies the central node, directly associated nodes, continuously connected downstream structural nodes, and manufacturing association records between the above nodes to the verification area, and retains the source node identifier and source record version information for each copied node, and retains the source relationship identifier for each copied relationship.

[0089] Local relational snapshots also include snapshot identifiers, snapshot creation time, and data model version.

[0090] Before formally submitting the cleaning task, the data processing server rereads the source record version information of the central node, related nodes, and the creation of related records.

[0091] When the version number of any source record changes, the cleansing transaction is stopped, the current local relation snapshot is discarded, and a new local relation snapshot is formed based on the updated source record.

[0092] If the source system does not provide a source record version number, compare the last update time and the content digest. If the last update time or the content digest changes, treat it as a change in the source record version information.

[0093] When the candidate cleaning set is of the candidate merging type, the incremental package generation unit creates a virtual standard node in the local relation snapshot and establishes a node mapping relationship between each candidate merging record and the virtual standard node.

[0094] When the candidate cleaning set is of the candidate split type, the incremental package generation unit divides the candidate split records into at least two record groups according to the correspondence between identity attribute keys, technical status, valid range and manufacturing-related records.

[0095] Candidate split records with different identity attribute keys are assigned to different record groups.

[0096] Candidate split records with the same identity attribute key but conflicting structural or process relationship keys are assigned to different record groups.

[0097] Candidate split records that have the same identity attribute key, structural relationship key, and process relationship key, and whose effective intervals intersect, are consecutive at the beginning and end, or are consecutively connected through a clear version inheritance effective time, are classified into the same record group.

[0098] The incremental package generation unit creates a virtual standard node for each record group and establishes a node mapping relationship between the manufacturing master data records in each record group and the corresponding virtual standard node.

[0099] Valid intervals within the same record group are arranged from earliest to latest according to their effective start time. Adjacent valid intervals are considered continuous when they intersect, the expiration time of the preceding valid interval is equal to the effective start time of the following valid interval, or they are connected consecutively through the succession effective times recorded in the version succession relationship table.

[0100] When all valid intervals in a record group can be continuously connected, the target valid interval of the virtual standard node is determined by the earliest effective start time and the latest failure termination time in the record group.

[0101] When there is a time interval between adjacent valid intervals and no version succession relationship exists, the manufacturing master data records on both sides of the time interval are not mapped to the same virtual standard node, but are assigned to different record groups.

[0102] Under the candidate split type, the original manufacturing master data record is converted into a historical alias node after formal cleaning. This historical alias node establishes a splitting and inheritance relationship with each virtual standard node.

[0103] Each split inheritance relationship record node mapping identifier, application scope, and valid interval. Subsequent new business operations select the corresponding standard manufacturing master data record based on the application scope and valid interval. Manufacturing execution reference relationships already formed before the cleansing still point to historical alias nodes.

[0104] For object attributes with consistent field values ​​in the candidate cleaning set, the incremental package generation unit directly writes the standardized field value into the virtual standard node.

[0105] For object attributes with different field values, the incremental package generation unit extracts the release record, cross-system reception record, and manufacturing business reference record corresponding to each candidate field value.

[0106] The publication record originates from the source system with master data publication authority and includes at least the object identifier, technical status, publication time, and publication result.

[0107] Cross-system reception records originate from the target system's interface reception receipts and include at least the receiving object identifier, reception time, and reception result.

[0108] Manufacturing business reference records originate from production work orders, quality inspection tasks, inventory business records, or assembly history, and at least include the business object identifier, the referenced object identifier, the reference time, and the execution status.

[0109] For the The integrity of the manufacturing business event chain for each candidate field value is determined by the following formula: in, Indicates the first The results of the integrity judgment of the manufacturing business event chain corresponding to the candidate field values; This represents the corresponding set of successful posting records; This represents the corresponding set of successful cross-system reception records; This represents the corresponding set of manufacturing business reference records; Indicates the time when the successful release record was created; This indicates the time when the cross-system reception success record was generated; This indicates the time when the manufacturing business record referenced was created.

[0110] when At that time, determine the first Each candidate field value corresponds to a complete manufacturing business event chain; when At that time, determine the first The candidate field values ​​did not form a complete manufacturing business event chain.

[0111] In addition to meeting the aforementioned time sequence, release records, cross-system reception records, and manufacturing business reference records should also meet the following requirements: continuous correspondence of object identifiers, identical technical status, and the valid interval of each record intersects with the target valid interval of the corresponding virtual standard node.

[0112] Only one candidate field value satisfies the condition for the same object property. When this happens, the candidate field value is written to the virtual standard node.

[0113] If there are two or more candidate field values ​​for the same object property that satisfy At that time, the complete manufacturing business event chain with the latest release time is selected.

[0114] If there is only one complete manufacturing business event chain with the latest release time, the candidate field value corresponding to that complete manufacturing business event chain is written into the virtual standard node.

[0115] If there are two or more complete manufacturing business event chains with the latest release time, and the candidate field values ​​corresponding to each complete manufacturing business event chain are the same, then write the same field value into the virtual standard node.

[0116] If there are two or more complete manufacturing business event chains with the latest release time, and the candidate field values ​​corresponding to each complete manufacturing business event chain are different, the candidate cleaning set will be marked as isolated data, the isolation reason will be recorded as object attribute conflict, and no corresponding cleaning incremental package will be generated.

[0117] The same object property does not exist. When the candidate field value is not found, the candidate cleaning set is marked as isolated data, and the isolation reason is recorded as the object attribute lacking a complete manufacturing business event chain.

[0118] For the bill of materials relationship and process applicability relationship connected to the manufacturing master data record in the candidate cleaning set, the incremental package generation unit determines the virtual standard node to which it should point based on the node mapping relationship, application scope and effective range.

[0119] If a valid range of a bill of materials relationship or process applicability relationship intersects only with the target valid range of a virtual standard node, and its structural relationship key or process relationship key can uniquely correspond to that virtual standard node, then the manufacturing association record is determined as an association to be migrated.

[0120] For candidate merge types, the bill of materials relationship and process applicability relationship of each candidate merge record all point to the same virtual standard node when there is no relationship conflict.

[0121] For candidate split types, the bill of materials relationship is mapped to the corresponding virtual standard node according to the identity attribute key, structural relationship key, and effective range; the process applicability relationship is mapped to the corresponding virtual standard node according to the identity attribute key, process relationship key, and effective range.

[0122] If a bill of materials relationship or process applicability relationship corresponds to two or more virtual standard nodes at the same time, and the unique pointer cannot be determined based on the structural relationship key, process relationship key, application scope, and effective interval, the candidate cleaning set will be marked as isolated data, and the reason for isolation will be recorded as the inability to uniquely migrate the associated relationship.

[0123] The data processing server reads the source status of the business object corresponding to the manufacturing execution reference relationship, and calls the business status mapping table according to the source system identifier to convert the source status into a unified business status.

[0124] When a business object is in a draft, unissued, or withdrawn state, and there is no execution record associated with the business object, the corresponding manufacturing execution reference relationship will be determined as a transferable reference relationship.

[0125] When a business object is in the issued, executing, or completed state, the corresponding manufacturing execution reference relationship will be determined as a frozen reference relationship.

[0126] If a business is in draft, unissued, or cancelled status, but there are production performance records, quality inspection results records, inventory turnover records, or assembly history records, the corresponding manufacturing execution reference relationship will still be determined as a frozen reference relationship.

[0127] When the source business status cannot be transformed through the business status mapping table, the corresponding manufacturing execution reference relationship will be determined as a frozen reference relationship.

[0128] An execution record has been created, meaning the source system has generated a business fact record with a unique record identifier, the creation time, and the identifier of the referenced object. Unsaved interface data, temporary calculation results, and plan results without unique record identifiers are not considered execution records.

[0129] Migrative references are written to the relation redirection set in the cleansing incremental package; frozen references are written to the set that retains their original pointers in the cleansing incremental package.

[0130] The cleaning incremental package includes node processing data, relationship processing data, and valid range processing data.

[0131] The node processing data includes candidate cleaning set identifier, candidate cleaning type, virtual standard node identifier, source node identifier of the original manufacturing master data record, node mapping relationship, and historical alias node conversion identifier.

[0132] The relation processing data includes the source relation identifier, relation type, original starting node identifier, original ending node identifier, proposed starting node identifier, proposed ending node identifier, and relation processing type.

[0133] Relationship processing types include relation redirection and keeping the original pointer.

[0134] The valid range processing data includes the valid range of the original manufacturing master data record, the target valid range of the virtual standard node, the retained valid range of the historical alias node, and the application scope corresponding to the split inheritance relationship.

[0135] The cleaning incremental package only records the differences between the local relational snapshot and the result to be cleaned, and does not copy all manufacturing master data in the formal master data storage area.

[0136] The shadow relation building unit loads the cleansing incremental package into the local relation snapshot.

[0137] For bill of materials relationships, process applicable relationships, and migrateable reference relationships with relationship processing type of relationship redirection, the starting node or ending node is replaced by the corresponding virtual standard node from the original manufacturing master data record according to the node mapping relationship.

[0138] For frozen references that maintain their original pointers, the source relation identifier, start node identifier, and end node identifier remain unchanged.

[0139] For candidate merge types, each original manufacturing master data record is marked as a historical alias node that has established a merge succession relationship with the same virtual standard node.

[0140] For candidate split types, the original manufacturing master data record is marked as a historical alias node, and a split inheritance relationship is established between the historical alias node and each virtual standard node.

[0141] The shadow relationship construction unit sets up virtual standard nodes, historical alias nodes, and effective intervals of inheritance relationships according to the effective interval data processing, forming a shadow cleaning relationship graph.

[0142] The shadow cleaning diagram represents the data structure to be formed after the cleaning increment package is written to the formal main data storage area. Write operations occurring during the formation of the shadow cleaning diagram are confined to the verification area and do not alter the data in the formal main data storage area.

[0143] Relationship constraint verification includes bill of materials cycle verification, technical status valid range conflict verification, manufacturing use compatibility verification, migrated reference uniqueness verification, and frozen reference pointer invariance verification.

[0144] Bill of Materials Cyclic Validation: The constraint verification unit traverses the shadow cleaning relationship graph from the parent node to the child node in the bill of materials relationship and establishes a path node set for the current traversal path.

[0145] If the node to be visited already exists in the current path node set, determine that a bill of materials loop is formed in the shadow cleaning relationship graph that starts from and returns to the node. The bill of materials loop validation fails.

[0146] If a node has been visited in another completed path but is not in the current path node set, it is uncertain whether a bill of materials cycle will be formed.

[0147] Technical status valid interval conflict check: Within the same application scope, if two technical states both point to the same virtual standard node, and the valid intervals of the two technical states intersect, but there is no version inheritance relationship between the two technical states, the conflict check of the valid intervals of the technical states fails.

[0148] If there is a version succession relationship between two technical states, and the failure termination time of the previous technical state is not later than the succession effective time recorded in the version succession relationship table, it is uncertain whether there is a conflict in the effective interval of the technical state.

[0149] Manufacturing application compatibility verification: The constraint verification unit verifies both the structural purpose and the technological purpose.

[0150] Within the same valid interval, if two bill of materials relationships that are in effect have the same structural relationship key, but their usage, unit of measurement, or substitute group identifiers are different, and there is no version succession relationship or substitute relationship, then the structural purpose is determined to be incompatible.

[0151] Within the same application scope and the same process location, if two effective process application relationships point to different process versions, there is no version inheritance relationship between the two process versions, and both process application relationships are marked as mandatory application, then the process uses are determined to be incompatible.

[0152] If the structural application or the process application is incompatible, the manufacturing application compatibility check fails.

[0153] If the same manufacturing master data record is repeatedly referenced in different parent nodes, different structural locations, different application scopes, or different bill of materials versions, it is uncertain whether it is due to incompatibility for manufacturing purposes.

[0154] Transferable reference uniqueness check: The constraint verification unit checks the effective termination node of each transferable reference relationship in the shadow cleaning relationship graph.

[0155] When the same manufacturing execution reference node points to only one active virtual standard node within the same valid range, the migration reference uniqueness check passes.

[0156] When the same manufacturing execution reference node points to two or more virtual standard nodes that are in effect within the same valid range, the uniqueness check of the migrated reference fails.

[0157] Frozen reference immutability check: The constraint verification unit matches the frozen reference relationships in the local relation snapshot with the frozen reference relationships in the shadow cleaning relation graph one by one, based on the source relation identifier.

[0158] When the source relation identifier, start node identifier, and end node identifier are all consistent, the freeze reference pointer invariance check passes.

[0159] If any identifier changes, or if the corresponding frozen reference relationship is missing in the shadow cleaning relationship graph, the frozen reference pointing invariance check fails.

[0160] If any of the above checks fails, the relation constraint check result is also failed, and the overall quality score is not used to offset the failure of a single check.

[0161] After the relation constraint verification passes, the transaction processing unit verifies the source record version information recorded in the local relation snapshot again.

[0162] If the source record version information has not changed, the following write operations will be encapsulated into the same wash transaction: Create standard manufacturing master data records in the formal master data storage area; Convert the original manufacturing master data record into a historical alias node; Redirect the relationships to be migrated to the corresponding standard manufacturing master data records; Maintain frozen references pointing to historical alias nodes; Write to the valid range of the standard manufacturing master data record and historical alias nodes; Write the merged succession relationship or the split succession relationship.

[0163] If any of the above write operations fail, the transaction processing unit will undo all write operations that have been completed in the same cleaning transaction, restore the main data storage area to the state before the transaction was executed, and form a cleaning transaction failure record.

[0164] The historical mapping unit establishes the historical mapping relationship only after the cleaning transaction is successfully committed.

[0165] Historical mapping relationships include cleaning transaction identifiers, source record identifiers, standard manufacturing master data record identifiers, historical alias node identifiers, and inheritance relationship identifiers.

[0166] The frozen reference relationship is associated with the corresponding historical alias node and stored in a way that does not change with subsequent changes to the standard manufacturing master data record.

[0167] If the relation constraint verification fails, the source record version information changes, or the cleansing transaction write fails, the manufacturing master data record and manufacturing association record in the formal master data storage area will not be changed.

[0168] Isolated data is a set of candidate cleansing data that has not entered an automatic cleansing transaction and is not used as a new manufacturing master data type.

[0169] The isolation record includes the candidate cleansing set identifier, snapshot identifier, source node identifier, source relation identifier, isolation reason, conflicting fields or relations, and formation time.

[0170] The reasons for isolation include incomplete identity fields, field type conversion failure, inability to map technical status, object attribute conflicts, missing complete manufacturing business event chains for object attributes, inability to uniquely migrate association relationships, failure to pass relationship constraint validation, changes in source record version information, and failure to write cleansing transactions.

[0171] After the candidate cleansing set is marked as isolated data, the manufacturing master data records and manufacturing association records in the formal master data storage area remain unchanged.

[0172] Component records in the product lifecycle management system and material records in the enterprise resource planning system have different object identifiers, but after standardization, they have the same identity attribute key.

[0173] The valid ranges of the two records intersect, and there are no conflicting bill of materials relationships or process applicability relationships within the same application scope. Therefore, they are classified into the same candidate cleaning set of the same candidate merging type.

[0174] The incremental packet generation unit creates a virtual standard node in the local relation snapshot and maps two records to that virtual standard node.

[0175] Among them, production plans that have not yet been issued have no execution records, and their manufacturing execution reference relationships are determined to be transferable reference relationships; inspection tasks that have already generated inspection results have execution records, and their manufacturing execution reference relationships are determined to be frozen reference relationships.

[0176] The shadow relationship construction unit temporarily redirects the manufacturing execution reference relationship that has not yet been issued to the virtual standard node, keeps the manufacturing execution reference relationship of the inspection task pointing to the original manufacturing master data record, and sets the original manufacturing master data record as a historical alias node.

[0177] After the shadow cleaning relationship graph passes the relationship constraint verification, a standard manufacturing master data record is created in the formal master data storage area. Production plans that have not yet been issued reference the standard manufacturing master data record, and inspection tasks continue to reference historical alias nodes.

[0178] The same object identifier corresponds to two sets of manufacturing master data records. The two sets of records have different identity attribute keys and are associated with different process versions that do not have a version inheritance relationship.

[0179] The candidate cleaning determination unit identifies the corresponding record as a candidate cleaning set of candidate split types and divides it into two record groups.

[0180] The incremental package generation unit creates a first virtual standard node and a second virtual standard node for the two record groups, and establishes a node mapping relationship between the original manufacturing master data record and the two virtual standard nodes.

[0181] Unissued work orders are uniquely mapped to the corresponding virtual standard nodes based on their process relationship keys and valid ranges; batch records that have already been produced continue to point to the original manufacturing master data records as frozen references.

[0182] After the shadow cleaning relationship graph passes the relationship constraint verification, two standard manufacturing master data records are created in the formal master data storage area. The original manufacturing master data record is converted into a historical alias node, and a splitting and inheritance relationship is established between it and the two standard manufacturing master data records.

[0183] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for cleaning and quality verification of master data in manufacturing enterprises, characterized in that, Includes the following steps: S1: Obtain manufacturing master data records from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, as well as manufacturing association records associated with the manufacturing master data records. Manufacturing association records include bill of materials relationships, process applicability relationships, and manufacturing execution reference relationships. S2: Determine the candidate cleaning set based on the object attributes, technical status, valid range, and manufacturing-related records of the manufacturing master data records, and determine whether the candidate cleaning set is a candidate merging type or a candidate splitting type; S3: Extract the manufacturing master data records and their directly related nodes and relationships from the candidate cleaning set to form a local relationship snapshot isolated from the formal master data storage area; When the candidate cleansing set is of the candidate merge type, a virtual standard node is created in the local relation snapshot; when the candidate cleansing set is of the candidate split type, at least two virtual standard nodes are created in the local relation snapshot. Establish a node mapping relationship between manufacturing master data records and virtual standard nodes, and generate cleaning incremental packages; S4: Based on whether the business object corresponding to the manufacturing execution reference relationship has already formed an execution record, the manufacturing execution reference relationship is divided into a migrateable reference relationship and a frozen reference relationship; S5: Load the cleaning incremental package into the local relation snapshot, temporarily redirect the migration-to-be-migrated association identified by the cleaning incremental package to the corresponding virtual standard node according to the node mapping relationship, keep the frozen reference relationship pointing to the original manufacturing master data record, and set the original manufacturing master data record as a historical alias node associated with the corresponding virtual standard node to form a shadow cleaning relation graph; S6: Perform relation constraint verification on the shadow cleaning relation graph; When the relation constraint verification passes, the merging or splitting of the master data records and the redirection of the association relationships are performed in the formal master data storage area according to the cleaning incremental package, while keeping the frozen reference relationships pointing to the corresponding historical alias nodes; when the relation constraint verification fails, the candidate cleaning set is marked as isolated data.

2. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 1, characterized in that, After obtaining the manufacturing master data record, the method further includes: Based on the source system identifier and manufacturing object type of the manufacturing master data record, the corresponding field mapping relationship is invoked to convert the source fields in different source systems into a unified field structure; Convert object attributes according to the data type corresponding to the unified field structure, convert the unit of measurement used in the object attributes to the benchmark unit of measurement of the corresponding manufacturing object type, and convert the technical status of different source systems to a unified technical status. The effective time and expiration time of the manufacturing master data record are converted into a valid interval defined by the start time and the end time, and the source system identifier and source record identifier are retained in the converted manufacturing master data record.

3. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 2, characterized in that, Determine the candidate cleaning set, including: The identity fields used to distinguish the identity of manufacturing objects are determined according to the data model corresponding to the type of manufacturing object. The standardized field values ​​of each identity field are combined in a fixed field order to form identity attribute keys. The node identifiers, relationship types, relationship directions, technical status, and valid ranges associated with the manufacturing master data records are combined in a fixed data order to form manufacturing application characteristics; Among manufacturing master data records with the same type of manufacturing objects, those manufacturing master data records with different object identifiers, the same identity attribute key, and no conflicting bill of materials relationships or process applicability relationships within the intersecting valid interval are identified as candidate merge records; Manufacturing master data records with the same object identifier but different identity attribute keys, or manufacturing master data records with the same object identifier and identity attribute keys but associated with different structural positions or process versions within the intersecting valid interval, are identified as candidate split records; Corresponding candidate merge records are grouped into a candidate cleansing set of the candidate merge type, and corresponding candidate split records are grouped into a candidate cleansing set of the candidate split type.

4. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 3, characterized in that, Create a snapshot of local relationships and generate a cleansing incremental package, including: The manufacturing master data records in the candidate cleaning set are identified as the central nodes. The associated nodes that are directly connected to the central nodes through the manufacturing association records are extracted, and the relationship identifier, relationship direction, technical status and effective range between the central nodes and the associated nodes are extracted. The central node, associated nodes, and their relationships are copied to a verification area isolated from the formal main data storage area, and the corresponding source node identifier and source relationship identifier are retained for the copied nodes and relationships to form a local relationship snapshot; For candidate merging types, each candidate merging record is mapped to the same virtual standard node; for candidate splitting types, each candidate splitting record is divided into at least two record groups according to the correspondence between identity attribute keys, technical status, valid range and manufacturing-related records, and each record group is mapped to the corresponding virtual standard node. Based on the node mapping relationship, determine the virtual standard node to be created, the manufacturing master data record to be converted into a historical alias node, the association relationship to be migrated, the frozen reference relationship to be kept, and the valid range to be adjusted, and write the corresponding node processing data, relationship processing data and valid range processing data into the cleaning incremental package.

5. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 4, characterized in that, Determine the object attributes of the virtual standard node, including: Extract the release records, cross-system reception records, and manufacturing business reference records corresponding to the object attributes with different field values ​​in the candidate cleaning set; Based on the record formation time, release records, cross-system reception records, and manufacturing business reference records with the same object identifier, the same technical status, and corresponding valid intervals are sequentially connected to form a manufacturing business event chain; A complete event chain is defined as a chain of manufacturing business events that sequentially includes release records, cross-system reception records, and manufacturing business reference records, with the formation time of each record increasing sequentially. When the same object attribute corresponds to only one complete event chain, the field value corresponding to that complete event chain is written to the virtual standard node; when the same object attribute corresponds to at least two complete event chains, the field value corresponding to the complete event chain with the latest release time within the target valid interval is written to the virtual standard node; when the same object attribute does not have a complete event chain, the corresponding candidate cleaning set is marked as isolated data, and the generation of the corresponding cleaning incremental package is stopped.

6. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 5, characterized in that, The manufacturing execution reference relationships are divided into migrateable reference relationships and frozen reference relationships, including: Read the business status of the business object corresponding to the manufacturing execution reference relationship, as well as the execution records associated with the business object; When the business status is draft, not issued, or withdrawn, and there is no execution record associated with the business object, the corresponding manufacturing execution reference relationship is determined as a transferable reference relationship; When the business status is issued, in progress, or completed, or when there are production performance records, quality inspection result records, inventory turnover records, or assembly history records associated with the business object, the corresponding manufacturing execution reference relationship will be determined as a frozen reference relationship. Write the source relation identifier, original pointing object identifier, and partitioning result of the migrateable and frozen reference relations into the relation processing data.

7. The method for cleaning and quality verification of master data in manufacturing enterprises according to claim 6, characterized in that, The relation constraint verification includes: Traverse the shadow cleaning relationship graph from the parent node to the child node in the bill of materials relationship. When the traversal path returns to a node that already exists in the current traversal path, it is determined that the bill of materials loop check has failed. When two technical states pointing to the same virtual standard node within the same application scope have intersecting valid intervals and there is no version inheritance relationship between the two technical states, the conflict check for the valid intervals of the technical states fails. When the same virtual standard node is located under two different parent nodes of the same bill of materials version within the same valid range, and the corresponding bill of materials relationship is not marked as a replacement relationship, or is simultaneously associated with two process versions that do not have a version inheritance relationship, the manufacturing use compatibility check is determined to have failed. When the same manufacturing execution reference node points to at least two virtual standard nodes that are in effect within the same valid range, the uniqueness check of the migrated reference is determined to have failed. If the source relation identifier, start node identifier, or end node identifier of any frozen reference relationship is inconsistent in the local relation snapshot and the shadow cleansing relation graph, it is determined that the frozen reference pointing invariance check has failed. When all verifications pass, the creation of the standard manufacturing master data record, the conversion of the original manufacturing master data record to the historical alias node, the redirection of the migration association and the adjustment of the effective range are encapsulated into the same cleaning transaction, and the cleaning transaction is written into the formal master data storage area.

8. A master data cleaning and quality verification system for manufacturing enterprises, applicable to the master data cleaning and quality verification method for manufacturing enterprises as described in any one of claims 1-7, characterized in that, It includes an interface component, a data processing server, a formal main data storage area, and a verification area isolated from the formal main data storage area. The data processing server is connected to the interface component, the formal main data storage area, and the verification area, respectively. The interface component obtains manufacturing master data records and manufacturing association records associated with the manufacturing master data records from the product lifecycle management system, enterprise resource planning system, and manufacturing execution system, and sends the obtained data to the data processing server; The data processing server determines the candidate cleaning set and candidate cleaning type based on the object attributes, technical status, valid range and manufacturing related records of the manufacturing master data records, and copies the manufacturing master data records and their directly related nodes and relationships in the candidate cleaning set to the verification area to form a local relationship snapshot. The data processing server creates virtual standard nodes in the local relational snapshot according to the candidate cleaning type, establishes a node mapping relationship between the manufacturing master data record and the virtual standard nodes, and generates cleaning incremental packages. Based on whether the business object corresponding to the manufacturing execution reference relationship has already formed an execution record, the manufacturing execution reference relationship is divided into migrateable reference relationship and frozen reference relationship; The data processing server temporarily redirects the migration-to-be-migrated associations identified by the cleaning incremental package to the corresponding virtual standard nodes according to the node mapping relationship, keeps the frozen reference relationship pointing to the original manufacturing master data record, and sets the original manufacturing master data record as a historical alias node, forming a shadow cleaning relationship graph; The data processing server performs relation constraint verification on the shadow cleaning relation graph. When the relation constraint verification passes, it updates the formal master data storage area according to the cleaning increment package. When the relation constraint verification fails, it marks the corresponding candidate cleaning set as isolated data.

9. The manufacturing enterprise master data cleaning and quality verification system according to claim 8, characterized in that, The data processing server includes a standardization processing unit, a candidate cleaning and determination unit, an incremental package generation unit, and a shadow relationship construction unit. The standardization processing unit calls the corresponding field mapping relationship according to the source system identifier and the manufacturing object type to convert the manufacturing master data record into a unified field structure, and unifies the data type, unit of measurement, technical status and valid range of the manufacturing master data record; The candidate cleaning determination unit generates identity attribute keys based on the data model corresponding to the manufacturing object type, generates manufacturing use features based on the manufacturing association records, technical status, and effective range of the manufacturing master data record, and determines the candidate cleaning set and candidate cleaning type based on the identity attribute keys and manufacturing use features. The incremental package generation unit establishes a node mapping relationship between manufacturing master data records and virtual standard nodes based on the candidate cleaning type, and generates a cleaning incremental package containing node processing data, relationship processing data, and effective interval processing data based on the node mapping relationship. The shadow relationship construction unit loads the cleaning incremental package into the local relationship snapshot, temporarily redirects the relationships to be migrated according to the relationship processing data, and retains the original pointers of the frozen reference relationships to form a shadow cleaning relationship graph.

10. A master data cleaning and quality verification system for manufacturing enterprises according to claim 9, characterized in that, The data processing server also includes a constraint verification unit, a transaction processing unit, and a history mapping unit; The constraint verification unit detects the bill of materials cycle, technical status valid interval conflict, manufacturing purpose conflict, non-unique pointer of transferable reference, and pointer change of frozen reference relationship in the shadow cleaning relationship diagram, and generates relationship constraint verification results. When the relation constraint verification passes, the transaction processing unit encapsulates the creation of the standard manufacturing master data record, the conversion of the original manufacturing master data record to the historical alias node, the redirection of the association to be migrated, and the adjustment of the effective range into the same cleaning transaction, and writes the cleaning transaction into the formal master data storage area. The historical mapping unit establishes a correspondence between the source record identifier, the standard manufacturing master data record identifier, the historical alias node identifier, and the cleaning transaction identifier, and stores the correspondence in association with the frozen reference relationship that retains its original pointer.

Images