Digital main line processing method, medium and system based on industrial data atlas

Abstract

The invention provides a digital main line processing method based on an industrial data atlas, a medium and a system, and belongs to the technical field of the industrial data atlas. Unified access of cross-system data is realized by constructing a multi-source data acquisition adapter, cleaning and standardization processing are performed on heterogeneous data, an integrated data set is constructed, and the data collection efficiency is improved. Data entities are abstracted as nodes, cross-system association relationships are abstracted as edges to construct an industrial data graph, a graph index optimization strategy and a hierarchical storage architecture are adopted to improve cross-system query performance, a graph adaptive evolution model is utilized to realize dynamic adjustment of a graph structure, and a cross-system deep association analysis function is realized through a graph traversal algorithm. And an access control and data security mechanism is established to complete system protection, so that the technical problem that the industrial data processing system is difficult to realize cross-system deep correlation analysis is solved.

Description

Technical Field

[0001] This invention belongs to the field of industrial data mapping technology, and more specifically, relates to a digital mainline processing method, medium, and system based on industrial data mapping. Background Technology

[0002] In the field of modern industrial intelligent manufacturing, enterprises collect and manage production data through multiple heterogeneous systems such as Enterprise Resource Planning (ERP), Manufacturing Execution Systems (MES), and Customer Relationship Management (CRM). Traditional data integration technologies mainly employ solutions such as data warehouses, ETL tools, and data buses to achieve unified storage and management of multi-source data. These methods play a crucial role in business scenarios such as production planning management, quality control, supply chain coordination, and equipment maintenance by establishing unified data models and standardized data interfaces. They can effectively integrate structured data from different systems and provide basic query and analysis functions. However, traditional data integration solutions have significant shortcomings in handling cross-system data correlation analysis. Due to the use of a relational data model based on table structures, data correlation can only be performed through predefined foreign key relationships. When faced with complex multi-level relationships, a large number of table join operations are required, resulting in low query efficiency and a tendency to generate Cartesian product problems. This makes it difficult to flexibly respond to dynamically changing correlation analysis needs. In current industrial data analysis applications, due to the structured constraints of traditional relational databases and the limitations of query language expression capabilities, enterprises struggle to achieve in-depth cross-system data correlation analysis and cannot effectively uncover complex business relationships hidden in multi-source data, severely restricting the effectiveness of data-driven decision-making. In other words, existing technologies present a technical problem: industrial data processing systems struggle to achieve deep cross-system correlation analysis. Summary of the Invention

[0003] In view of this, the present invention provides a digital mainline processing method, medium and system based on industrial data graphs, which can solve the technical problem that existing industrial data processing systems are unable to achieve deep cross-system correlation analysis.

[0004] This invention is implemented as follows: The first aspect of this invention provides a digital mainline processing method based on industrial data graphs. This method constructs an industrial data relationship network by abstracting data entities as nodes and relationships as edges, and uses a graph traversal algorithm to achieve deep cross-system correlation analysis. The method includes the following steps: constructing a multi-source data acquisition adapter to collect data in real time from database tables in enterprise resource planning systems, manufacturing execution systems, and customer relationship management systems; establishing a data source identifier mapping table to achieve unified access to heterogeneous data sources; cleaning and standardizing the collected multi-source heterogeneous data; constructing an integrated dataset; establishing a data quality assessment vector; and completing data preprocessing and quality control; and constructing an industrial data graph based on the integrated dataset, abstracting data entities as nodes and relationships as edges. Relationships are abstracted as edges, and the attribute information of nodes and edges is stored using a graph database. A graph index optimization strategy and hierarchical storage architecture are constructed, a multi-dimensional index is established, and a hot data caching mechanism and a query result pre-computation system are adopted. A graph adaptive evolution model is used to monitor business changes in real time. When the Euclidean distance between the evolution feature vector and the baseline evolution vector in the multi-dimensional space is >0.85, the pattern matching algorithm is activated to automatically adjust the graph structure. Data traceability and exploration service functions are implemented. Intelligent data traceability function is implemented based on graph traversal algorithm. When the query hysteresis error is >0.15, the hysteresis compensation algorithm is activated. If the hysteresis error ∈ [0, 0.05], the standard query processing mode is adopted. A role-based access control and data security guarantee mechanism is established.

[0005] Among them, the multi-source data acquisition adapter adopts a protocol-adaptive connection method, establishes corresponding data access drivers for different database types, and supports multiple data access protocols such as Structured Query Language, Open Database Connectivity, and Application Programming Interface.

[0006] The data source identifier mapping table records the unique identifier, database connection parameters, table structure information and field mapping relationships of each data source, and records the access path and update frequency of each system's data source. It is used to establish a unified data access interface. The unique identifier comes from the system code of the Enterprise Resource Planning System, Manufacturing Execution System and Customer Relationship Management System.

[0007] The data quality assessment vector comprises five dimensions: data integrity coefficient, accuracy coefficient, consistency coefficient, timeliness coefficient, and validity coefficient. It is constructed by performing quality checks and scoring on the original data to monitor data integrity.

[0008] In the industrial data graph, nodes represent data entities, and edges represent the relationships between entities. Each node and edge contains rich attribute information. Node attributes include entity identifier, entity type, business attributes, and timestamp, while edge attributes include relationship type, relationship strength, relationship direction, and relationship weight.

[0009] Among them, the graph index optimization strategy includes three types: node attribute index, relation index and composite index. The node attribute index is built based on the key attributes of the node, the relation index is built based on the type and weight of the edge, and the composite index combines multiple attributes to build a joint index.

[0010] The tiered storage architecture stores hot data in the memory layer, warm data in the solid-state drive layer, and cold data in the hard disk drive layer, dynamically adjusting the storage tiers based on data access frequency.

[0011] The query performance optimization vector includes five metrics: index hit rate, cache hit rate, query response time, concurrent processing capability, and resource utilization. These metrics are used to describe the effectiveness of query optimization strategies, evaluate the optimization results, and complete the optimization configuration for query performance.

[0012] Among them, the graph adaptive evolution model adopts a multi-scale representation integration mechanism based on hierarchical feature fusion and a sequence dependency modeling algorithm based on cyclic attention. When changes in business requirements or the access of new systems are detected, the model automatically analyzes changes in data patterns, recommends appropriate graph structure adjustment schemes, and ensures the backtrackability of the evolution process through a version control mechanism.

[0013] The evolutionary feature vector includes five dimensions: the rate of change of the number of nodes, the rate of change of the number of edges, the rate of change of structural density, the rate of change of connectivity, and the rate of change of level depth. The baseline evolutionary vector includes five baseline values: standard node density, standard edge density, standard structural complexity, standard connectivity, and standard number of levels.

[0014] The data tracing function is based on a graph traversal algorithm. After the user inputs the tracing target, the system performs a depth-first search or a breadth-first search along the path of the relationship in the graph to generate a complete data tracing link. Path pruning and heuristic search strategies are used to improve search efficiency during the tracing process.

[0015] Among them, the hysteresis compensation algorithm optimizes repeated queries by establishing a query history record and result caching mechanism. When the hysteresis effect error is controlled within 0.05, the standard query processing mode is adopted. When the error exceeds the threshold, the compensation algorithm is activated to correct the error, eliminate the impact of backhaul error, and complete the construction of the data traceability link.

[0016] Among them, the role-based access control mechanism assigns different data access permissions to different user roles, encrypts the data transmission process, adopts the Secure Sockets Layer protocol to prevent data from being stolen or tampered with during transmission, and sets up a data backup strategy to complete the security protection and data protection of the digital mainline system.

[0017] The graph adaptive evolution model consists of an input layer that receives evolutionary feature vectors and business change information, a feature extraction layer that extracts structural features through a convolutional neural network, an attention layer that establishes relationships between features using a multi-head attention mechanism, a decision layer that outputs evolutionary strategy suggestions through a fully connected neural network, and an output layer that generates a graph structure adjustment scheme.

[0018] The evolutionary feature fusion function integrates multi-dimensional evolutionary feature information. Its inputs include structural change feature vectors, business demand change vectors, and system performance index vectors, and its output is a comprehensive evolutionary feature vector.

[0019] The query optimization weight function is used to calculate the weight allocation of different optimization strategies. The inputs include query type identifier, historical performance data and system resource status, and the output is the optimization strategy weight vector.

[0020] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned digital mainline processing method based on industrial data graphs.

[0021] A third aspect of the present invention provides a digital mainline processing system based on industrial data maps, comprising the aforementioned computer-readable storage medium. The system is any one of a computer, a server, or a microcontroller. The computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0022] This invention constructs an industrial data relationship network based on a graph structure, transforming the traditional tabular data organization into a node-edge graph structure representation. This allows cross-system data entities to form a complete association network through edge connections, completely overcoming the limitations of traditional relational databases in expressing complex relationships. This invention replaces traditional SQL join queries with graph traversal algorithms, achieving multi-level association analysis through depth-first search and breadth-first search strategies. This avoids the overhead of multi-table joins in traditional methods, significantly improving the efficiency and accuracy of cross-system data analysis. It transforms join queries that previously required complex SQL statements into an intuitive graph path search process. Through graph index optimization and adaptive evolution mechanisms, this invention ensures the system can efficiently handle large-scale cross-system association analysis tasks while possessing the ability to dynamically adjust the graph structure to adapt to constantly changing business association needs, fundamentally solving the technical limitations of traditional data processing systems in deep cross-system association analysis. In summary, this invention solves the technical problem mentioned in the background art of the difficulty in implementing deep cross-system association analysis in industrial data processing systems. Attached Figure Description

[0023] Figure 1 This is a flowchart of the method of the present invention.

[0024] Figure 2 This is a graph showing the trend of changes in the spectral structure evolution characteristics in Example 2.

[0025] Figure 3 This is a comparison chart of the waiting time distribution for different query types in Example 2. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0027] like Figure 1 The diagram shown is a flowchart of a digital mainline processing method based on industrial data maps provided by the first aspect of the present invention. This method includes the following steps:

[0028] S01. Construct a multi-source data acquisition adapter to collect data in real time from database tables in the enterprise resource planning system, manufacturing execution system, and customer relationship management system. Establish a data source identifier mapping table to record the access path and update frequency of data sources in each system, and complete the unified access of heterogeneous data sources.

[0029] S02. Clean and standardize the collected multi-source heterogeneous data to eliminate data format differences and noise, construct an integrated dataset with a unified data format, establish a data quality assessment vector to monitor data integrity, and complete data preprocessing and quality control.

[0030] S03. Construct an industrial data graph based on the integrated dataset, abstract data entities as nodes and relationships as edges, use a graph database to store the attribute information of nodes and edges, form a complete industrial data relationship network, and complete the initial construction of the graph model.

[0031] S04. Construct graph index optimization strategies and hierarchical storage architecture, establish multi-dimensional indexes for node attributes in the graph database, adopt hot data caching mechanisms and query result pre-computation systems, construct query performance superposition optimization vectors to evaluate the optimization effect, and complete the query performance optimization configuration.

[0032] S05. Use the graph adaptive evolution model to monitor business changes in real time. When the Euclidean distance between the evolution feature vector and the baseline evolution vector in the multidimensional space is >0.85, start the pattern matching algorithm to automatically adjust the graph structure and complete the adaptive evolution of the graph structure.

[0033] S06. Implement data traceability and exploration service functions. Implement intelligent data traceability function based on graph traversal algorithm. When the query hysteresis error > 0.15, start the hysteresis compensation algorithm to eliminate the influence of backhaul error. If the hysteresis error ∈ [0, 0.05], adopt the standard query processing mode to complete the construction of data traceability link.

[0034] S07. Establish a role-based access control and data security mechanism, encrypt data transmission, set data backup strategies, and complete the security protection and data protection of the digital mainline system.

[0035] The multi-source data acquisition adapter employs a protocol-adaptive connection method, establishing corresponding data access drivers for different database types and supporting various data access protocols such as Structured Query Language, Open Database Connectivity, and Application Programming Interfaces. The data source identifier mapping table records the unique identifier, database connection parameters, table structure information, and field mapping relationships for each data source, used to establish a unified data access interface. The unique identifiers in the data source identifier mapping table are derived from the system codes of the Enterprise Resource Planning System, Manufacturing Execution System, and Customer Relationship Management System. The database connection parameters are used for subsequent connection establishment of the data acquisition adapter.

[0036] The data quality assessment vector comprises five dimensions: data integrity coefficient, accuracy coefficient, consistency coefficient, timeliness coefficient, and validity coefficient. It is constructed by performing quality checks and scoring on the original data. The data integrity coefficient represents the completeness of data records; the accuracy coefficient represents the correctness of data values; the consistency coefficient represents the degree of uniformity between data from different data sources; the timeliness coefficient represents the timeliness of data updates; and the validity coefficient represents the degree to which data conforms to business rules. The data integrity coefficient is derived from the integrity statistics of centralized records in the integrated dataset, while the accuracy coefficient is used for subsequent accuracy verification of the node attributes in the industrial data graph.

[0037] In the industrial data graph, nodes represent data entities, and edges represent the relationships between entities. Each node and edge contains rich attribute information. Node attributes include entity identifier, entity type, business attributes, and timestamps, while edge attributes include relationship type, relationship strength, relationship direction, and relationship weight. The graph database uses an attribute graph model to store data and supports graph query and traversal operations. The entity identifier is derived from the primary key field in the integrated dataset, and the relationship weights are used for path calculation in subsequent graph traversal algorithms.

[0038] The graph index optimization strategy includes three types: node attribute index, relationship index, and composite index. Node attribute indexes are built based on the key attributes of nodes, relationship indexes are built based on the type and weight of edges, and composite indexes combine multiple attributes to create a joint index. The hierarchical storage architecture stores hot data in memory, warm data in solid-state drives (SSDs), and cold data in hard disk drives (HDDs), dynamically adjusting the storage hierarchy based on data access frequency. The node attribute index is built based on the business attributes of nodes in the industrial data graph, and the relationship index is used for subsequent relationship lookup optimization in graph traversal algorithms.

[0039] The query performance optimization vector is used to describe the effectiveness evaluation of the query optimization strategy, and includes five indicators: index hit rate, cache hit rate, query response time, concurrency processing capability, and resource utilization. Index hit rate represents the proportion of queries that use the index; cache hit rate represents the proportion of query results that are hit in the cache; query response time represents the average execution time of a query operation; concurrency processing capability represents the system's ability to process query requests simultaneously; and resource utilization represents the efficiency of system resource utilization. The index hit rate is derived from the execution statistics of the graph index optimization strategy, and the query response time is used for threshold determination in subsequent latency compensation algorithms.

[0040] The graph adaptive evolution model employs a multi-scale representation integration mechanism based on hierarchical feature fusion and a sequence dependency modeling algorithm based on recurrent attention to achieve graph structure evolution. When changes in business requirements or the integration of new systems are detected, the model automatically analyzes changes in data patterns, recommends appropriate graph structure adjustment schemes, and ensures the backtracking of the evolution process through a version control mechanism. The input of the graph adaptive evolution model comes from business change monitoring data, and the graph structure adjustment schemes are used for subsequent structural updates of the industrial data graph.

[0041] The evolutionary congestion risk matrix is ​​used to assess the potential risks during the evolution of the graph structure. The matrix dimensions include structural complexity, data consistency, performance impact, compatibility, and rollback difficulty. Structural complexity represents the complexity of the new graph structure; data consistency represents the degree to which data remains consistent during evolution; performance impact represents the degree to which evolution affects system performance; compatibility represents the degree of compatibility between the new structure and the existing system; and rollback difficulty represents the difficulty of restoring to the original state in the event of evolution failure. The structural complexity is derived from graph structure analysis and calculation, and the rollback difficulty is used for risk assessment in subsequent evolutionary decisions.

[0042] The query wait time distribution vector records the wait time distribution characteristics of different types of query operations, including the average wait time for five query types: simple node query, relation traversal query, multi-hop path query, aggregation statistical query, and union query. This vector is used to analyze system query performance bottlenecks and guide the formulation and adjustment of query optimization strategies. The wait time for simple node queries is derived from query log statistics of the graph database, and the query wait time distribution vector is used in the subsequent calculation of the query performance superposition optimization vector.

[0043] The performance gain evaluation matrix is ​​used to quantify the performance improvement effect brought about by the graph index optimization strategy. The matrix rows represent different query types, the columns represent different optimization strategies, and the matrix elements represent the performance gain coefficient of the query type under the optimization strategy. The performance gain coefficient is calculated by comparing the query response time before and after optimization and is used to evaluate the effectiveness of the optimization strategy. The performance gain coefficient is derived from the comparative calculation of query response time, and the performance gain evaluation matrix is ​​used for subsequent selection and adjustment of query optimization strategies.

[0044] The system load fluctuation monitoring vector records the fluctuation characteristics of system resource usage in real time, including five dimensions: processor utilization fluctuation, memory utilization fluctuation, disk I / O fluctuation, network bandwidth fluctuation, and concurrent connection fluctuation. This vector is used to monitor the system's operating status, triggering resource adjustment and load balancing mechanisms when the fluctuation exceeds a preset threshold. The processor utilization fluctuation originates from system monitoring data collection, and the system load fluctuation monitoring vector is used for subsequent dynamic allocation of system resources.

[0045] The data tracing function is implemented based on a graph traversal algorithm. After the user inputs the tracing target, the system performs a depth-first search or breadth-first search along the relational paths in the graph to generate a complete data tracing link. Path pruning and heuristic search strategies are employed during the tracing process to improve search efficiency, while simultaneously recording the weight and confidence information of the tracing path. The tracing target originates from the user's input request, and the data tracing link is used for subsequent business analysis and problem localization.

[0046] The hysteresis compensation algorithm is used to eliminate the impact of backhaul errors during graph queries. It optimizes duplicate queries by establishing query history records and a result caching mechanism. When the hysteresis error is controlled within 0.05, the standard query processing mode is adopted; when the error exceeds a threshold, the compensation algorithm is activated to correct the error. The query history records are derived from the graph database's operation logs, and the hysteresis compensation algorithm ensures the accuracy of subsequent query results.

[0047] The role-based access control mechanism assigns different data access permissions to different user roles, ensuring data security and confidentiality. Data transmission is encrypted using the Secure Sockets Layer (SSL) protocol to prevent data theft or tampering during transmission. A data backup strategy regularly backs up industrial data maps and related data to prevent data loss and facilitate disaster recovery. The user roles are defined based on the enterprise organizational structure, and the data access permissions are used for subsequent data query and operation access control.

[0048] The evolutionary feature vector describes the characteristic information of graph structure changes, including five dimensions: the rate of change in the number of nodes, the rate of change in the number of edges, the rate of change in structural density, the rate of change in connectivity, and the rate of change in hierarchy depth. The rate of change in the number of nodes indicates the degree of change in the number of nodes in the graph; the rate of change in the number of edges indicates the degree of change in the number of edges in the graph; the rate of change in structural density indicates the change in the density of the graph structure; the rate of change in connectivity indicates the degree of change in the connectivity of the graph; and the rate of change in hierarchy depth indicates the degree of change in the hierarchy depth of the graph structure. The rate of change in the number of nodes is derived from the structural statistics of industrial data graphs, and the evolutionary feature vector is used as input for subsequent graph adaptive evolution models.

[0049] The baseline evolution vector provides a reference standard for graph structure evolution, comprising five baseline values: standard node density, standard edge density, standard structural complexity, standard connectivity, and standard number of levels. Standard node density represents the node distribution density under normal operational conditions; standard edge density represents the relation distribution density under normal operational conditions; standard structural complexity represents the standard complexity of the graph structure; standard connectivity represents the standard connectivity level of the graph; and standard number of levels represents the standard level depth of the graph. The standard node density is derived from statistical analysis of the graph structure during historical stable periods, and the baseline evolution vector is used for distance calculation of subsequent evolutionary feature vectors.

[0050] The structure of the graph adaptive evolution model is as follows: the input layer receives evolution feature vectors and business change information; the feature extraction layer extracts structural features through a convolutional neural network; the attention layer establishes the correlation between features using a multi-head attention mechanism; the decision layer outputs evolution strategy suggestions through a fully connected neural network; and the output layer generates a graph structure adjustment scheme.

[0051] The steps for establishing the training dataset for the graph adaptive evolution model include collecting historical graph evolution cases, labeling the differences in graph structure before and after evolution, constructing feature vectors to describe the evolution pattern, establishing an evaluation standard for evolution effect, forming a training sample set containing input features and expected output, and dividing the training set, validation set, and test set in an 8:1:1 ratio.

[0052] The training steps of the graph adaptive evolution model include initializing model parameters, training the model using the stochastic gradient descent algorithm, updating network weights using the backpropagation algorithm, evaluating model performance using a validation set, employing an early stopping mechanism to prevent overfitting, and finally validating the model's generalization ability on a test set and saving the trained model parameters for actual deployment.

[0053] The multi-scale representation integration mechanism based on hierarchical feature fusion enhances the model's ability to perceive changes in the graph structure by weightedly fusing features from different network layers and combining low-level detail features with high-level semantic features. This improves the accuracy and completeness of evolutionary strategy recommendations. The mechanism enables the model to simultaneously capture local structural changes and global pattern evolution, providing more accurate analysis and prediction for graph evolution scenarios.

[0054] The proposed inter-sequence dependency modeling algorithm based on cyclic attention establishes correspondences between evolutionary sequences through cyclic attention calculations when processing the temporal features of graph evolution. This accurately captures long-term dependencies and evolutionary patterns during the evolutionary process, significantly improving the model's prediction accuracy of evolutionary trends and its ability to identify anomalous evolutionary patterns. The algorithm enables the model to understand the intrinsic laws of the evolutionary process, providing decision support for adaptive evolution.

[0055] The specific implementation methods of the above steps are described in detail below.

[0056] The specific implementation of step S01 involves constructing a multi-source data acquisition adapter to achieve unified access to heterogeneous data sources. First, a data access driver based on Structured Query Language (SCL) is established for the Enterprise Resource Planning (ERP) system. This driver parses the database table structure of the ERP system and extracts metadata information from the table fields. Then, a data access driver based on Open Database Connectivity (ODB) is established for the Manufacturing Execution System (MES). Communication with the MES database is achieved by configuring the data source name and connection string. Next, a data access driver based on the Application Programming Interface (API) is established for the Customer Relationship Management (CRM) system. This driver calls the data interface provided by the CRM system to obtain business data. Finally, a data source identifier mapping table is created to record the unique identifiers of each system. These unique identifiers use the system code as the primary key to ensure the mapping relationship. The process involves ensuring uniqueness and recording database connection parameters, including host address, port number, database name, and user credentials. It also requires recording table structure information, such as table name, field names, data types, and primary / foreign key relationships. Furthermore, it establishes field mapping relationships to associate fields with the same business meaning across different systems. Finally, it sets the data update frequency parameter: for manufacturing execution systems with high real-time requirements, the update frequency is set to 1 to 5 seconds; for enterprise resource planning systems, it's set to 10 to 30 seconds; and for customer relationship management systems, it's set to 30 to 60 seconds. This step aims to resolve protocol differences and inconsistent access methods between different data sources, achieving transparent access to data sources through protocol adaptation technology and providing a unified data acquisition interface for subsequent data processing.

[0057] The specific implementation of step S02 involves cleaning and standardizing the collected multi-source heterogeneous data. First, missing value detection is performed on the raw data, and the missing percentage for each field is calculated. When the missing percentage is less than 5%, mean or mode imputation methods are used for data completion. When the missing percentage is between 5% and 20%, a regression-based predictive imputation method is used. When the missing percentage is greater than 20%, the field is marked as low-quality data and its weight is reduced in subsequent processing. Next, the data format is standardized by converting date and time formats to the ISO 8601 format, normalizing numerical units, and converting text encoding to a universal character set conversion format (8-bit encoding). Then, outlier detection is performed using box plots to identify outlier data points exceeding 1.5 times the interquartile range. For identified outliers, it is determined whether they are reasonable extreme values ​​according to business rules. If they are data noise, they are removed or corrected. Then, a data quality assessment vector is constructed. The data integrity coefficient is calculated by dividing the number of complete records by the total number of records. The accuracy coefficient is calculated by comparing with the reference dataset to obtain the proportion of correct data. The consistency coefficient is calculated by detecting the consistency of data of the same entity between different data sources. The timeliness coefficient is calculated by dividing the real-time data volume by the total data volume. The validity coefficient is calculated by verifying whether the data conforms to the predefined business rules. Finally, an integrated dataset with a unified data format is formed. This step uses statistical analysis methods and data mining techniques to improve the quality of the original data and ensure the reliability of the data foundation for subsequent graph construction. When the coefficients of each dimension of the data quality assessment vector are all greater than 0.8, the data quality is considered to have reached the usable standard.

[0058] The specific implementation of step S03 is to construct an industrial data graph based on an integrated dataset. First, data entity identification is performed by traversing all data records in the integrated dataset and extracting unique entity identifiers based on the primary key field. Business objects such as product information, work order information, equipment information, and personnel information are abstracted as graph nodes. Each node is assigned an entity type attribute to distinguish different categories of business entities. Business attribute information, including product number, specifications, and production batch values, is recorded. A timestamp attribute is added to record the node's creation and update times. Then, relationship extraction is performed, analyzing foreign key constraints between data tables to identify direct relationships between entities. Implicit relationships, such as supply relationships between products and suppliers and usage relationships between work orders and equipment, are mined through business rules. The identified relationships are abstracted as graph edges, and a relationship type attribute, such as containment, is set for each edge. Relationships such as dependencies and associations are calculated by determining the frequency of association occurrences. Relationship directions are set to represent the starting and target nodes. Normalized association strength values ​​are used to calculate relationship weights. A graph database is selected as the storage engine, and an attribute graph model is used for data organization. Nodes, edges, and their attribute information are written into the graph database. Node and edge indexes are created to support efficient graph query operations. Graph query language is used to define graph traversal rules and query patterns, ultimately forming a complete industrial data relationship network. The purpose of this step is to transform traditional relational data into graph-structured data, utilizing the graph's topological structure to naturally express complex relationships between entities, providing a graph computational foundation for subsequent data tracing and relationship analysis. The uniqueness of entity identifiers is ensured by calculating the digest value of the primary key field using a hash algorithm. The numerical range of relationship weights is set between 0.1 and 1.0.

[0059] The specific implementation of step S04 involves constructing a graph index optimization strategy and a hierarchical storage architecture to improve query performance. First, multidimensional indexes are created for nodes in the graph database. Category indexes are created for the entity type attributes of nodes, and exact match indexes are created for key fields in business attributes such as product number and work order number. A range query index is created for the timestamp attribute to support time-period queries. A B-tree index structure is used to quickly locate node attributes. Then, relationship indexes are created for edges. Category indexes are created based on the relationship type of the edges to facilitate filtering by relationship type. A sorting index is created based on the relationship weight to support queries sorted by weight. A composite index is created to combine relationship types and weights to create a joint index, improving the efficiency of complex queries. Next, a hot data caching mechanism is established. Historical query logs are analyzed to identify frequently accessed nodes and relationships. The top 10% of accessed data are marked as hot data and loaded into the memory cache. The Least Recently Used (LRU) algorithm is used to manage the cache space; when the cache is full, the least recently accessed data is evicted. A query result pre-calculation system is implemented. For common query patterns such as fixed-path traversal queries, results are pre-calculated and cached. When the same or similar queries are received, the pre-calculated results are directly returned. A hierarchical storage architecture is constructed to store data according to access... The query performance is divided into three layers: hot data, warm data, and cold data. Hot data is stored in memory for sub-millisecond access, warm data is stored in SSDs for millisecond access, and cold data is stored in HDDs for second-level access. The storage layer is dynamically adjusted based on the data access frequency. When the access frequency of warm data exceeds a threshold, it is promoted to the hot data layer; when the access frequency of hot data falls below the threshold, it is demoted to the warm data layer. The thresholds are set as follows: hot data access frequency greater than 100 times per second, warm data access frequency between 10 and 100 times per second, and cold data access frequency less than 10 times per second. Finally, a query performance optimization mechanism is built. The system performs performance evaluation by calculating the index hit rate (number of queries using the index divided by the total number of queries), the cache hit rate (number of queries returning cached results divided by the total number of queries), the average execution time of all queries, and the concurrent processing capability by stress testing to determine the maximum number of queries the system can handle simultaneously. Resource utilization is calculated by dividing the actual amount of resources used by the total amount of available resources. This step employs a multi-level caching strategy and tiered storage technology to significantly reduce data access latency. Index optimization reduces the query scan range, resulting in an overall query performance improvement of 3 to 5 times compared to the unoptimized state.

[0060] The specific implementation of step S05 involves using a graph adaptive evolution model to monitor and respond to business changes in real time. First, an evolutionary feature vector is constructed. The total number of nodes in the current graph is counted, and the rate of change in the number of nodes is calculated as the current number of nodes minus the baseline number of nodes, then divided by the baseline number of nodes. The total number of edges in the current graph is counted, and the rate of change in the number of edges is calculated as the current number of edges minus the baseline number of edges, then divided by the baseline number of edges. The rate of change in structural density is calculated as the current number of edges divided by the maximum possible number of edges, then the difference between this and the baseline density, divided by the baseline density. The number of connected components in the graph is analyzed to calculate the rate of change in connectivity. The maximum depth of the graph is measured to calculate the layer. The evolutionary feature vector is composed of the change rates of the five dimensions. A baseline evolutionary vector is then obtained. The standard node density is statistically derived from the graph structure during historical stable periods. The standard edge density is also derived from historical data analysis. The standard structural complexity is obtained by calculating the average cyclic complexity of the historical graph. The standard connectivity is the reciprocal of the average connected components of the historical graph. The standard number of levels is the average maximum depth of the historical graph. Next, the Euclidean distance between the evolutionary feature vector and the baseline evolutionary vector in five-dimensional space is calculated. This is done by squared differences in corresponding dimensions of the two vectors, summing them, and then taking the square root. When the calculated distance value is greater than... A value of 0.85 indicates a significant change in the graph structure. At this point, a pattern matching algorithm is initiated to analyze the change pattern. The algorithm first extracts the weights of each dimension of the evolutionary feature vector to identify which dimensions show the most significant changes. If the rate of change in the number of nodes is the largest, it is identified as an entity expansion pattern; if the rate of change in the number of edges is the largest, it is identified as a relation enhancement pattern; and if the rate of change in the structure density is the largest, it is identified as a structure reorganization pattern. Based on the identified evolutionary patterns, a suitable graph structure adjustment scheme is selected from a predefined adjustment scheme library. The adjustment schemes include adding new node types and relation types, modifying node attributes, and adjusting relation weights. The operation involves recording the adjustment plan through a version control mechanism to ensure that the evolution process is traceable and rollbackable. The purpose of this step is to enable the graph to automatically adapt to business changes and avoid the lag of manual maintenance. The multi-scale representation integration mechanism based on hierarchical feature fusion can simultaneously capture local detail changes and global structural evolution. The inter-sequence dependency modeling algorithm based on recurrent attention can accurately capture the temporal pattern of the evolution process. The Euclidean distance threshold of 0.85 is the optimal judgment point obtained through experimental verification. Changes below this threshold are normal fluctuations and do not require adjustment, while changes above this threshold require triggering the adaptive evolution mechanism.

[0061] The specific implementation of step S06 is to realize the data traceability and exploration service function. First, the traceability target input by the user is received. The traceability target can be a business entity identifier such as product number, work order number, or batch number. The starting node corresponding to the traceability target is located in the graph database. The depth-first search algorithm or breadth-first search algorithm is used to traverse along the edges of the graph. The depth-first search algorithm is suitable for scenarios with long traceability links, which prioritize exploring a single complete path. The breadth-first search algorithm is suitable for scenarios that require comprehensive exploration, which expands the search range hierarchically. During the traversal, a path pruning strategy is applied. When the traversal depth exceeds the preset maximum depth, the search for that path is stopped. When the cumulative path weight is lower than the threshold, the path is pruned. The cumulative path weight is obtained by multiplying the relation weights of all edges on the path. A heuristic search strategy is used to prioritize exploring edges with higher weights. The nodes to be explored are processed in descending order of edge weight by maintaining a priority queue. All paths, their weights, and confidence information are recorded during the traversal. The confidence is calculated based on the path weight and path length. A complete data traceability link is generated and displayed to the user in a graphical form. Then, query lag effect detection is performed, the response time of the query operation is statistically analyzed, and the lag effect error is calculated to be equal to the actual response time. The response time is calculated by subtracting the theoretical optimal response time from the response time and then dividing by the theoretical optimal response time. When the hysteresis error is greater than 0.15, it indicates that the system has a significant backhaul error. At this point, the hysteresis compensation algorithm is activated. The hysteresis compensation algorithm stores recent query requests and results by establishing a query history table. For duplicate or similar queries, results are returned directly from the history table. A result caching mechanism is established to store the results of high-frequency queries in memory. The cache expiration time is set to 300 to 600 seconds, adjusted according to the data update frequency. For first-time queries or queries where the cache has expired, real-time calculations are performed and the cache is updated. When the hysteresis error is controlled within... A value between 0 and 0.05 indicates good query performance. In this case, the standard query processing mode is used to directly execute the graph traversal operation and return the results. This step utilizes the natural relational structure of the graph to achieve efficient data tracing. Compared with traditional relational database multi-table join queries, the graph traversal algorithm can significantly reduce query complexity. Path pruning and heuristic search further improve tracing efficiency. The hysteresis compensation algorithm effectively eliminates the performance loss caused by repeated queries. The maximum search depth reference value is set to 10 to 15 layers, the path weight threshold is set to 0.1, and the hysteresis effect error threshold of 0.15 corresponds to an additional delay of approximately 150 milliseconds.

[0062] The specific implementation of step S07 involves establishing a role-based access control and data security mechanism. First, user roles are defined according to the enterprise's organizational structure, categorizing users into different levels such as system administrator, data analyst, business operator, and visitor roles. Data access permissions are configured for each role: system administrators have read / write permissions for all data and system configuration permissions; data analysts have read-only and query permissions for data; business operators have read / write permissions for specific business domain data; and visitors only have read-only permissions for public data. An access control matrix is ​​established to record the access relationship between roles and data resources. Before a user performs a data operation, permission verification is performed. The access control matrix is ​​queried to determine if the current user role has the corresponding operation permissions. If permissions are insufficient, the operation is rejected and an access log is recorded. Then, the data transmission process is encrypted. An encrypted channel is established using Secure Sockets Layer (SSL) or Transport Layer Security (TLS) protocols. Advanced Encryption Standard (ALS) algorithms are used for symmetric encryption of the transmitted data, with a key length of 256 bits to ensure encryption strength. For sensitive data such as passwords and personal information, asymmetric encryption algorithms are used, encrypting the data using the recipient's public key. The original data is obtained by decrypting the private key. Sensitive fields are encrypted and stored at the data storage level. A hash algorithm is used for one-way encryption of the password. Then, a data backup strategy is set up, combining full and incremental backups of the industrial data map. Full backups are performed weekly to export and store the complete graph database, while incremental backups are performed daily to back up only the data added and modified that day. Backup data is stored in an off-site disaster recovery center to achieve geographical redundancy. Backup retention periods are set: recent backups are retained for 30 days, and historical backups are retained monthly for 12 months. Backup data exceeding the retention period is automatically cleaned up. A disaster recovery plan is established so that the system can quickly recover from the backup data when a failure occurs. Recovery drills are performed regularly to verify the availability of the backup data. This step ensures the confidentiality, integrity, and availability of data through a multi-layered security protection mechanism. Role-based access control enables fine-grained permission management. Encryption technology prevents data from being stolen or tampered with during transmission and storage. The backup strategy ensures the persistence and recoverability of data. It is recommended to use Secure Sockets Layer protocol version 3.0 or higher and Transport Layer Security protocol version 1.2 or higher. The encryption key for backup data should be managed separately from the production environment.

[0063] Further explanation is needed regarding the detailed structure of the graph adaptive evolution model, which comprises five main parts: an input layer, a feature extraction layer, an attention layer, a decision layer, and an output layer. The input layer receives two types of data: the first is evolutionary feature vectors, containing numerical values ​​across five dimensions: the rate of change in node number, the rate of change in edge number, the rate of change in structure density, the rate of change in connectivity, and the rate of change in hierarchy depth; the second is business change information, including discrete features such as new system access identifiers, business process change identifiers, and data pattern change identifiers. The input layer normalizes continuous features, mapping their values ​​to the 0-1 range, and performs one-hot encoding on discrete features. The encoding is converted into a vector representation. The feature extraction layer adopts a convolutional neural network architecture. The first convolutional layer uses 64 convolutional kernels to extract local features, with a kernel size of 3x3 and a stride of 1. A modified linear unit activation function is used to introduce non-linearity. The second convolutional layer uses 128 convolutional kernels to extract higher-level features. Max pooling is used to downsample and reduce the feature dimensionality, with a pooling window size of 2x2. The third convolutional layer uses 256 convolutional kernels to extract global semantic features. Residual connections are used to fuse shallow and deep features, achieving multi-scale representation integration based on hierarchical feature fusion. The attention layer uses a multi-head architecture. The attention mechanism divides the feature vector into eight attention heads, each independently calculating the query vector, key vector, and value vector. It calculates the association weights between features through scaled dot product attention. The outputs of multiple attention heads are concatenated and linearly transformed to obtain a comprehensive feature representation. A recurrent attention structure is introduced to cyclically associate the features of the current time step with those of historical time steps. Gated recurrent units are used to capture the temporal dependencies in the evolutionary process, achieving sequential dependency modeling based on recurrent attention. The decision layer contains three fully connected neural network layers. The first fully connected layer maps the feature vector output by the attention layer to a 512-dimensional hidden representation. The second fully connected layer further maps it to 256 dimensions, and the third fully connected layer maps it to the number of evolutionary strategy categories. The Softmax function is used to convert the output into a probability distribution, representing the recommendation probability of different evolutionary strategies. The output layer generates a graph structure adjustment scheme. Based on the probability distribution output by the decision layer, the evolutionary strategy with the highest probability is selected. This strategy corresponds to a set of specific adjustment operations, including node type addition, relation type addition, attribute modification, and weight adjustment operations. The adjustment scheme is output in a structured format, including information such as operation type, target object, modified content, and execution order.

[0064] The detailed steps for establishing the training dataset for the graph adaptive evolution model include four stages: historical data collection, case annotation and feature construction, and dataset partitioning. The historical data collection stage extracts the graph evolution history from the past 2 to 3 years from the production environment. Each record includes a snapshot of the graph structure before evolution, an evolution trigger event, a snapshot of the graph structure after evolution, and an evolution effect evaluation result. The graph structure snapshot is stored in a graph serialization format, containing the set of nodes and edges and all their attribute information. Evolution trigger events record the business reasons that led to graph changes, such as the introduction of new product lines, production process optimization, and system function expansion. The evolution effect evaluation result is quantitatively evaluated using system performance indicators such as query response time, data throughput, and resource utilization. To assess the merits and demerits of each evolutionary case, manual annotation is performed during the case annotation phase. First, the structural differences in the graph before and after evolution are calculated, and the number of newly added, deleted, and modified nodes, as well as the number of newly added, deleted, and modified edge weights, are counted. Then, the evolution type is labeled, categorizing it into four modes: expansion evolution, contraction evolution, restructuring evolution, and optimization evolution. Expansion evolution corresponds to scenarios of business scale growth, contraction evolution to scenarios of business simplification, restructuring evolution to scenarios of process reorganization, and optimization evolution to scenarios of performance improvement. The evolution effect is then annotated, categorized into four levels based on the degree of system performance improvement after evolution: excellent, good, average, or poor. Excellent indicates a performance improvement of over 30%, good indicates a performance improvement of over 30%, and average indicates a performance improvement of over 30%. Performance improvement is indicated as 10% to 30%, generally 0% to 10%, and poor performance indicates a decrease. In the feature construction phase, training samples are built based on labeled evolutionary cases. For each evolutionary case, an evolutionary feature vector is extracted as input features. The rate of change in the number of nodes is calculated as the number of nodes after evolution minus the number of nodes before evolution, then divided by the number of nodes before evolution. The rate of change in the number of edges is calculated as the number of edges after evolution minus the number of edges before evolution, then divided by the number of edges before evolution. The rate of change in structural density is calculated by comparing the density index of the graph before and after evolution. The rate of change in connectivity is calculated by comparing the number of connected components in the graph before and after evolution. The rate of change in layer depth is calculated by comparing the maximum depth of the graph before and after evolution. Business change information is encoded into feature vectors. Evolutionary types and their effects are labeled as tags. A correspondence between input feature vectors and output labels is constructed to form training samples. During the dataset partitioning phase, all training samples are randomly divided into training, validation, and test sets in an 8:1:1 ratio. The training set (containing 80% of the samples) is used for model parameter learning, the validation set (containing 10% of the samples) is used for hyperparameter tuning and model selection, and the test set (containing 10% of the samples) is used for final performance evaluation. This ensures a consistent sample distribution across the three datasets. For evolutionary types with limited sample sizes, data augmentation techniques are used to expand the sample size. Small perturbations to the graph structure, such as randomly adding or deleting a few edges, generate similar evolutionary cases, ultimately forming a training dataset containing 5000 to 10000 evolutionary cases.The dataset is stored in a standard format for easy batch loading during model training. After the training dataset is built, a quality check is performed to verify the completeness of the input features and the accuracy of the labels, and outlier samples are removed to ensure dataset quality.

[0065] It should be noted that the multi-scale representation integration mechanism based on hierarchical feature fusion weightedly fuses features from different layers of the convolutional neural network, enabling the model to simultaneously obtain local detail information and global semantic information. Compared with traditional methods that only use single-level features, this mechanism significantly enhances the model's ability to perceive changes in the graph structure. Shallow features retain the fine-grained structural characteristics of the graph, such as the direct connections between nodes, while deep features extract high-level semantic features of the graph, such as community structure and modularity. The fusion of the two allows the model to capture both subtle local changes and understand global evolutionary trends. In complex industrial scenarios, graph evolution often involves changes at multiple scales. This mechanism, through multi-scale representation integration, can comprehensively analyze evolutionary patterns, providing richer feature information for evolutionary strategy recommendations and improving the accuracy and completeness of evolutionary decisions.

[0066] Furthermore, this invention addresses the technical problem of industrial data processing systems lacking adaptive evolution capabilities. Existing industrial data processing systems typically employ static data structures and fixed processing flows. When enterprise business processes are adjusted, new systems are integrated, or data models change, the system cannot automatically adapt to these changes, requiring significant manual configuration and system reconstruction. This not only increases maintenance costs but may also introduce data inconsistencies and system instability during the adjustment process. This invention establishes a graph adaptive evolution model, using evolutionary feature vectors to monitor graph structure changes. When the Euclidean distance between the evolutionary feature vector and the baseline evolutionary vector exceeds 0.85, structural adjustments are automatically initiated. Utilizing a multi-scale representation integration mechanism based on hierarchical feature fusion and a sequence dependency modeling algorithm based on cyclic attention, intelligent perception of business changes and automatic optimization of the graph structure are achieved. A version control mechanism ensures the traceability and rollbackability of the evolution process, completely solving the technical problem of traditional systems lacking adaptive capabilities.

[0067] A second aspect of the present invention provides a computer-readable storage medium storing program instructions that, when executed in a computer, perform the aforementioned digital mainline processing method based on industrial data graphs.

[0068] A third aspect of the present invention provides a digital mainline processing system based on industrial data maps, comprising the aforementioned computer-readable storage medium. The system is any one of a computer, a server, or a microcontroller. The computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

[0069] Specifically, the principle of this invention is as follows: The fundamental principle that enables this invention to solve the core technical problem of the difficulty in achieving deep cross-system correlation analysis in industrial data processing systems lies in the adoption of a graph data model to reorganize and manage cross-system industrial data. Traditional relational databases use a two-dimensional table structure to store data and express the relationships between data through foreign key constraints. While this structured organization method has advantages in data management within a single system, it faces serious limitations when handling complex cross-system correlation analysis. It requires multi-table join operations to obtain correlation information, resulting in high query complexity and difficulty in expressing multi-level indirect relationships. This invention constructs a unified industrial data relationship network by uniformly abstracting data entities from different systems as nodes in a graph and abstracting cross-system relationships as edges in a graph. This graph structure is naturally suitable for expressing and processing complex relationships, making deep cross-system correlation analysis a problem of searching and traversing paths in the graph. The application of graph traversal algorithms is the key technical means for achieving deep cross-system correlation analysis in this invention. Depth-first search can discover indirect relationship paths between data entities, and breadth-first search can analyze the scope and degree of influence of data entity relationships. These analytical capabilities are difficult to achieve with traditional relational queries. The graph indexing optimization strategy of this invention significantly improves the efficiency of graph traversal by establishing node attribute indexes and relationship indexes, making large-scale cross-system correlation analysis possible. The adaptive evolution model ensures that the graph structure can dynamically adjust as cross-system business relationships change, maintaining the system's adaptability to complex correlation analysis requirements. This graph-based data organization and analysis method solves the fundamental shortcomings of traditional methods at the data model level; therefore, the technical solution of this invention is logically sound and reasonable.

[0070] The following provides a specific embodiment 1 of the present invention, and the specific implementation of each step in this embodiment 1 is described in detail below.

[0071] The specific implementation of step S01 is to achieve unified access to heterogeneous data sources by constructing a multi-source data acquisition adapter, wherein the data update frequency optimization function is expressed as follows:

[0072] ;

[0073] In the formula, The real-time data demand coefficient is calculated by counting the number of data queries per unit time. The baseline value is updated to reflect the standard data; it is typically set to 1. This refers to the system response time, in seconds. This is the baseline response time, expressed in seconds, typically 1 second. The system type weighting coefficient is 0.8 for Manufacturing Execution Systems, 0.5 for Enterprise Resource Planning Systems, and 0.3 for Customer Relationship Management Systems. This is the latency sensitivity factor, with a default value of 0.6. This is the network fluctuation error term, with an empirical value of 0.05 to 0.15.

[0074] The specific implementation of step S02 involves cleaning and standardizing the collected multi-source heterogeneous data. The formula for calculating the data quality assessment vector is as follows:

[0075] ;

[0076] Among them, data integrity coefficient Accuracy coefficient Consistency coefficient Timeliness coefficient , effectiveness coefficient In the formula, For the complete record count, This represents the total number of records. For the correct number of data, To verify the total number of data, For the first Data values ​​from one data source, For reference data values, For the largest difference in data values, For the number of data sources, For real-time data volume, To comply with business rules regarding data volume, To check the total amount of data.

[0077] The specific implementation of step S03 is to construct an industrial data map based on the integrated dataset. The formula for calculating the relationship weight is as follows:

[0078] ;

[0079] In the formula, For the frequency of occurrence of association, For the maximum association frequency, This represents the relation type weight coefficient: 0.9 for containment relations, 0.7 for dependency relations, and 0.5 for association relations. The semantic distance between entities is calculated using the cosine distance of word vectors. This is the distance threshold, typically set to 5. This is a semantic similarity enhancement term, with an empirical value of 0.1 to 0.3.

[0080] The specific implementation of step S04 involves constructing a graph index optimization strategy and a hierarchical storage architecture. The formula for calculating the query performance superposition optimization vector is as follows:

[0081] ;

[0082] Among them, index hit rate Cache hit rate Normalized query response time Normalized concurrent processing capability resource utilization rate In the formula, The number of queries that used the index. This represents the total number of queries. To cache the number of hit queries, For the first The execution time of this query is in milliseconds. This represents the total number of queries. This is the baseline response time, measured in milliseconds, typically 10 milliseconds. To the maximum number of concurrent queries, This is the baseline concurrency level, typically set to 100. This represents the amount of resources already used. This represents the total amount of available resources.

[0083] The specific implementation of step S05 involves using a graph adaptive evolution model to monitor business changes in real time. The formulas for calculating the rate of change of each dimension are as follows:

[0084] Rate of change in the number of nodes ;

[0085] Rate of change of edge number ;

[0086] Structural density change rate ;

[0087] rate of change of connectivity ;

[0088] Hierarchical depth change rate ;

[0089] In the formula, This represents the current node number. The number of reference nodes, This is the current edge number. As the baseline number of edges, For an undirected graph, the maximum possible number of edges is... For directed graphs , As the baseline density, This represents the current number of connected components. The number of baseline connected components, At the current maximum depth, The maximum depth is the baseline. The Euclidean distance between the evolutionary eigenvector and the baseline evolutionary vector is calculated using the following formula:

[0090] ;

[0091] In the formula, For the first The rate of change of evolutionary characteristics, in order of frequency, is: , , , , , For the first The historical maximum value of the rate of change of dimension For the first The baseline evolution values, in order, are baseline node density, baseline edge density, baseline structural complexity, baseline connectivity, and baseline number of levels. For the first The maximum value of the baseline evolution. The evolutionary feature fusion function is expressed as follows:

[0092] ;

[0093] In the formula, The structural change feature vector has a magnitude of... , The vector represents the changes in business requirements, and its magnitude is... ,in For the new system access metrics, For business process change metrics, For functional expansion metrics, Let the system performance index vector have a magnitude of ,in For processor performance metrics, For memory performance metrics, Input / output performance metrics , , These are the historical maximum values ​​of the corresponding vector magnitudes. , , The weighting coefficients and experience points , , .

[0094] The specific implementation of step S06 is to realize the data tracing and exploration service functions, wherein the formula for calculating the cumulative path weight is expressed as follows:

[0095] ;

[0096] The confidence level calculation formula is expressed as follows:

[0097] ;

[0098] In the formula, For the first on the path The relation weight of each edge. This represents the number of edges contained in the path. The path length is equal to the number of edges. , This is the standard path length, typically set to 5. This is a semantic consistency enhancement term, with an empirical value of 0.05 to 0.2. The formula for calculating the hysteresis effect error is as follows:

[0099] ;

[0100] In the formula, This is the actual response time, in milliseconds. This is the theoretically optimal response time, in milliseconds. To calculate the complexity coefficient, it is obtained by statistically analyzing the number of nodes and edges involved in the query. ,in To query the number of nodes involved, To query the number of edges involved, and These are complexity weighting coefficients, with empirical values ​​of 0.3 and 0.7 respectively. This is the standard complexity baseline value, typically taken as 100. The path complexity factor is determined based on path depth and number of branches, with an empirical value ranging from 0.8 to 1.2. The query optimization weight function is expressed as follows:

[0101] ;

[0102] In the formula, The coefficients for query types are 0.3 for simple node queries, 0.6 for relational traversal queries, and 0.9 for multi-hop path queries. The historical performance coefficient is calculated as follows: ,in The historical average response time This is the current average response time. The resource status coefficient is calculated as follows: , This is the current response time, in milliseconds. This is the response time threshold, in milliseconds, with a default value of 100 milliseconds.

[0103] The specific implementation method of step S07 is the same as described above, and will not be repeated in detail here.

[0104] The formula for calculating the evolutionary blocking risk matrix is ​​as follows:

[0105] ;

[0106] In the formula, The structural complexity is derived by calculating the number of cycles and the variance of node degree in the graph. The data consistency coefficient. This is the performance impact coefficient. For compatibility coefficients, The difficulty level of the rollback is... , , , , These are the historical maximum values ​​for the corresponding indicators.

[0107] The formula for calculating the query wait time distribution vector is as follows:

[0108] ;

[0109] In the formula, The average wait time for a simple node query. The average wait time for relation traversal queries. The average wait time for multi-hop path queries. To aggregate the average wait time for statistical queries, This represents the average wait time for join queries, in milliseconds. This is the baseline waiting time, typically set to 10 milliseconds.

[0110] The formula for calculating the performance gain evaluation matrix is ​​as follows:

[0111] ;

[0112] Among them, performance gain coefficient , For the first Average response time for each query type before optimization For the first The first query type Average response time under various optimization strategies To query the number of types, To optimize the number of strategies.

[0113] The formula for calculating the system load fluctuation monitoring vector is as follows:

[0114] ;

[0115] In the formula, , , , , These are the standard deviations of processor utilization, memory utilization, disk I / O rate, network bandwidth utilization, and concurrent connections, respectively. , , , , These are the mean values ​​of the corresponding indicators.

[0116] The principle and effect of the above formula are explained as follows: Data update frequency optimization function By comprehensively considering real-time data requirements, system response characteristics, and the impact of network fluctuations, this function dynamically adjusts the update frequency for different data sources. Compared to a fixed-frequency update mode, this function can adaptively adjust based on actual business needs and system performance, effectively reducing system resource consumption while ensuring data timeliness, and improving data collection efficiency by an average of 25% to 35%. The data quality assessment vector quantifies data quality across multiple dimensions, including the consistency coefficient. This paper uses normalized Manhattan distance to measure the degree of difference between different data sources, transforming the abstract concept of data quality into a calculable numerical indicator. This provides a quality benchmark for subsequent data processing. Compared with traditional subjective evaluation methods, this vector calculation method is objective and reproducible, and can monitor data quality trends in real time, providing precise quality guidance for data cleaning and preprocessing. The relation weight calculation formula combines multiple factors such as association frequency, relation type, and semantic distance, with an exponential decay term... This method ensures that entities with greater distances have lower relationship weights, accurately quantifying the strength of relationships between entities. Compared to simple frequency statistics, this formula introduces semantic distance decay and semantic similarity enhancement terms, more accurately reflecting the true degree of relationship between entities. This provides accurate path weight information for graph traversal algorithms, significantly improving the accuracy and efficiency of data tracing. The query performance optimization vector comprehensively reflects the effectiveness of query optimization strategies through the integrated evaluation of multiple performance indicators. The interrelationships of each dimension of this vector form a closed-loop feedback mechanism for performance optimization, identifying system performance bottlenecks and guiding adjustments to optimization strategies. Compared to single performance indicator evaluation, this vector method avoids overall performance degradation caused by local optimization, achieving a comprehensive improvement in query performance. The evolutionary feature change rate calculation formula group uses standardized relative change... Five dimensions are used to systematically quantify the degree of change in the map structure. These formulas use relative rates of change to eliminate the influence of differences in baseline values, ensuring the comparability of changes across different dimensions and providing accurate quantitative basis for evolutionary decision-making. Euclidean distance calculation formula. The geometric distance in five-dimensional space quantifies the degree of change in the map structure, providing an accurate criterion for triggering adaptive evolution. This formula uses normalization to eliminate the influence of differences in dimensions. Compared to simple threshold comparison methods, geometric distance can comprehensively consider the combined effects of multi-dimensional changes, improving the accuracy and timeliness of evolutionary decisions. The evolutionary feature fusion function weightedly fuses information from three dimensions: structural change, business requirements, and system performance, forming a comprehensive evolutionary feature representation. This function uses vector magnitude calculation and normalization to balance the influence of different types of features, providing comprehensive input information for the adaptive evolutionary model. Compared to single feature input, feature fusion improves the accuracy and adaptability of evolutionary strategy recommendations. Path weight accumulation formula. By using multiplication, it is ensured that a lower weight on any side of the path significantly affects the overall path weight, effectively filtering high-quality tracing paths. (Confidence calculation formula) Taking into account the combined effects of path weight and path length, the exponential decay term suppresses the confidence of excessively long paths, ensuring the reliability of the tracing results. Compared to the simple weighted averaging method, this formula can more accurately assess the reliability of the tracing path. The hysteresis error calculation formula quantifies the backhaul error during the query process, providing triggering conditions and compensation basis for the hysteresis compensation algorithm. This formula combines the effects of actual response time deviation and computational complexity. This method can accurately identify system performance anomalies. Compared to fixed threshold methods, this formula considers the dynamic impact of path complexity, providing accurate performance evaluations across different query scenarios and effectively improving query response stability. The query optimization weight function adopts an exponential decay form. By incorporating the influence of multiple factors, dynamic allocation of weights for different optimization strategies is achieved. The exponential term in this function effectively suppresses the weight of queries with high response times, prioritizing the performance of low-latency queries. Compared to linear weight allocation methods, the exponential decay mechanism better balances query performance and resource utilization, improving the overall query service quality. Other matrix and vector formulas, through structured mathematical representations, transform complex system states and performance indicators into calculable and comparable numerical forms, providing a quantitative basis for system monitoring, performance evaluation, and decision support. Compared to qualitative analysis methods, these mathematical models are more accurate and operable, supporting automated system management and optimization adjustments.

[0117] To better understand and implement this invention, the following is a specific application scenario of this invention, Example 2:

[0118] A certain enterprise possesses a complete production and manufacturing system, including an Enterprise Resource Planning (ERP) system, a Manufacturing Execution System (MES) system, and a Customer Relationship Management (CRM) system. The technical team employed the technical solution of this invention to construct a complete industrial data mapping digital masterline processing system.

[0119] First, the technical team built a multi-source data acquisition adapter to collect real-time data from the company's three existing core systems. The ERP system contains 2847 database tables, the MES system contains 1925 database tables, and the CRM system contains 856 database tables. The team established a data source identifier mapping table to record the access paths and update frequencies of each system. The ERP system's data update frequency is set to once per hour, the MES system's real-time production data update frequency is once every 5 minutes, and the CRM system's customer data update frequency is once every 30 minutes. The multi-source data acquisition adapter supports multiple data access protocols such as SQL, ODBC, and API, ensuring unified access to heterogeneous data sources.

[0120] During the data preprocessing stage, the technical team cleaned and standardized the collected multi-source heterogeneous data. The total amount of raw data reached 67TB, and after data cleaning, the effective data volume was 52TB. The technical team constructed a data quality assessment vector, which includes quality indicators in five dimensions, as shown in Table 1.

[0121] Table 1 Data Quality Assessment Vector Indicators

[0122] The data quality assessment results show that all indicators have reached the preset threshold standards, laying a good foundation for the subsequent construction of industrial data maps.

[0123] Based on an integrated dataset, the technical team constructed an industrial data graph. The graph contains 156,892 data nodes and 284,567 connecting edges. Node types include product entity nodes, process flow nodes, equipment status nodes, quality inspection nodes, and order information nodes. Edge types include process associations, equipment dependencies, quality correlations, and temporal relationships. Each node contains entity identifier, entity type, business attributes, and timestamp information, while each edge contains relationship type, relationship strength, relationship direction, and relationship weight information. The graph database uses the Neo4j attribute graph model to store data, supporting complex graph query and traversal operations.

[0124] To optimize query performance, the technical team implemented a graph index optimization strategy and a tiered storage architecture. They established 89 node attribute indexes, 156 relationship indexes, and 43 composite indexes. The tiered storage architecture stores frequently accessed data (higher than 80% access frequency) in a 256GB memory layer, warm data (accessed between 20% and 80% access frequency) in an 8TB SSD layer, and cold data (accessed less than 20% access frequency) in a 32TB hard disk layer. The five metrics for optimizing query performance are shown in Table 2.

[0125] Table 2 Query Performance Optimization Metrics

[0126] The technical team implemented an adaptive graph evolution model. This model employs a multi-scale representation integration mechanism based on hierarchical feature fusion and a sequence dependency modeling algorithm based on recurrent attention. The evolutionary feature vector includes five dimensions: the rate of change of node number, the rate of change of edge number, the rate of change of structural density, the rate of change of connectivity, and the rate of change of layer depth. The baseline evolutionary vector is defined with standard node density of 0.745, standard edge density of 1.287, standard structural complexity of 2.156, standard connectivity of 0.923, and standard number of layers of 4.2. When the Euclidean distance between the evolutionary feature vector and the baseline evolutionary vector exceeds 0.85, the system automatically initiates a pattern matching algorithm to adjust the graph structure.

[0127] During system operation, the technical team detected a significant business change event. The company added an automated production line, causing a significant change in the evolutionary feature vector. For example... Figure 2 As shown, the rate of change of the number of nodes increased from 0.123 to 0.456, the rate of change of the number of edges increased from 0.089 to 0.378, and the rate of change of the structure density increased from 0.067 to 0.234. The calculated Euclidean distance between the evolved eigenvector and the baseline evolved vector was 0.912, exceeding the set threshold of 0.85. The system automatically initiated graph structure adjustment, adding 2847 nodes and 4692 edges, completing the adaptive evolution of the graph structure.

[0128] The data traceability and exploration service is implemented based on a graph traversal algorithm. The technical team designed an intelligent data traceability function that supports tracing from any data node along related paths. In a quality issue traceability case, starting from the non-conforming product node, the system used a depth-first search algorithm to traverse 156 relevant nodes and 289 related edges, generating a complete data traceability link. During the traceability process, a query lag effect error of 0.187 was found, exceeding the set threshold of 0.15. The system automatically activated a lag compensation algorithm, eliminating the impact of backhaul errors through query history and result caching mechanisms, ultimately reducing the lag effect error to 0.043.

[0129] like Figure 3 As shown, the system's query wait time distribution vector records the wait time characteristics of different types of query operations. The average wait time for simple node queries is 23.7 seconds. The relation traversal query result was 89.4. The multi-hop path query result is 167.8. The aggregated statistics query result was 245.3. The combined query result is 312.6. This data provides important reference for query performance optimization.

[0130] The technical team established a role-based access control and data security mechanism. The system defines five user roles: system administrator, data analyst, production engineer, quality inspector, and general user. Different roles have different data access permissions: system administrators have read and write permissions for all data; data analysts have read permissions for statistical analysis data; production engineers have read and write permissions for production-related data; quality inspectors have read and write permissions for quality data; and general users can only view basic statistical information. Data transmission uses SSL encryption with a 256-bit encryption strength. The data backup strategy is set to daily incremental backups and weekly full backups, with backup data stored in a separate storage system.

[0131] The system load fluctuation monitoring vector records system resource usage in real time. Processor utilization fluctuates between 45% and 78%, memory utilization between 52% and 85%, disk I / O between 156 and 892 MB / s, and network bandwidth between 23 and 156 MB / s. The number of concurrent connections fluctuates between 89 and 267. When the system load fluctuation exceeds the preset threshold, the system automatically triggers resource adjustment and load balancing mechanisms to ensure stable system operation.

[0132] The evolutionary congestion risk matrix assessed the potential risks during the evolution of the graph structure. The risk coefficients were: structural complexity (0.234), data consistency (0.156), performance impact (0.089), compatibility (0.067), and rollback difficulty (0.123). The overall risk assessment result was 0.134, falling within the low-risk range and ensuring the safety of the evolutionary process.

[0133] The performance gain evaluation matrix quantifies the performance improvement effects of different optimization strategies. The node attribute indexing strategy has a performance gain coefficient of 2.34 for simple node queries and 1.67 for relational traversal queries. The relational indexing strategy has a performance gain coefficient of 3.12 for multi-hop path queries and 2.78 for aggregate statistical queries. The composite indexing strategy has a performance gain coefficient of 4.25 for join queries, significantly improving the execution efficiency of complex queries.

[0134] In practical applications, evolutionary feature fusion functions integrate multidimensional evolutionary feature information. The maximum value of the structural change feature vector. The maximum value of the business requirement change vector is 2.456. The maximum value of the system performance index vector is 1.789. The value is 3.123. The weighting coefficient is set to... =0.4, =0.35, =0.25, which satisfies the condition. Constraints. Comprehensive evolutionary eigenvectors. The calculated result is 0.867, providing a quantitative basis for map evolution decision-making.

[0135] The query optimization weight function calculates the weight allocation for different optimization strategies. Query type coefficient. Based on query complexity, it is divided into 5 levels, with historical performance coefficients. Based on performance statistics from the past 30 days, resource status coefficient Reflects current system resource usage. Response time threshold. Set to 200 When the query response time exceeds the threshold, the weight of the optimization strategy decays exponentially.

[0136] This invention represents a significant technological advancement over traditional data processing methods. Traditional methods typically use relational databases to store multi-source data, making it difficult to efficiently represent and query the complex relationships between data. This invention, however, constructs an industrial data graph, storing data entities and relationships in a graph structure. This allows for a clear and intuitive representation of complex relationship networks, greatly improving the efficiency of relational queries and data traceability. Traditional methods require manual adjustments to data models and query logic when facing business changes. This invention's adaptive evolutionary model automatically detects business changes and adjusts the graph structure accordingly, significantly enhancing the system's adaptability. Traditional query optimization relies primarily on database indexes and SQL optimization. This invention, through graph index optimization strategies, a hierarchical storage architecture, and combined query performance optimization, forms a multi-dimensional performance optimization system, better addressing the complex query needs of industrial data.

[0137] It should be noted that the variables involved in this invention are explained in detail in Tables 3, 4, and 5 below.

[0138] Table 3. Variable Explanation Table (Part 1)

[0139] Table 4. Variable Explanation Table (Part Two)

[0140] Table 5. Variable Explanation Table (Part 3)

[0141] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A digital thread processing method based on industrial data maps, characterized in that, An industrial data relationship network is constructed by abstracting data entities as nodes and relationships as edges. A graph traversal algorithm is used to achieve deep cross-system correlation analysis, including the following steps: Constructing a multi-source data acquisition adapter to collect data in real time from database tables in Enterprise Resource Planning (ERP), Manufacturing Execution System (MES), and Customer Relationship Management (CRM) systems; establishing a data source identifier mapping table to achieve unified access to heterogeneous data sources; cleaning and standardizing the collected multi-source heterogeneous data to construct an integrated dataset; establishing a data quality assessment vector to complete data preprocessing and quality control; and constructing an industrial data graph based on the integrated dataset, abstracting data entities as nodes and relationships as edges, and using a graph database to store nodes and edges. Edge attribute information; construct graph index optimization strategy and hierarchical storage architecture, establish multi-dimensional index, adopt hot data caching mechanism and query result pre-computation system; use graph adaptive evolution model to monitor business changes in real time, when the Euclidean distance between the evolution feature vector and the baseline evolution vector in multi-dimensional space is >0.85, start the pattern matching algorithm to automatically adjust the graph structure; realize data traceability and exploration service functions, realize intelligent data traceability function based on graph traversal algorithm, when the query hysteresis error is >0.15, start the hysteresis compensation algorithm, if the hysteresis error ∈ [0, 0.05], adopt the standard query processing mode; establish role-based access control and data security guarantee mechanism.

2. The method according to claim 1, characterized in that, The multi-source data acquisition adapter adopts a protocol-adaptive connection method, establishes corresponding data access drivers for different database types, and supports multiple data access protocols such as Structured Query Language, Open Database Connectivity, and Application Programming Interface.

3. The method according to claim 2, characterized in that, The data source identifier mapping table records the unique identifier, database connection parameters, table structure information, and field mapping relationships of each data source. It also records the access path and update frequency of each system's data source, and is used to establish a unified data access interface. The unique identifier is derived from the system codes of the Enterprise Resource Planning System, Manufacturing Execution System, and Customer Relationship Management System.

4. The method according to claim 3, characterized in that, The data quality assessment vector comprises five dimensions: data integrity coefficient, accuracy coefficient, consistency coefficient, timeliness coefficient, and validity coefficient. It is constructed by performing quality checks and scoring on the original data to monitor data integrity.

5. The method according to claim 4, characterized in that, In an industrial data graph, nodes represent data entities, and edges represent the relationships between entities. Each node and edge contains rich attribute information. Node attributes include entity identifier, entity type, business attributes, and timestamp, while edge attributes include relationship type, relationship strength, relationship direction, and relationship weight.

6. The method according to claim 5, characterized in that, Graph index optimization strategies include three types: node attribute index, relation index, and composite index. Node attribute index is built based on the key attributes of nodes, relation index is built based on the type and weight of edges, and composite index combines multiple attributes to create a joint index.

7. The method according to claim 6, characterized in that, The tiered storage architecture stores hot data in the memory tier, warm data in the solid-state drive tier, and cold data in the hard disk drive tier, dynamically adjusting the storage tiers based on data access frequency.

8. The method according to claim 7, characterized in that, The query performance optimization vector includes five metrics: index hit rate, cache hit rate, query response time, concurrency processing capability, and resource utilization. These metrics are used to describe the effectiveness of query optimization strategies, evaluate the optimization results, and complete the configuration for query performance optimization.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores program instructions, which, when executed in a computer, are used to perform the digital mainline processing method based on industrial data maps as described in any one of claims 1-8.

10. A digital mainline processing system based on industrial data maps, characterized in that, The system includes the computer-readable storage medium of claim 9, wherein the system is any one of a computer, a server, or a microcontroller, the computer-readable storage medium is disposed within the system, and the system is provided with a microprocessor that executes the program instructions stored in the computer-readable storage medium.

Images