Information model construction method and device, equipment and storage medium

Abstract

The application discloses an information model construction method and device, equipment and a storage medium, and relates to the field of industrial automation. The method comprises the following steps: performing field analysis and normalization processing on a received multi-source heterogeneous industrial point table to generate a standardized point table object with a unified structure; performing semantic analysis on the standardized point table object, extracting multi-dimensional features of each data point, and performing joint semantic analysis based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object; converting the industrial semantic object into a semantic node, constructing an edge based on the association relationship between the industrial semantic objects, and generating a protocol-independent industrial semantic graph; and mapping each semantic node to an OPC UA node in a traversal manner to construct an OPC UA information model. By introducing a protocol-independent semantic intermediate layer, automatic conversion from an industrial point table to an information model is realized, the modeling result can be kept consistent in semantics and standardized in structure among devices with different protocols, and the generality and generalization capability of the information model are improved.

Description

Technical Field

[0001] This application relates to the field of industrial automation, and in particular to an information model construction method, apparatus, equipment and storage medium. Background Technology

[0002] In industrial automation systems, industrial equipment such as PLCs, DCSs, RTUs, smart meters, and industrial controllers typically describe their internal data objects using point lists. With the increasing demand for data interconnection among numerous industrial devices, OPC UA (OPC Unified Architecture) has become a universal data integration standard in the industrial automation field. As a cross-platform, object-oriented industrial communication standard, OPC UA encapsulates the functions, data, and internal logical relationships of devices into a unified information model. OPC UA not only defines a unified data communication protocol but also establishes a standard information model, describing industrial equipment and its data structure through objects, variables, properties, and references, enabling different systems to exchange data based on this unified information model. Therefore, during the industrial equipment access process, the point lists must first be converted into an information model conforming to the OPC UA specification before the OPC UA server provides standardized data access capabilities.

[0003] In engineering practice, the construction of industrial data point tables into OPC UA information models has long relied on manual modeling. Implementers need to parse the data point table content line by line, manually create object nodes and variable nodes in the OPC UA server development environment, configure the hierarchy and reference relationships between nodes, and finally export a deployable model file. Manual modeling relies on human experience and is inefficient and error-prone, making it a major efficiency bottleneck in the process of industrial data interconnection.

[0004] To improve modeling efficiency, automated modeling methods for OPC UA information models have emerged. Existing automated OPC UA information model construction methods typically use predefined templates, configuration files, or scripts to directly map fields in industrial point tables to OPC UA nodes and automatically generate corresponding NodeSet XML model files. While this reduces manual configuration and improves modeling efficiency to some extent, it is essentially still a field-level mapping method, directly generating OPC UA nodes based on point table fields. On the one hand, it lacks semantic expression capabilities for industrial data. The same industrial data may be described with different field names or organization methods under different devices and protocols, causing the system to only perform mechanical mapping based on field names, making it difficult to understand the business meaning expressed by the industrial data itself, resulting in poor universality of the constructed information model. On the other hand, the modeling process is highly coupled with industrial communication protocols. Different industrial protocols use different data organization structures, and existing modeling solutions usually develop corresponding conversion rules for specific protocols, causing the modeling process to depend on specific protocol implementations. When a new protocol or device is added, adaptation logic needs to be redeveloped, resulting in poor universality and generalization capabilities.

[0005] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0006] The main objective of this application is to provide an information model construction method, apparatus, device, and storage medium, which aims to solve the technical problems that existing automatic information model construction schemes lack the semantic expression capability of industrial data and are highly coupled with industrial communication protocols, resulting in poor universality and generalization ability of the constructed information models.

[0007] To achieve the above objectives, this application proposes an information model construction method, which includes: Receive multi-source heterogeneous industrial point tables, perform field parsing and normalization on the industrial point tables, and generate a standardized point table object with a unified structure; Semantic parsing is performed on the standardized point table object to extract multi-dimensional features of each data point in the standardized point table object, and joint semantic parsing is performed on the multi-dimensional features based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The industrial semantic objects are converted into semantic nodes, and edges are constructed based on the relationships between the industrial semantic objects to generate an industrial semantic graph that is independent of industrial protocols. Traverse the industrial semantic graph and map each semantic node to an OPC UA node to construct an OPC UA information model.

[0008] In one embodiment, the step of parsing and normalizing the industrial point table to generate a standardized point table object with a unified structure includes: The table fields of the industrial point table are parsed to extract basic industrial information; the basic industrial information includes at least variable name, variable description, data type, address information, access permissions, collection cycle, and equipment identifier; The data in different formats in the basic industrial information are uniformly parsed to obtain the original field set; The fields in the original field set are mapped to a pre-established standard field model to obtain a standardized field object set. Furthermore, based on a pre-built address abstraction model, the industrial addresses in the original field set are converted into a standard address object set. Based on a pre-built object model, the standardized field object set and the standardized address object set are integrated into a standardized industrial point table object; the standardized industrial point table object includes device identifier, variable name, data type, standard address object, access permission, and extended attributes.

[0009] In one embodiment, the step of performing semantic parsing on the standardized point table object, extracting multi-dimensional features of each data point in the standardized point table object, and performing joint semantic parsing on the multi-dimensional features based on a multi-constraint consistency determination mechanism to obtain an industrial semantic object includes: Semantic parsing is performed on the standardized point table object to extract multidimensional features of each data point in the standardized point table object, resulting in a feature vector; the multidimensional features in the feature vector include at least name features, data type features, address features, behavior features, and context features; Based on the multi-constraint consistency determination mechanism, the feature vector is jointly semantically determined with multiple constraints. The industrial semantic category corresponding to the feature vector is determined according to the determination result, resulting in an industrial semantic object that includes equipment semantic objects, variable semantic objects, attribute semantic objects and address semantic objects.

[0010] In one embodiment, the step of determining the industrial semantic category corresponding to the feature vector based on the determination result includes: If the determination result indicates that the feature vector satisfies all the constraints of the multiple constraints, then the industrial semantic category corresponding to the feature vector is determined according to the determination result of the multiple constraints. If the determination result indicates that the feature vector satisfies some of the multiple constraints, then a candidate semantic set is determined based on the target conditions satisfied by the feature vector among the multiple constraints. According to the preset scoring rules, each candidate semantic in the candidate semantic set is scored and sorted, and the industrial semantic category corresponding to the feature vector is determined according to the sorting order.

[0011] In one embodiment, the step of converting the industrial semantic objects into semantic nodes, constructing edges based on the relationships between the industrial semantic objects, and generating an industrial semantic graph independent of industrial protocols includes: The industrial semantic object is converted into a semantic node; the node attribute information of the semantic node includes node identifier, node type, semantic tag, source device, and semantic feature vector; the node type includes device node, variable node, function node, and attribute node; Based on the associations between the industrial semantic objects, the semantic relationships between the corresponding semantic nodes are determined, and edges between the semantic nodes are constructed based on the semantic relationships to generate an initial semantic graph; the associations include device hierarchy relationships, composition relationships, reference relationships, address mapping relationships, and business relationships; the semantic relationships include structural relationships, functional relationships, address relationships, acquisition behavior relationships, and context behavior relationships; The initial semantic graph is subjected to protocol decoupling and structural optimization to obtain an initial industrial semantic graph. Protocol decoupling includes removing protocol-specific hierarchical fields, eliminating protocol tree structures, merging equivalent semantic nodes, and unifying semantic connection paths. The structural optimization strategy includes node redundancy removal, edge weight convergence, and a stable graph structure. Node redundancy removal involves merging semantic nodes with the same semantic meaning but different origins. Edge weight convergence represents the normalization of multiple associations of the same semantic node. The stable graph structure represents a global consistency adjustment of the semantic relationships. A unified cross-protocol mapping is performed on the initial industrial semantic graph, and the industrial protocol is marked as the data source attribute of the semantic node to obtain an industrial semantic graph that is independent of the industrial protocol.

[0012] In one embodiment, the step of traversing the industrial semantic graph, mapping each semantic node to an OPC UA node, and constructing an OPC UA information model includes: Traverse the industrial semantic graph and map each semantic node to an OPC UA node; Based on the semantic relationships between the semantic nodes, the reference relationships between the OPC UA nodes are generated; The node attribute information of the semantic node is hashed to obtain the node identifier of the OPC UA node; Based on the OPC UA node, the reference relationship, and the node identifier, a node hierarchy is constructed and the reference relationship is bound to obtain the OPC UA information model.

[0013] In one embodiment, after the step of traversing the industrial semantic graph, mapping each semantic node to an OPC UA node, and constructing the OPC UA information model, the method further includes: The OPC UA information model is subjected to consistency verification. If the verification passes, the OPC UA address space is constructed based on the OPC UA information model. The OPC UA address space is serialized into a standard model file for model deployment on the OPC UA server.

[0014] Furthermore, to achieve the above objectives, this application also proposes an information model building apparatus, which includes: The point table standardization module is used to receive multi-source heterogeneous industrial point tables, perform field parsing and normalization processing on the industrial point tables, and generate standardized point table objects with a unified structure. The semantic parsing module is used to perform semantic parsing on the standardized point table object, extract the multi-dimensional features of each data point in the standardized point table object, and perform joint semantic parsing on the multi-dimensional features based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The semantic modeling module is used to convert the industrial semantic objects into semantic nodes, construct edges based on the association relationships between the industrial semantic objects, and generate an industrial semantic graph that is independent of industrial protocols. The model building module is used to traverse the industrial semantic graph, map each semantic node to an OPC UA node, and build an OPC UA information model.

[0015] In addition, to achieve the above objectives, this application also proposes an information model building device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the information model building method as described above.

[0016] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the information model construction method described above.

[0017] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the information model construction method described above.

[0018] One or more technical solutions proposed in this application have at least the following technical effects: By parsing and normalizing fields from heterogeneous industrial point tables, a standardized point table object with a unified structure is generated. Multi-dimensional features are then extracted from the data points in this object, and joint semantic parsing is performed on these features using a multi-constraint consistency judgment mechanism to obtain industrial semantic objects. These industrial semantic objects are then converted into semantic nodes, and edges are constructed between nodes based on the business relationships between them, generating an industrial semantic graph independent of industrial protocols. This graph is then traversed to map each semantic node to an OPC UA node, thus constructing an OPC UA information model. This unified structural transformation of the input point table decouples the subsequent parsing and modeling processes from the communication protocol, ensuring cross-protocol adaptability of the modeling process. Furthermore, by leveraging joint semantic parsing of multi-dimensional features, the business attributes of data points can be comprehensively defined from multiple dimensions. This gives the generated semantic objects a structured descriptive ability to represent the inherent meaning of industrial data. The semantic graph's topology is constructed based on relationships rather than the protocol's native data hierarchy, removing the private organizational structure of the communication protocol from this intermediate representation layer and retaining only cross-device logical connections. Based on protocol-independent industrial semantic graph execution node mapping, the constructed OPC UA information model can reflect the functional division and data association of devices. It can maintain semantic consistency and structural standardization of modeling results among devices with different protocols, thereby improving the universality and generalization ability of the information model. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart illustrating the information model construction method provided in this application embodiment; Figure 2 This is a schematic diagram of the information model construction process provided in the embodiments of this application; Figure 3 This is a schematic diagram of the point table standardization process provided in the embodiments of this application; Figure 4 This is a schematic diagram of the semantic parsing process provided in an embodiment of this application; Figure 5 A schematic diagram of a protocol-independent industrial semantic graph modeling process provided for embodiments of this application; Figure 6A schematic diagram of the modeling process for the information model provided in the embodiments of this application; Figure 7 This is a schematic diagram of the module structure of the information model construction device according to an embodiment of this application; Figure 8 This is a schematic diagram of the device structure of the hardware operating environment involved in the information model construction method in this application embodiment.

[0022] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0024] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0025] The key technical terms involved in the various embodiments of this application mainly include: OPC UA: Open Platform Communications Unified Architecture. It defines a unified data communication protocol and describes industrial equipment and its data structure through objects, variables, properties, and references. It establishes a standard information model, enabling different systems to exchange data based on the unified information model. ISG: Industrial Semantic Graph; NodeSet XML: Node set XML file, the installation and deployment configuration file for the OPC UA model; Point List: An industrial point list that describes data objects within equipment.

[0026] Existing automatic modeling schemes for OPC UA information models have limitations. On the one hand, they lack the ability to semantically express industrial data. On the other hand, the modeling process is highly coupled with industrial communication protocols, usually relying on the data format of specific industrial protocols. It is difficult to achieve unified modeling between different protocols. The data organization methods of different protocols are quite different, making it difficult for automatic modeling schemes to adapt to the multi-protocol equipment environment in complex industrial sites.

[0027] A comprehensive analysis of existing automated modeling solutions reveals the following main shortcomings: The lack of a unified semantic expression capability for industrial data stems from the direct mapping based on fields without establishing a unified semantic expression layer. The same industrial data may be described using different field names or organization methods under different devices and protocols, leading the system to perform only mechanical mapping based on field names. This makes it difficult to understand the business meaning expressed by the industrial data itself, resulting in poor universality of the constructed information model.

[0028] The modeling process is highly coupled with industrial communication protocols: different industrial protocols use different data organization structures, and existing modeling solutions typically develop corresponding conversion rules for specific protocols, causing the modeling process to depend on specific protocol implementations. When a new protocol or device is added, adaptation logic needs to be redeveloped, resulting in poor versatility, scalability, and generalization capabilities.

[0029] OPC UA information models still rely on manual configuration: Although some tools for automatic modeling exist, most solutions still require manual maintenance of configuration files such as field mapping relationships, node levels, and reference relationships, making it difficult to truly achieve automatic construction of industrial information models, resulting in high engineering implementation costs and low levels of automatic modeling.

[0030] It is evident that existing industrial point table modeling schemes suffer from insufficient semantic expression capabilities, weak cross-protocol adaptability, and low degree of automatic modeling, making it difficult to meet the needs of rapid access, unified modeling, and continuous evolution of a large number of devices in the context of industrial internet and intelligent manufacturing.

[0031] To address, or at least partially address, the aforementioned problems, this application provides a solution. This solution proposes an information model construction method that employs semantic-driven OPC UA information model construction based on industrial point tables. Instead of directly mapping industrial point tables to OPC UA nodes, it introduces a protocol-independent industrial semantic intermediate layer between the industrial point tables and the OPC UA information model. By establishing a unified industrial semantic expression space, it achieves the transformation of industrial data from field-level descriptions to semantic-level descriptions, and then automatically generates an information model conforming to the OPC UA specification based on the unified semantic model.

[0032] Specifically, the system first receives industrial point table data from different industrial equipment or protocols, and eliminates data structure differences between protocols through standardization. Then, it constructs industrial data semantic objects based on point table field features and uses a multi-constraint consistency resolution mechanism to complete semantic recognition of the industrial data. Next, it constructs a protocol-independent industrial semantic graph model to uniformly abstract and express equipment objects, data objects, and their logical relationships. Finally, it automatically generates an OPC UA information model based on the protocol-independent industrial semantic graph (ISG) model and outputs a NodeSet XML standard model file conforming to the OPC UA standard specification. Through the overall approach of data standardization, semantic understanding, unified modeling, and standard output, the construction process of the industrial information model is freed from specific industrial protocols and manual configuration, achieving automatic construction and continuous evolution of the industrial information model.

[0033] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or information model building device capable of performing the above functions. The following description uses an information model building device as an example to illustrate this embodiment and the subsequent embodiments.

[0034] Specifically, embodiments of this application provide an information model construction method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the information model construction method provided in an embodiment of this application.

[0035] like Figure 1 As shown, in this embodiment, the information model construction method includes steps S10 to S40: Step S10: Receive multi-source heterogeneous industrial point tables, perform field parsing and normalization processing on the industrial point tables, and generate a standardized point table object with a unified structure. Step S20: Semantic parsing is performed on the standardized point table object to extract the multi-dimensional features of each data point in the standardized point table object, and joint semantic parsing is performed on the multi-dimensional features based on the multi-constraint consistency judgment mechanism to obtain the industrial semantic object. Step S30: Convert the industrial semantic objects into semantic nodes, construct edges based on the business relationships between the industrial semantic objects, and generate an industrial semantic graph that is independent of industrial protocols. Step S40: Traverse the industrial semantic graph, map each semantic node to an OPC UA node, and construct the OPC UA information model.

[0036] First, receive multi-source heterogeneous industrial point tables, which are the original industrial point tables output by industrial equipment. Multi-source heterogeneous industrial point tables are industrial point tables output by different equipment and / or different industrial protocols.

[0037] Because different industrial equipment uses different industrial protocols and different point table organization methods, there are significant differences in field names, address representation methods and data organization formats. Therefore, it is necessary to establish a unified data representation method.

[0038] By standardizing the heterogeneous industrial point tables from multiple sources, a unified data representation method is proposed. In this embodiment, the standardization process for the industrial point tables includes field parsing and normalization. Basic industrial information such as variable names, register addresses, data types, access permissions, and units are extracted from the industrial point tables. Based on the extracted basic industrial information, fields in different protocols are uniformly mapped according to predefined data structure specifications. For example, Modbus register addresses, S7 DB addresses, and OPC UA node addresses are uniformly converted into system internal standard address objects.

[0039] After unifying the fields, a standardized point table object is generated. Each data point is described using a unified data structure, including fields such as device identifier, variable name, address information, data type, and access attributes. After normalization, data generated by different protocols has a unified data structure, providing consistent and standardized data input for subsequent semantic parsing.

[0040] Standardized point table objects are in-memory data instances with a unified data structure and access interface, such as generic data class objects containing fixed header fields. The purpose of normalization is to eliminate differences in data formats, naming conventions, and units of measurement between different industrial protocols or equipment manufacturers, enabling subsequent data processing to be based on a unified standard.

[0041] After generating a standardized point table object with a unified structure, semantic parsing is performed on the standardized point table object to extract multi-dimensional features of each data point in the standardized point table object. Based on a multi-constraint consistency mechanism, joint semantic parsing is performed on the extracted multi-dimensional features to obtain an industrial semantic object.

[0042] Among them, multidimensional features are a set of attributes that characterize the purpose of data points, such as semantic fragments in variable names, physical area types of data storage, etc. The role of the multi-constraint consistency judgment mechanism is to avoid ambiguity caused by single feature recognition by introducing multiple dimensions of logical judgment conditions, thereby improving the accuracy and robustness of semantic parsing.

[0043] Unlike traditional field mapping schemes that rely solely on fields to directly generate nodes, this approach constructs a unified industrial semantic space to achieve semantic data transformation. For example, a temperature acquisition point is identified not only because its name contains "Temperature," but also by combining its floating-point type, analog address, and periodic acquisition behavior to determine that it belongs to an analog measurement object. Ultimately, each point table field is converted into a unified semantic object, providing a unified semantic foundation for subsequent protocol-independent semantic modeling.

[0044] Protocol-independent semantic modeling is performed on industrial semantic objects obtained from semantic parsing. Specifically, industrial semantic objects are converted into semantic nodes, semantic edges are constructed based on the business relationships between industrial semantic objects, and an industrial semantic graph independent of industrial protocols is generated.

[0045] Industrial semantic graphs are topological networks that use graph theory to represent industrial entities and their relationships. For example, they are directed attribute graphs containing equipment nodes, variable nodes, and the connections between them. Industrial semantic objects are instantiated objects with clear industrial physical meaning and equipment-level attributes, such as equipment objects representing pumps or variable objects representing temperature readings. Business relationships are the actual control or measurement links between equipment in the industrial field.

[0046] By unifying the parsing and deep semantic mining of multi-source heterogeneous industrial point tables and transforming them into industrial semantic graphs independent of underlying protocols, and finally mapping them to a standard OPC UA information model, the entire process of automated conversion from heterogeneous industrial data to a standardized interconnection model is realized. This breaks down data silos in the industrial field and significantly reduces the difficulty of cross-system data integration and the cost of manual modeling. By stripping away the intermediate graph structure of protocol-specific layers to complete the mapping to standard nodes, the output information model structurally reflects the real functional relationships of devices, rather than the private organizational methods of communication protocols. This enables the automatic generation of cross-protocol, highly consistent information models without manual configuration.

[0047] In this embodiment, by parsing and normalizing the fields of multi-source heterogeneous industrial point tables, a standardized point table object with a unified structure is generated. Multi-dimensional features are then extracted from the data points in this object, and joint semantic parsing is performed on these multi-dimensional features using a multi-constraint consistency judgment mechanism to obtain industrial semantic objects. These industrial semantic objects are then converted into semantic nodes, and edges are constructed between nodes based on the business relationships between semantic objects, generating an industrial semantic graph independent of industrial protocols. This graph is then traversed to map each semantic node to an OPC UA node, thereby constructing an OPC UA information model. By unifying the structure of the input point tables, the subsequent parsing and modeling processes are decoupled from the communication protocol, ensuring cross-protocol adaptability of the modeling process. Furthermore, by leveraging the joint semantic parsing of multi-dimensional features, the business attributes of data points can be comprehensively defined from multiple dimensions, enabling the generated semantic objects to possess a structured description capability of the inherent meaning of industrial data. The topology of the semantic graph is constructed based on business relationships rather than the protocol's native data hierarchy, removing the private organizational structure of the communication protocol from this intermediate expression layer of the industrial semantic graph, retaining only cross-device common logical connections. Based on protocol-independent industrial semantic graph execution node mapping, the constructed OPC UA information model can reflect the functional division and data association of devices. It can maintain semantic consistency and structural standardization of modeling results among devices with different protocols, thereby improving the universality and generalization ability of the information model.

[0048] Reference Figure 2 The information model construction process shown in this embodiment mainly includes industrial point table input, industrial point table standardization processing, industrial semantic parsing, protocol-independent semantic modeling, automatic construction of OPC UA information model, and generation of NodeSet XML file.

[0049] For industrial point tables as input, the received industrial point tables are raw point tables from the industrial field. Specifically, they can originate from PLCs, DCSs, RTUs, edge gateways, industrial controllers, or other industrial equipment, or from equipment configuration files, engineering databases, or exported files in formats such as Excel, CSV, and XML. Because different manufacturers and different industrial protocols employ significantly different data organization methods, the specific source of the industrial point tables is not limited; instead, industrial point tables of various formats are uniformly used as the raw input object. Industrial point tables typically contain variable names, register addresses, data types, access permissions, units, descriptive information, and equipment identifiers, which collectively describe the data organization structure of the industrial equipment. The received industrial point table data serves as a unified data input object, providing the raw data foundation for subsequent standardization processing.

[0050] Understandably, for the original industrial point table containing abnormal data such as missing fields, null values, illegal characters, garbled characters, and invalid measurement points, preprocessing is performed using preset data cleaning rules: data points missing key fields (such as equipment ID, address, and data type) are directly filtered and written to the exception log; data points missing only non-core fields such as unit and description are automatically filled with default empty attributes; variable names containing illegal characters are subject to character filtering and cleaning, and data points that fail to be cleaned are marked as exception entries for manual screening and removal.

[0051] Furthermore, referring to Figure 3 The standardization process for industrial point tables mainly includes industrial point table field parsing, field structure standardization, industrial address normalization, and generation of standard industrial point table objects. Based on this, step S10 may include steps S11~S14: Step S11: Parse the fields of the industrial point table to extract the basic industrial information from the industrial point table; the basic industrial information includes at least variable name, variable description, data type, address information, access permissions, collection cycle and device identifier; Step S12: Perform unified parsing on the data of different formats in the basic industrial information to obtain the original field set; Step S13: Map the fields in the original field set to a pre-established standard field model to obtain a standardized field object set; and convert the industrial addresses in the original field set into a standard address object set based on a pre-built address abstract model. Step S14: Based on the pre-built object model, integrate the standardized field object set and the standardized address object set into a standardized industrial point table object; the standardized industrial point table object includes device identifier, variable name, data type, standard address object, access permission and extended attributes.

[0052] When parsing and normalizing the industrial point table, the first step is to parse the table fields of the industrial point table to extract the basic industrial information used to describe the data organization structure of industrial equipment. This basic industrial information is the basic data that reflects the physical characteristics and communication attributes of the measurement points of industrial field equipment. It includes at least variable name, variable description, data type, address information, access permissions, collection cycle and equipment identifier.

[0053] The data of different formats extracted from the basic industrial information are uniformly parsed to obtain the original field set. This original field set is a field dataset that has been cleaned up in basic format but has not undergone deep semantic processing, such as a set of key-value pairs stored in the form of an array or hash table.

[0054] The fields in the original field set are mapped to a pre-established standard field model to obtain a standardized field object set. Additionally, based on a pre-built address abstraction model, the industrial addresses in the original field set are converted into a standard address object set.

[0055] The standard field model serves to provide a universal set of baseline field definition specifications to eliminate differences in the expression of synonyms or abbreviations between point tables from different sources. The standard address object set is a collection of universal addressing description structure objects that abandon specific communication protocol syntax; for example, it includes arrays of structures with independent dimensions such as logical storage areas, byte offsets, and bit offsets.

[0056] The purpose of the address abstraction model is to decompose and reassemble the address strings specific to various underlying protocols into general logical addressing parameters independent of the communication protocol, thereby achieving protocol independence. Standardized industrial point table objects are data entities with strict type constraints and complete attribute fields.

[0057] Furthermore, based on a pre-built object model, the standardized field object set and the standardized address object set are integrated into a standardized industrial point table object, which includes device identifier, variable name, data type, standard address object, access permission, and extended attributes.

[0058] Because different industrial protocols and equipment from different manufacturers use different data representation methods, the same business object may have different field names, different address formats, and different organizational structures. Therefore, the received industrial point table object is parsed and normalized to convert it into a unified data representation.

[0059] Specifically, the industrial point table is first standardized through field parsing. For the received raw industrial point table, each field is parsed to identify basic industrial information such as variable names, address information, data types, access permissions, and device identifiers. Then, based on predefined data specifications, address representations from different protocols are uniformly mapped. For example, Modbus register addresses, S7 DB addresses, and OPC UA node addresses are uniformly converted into predefined standard address objects; and data type definitions from different manufacturers are uniformly converted into standard data type expressions. After field parsing, a unified data object structure is further established, encapsulating all fields into standard point table objects, eliminating the need for subsequent processing to consider specific industrial protocols. This standardization process eliminates data representation differences between different industrial protocols, providing a unified data foundation for industrial semantic parsing.

[0060] By introducing standard field models and address abstraction models, heterogeneous point table data, which originally depended heavily on the habits of specific equipment manufacturers and specific communication protocols, was structurally transformed into a standardized and modular collection of objects. This greatly improved the compatibility with massive heterogeneous data and laid a data quality foundation for subsequent accurate semantic parsing.

[0061] For field parsing of industrial point tables, since these tables may originate from different protocols or devices, their field structures often differ significantly. Therefore, upon receiving an industrial point table, field parsing is the first step. The core task of field parsing is to extract basic industrial information from the raw data of the industrial point table, including variable names (Tag Name), variable descriptions (Description), data types (Data Type), address information (Address), access permissions (Read / Write), scan cycle (Scan Cycle), and device ID (Device ID). Data of different formats is parsed uniformly using regular expression matching, structured parsing, or a protocol parser. For example, for Excel format point tables, fields are extracted using column mapping; for PLC exported files, register definitions are identified using a protocol parser; for JSON format data, structural information is extracted using key-value parsing to obtain the original field set for subsequent standardization processing. For field structure standardization, even after field parsing, inconsistencies in structure still exist between data from different sources, such as differences in field naming (temperature variables are named Temp and Temperature), unit representation (°C and Celsius), and address representation. Therefore, field structure standardization is necessary. First, a unified standard field model is pre-established, including standard variable names, standard data types, standard address structures, and standard access attributes. Then, fields from different sources are mapped to this unified standard field model. For example, the original field Temp is mapped to the standardized field Temperature, the original field REAL is mapped to the standardized field Float32, and the original field DB1.DBW0 is mapped to the standardized field StandardAddress. Through field structure standardization, the heterogeneous industrial point table fields from multiple sources are transformed into a unified structural expression. The key role of field structure standardization is to eliminate field-level differences, resulting in a standardized set of field objects that provides a structurally consistent data foundation for semantic parsing.

[0062] For industrial address normalization, different industrial protocols use different address systems. For example, Modbus uses register addresses; S7 uses block addresses; OPC UA uses NodeId; and EtherNet / IP uses tag paths. These addresses differ in logical expression, thus requiring unified normalization. By constructing a unified address abstraction model, all addresses are converted into standard industrial address objects. These standard industrial address objects contain address type, address space identifier, address offset, and access path. For example, Modbus addresses are converted to RegisterAddress objects, S7 addresses to BlockAddress objects, and OPC UA addresses to NodeAddress objects. This process eliminates the structural differences at the address level between different protocols, resulting in a set of standard address objects that give all data points a unified and computable structure.

[0063] For the generation of standard industrial point table objects, after standardizing the field structure and normalizing the address, all information is integrated into a unified standardized industrial point table object. This standardized industrial point table object includes device identifier (Device_ID), variable name (Variable_Name), data type (Data_Type), standard address object (Address_Object), access permission (Access_Permission), and extended attributes (Metadata). Through object encapsulation, all industrial point table data is converted into a unified object model. The core function is to convert unstructured industrial point tables into structured industrial data objects, resulting in a set of standard industrial point table objects, thus providing a unified input for semantic parsing. If multiple input industrial point tables correspond to the same physical device and there are multiple address records pointing to the same physical acquisition quantity, deduplication can be automatically performed. Data points are merged based on device ID and variable service name, and multiple sets of protocol addresses are uniformly encapsulated into the multi-address set attribute of the same standard address object, avoiding the generation of duplicate standardized point table objects.

[0064] Through field parsing and normalization, the industrial point table was transformed from multi-protocol, multi-format data into a unified structured industrial object. This imbues the original industrial data with a unified field structure and standard address representation capabilities, thus resolving the issue of inconsistent data structures between different industrial protocols. The reason for combining field structure standardization with address abstraction modeling is that industrial field data sources are complex, and relying solely on single-field mapping cannot eliminate protocol differences. A two-layer processing mechanism of unified field semantics and abstract address structure can simultaneously address both field inconsistencies and address system inconsistencies. The resulting standard industrial object provides a unified input foundation for subsequent semantic parsing, eliminating reliance on specific industrial protocols or original data formats. This provides a structurally consistent data foundation for multi-dimensional feature extraction and semantic judgment.

[0065] In one embodiment, refer to Figure 4 The semantic parsing process mainly includes multi-dimensional feature extraction, multi-constraint consistency semantic determination, semantic determination and convergence, and industrial semantic object generation. For example... Figure 4 As shown, for the semantic determination of multi-constraint consistency, if all constraints pass, an industrial semantic object is directly generated. If not all constraints pass, a semantic determination convergence determination is performed on the standardized industrial point table object to determine whether the comprehensive score is lower than the preset minimum matching threshold. If the comprehensive score is lower than the preset minimum matching threshold, the standardized industrial point table object is marked as an unknown semantic node, and after manually supplementing the semantic labels, multi-dimensional feature extraction is performed again, and then the semantic parsing process is re-executed. If the comprehensive score is greater than or equal to the preset minimum matching threshold, an industrial semantic object is also directly generated.

[0066] Based on this, step S20 may include steps S21 to S22: Step S21: Perform semantic parsing on the standardized point table object, extract multi-dimensional features of each data point in the standardized point table object, and obtain a feature vector; the multi-dimensional features in the feature vector include at least name features, data type features, address features, behavior features, and context features; Step S22: Based on the multi-constraint consistency determination mechanism, the feature vector is jointly semantically determined with multiple constraints. The industrial semantic category corresponding to the feature vector is determined according to the determination result, resulting in an industrial semantic object that includes equipment semantic objects, variable semantic objects, attribute semantic objects and address semantic objects.

[0067] In the semantic parsing stage, the standardized point table object is semantically parsed to extract the multidimensional features of each data point in the standardized point table object and obtain the feature vector. The multidimensional features in the feature vector include at least name features, data type features, address features, behavior features and context features.

[0068] Furthermore, based on the multi-constraint consistency determination mechanism, the feature vector is jointly semantically determined with multiple constraints, and the industrial semantic category corresponding to the feature vector is determined according to the determination result, thus obtaining an industrial semantic object containing equipment semantic objects and variable semantic objects.

[0069] It should be noted that multidimensional features are a set of features that can jointly characterize the essential attributes of an industrial measurement point from different perspectives, such as feature clusters that combine lexical, syntactic, spatial, and temporal attributes.

[0070] The purpose of the multi-constraint consistency determination mechanism is to prevent misjudgment due to a single feature (such as the failure of name features due to non-standard naming). By cross-comparison and joint determination of multiple constraints, it ensures the global rationality and accuracy of semantic parsing results.

[0071] Industrial semantic categories are specific hierarchical nodes in a predefined logical classification system of industrial entities, such as motor equipment, ambient temperature variables, or control command variables.

[0072] Equipment semantic objects are logical abstract objects that represent physical equipment in an industrial setting, such as an object representing a specific model of centrifugal pump; variable semantic objects are logical abstract objects that represent specific operating parameters of equipment, such as a numerical object representing the rotational speed of the centrifugal pump.

[0073] In some embodiments, step S22 may further include steps A1 to A3: Step A1: If the determination result indicates that the feature vector satisfies all the conditions of the multiple constraints, then determine the industrial semantic category corresponding to the feature vector based on the determination result of the multiple constraints. Step A2: If the determination result indicates that the feature vector satisfies some of the multiple constraints, then determine the candidate semantic set according to the target conditions satisfied by the feature vector among the multiple constraints. Step A3: According to the preset scoring rules, score and sort each candidate semantic in the candidate semantic set, and determine the industrial semantic category corresponding to the feature vector according to the sorting order.

[0074] When determining the industrial semantic category corresponding to the feature vector based on the determination result of the consistency of multiple constraints, if the determination result represents that the feature vector satisfies all constraints of multiple constraints, then the unique industrial semantic category corresponding to the feature vector is determined based on the determination result of multiple constraints.

[0075] If the judgment result indicates that the feature vector satisfies some of the constraints of multiple constraints, that is, at least one of the multiple constraints is not satisfied, then the candidate semantic set is determined according to the target conditions satisfied by the feature vector among the multiple constraints; then, the candidate semantics in the candidate semantic set are scored and ranked according to the preset scoring rules; finally, the industrial semantic category corresponding to the feature vector is determined according to the ranking order, and the industrial semantic category is the candidate semantic category with the highest score in the candidate semantic set.

[0076] Traditional methods typically generate OPC UA nodes directly based on field names. However, this application's embodiment first performs semantic understanding on standardized industrial point tables, elevating industrial data from field-level descriptions to semantic-level expressions. By reading standardized industrial point table objects, multiple semantic features are extracted for each data point, including field name features, data type features, address organization features, access behavior features, and contextual relationship features. Then, based on a pre-established industrial semantic rule base, joint analysis is performed on these multiple features, and a multi-constraint consistency determination mechanism is used to determine the industrial semantic category corresponding to the field. For example, when a field name contains "Temperature," and its data type is floating-point, its address is located in the analog acquisition area, and it has periodic acquisition attributes, multiple constraints are considered, and it is determined to be a temperature measurement semantic object, rather than solely based on the field name. Finally, a unified set of industrial semantic objects is generated, including equipment semantic objects, variable semantic objects, attribute semantic objects, and address semantic objects. By establishing a unified semantic expression, the differences caused by different equipment field naming methods are effectively eliminated, improving the semantic understanding capability of industrial data.

[0077] Based on the standardized industrial point table, an industrial semantic parsing mechanism based on multi-constraint consistency is introduced to realize the transformation of industrial data from structured description to semantic expression.

[0078] Specifically, the standard industrial point table objects are first subjected to multi-dimensional feature extraction. Unlike the traditional approach that relies solely on field names, this method constructs an industrial data feature space from multiple dimensions, including name features, type features, address features, behavior features, and context features. For example, Temperature_1 not only analyzes the name but also combines its data type, analog address, periodic sampling behavior, and contextual relationship with the same equipment to form a complete feature vector representation. This transforms industrial data from field descriptions into a feature space expression, resulting in feature vectors.

[0079] Furthermore, to avoid misjudgments caused by relying solely on variable names or single rules for semantic recognition, the semantic recognition process for industrial data is divided into multiple independent constraint dimensions. Each constraint corresponds to a characteristic attribute of the industrial data. First, it is calculated whether each constraint meets the corresponding semantic requirements. Then, the final semantic category is determined by combining the judgment results of all constraints. For example, when determining whether a variable belongs to a temperature measurement point, it is not only judged whether the variable name contains keywords such as "Temp" or "Temp," but also whether the data type is floating-point, whether the address is located in the analog signal acquisition area, whether the acquisition period meets the analog signal sampling characteristics, and whether there is a temperature control object in the equipment context. Only when multiple constraints are satisfied simultaneously is the variable considered to have the semantics of a temperature measurement point, thereby effectively avoiding misjudgment problems caused by non-standard variable naming.

[0080] The multi-constraint consistency determination mechanism introduces five types of constraints: semantics, data type, address, acquisition behavior, and context. The semantic determination rule is as follows: when all constraints are simultaneously satisfied, a unique semantic category is directly output; if any key constraint is not satisfied, multiple candidate semantic sets are generated, and the most suitable industrial semantic category is selected through semantic determination and convergence. The formal expression is as follows: (1) in, Identifying data from a single standardized industrial measurement point The semantic result output after constraint verification; this semantic result is a piecewise function, the first segment. , , ,...., Corresponding in sequence There are several different constraints, including but not limited to name constraints, data type constraints, address constraints, acquisition behavior constraints, and context constraints. If all constraints are satisfied, the output will be... The verification result is the uniquely determined standardized industrial semantic category.

[0081] Second segment If at least one of the multiple constraints is not met, then the output is a set of multiple candidate industrial semantic categories. When it is impossible to fully satisfy all hard constraints or directly lock in a unique semantic, a list of matched candidate semantics is generated, and a weighted convergence selection is performed.

[0082] Based on this, a comprehensive matching score is calculated for each candidate industrial semantic category in the generated candidate semantic set. The score is obtained by weighting multiple dimensions such as the degree of comprehensive semantic matching, the degree of consistency of hierarchical structure, the degree of address matching, and the degree of consistency of collection behavior. The candidate industrial semantic categories are sorted according to the comprehensive score, and the highest score is selected as the final recognition result according to the sorting order.

[0083] For example, if a variable name matches both the "Temperature" and "ProcessValue" candidate industrial semantic categories, and the variable is an analog address, has a 100ms acquisition period, and belongs to a temperature control device, then "Temperature" has a higher overall score and is classified as the "Temperature" industrial semantic category. If the candidate semantic set contains mutually exclusive semantic tags such as "temperature" and "pressure," a mutually exclusive semantic priority rule is enabled: equipment and process context are used as high-priority auxiliary judgment criteria, address and acquisition behavior features are given priority, and the ranking weight is reduced based solely on candidate semantics with name matching to minimize semantic confusion. It should be noted that equipment and process context are only used as hard judgment constraints here and do not participate in the weighted calculation of the overall score. They are only used as auxiliary priority reference conditions when mutually exclusive candidate industrial semantic categories appear.

[0084] The comprehensive scoring expression can be represented as: (2) in, The scoring factors include semantic matching degree. Consistency of equipment hierarchy structure Address matching degree Consistency with data collection behavior ; , , and These are the weighting coefficients of each scoring factor, and the semantic category with the highest score is ultimately selected as the unique industrial semantic category. The weighting coefficients can be configured according to the actual industrial scenario.

[0085] Since some industrial fields may have semantic conflicts (such as inconsistent names and addresses), a multi-constraint consistency scoring mechanism is introduced to score and rank candidate industrial semantic categories, and resolve the multi-possible multi-constraint consistency semantics into a unique and definite semantic, thus obtaining a deterministic semantic object.

[0086] If the scores of all candidate semantics after comprehensive scoring are all lower than the preset minimum matching threshold, it is determined that there is no matching standard semantics for the data point, and it is marked as an unknown semantic node. All information such as variable names, original addresses, data types and access permissions of the original point table are completely retained as built-in extended attributes of the semantic node. After manually annotating the semantic labels, the multi-constraint consistency semantic parsing process is re-executed.

[0087] Finally, based on the deterministic semantic parsing results, standard industrial semantic objects are generated. These objects include Device Semantic Objects, Variable Semantic Objects, Function Semantic Objects, and Property Semantic Objects. Each semantic object contains not only a semantic category but also its structural attributes and relationships. This upgrades field-level data to semantic-level industrial objects.

[0088] By employing semantic parsing and multi-constraint consistency determination, the transformation of industrial data from structured industrial objects to semantic industrial objects is achieved. This enables semantic recognition of industrial variables from multiple dimensions and outputs a unique and definite industrial semantic category. The reason for adopting a multi-constraint consistency determination mechanism is that industrial sites often exhibit inconsistent variable naming and heterogeneous data sources. A single rule or feature can easily lead to semantic misjudgment. Therefore, joint determination using multi-dimensional constraints such as name, type, address, behavior, and context can significantly improve the accuracy and stability of semantic recognition. The resulting industrial semantic objects not only contain the semantic categories of variables but also their corresponding structural and behavioral feature information, providing a semantic node foundation for the construction of the Industrial Semantic Graph (ISG) and enabling modeling within a unified semantic space.

[0089] For the construction of industrial semantic graphs, refer to Figure 5 The illustrated industrial semantic graph construction process mainly includes semantic node generation, semantic relation generation, medial decoupling graph structure construction, semantic graph structure optimization and convergence, and cross-protocol same mapping. Based on this, step S30 may include steps S31-S34: Step S31: Convert the industrial semantic object into a semantic node; the node attribute information of the semantic node includes node identifier, node type, semantic tag, source device, and semantic feature vector; the node type includes device node, variable node, function node, and attribute node; Step S32: Based on the association relationships between the industrial semantic objects, determine the semantic relationships between the corresponding semantic nodes, construct the edges between the semantic nodes based on the semantic relationships, and generate an initial semantic graph; the association relationships include device hierarchy relationships, composition relationships, reference relationships, address mapping relationships, and business association relationships; the semantic relationships include structural relationships, functional association relationships, address association relationships, acquisition behavior relationships, and context behavior relationships; Step S33: Perform protocol decoupling and structural optimization on the initial semantic graph to obtain an initial industrial semantic graph. The protocol decoupling includes removing protocol-specific hierarchical fields, eliminating protocol tree structures, merging equivalent semantic nodes, and unifying semantic connection paths. The optimization strategy of the structural optimization includes node redundancy removal, edge weight convergence, and stable graph structure. Node redundancy removal involves merging semantic nodes with the same semantics but different origins. Edge weight convergence represents the normalization of multiple associations of the same semantic node. The stable graph structure represents the global consistency adjustment of the semantic relationships. Step S34: Perform a unified cross-protocol mapping on the initial industrial semantic graph, and mark the industrial protocol as the data source attribute of the semantic node to obtain an industrial semantic graph that is independent of the industrial protocol.

[0090] Industrial semantic objects are converted into semantic nodes. The node attribute information of the semantic node includes node identifier, node type, semantic tag, source device and semantic feature vector. The node type further includes device node, variable node, function node and attribute node.

[0091] Based on the business relationships between industrial semantic objects, the semantic relationships between corresponding semantic nodes are determined, and semantic edges between semantic nodes are constructed based on these semantic relationships to generate an initial semantic graph. The relationships between industrial semantic objects include equipment hierarchy relationships, composition relationships, reference relationships, address mapping relationships, and business relationships; the semantic relationships include structural relationships, functional relationships, address relationships, data acquisition behavior relationships, and contextual relationships. Among these, structural relationships can be used to represent equipment hierarchy relationships.

[0092] The initial semantic graph is subjected to protocol decoupling and structural optimization to obtain the initial industrial semantic graph. Protocol decoupling includes removing protocol-specific hierarchical fields, eliminating protocol tree structures, merging equivalent semantic nodes, and unifying semantic connection paths. The structural optimization strategy includes node redundancy removal, edge weight convergence, and stable graph structure. Node redundancy removal involves merging semantic nodes with the same semantics but different origins; edge weight convergence normalizes multiple associations of the same semantic node; and stable graph structure represents a global consistency adjustment of semantic relationships.

[0093] Finally, a unified cross-protocol mapping is performed on the initial industrial semantic graph, marking industrial protocols as data source trees for semantic nodes, resulting in an industrial semantic graph independent of industrial protocols. Specifically, the industrial protocols marked as data source trees for semantic nodes refer to the industrial protocols of the data points corresponding to those semantic nodes.

[0094] To achieve unified cross-protocol modeling, a protocol-independent industrial semantic graph (ISG) is constructed. Specifically, based on the industrial semantic objects obtained from semantic parsing, each semantic object is converted into a semantic node. Connections between semantic nodes are established according to relationships such as equipment structure, data organization, and business logic. When establishing these connections, multiple factors are considered, including the subordinate, composition, reference, address continuation, and business relationships of the industrial semantic objects, ensuring that the ISG accurately reflects the data organization structure of industrial equipment. Then, ISG optimization is further performed, merging duplicate nodes, reorganizing hierarchical relationships, and validating connections, ultimately forming a protocol-independent industrial semantic graph model with a unified hierarchical structure. Because this ISG has transcended the data organization methods of specific industrial protocols, devices using different protocols can be uniformly modeled in a unified semantic space, providing a unified data structure for the subsequent automatic construction of information models.

[0095] In one specific implementation, protocol-agnostic industrial semantic graph (ISG) modeling is used to establish a unified semantic representation structure across different industrial protocols, thereby achieving consistent cross-protocol modeling of industrial data. The ISG modeling process takes industrial semantic objects as input, and through node construction, relation generation, graph structure optimization, and consistency convergence, forms a unified ISG model. This model does not depend on any specific industrial protocol but is based on industrial semantic relationships, thus achieving a unified expression across protocols.

[0096] Specifically, the process begins by reading the set of industrial semantic objects. Each industrial semantic object is converted into a node in the ISG, i.e., a semantic node. Node types include, but are not limited to, Device Nodes, Variable Nodes, Function Nodes, and Property Nodes. Node definitions contain structural information such as a unique identifier (Node_ID), node type (Node_Type), semantic label (Semantic_Label), source device (Source_Device), and semantic feature vector (Feature_Vector), which constitute the node's attribute information. For example, a temperature measurement point semantic object will be mapped to generate a variable node with the semantic label "Temperature," while a PLC device semantic object will be mapped to generate a device node. This process converts the industrial semantic objects into computable graph nodes, resulting in a set of semantic nodes.

[0097] After generating semantic nodes, semantic edges between nodes are further constructed. Unlike traditional modeling that relies on only a single-level relationship, ISG modeling constructs a multi-class complete semantic relationship system. The definitions of each type of relationship are as follows: Structural relationships: used to characterize the subordinate or affiliation relationship between equipment and internal variables and functions. A typical scenario is that PLC equipment has various built-in acquisition variables and control functions.

[0098] Functional association: This is used to characterize the association between data points and corresponding business logic, such as temperature measurement points matching temperature control logic, and pressure measurement points matching over-limit alarm logic.

[0099] Address association: used to characterize the corresponding mapping of the same physical measurement point in different protocol addresses, such as Modbus register points and S7 DB storage points being associated with the same acquisition semantic variable.

[0100] Data acquisition behavior relationship: Used to characterize similar measurement points with consistent data acquisition configurations, such as a set of measurement points that share a scanning cycle and belong to the same data acquisition data source.

[0101] Contextual relationships: This is used to characterize the business binding relationship of all measurement points under the same process unit and the same equipment group. Temperature, flow and pressure measurement points within the same process flow will establish contextual relationships, which facilitates unified modeling and analysis at the subsequent process level.

[0102] After constructing the semantic nodes, it is necessary to further determine whether there are business relationships between different nodes. Since industrial data may simultaneously have structural relationships at the device level, functional relationships, address mapping relationships, and data collection behavior relationships, relying on only one type of relationship can easily lead to an incomplete graph structure. Therefore, in this embodiment, the degree of association on multiple dimensions is calculated separately, and then the association weights between nodes are formed by combining them.

[0103] For example, two semantic nodes representing a temperature acquisition point and a temperature alarm point, respectively, can still establish a semantic relationship between them because they belong to the same device, have the same acquisition behavior, and are related in business semantics, even though they correspond to different data addresses. However, for two data points with adjacent addresses but completely different semantics, no semantic relationship will be established.

[0104] To express the correlation between the same equipment or the same process segment, the correlation strength between semantic nodes can be calculated based on the feature vector of the industrial semantic object. The correlation weight used to characterize this correlation strength can be expressed as: (3) in, to The weights are configurable; and satisfy the following conditions: The components in Formula 3 are used to characterize the relationships between semantic nodes in different dimensions, where, Indicates the degree of consistency of semantic tags. This indicates the consistency of semantic nodes within the device hierarchy. Indicates the degree of correlation in industrial address spaces. Indicates the degree of consistency in data collection behavior patterns. It indicates the degree of consistency of the industrial context in which the semantic node is located, such as the equipment it belongs to, the process unit it belongs to, the production process it belongs to, or the upstream and downstream logical relationships.

[0105] A comprehensive scoring mechanism maps multi-dimensional relationships to a single weight value to determine whether a semantic connection has been established between nodes. A preset association weight threshold T (e.g., T=0.3, configurable) is set, and the connection is only established when... When semantic edges are established between semantic nodes, if If no direct business relationship is determined, no semantic edge is generated.

[0106] Based on semantic nodes and the semantic edges between them, an initial semantic graph, denoted as G0, is generated. G0 inherits the hierarchical constraints from the original industrial protocol and still carries a protocol-specific organizational structure, making it impossible to achieve unified modeling across protocols. Therefore, it is necessary to perform protocol decoupling and structural optimization on the initial semantic graph G0, removing the protocol-specific organizational structure and retaining only the business semantic relationships.

[0107] Specific optimization operations include removing protocol-specific hierarchical fields, eliminating protocol tree structures such as DB blocks / registers, merging equivalent semantic nodes from different protocols pointing to the same physical measurement point, and reorganizing the unified semantic connection path. Taking a temperature measurement point as an example, regardless of whether the measurement point originates from a Modbus register or an S7 DB block, it will be converged into the same temperature semantic node, no longer distinguishing the underlying protocol source. After protocol decoupling optimization, a standardized initial industrial semantic graph G1 is obtained.

[0108] Protocol decoupling does not mean deleting industrial protocol information, but rather separating protocol-related data organization from the semantic model, allowing industrial objects to retain only their business semantic attributes. For example, the same temperature measurement point may be represented by different addresses in different industrial protocols. Although the address representations are completely different, they all represent the same temperature variable. Therefore, they are uniformly mapped to the same Temperature semantic node, retaining only the device origin as a node attribute, and no longer retaining the specific protocol hierarchy, thereby achieving unified modeling across protocols.

[0109] The protocol decoupling process for the initial semantic graph only removes the hierarchical nesting structure caused by the protocol, without deleting the original protocol address information; the underlying original address data such as Modbus register address, S7 DB block address and EtherNet / IP label path are uniformly and persistently stored as extended property attributes of the corresponding semantic nodes, so as to complete the cross-protocol unified semantic modeling while fully retaining the underlying address mapping information required for device data acquisition.

[0110] Furthermore, since the initial graph structure may contain redundant nodes or repeated semantic relationships, structural optimization of the initial semantic intent is also required. Optimization strategies include: Node redundancy removal: Merging nodes with the same semantics but different origins.

[0111] Edge weight convergence: The weights of multiple relationships are normalized. Standardized weights are obtained through normalization operations to eliminate the weight superposition bias caused by multiple associated edges and ensure that the node association strength has a unified comparison standard.

[0112] Graph structure stabilization: Through constraint propagation algorithms and other methods, semantic relationships are made globally consistent, with no isolated key nodes, no conflicting relationship paths, no cyclic semantic dependency conflicts, and a stable ISG formed after optimization. Graph structure optimization does not change the original business relationships between industrial semantic objects, but rather, while maintaining semantic consistency, it uniformly organizes duplicate nodes, redundant relationships, and conflicting connections, so that the final generated ISG has uniqueness, stability, and consistency, providing a unified data foundation for subsequent information model generation. For isolated semantic nodes without associated edges, they are not directly deleted during graph structure optimization. Instead, independent nodes are retained and attached to their respective device nodes. The attachment is completed through basic hierarchical relationships, ensuring that all industrial semantic objects can access the complete ISG hierarchical structure, and there are no wandering, unaffiliated nodes.

[0113] Finally, a cross-protocol unified mapping mechanism is adopted to construct a protocol-independent industrial semantic graph. The core mechanism is that industrial protocols are not used as modeling dimensions, but are only marked as data source attributes. Modbus, S7, OPC UA, and private protocol devices can all be mapped to the same ISG after processing. Device A (Modbus), Device B (S7), and Device C (OPC) ultimately share the same semantic structure.

[0114] Through protocol-independent semantic modeling, the transformation of industrial semantic objects into ISG structures was achieved, upgrading industrial data from field-based representations to a semantic relationship network structure, thereby enabling unified modeling and representation of data from different industrial protocols. The reason for adopting a multi-type semantic relationship modeling mechanism is that industrial data not only exhibits hierarchical relationships but also complex relationships such as functional, behavioral, and address associations. Single-relationship modeling cannot fully represent the true structure of an industrial system. Therefore, constructing an ISG through multi-dimensional relationship fusion can more comprehensively characterize the semantic structure of the industrial system. The final constructed ISG eliminates structural differences between different industrial protocols, enabling all industrial data to be expressed in a unified semantic space, providing a standardized graph structure foundation for the subsequent automatic generation of OPC UA information models.

[0115] Based on the constructed protocol-independent industrial semantic graph, an OPC UA information model is built. In one embodiment, Figure 6 An exemplary construction process for an OPC UA information model is provided, which includes semantic node to OPC UA node mapping, friendship relationship to OPC UA reference relationship generation, node identifier generation and uniqueness control, OPC UA address space construction, OPC UA information model consistency verification, and NodeSet XML file generation. For the construction of the OPC UA information model, step S40 may include steps S41-S44: Step S41: Traverse the industrial semantic graph and map each semantic node to an OPC UA node; Step S42: Based on the semantic relationships between the semantic nodes, generate the reference relationships between the OPC UA nodes; Step S43: Perform hash calculation on the node attribute information of the semantic node to obtain the node identifier of the OPC UA node; Step S44: Based on the OPC UA node, the reference relationship, and the node identifier, construct a node hierarchy and bind the reference relationship to obtain the OPC UA information model.

[0116] When constructing the OPC UA information model, the industrial semantic graph is first traversed, mapping each semantic node to an OPC UA node. Then, based on the semantic relationships between semantic nodes, reference relationships between OPC UA nodes are generated, and the node attribute information of the semantic nodes is hashed to obtain the node identifier of the OPC UA node. Finally, based on the reference relationships between OPC UA nodes and OPC UA nodes, as well as the node identifiers obtained through hash calculation, a node hierarchy is constructed and the reference relationships are bound to obtain the OPC UA information model.

[0117] Specifically, based on the protocol-independent industrial semantic graph, an OPC UA information model is automatically generated. First, all semantic nodes in the ISG are traversed, and the corresponding OPC UA node type is automatically determined according to the semantic category of the node, mapping the semantic node to an OPC UA node.

[0118] For example, device semantic objects are mapped to Object nodes, variable semantic objects to Variable nodes, and attribute semantic objects to Property nodes. Then, reference relationships are automatically generated based on the semantic relationships between semantic nodes in the ISG, and node attributes are automatically configured. The entire modeling process is driven by the ISG, requiring no manual configuration of node hierarchy and reference relationships, and can automatically generate an information model conforming to the OPC UA standard.

[0119] In the specific implementation process, for the mapping from semantic nodes to OPC UA nodes, the Protocol Independent Semantic Graph (ISG) is read first. Each semantic node in the ISG will be mapped to a corresponding node in the OPC UA information model. The mapping rules include device semantic node mapping, variable semantic node mapping, function semantic node mapping, and attribute semantic node mapping.

[0120] Device semantic node mapping: Device semantic nodes correspond to the generation of OPC UA device object nodes, which are used to represent the physical body of industrial equipment such as PLCs, sensors, and actuators. The attribute content configured for this node includes the object type adopting a device-specific type, the browsing name being the corresponding device name, and being assigned a globally unique node identifier.

[0121] Variable semantic node mapping: Variable semantic nodes are mapped to OPC UA Variable Nodes to represent collected data points, such as temperature, pressure, and current. Node attributes include data type (Float, Int, and Boolean, etc.), access level (Read / Write), and acquisition period.

[0122] Functional semantic node mapping: Functional semantic nodes are mapped to OPC UA Method Nodes to represent control logic or functional behavior, such as start / stop control, alarm triggering, etc.

[0123] Attribute semantic node mapping: Attribute semantic nodes are mapped to OPC UA Property Nodes, which are used to describe additional attributes such as units, precision, and range.

[0124] The OPC UA node set is obtained through deterministic mapping from semantic nodes to OPC UA node types, and OPC UA node attribute configuration based on semantic node associations and node attribute information.

[0125] If a semantic object carries a non-standard data type not defined by the OPC UA standard, it is automatically mapped to the general AnyType basic type, and the original non-standard type name is stored in the node Property. If there are multiple conflicting access permission records for the same variable, the highest permission rule is adopted, and all original access permission information is recorded in the node extended properties.

[0126] For the generation of semantic relationships to OPC UA reference relationships, within the OPC UA information model, nodes are associated and expressed through references. Based on various semantic edges within the industrial semantic graph, corresponding OPC UA reference types are automatically generated. The correspondence rules between various semantic relationships and OPC UA reference relationships are as follows: the dependency relationship between a device and a variable matches a HasComponent reference; the association relationship between a variable and its subordinate properties matches a HasProperty reference; the hierarchical organization relationship of device layering matches an Organizes reference; the inheritance relationship between parent and child types matches a HasSubtype reference; and the control function call relationship of a device matches a HasMethod reference.

[0127] Through node mapping, reference generation, and attribute configuration, the semantic relationship graph is converted into a standard OPC UA relationship structure. The generation of references not only involves static mapping based on node types but also comprehensively considers the semantic relationships already established between nodes in the ISG. When the association weight between semantic nodes exceeds a preset weight threshold, i.e.... Reference relationships are generated only when the time is right. The generation of a reference relationship can be represented in the following form: (4) Only semantic nodes with explicit relationships in the ISG will generate corresponding OPC UA references between their corresponding nodes. This ensures that the generated information model accurately reflects the business relationships between industrial objects, rather than simply constructing a fixed hierarchical structure. The generation of references depends not only on the node type but also on the existing connections within the ISG. A corresponding OPC UA reference is only generated when two nodes have explicit semantic edges within the ISG, thus guaranteeing consistency between the information model structure and the ISG structure.

[0128] To ensure the global uniqueness of the OPC UA information model, a standardized NodeId generation mechanism is used to generate the node identifier for OPC UA nodes. The generation logic of this node identifier is based on a ternary hash structure, which is obtained by hashing the device identifier, semantic type, and variable name in the attribute information of the semantic node corresponding to the OPC UA node. The generation constraints include global uniqueness, traceability of source, and reproducibility of input. The NodeId registry is maintained synchronously to avoid duplicate allocation of node IDs.

[0129] The uniqueness control of OPC UA node identifiers does not aim to generate random identifiers, but relies on stable attributes such as device identifiers, semantic types, and variable names for hash calculation. The same industrial variable can output a completely consistent NodeId in multiple modeling operations. Node ID conflicts between different devices can be naturally avoided. At the same time, it has built-in traceability capabilities, which facilitates the docking and mapping between operation and maintenance and third-party industrial platforms.

[0130] Furthermore, such as Figure 6 As shown, the constructed OPC UA information model needs to be verified for consistency. If the verification fails, the node mapping is re-performed until the verification passes. Then, an OPC UA address space is constructed based on the OPC UA information model, and the OPC UA address space is serialized into a standard model file that can be directly used for OPC UA server deployment. This standard model file is a NodeSet XML file.

[0131] Based on this, after step S44, steps S45-S46 may be further included: Step S45: Perform a consistency check on the OPC UA information model. If the check passes, construct the OPC UA address space based on the OPC UA information model. Step S46: Serialize the OPC UA address space into a standard model file for model deployment on the OPC UA server.

[0132] For the constructed OPC UA information model, a consistency check is performed. If the check fails, the process returns and executes step S41 again until the check passes. If the check passes, an OPC UA address space is constructed based on the OPC UA information model, and the OPC UA address space is serialized into a NodeSet XML file for model deployment of OPC UA services.

[0133] For generating NodeSet XML files, a standard NodeSet XML file can be automatically generated based on the OPC UA information model. Specifically, namespace information is first established, and then object nodes, variable nodes, attribute nodes, and reference relationship nodes are generated sequentially according to a preset standard format. During the generation process, node uniqueness verification, reference integrity verification, namespace consistency verification, and data type consistency verification are performed simultaneously to ensure that the output information model conforms to the OPC UA standard specification, ultimately forming a standard NodeSet XML file that can be directly imported into servers that support the OPC UA standard, enabling rapid deployment of industrial equipment information models.

[0134] This automated modeling process for industrial data points, using industrial data points as input and standard NodeSet XML files as output, incorporates industrial semantic parsing and protocol-independent semantic modeling. This enables the automatic conversion of industrial data from raw data points to a standardized semantic model and then to an OPC UA information model. The entire process, through multiple collaborative steps including standardization, semantic parsing, semantic modeling, and automated modeling, not only solves problems such as the large workload of manual modeling, severe protocol coupling, difficulty in model reuse, and operational and maintenance difficulties, but also provides a general and scalable solution for the rapid access and unified modeling of multi-protocol devices in the Industrial Internet environment.

[0135] Furthermore, the construction of the OPC UA address space is based on the generated set of nodes and their reference relationships, resulting in a complete OPC UA address space. The specific construction process includes node hierarchy construction, reference relationship binding, and namespace management.

[0136] The node hierarchy is built based on the ISG structure, establishing a hierarchical organization that includes the Device layer, Variable layer, and Property layer. Reference binding embeds reference relationships between nodes, forming a complete topology. Namespace management automatically generates Uniform Resource Identifiers (URIs), including factory-level, device-level, and model-level namespaces, ultimately forming a complete OPC UA information model structure tree.

[0137] Consistency checks on the OPC UA information model are performed after the OPC UA address space is generated. The checks include: node uniqueness verification to ensure no duplicate NodeIds; reference loop detection to prevent illegal circular references; type consistency verification, ensuring that Variable DataType matches the semantic type; and hierarchy consistency verification, ensuring that the device-variable structure conforms to the ISG structure.

[0138] If the verification fails, an anomaly log is automatically generated containing at least one of the conflict node ID, conflict type, and conflict reason, and all abnormal semantic nodes are marked. Two processing modes are provided for abnormal semantic nodes. The first is the automatic processing mode, which automatically rolls back to the ISG structure optimization step to re-execute node merging and relationship convergence. The second is the manual intervention mode, which jumps to the semantic rule base or the original standardized point table editing interface, and after manually correcting the point table fields and semantic constraint rules, the modeling process is re-executed.

[0139] After completing the OPC UA address space construction, a NodeSet XML file is automatically generated. The specific generation process includes: node serialization, reference serialization, namespace writing, and data type definition writing. The output is an OPC UA standard model file that conforms to the preset standard specifications. This file can be directly imported into an OPC UA server or industrial platform.

[0140] As an example, when generating NodeSet XML files, the uniqueness of the namespace URI is automatically verified, and duplicate namespaces are automatically appended with a unique device identifier to distinguish them. For large-scale models with more than a preset threshold of measurement points, it supports the generation of multiple independent NodeSet XML files in chunks. Each file carries a unified root namespace identifier and can be imported into the OPC UA server in batches to complete the merging and loading.

[0141] By mapping node types, generating reference relationships, constructing address spaces, and verifying consistency, an OPC UA address space conforming to a preset standard is automatically generated and serialized into a NodeSet XML file. Unlike generating OPC UA nodes based on field mapping, this embodiment uses ISG as the driver to achieve automatic conversion from semantic relationship structure to standard industrial information model.

[0142] By mapping nodes, generating reference relationships, and configuring node attributes, an automatic conversion from industrial ISG to the standard OPC UA information model is achieved. This enables the industrial semantic structure to be mapped to a standard-compliant OPC UA model, ultimately generating a NodeSet XML file that can be directly deployed and used. The semantic-driven mapping mechanism is adopted because traditional OPC UA modeling schemes rely on manual configuration of node relationships, which is difficult to adapt to the needs of large-scale industrial equipment access. The ISG-driven approach automatically completes node type identification and reference relationship generation, effectively reducing the workload of manual modeling while improving the structural consistency and standardization of the generated information model. The final OPC UA information model achieves a unified mapping between industrial semantics and the standard industrial communication model, providing fundamental support for the publication and interoperability of industrial data on standardized industrial platforms.

[0143] For example, taking a temperature acquisition PLC device equipped with the Modbus RTU protocol in an industrial field as an example, the device exports an original industrial point table in Excel format, which includes variables such as temperature measuring point Temp1 and pressure measuring point Press1. First, industrial point table standardization is performed, mapping the original field Temp to the standard variable name Temperature, converting the 40001 class register address to a standard industrial address object, and generating a standardized industrial point table object. Then, through a multi-constraint consistency semantic parsing mechanism, combining multi-dimensional features such as variable name floating-point data type, analog address, and periodic acquisition behavior, the temperature measurement semantic object corresponding to Temp1 is identified. Based on all semantic objects, a protocol-independent industrial semantic graph (ISG) is constructed, establishing the subordinate semantic relationship between the device and each measurement point, and completing the graph structure optimization convergence. Based on the protocol-independent industrial semantic graph, OPC UA Object device nodes and Variable measurement point nodes are automatically mapped and generated, and HasComponent subordinate reference relationships are automatically generated. After completing the consistency verification of the information model, a NodeSet XML file conforming to the OPC UA standard is output, which can be directly imported into a standard OPC UA server to complete device modeling and publishing. The entire process requires no manual configuration and is compatible with multiple industrial protocols.

[0144] In this embodiment, to address the problems of inconsistent multi-protocol data structures, difficulty in stable identification of industrial data semantics, and reliance on manual configuration of industrial information models in existing industrial systems, a semantic-driven automatic construction scheme for OPC UA information models based on industrial point tables is proposed. This scheme does not simply perform format conversion or rule mapping on industrial data, but constructs a complete technical chain including standardized processing of industrial point tables, multi-constraint consistent semantic parsing, protocol-independent semantic graph modeling, and automatic generation of OPC UA information models. This achieves a step-by-step abstraction and unification of industrial data from underlying field expression to semantic structure expression and then to standard information model expression.

[0145] During the standardization phase of industrial point tables, field normalization and address abstraction modeling are used to transform heterogeneous data from different industrial protocols into unified structured industrial objects, thereby solving the problem that cross-protocol data cannot be directly compared and processed.

[0146] In the multi-constraint semantic parsing stage, a multi-dimensional constraint collaborative judgment mechanism is introduced, including name, data type, address structure, acquisition behavior, and context. By determining the semantic category of industrial data through multi-constraint consistency, the semantic misjudgment caused by the traditional reliance on single-field matching is avoided, thereby improving the determinism and stability of industrial semantic recognition.

[0147] In the protocol-independent semantic graph construction phase, a multi-type semantic relationship network is built based on semantic objects. Through a unified modeling approach of structural relationships, functional relationships, address associations, and behavioral associations, the influence of different industrial protocols on the data organization structure is eliminated, enabling industrial data to be expressed in the form of semantic graphs in a unified semantic space, thus achieving unified modeling of cross-protocol industrial data.

[0148] During the OPC UA information model generation stage, the OPC UA address space is automatically generated based on the semantic graph structure. Through semantic node mapping, reference relationship generation, and NodeSet XML file generation, the automatic conversion from semantic structure to standard industrial information model is realized, avoiding the problem of relying on manual configuration of nodes and relationships in the traditional method.

[0149] By constructing a semantically driven industrial data modeling system, industrial data processing is upgraded from the traditional protocol- or field rule-based processing method to a unified modeling method based on semantic consistency. This enables unified cross-protocol expression of industrial data and automatic generation of standardized information models, eliminating the dependence of traditional solutions on single protocol templates. New device types only need to be adapted by extracting basic fields to automatically integrate into the unified information model system. The generated OPC UA information model has both semantic accuracy and topological standardization, which can support upper-layer applications to intuitively understand device business and seamlessly connect to the industrial standard communication stack, providing a scalable technical foundation for unified access and centralized management of large-scale heterogeneous device groups.

[0150] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the information model construction method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0151] This application also provides an information model building apparatus, please refer to... Figure 7 The information model building apparatus includes: The point table standardization module 10 is used to receive multi-source heterogeneous industrial point tables, perform field parsing and normalization processing on the industrial point tables, and generate a standardized point table object with a unified structure. The semantic parsing module 20 is used to perform semantic parsing on the standardized point table object, extract the multi-dimensional features of each data point in the standardized point table object, and perform joint semantic parsing on the multi-dimensional features based on the multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The semantic modeling module 30 is used to convert the industrial semantic objects into semantic nodes, construct edges based on the association relationships between the industrial semantic objects, and generate an industrial semantic graph that is independent of industrial protocols. The model building module 40 is used to traverse the industrial semantic graph, map each semantic node to an OPC UA node, and build an OPC UA information model.

[0152] In one embodiment, the point table standardization module 10 is further configured to: The table fields of the industrial point table are parsed to extract basic industrial information; the basic industrial information includes at least variable name, variable description, data type, address information, access permissions, collection cycle, and equipment identifier; The data in different formats in the basic industrial information are uniformly parsed to obtain the original field set; The fields in the original field set are mapped to a pre-established standard field model to obtain a standardized field object set. Furthermore, based on a pre-built address abstraction model, the industrial addresses in the original field set are converted into a standard address object set. Based on a pre-built object model, the standardized field object set and the standardized address object set are integrated into a standardized industrial point table object; the standardized industrial point table object includes device identifier, variable name, data type, standard address object, access permission, and extended attributes.

[0153] In one embodiment, the semantic parsing module 20 is further configured to: Semantic parsing is performed on the standardized point table object to extract multidimensional features of each data point in the standardized point table object, resulting in a feature vector; the multidimensional features in the feature vector include at least name features, data type features, address features, behavior features, and context features; Based on the multi-constraint consistency determination mechanism, the feature vector is jointly semantically determined with multiple constraints. The industrial semantic category corresponding to the feature vector is determined according to the determination result, resulting in an industrial semantic object that includes equipment semantic objects, variable semantic objects, attribute semantic objects and address semantic objects.

[0154] In one embodiment, the semantic parsing module 20 is further configured to: If the determination result indicates that the feature vector satisfies all the constraints of the multiple constraints, then the industrial semantic category corresponding to the feature vector is determined according to the determination result of the multiple constraints. If the determination result indicates that the feature vector satisfies some of the multiple constraints, then a candidate semantic set is determined based on the target conditions satisfied by the feature vector among the multiple constraints. According to the preset scoring rules, each candidate semantic in the candidate semantic set is scored and sorted, and the industrial semantic category corresponding to the feature vector is determined according to the sorting order.

[0155] In one embodiment, the semantic modeling module 30 is further configured to: The industrial semantic object is converted into a semantic node; the node attribute information of the semantic node includes node identifier, node type, semantic tag, source device, and semantic feature vector; the node type includes device node, variable node, function node, and attribute node; Based on the associations between the industrial semantic objects, the semantic relationships between the corresponding semantic nodes are determined, and edges between the semantic nodes are constructed based on the semantic relationships to generate an initial semantic graph; the associations include device hierarchy relationships, composition relationships, reference relationships, address mapping relationships, and business relationships; the semantic relationships include structural relationships, functional relationships, address relationships, acquisition behavior relationships, and context behavior relationships; The initial semantic graph is subjected to protocol decoupling and structural optimization to obtain an initial industrial semantic graph. Protocol decoupling includes removing protocol-specific hierarchical fields, eliminating protocol tree structures, merging equivalent semantic nodes, and unifying semantic connection paths. The structural optimization strategy includes node redundancy removal, edge weight convergence, and a stable graph structure. Node redundancy removal involves merging semantic nodes with the same semantic meaning but different origins. Edge weight convergence represents the normalization of multiple associations of the same semantic node. The stable graph structure represents a global consistency adjustment of the semantic relationships. A unified cross-protocol mapping is performed on the initial industrial semantic graph, and the industrial protocol is marked as the data source attribute of the semantic node to obtain an industrial semantic graph that is independent of the industrial protocol.

[0156] In one embodiment, the model building module 40 is further configured to: Traverse the industrial semantic graph and map each semantic node to an OPC UA node; Based on the semantic relationships between the semantic nodes, the reference relationships between the OPC UA nodes are generated; The node attribute information of the semantic node is hashed to obtain the node identifier of the OPC UA node; Based on the OPC UA node, the reference relationship, and the node identifier, a node hierarchy is constructed and the reference relationship is bound to obtain the OPC UA information model.

[0157] In one embodiment, the model building module 40 is further configured to: The OPC UA information model is subjected to consistency verification. If the verification passes, the OPC UA address space is constructed based on the OPC UA information model. The OPC UA address space is serialized into a standard model file for model deployment on the OPC UA server.

[0158] The information model building apparatus provided in this application, employing the information model building method described in the above embodiments, can solve the technical problems of existing automatic information model building schemes having insufficient semantic expression capabilities for industrial data, and the high coupling between the modeling process and industrial communication protocols, resulting in poor versatility and generalization ability of the constructed information model. Compared with the prior art, the beneficial effects of the information model building apparatus provided in this application are the same as those of the information model building method provided in the above embodiments, and other technical features in the information model building apparatus are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0159] This application provides an information model building device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the information model building method in Embodiment 1 above.

[0160] The following is for reference. Figure 8The diagram illustrates a structural schematic suitable for implementing the information model building device of the embodiments of this application. The information model building device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 8 The information model building device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0161] like Figure 8 As shown, the information model building device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the information model building device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the information modeling device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows an information modeling device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented alternatively.

[0162] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0163] The information model building device provided in this application, employing the information model building method described in the above embodiments, can solve the technical problems of existing automatic information model building schemes having insufficient semantic expression capabilities for industrial data, and the high coupling between the modeling process and industrial communication protocols, resulting in poor versatility and generalization ability of the constructed information model. Compared with the prior art, the beneficial effects of the information model building device provided in this application are the same as those of the information model building method provided in the above embodiments, and other technical features in this information model building device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0164] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0165] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0166] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the information model construction method in the above embodiments.

[0167] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0168] The aforementioned computer-readable storage medium may be included in the information modeling device; or it may exist independently and not be assembled into the information modeling device.

[0169] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the information modeling device, cause the information modeling device to: Receive multi-source heterogeneous industrial point tables, perform field parsing and normalization on the industrial point tables, and generate a standardized point table object with a unified structure; Semantic parsing is performed on the standardized point table object to extract multi-dimensional features of each data point in the standardized point table object, and joint semantic parsing is performed on the multi-dimensional features based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The industrial semantic objects are converted into semantic nodes, and edges are constructed based on the relationships between the industrial semantic objects to generate an industrial semantic graph that is independent of industrial protocols. Traverse the industrial semantic graph and map each semantic node to an OPC UA node to construct an OPC UA information model.

[0170] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0171] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0172] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0173] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described information model construction method. This addresses the technical problems of existing automatic information model construction schemes lacking semantic expression capabilities for industrial data and having poor versatility and generalization ability due to the high coupling between the modeling process and industrial communication protocols. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the information model construction method provided in the above embodiments, and will not be repeated here.

[0174] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the information model construction method described above.

[0175] The computer program product provided in this application can solve the technical problems of existing automatic information model construction schemes having insufficient semantic expression capabilities for industrial data, and the high coupling between the modeling process and industrial communication protocols, resulting in poor versatility and generalization ability of the constructed information models. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the information model construction method provided in the above embodiments, and will not be repeated here.

[0176] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for constructing an information model, characterized in that, The information model construction method includes: Receive multi-source heterogeneous industrial point tables, perform field parsing and normalization on the industrial point tables, and generate a standardized point table object with a unified structure; Semantic parsing is performed on the standardized point table object to extract multi-dimensional features of each data point in the standardized point table object, and joint semantic parsing is performed on the multi-dimensional features based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The industrial semantic objects are converted into semantic nodes, and edges are constructed based on the relationships between the industrial semantic objects to generate an industrial semantic graph that is independent of industrial protocols. Traverse the industrial semantic graph and map each semantic node to an OPC UA node to construct an OPC UA information model.

2. The information model construction method as described in claim 1, characterized in that, The step of parsing and normalizing the industrial point table to generate a standardized point table object with a unified structure includes: The table fields of the industrial point table are parsed to extract basic industrial information; the basic industrial information includes at least variable name, variable description, data type, address information, access permissions, collection cycle, and equipment identifier; The data in different formats in the basic industrial information are uniformly parsed to obtain the original field set; The fields in the original field set are mapped to a pre-established standard field model to obtain a standardized field object set. Furthermore, based on a pre-built address abstraction model, the industrial addresses in the original field set are converted into a standard address object set. Based on a pre-built object model, the standardized field object set and the standardized address object set are integrated into a standardized industrial point table object; the standardized industrial point table object includes device identifier, variable name, data type, standard address object, access permission, and extended attributes.

3. The information model construction method as described in claim 1, characterized in that, The steps of performing semantic parsing on the standardized point table object, extracting multi-dimensional features of each data point in the standardized point table object, and performing joint semantic parsing on the multi-dimensional features based on a multi-constraint consistency determination mechanism to obtain an industrial semantic object include: Semantic parsing is performed on the standardized point table object to extract multidimensional features of each data point in the standardized point table object, resulting in a feature vector; the multidimensional features in the feature vector include at least name features, data type features, address features, behavior features, and context features; Based on the multi-constraint consistency determination mechanism, the feature vector is jointly semantically determined with multiple constraints. The industrial semantic category corresponding to the feature vector is determined according to the determination result, resulting in an industrial semantic object that includes equipment semantic objects, variable semantic objects, attribute semantic objects and address semantic objects.

4. The information model construction method as described in claim 3, characterized in that, The step of determining the industrial semantic category corresponding to the feature vector based on the determination result includes: If the determination result indicates that the feature vector satisfies all the constraints of the multiple constraints, then the industrial semantic category corresponding to the feature vector is determined according to the determination result of the multiple constraints. If the determination result indicates that the feature vector satisfies some of the multiple constraints, then a candidate semantic set is determined based on the target conditions satisfied by the feature vector among the multiple constraints. According to the preset scoring rules, each candidate semantic in the candidate semantic set is scored and sorted, and the industrial semantic category corresponding to the feature vector is determined according to the sorting order.

5. The information model construction method as described in claim 1, characterized in that, The step of converting the industrial semantic objects into semantic nodes, constructing edges based on the relationships between the industrial semantic objects, and generating an industrial semantic graph independent of industrial protocols includes: The industrial semantic object is converted into a semantic node; the node attribute information of the semantic node includes node identifier, node type, semantic tag, source device, and semantic feature vector; the node type includes device node, variable node, function node, and attribute node; Based on the associations between the industrial semantic objects, the semantic relationships between the corresponding semantic nodes are determined, and edges between the semantic nodes are constructed based on the semantic relationships to generate an initial semantic graph; the associations include device hierarchy relationships, composition relationships, reference relationships, address mapping relationships, and business relationships; the semantic relationships include structural relationships, functional relationships, address relationships, acquisition behavior relationships, and context behavior relationships; The initial semantic graph is subjected to protocol decoupling and structural optimization to obtain an initial industrial semantic graph. Protocol decoupling includes removing protocol-specific hierarchical fields, eliminating protocol tree structures, merging equivalent semantic nodes, and unifying semantic connection paths. The structural optimization strategy includes node redundancy removal, edge weight convergence, and a stable graph structure. Node redundancy removal involves merging semantic nodes with the same semantic meaning but different origins. Edge weight convergence represents the normalization of multiple associations of the same semantic node. The stable graph structure represents a global consistency adjustment of the semantic relationships. A unified cross-protocol mapping is performed on the initial industrial semantic graph, and the industrial protocol is marked as the data source attribute of the semantic node to obtain an industrial semantic graph that is independent of the industrial protocol.

6. The information model construction method as described in claim 1, characterized in that, The step of traversing the industrial semantic graph, mapping each semantic node to an OPC UA node, and constructing the OPC UA information model includes: Traverse the industrial semantic graph and map each semantic node to an OPC UA node; Based on the semantic relationships between the semantic nodes, the reference relationships between the OPC UA nodes are generated; The node attribute information of the semantic node is hashed to obtain the node identifier of the OPC UA node; Based on the OPC UA node, the reference relationship, and the node identifier, a node hierarchy is constructed and the reference relationship is bound to obtain the OPC UA information model.

7. The information model construction method as described in claim 6, characterized in that, After the step of traversing the industrial semantic graph, mapping each semantic node to an OPC UA node, and constructing the OPC UA information model, the method further includes: The OPC UA information model is subjected to consistency verification. If the verification passes, the OPC UA address space is constructed based on the OPC UA information model. The OPC UA address space is serialized into a standard model file for model deployment on the OPC UA server.

8. An information model building apparatus, characterized in that, The information model building device includes: The point table standardization module is used to receive multi-source heterogeneous industrial point tables, perform field parsing and normalization processing on the industrial point tables, and generate standardized point table objects with a unified structure. The semantic parsing module is used to perform semantic parsing on the standardized point table object, extract the multi-dimensional features of each data point in the standardized point table object, and perform joint semantic parsing on the multi-dimensional features based on a multi-constraint consistency judgment mechanism to obtain an industrial semantic object. The semantic modeling module is used to convert the industrial semantic objects into semantic nodes, construct edges based on the association relationships between the industrial semantic objects, and generate an industrial semantic graph that is independent of industrial protocols. The model building module is used to traverse the industrial semantic graph, map each semantic node to an OPC UA node, and build an OPC UA information model.

9. An information model building device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the information model construction method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the information model construction method as described in any one of claims 1 to 7.

Images