Stainless steel product intelligent factory model construction method based on ontology

Abstract

The invention belongs to the field of intelligent manufacturing application, and relates to a stainless steel product intelligent factory model construction method based on ontology, which comprises the following steps: collecting and preprocessing heterogeneous data sources of a stainless steel product intelligent factory; concepts in the field of the stainless steel product intelligent factory are extracted from the preprocessed data, and a trunk concept ontology and a plurality of branch ontologies associated with the trunk concept ontology are constructed; constructing an ontology model based on the trunk concept ontology and the branch ontology; and carrying out iterative optimization on the ontology model. Through construction of the trunk concept ontology and the branch ontology, information from multiple data sources can be efficiently managed and integrated, and the problems that in the field of stainless steel product intelligent factories, a concept set is huge and complex, the number of core concepts is small, the similarity between the concepts is low, and the hierarchical relation is not complex are solved; the ontology model is integrated into an information system of a stainless steel product intelligent factory, data consistency and sharing are achieved, and intelligent decision support is provided.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent manufacturing applications, and relates to a method for constructing an intelligent factory model of stainless steel products based on ontology. Background Art

[0002] With the rapid development of information technology, the manufacturing industry is undergoing unprecedented changes. As the core component of Industry 4.0, the concept of an intelligent factory is to achieve the intelligence, automation, and networking of the production process by integrating advanced technologies such as the Internet of Things, artificial intelligence, and big data analysis. The main feature of an intelligent factory is the need to manage multiple data sources simultaneously. The goal of an intelligent factory is to improve production efficiency, reduce costs, enhance product quality, and achieve a flexible production process to meet the personalized and customized market demands.

[0003] In the construction of an intelligent factory, the integration and application of information technology are the key. However, traditional information systems often suffer from problems such as data islands, information inconsistency, and insufficient decision support. These problems lead to a waste of information resources and affect the improvement of production efficiency and product quality. To solve these problems, a new technical method is needed to construct a unified knowledge model to achieve information integration and intelligent decision support.

[0004] Currently, the application of ontology in the field of intelligent manufacturing. As a knowledge representation method in the field of artificial intelligence, ontology has shown its powerful knowledge organization and information integration capabilities in multiple fields. By defining concepts, attributes, relationships, etc. within a domain, ontology provides a standardized framework for information expression and reasoning. In the context of an intelligent factory, the application of ontology can achieve the unified representation and intelligent processing of knowledge in various aspects such as production, management, and maintenance.

[0005] It is still in its infancy. Although there have been some research and practical explorations on the application of ontology in intelligent factories, these studies are often limited to specific application scenarios and lack a general ontology construction method and framework. In addition, the implementation and application of ontology also face technical challenges such as ontology design, editing, reasoning, and integration.

[0006] Such as Figure 1 shown is the conventional workflow of a stainless steel products factory. In the field of intelligent factories for stainless steel products, traditional knowledge management systems face the following technical problems: Diversity of data sources: An intelligent factory for stainless steel products needs to process information from multiple data sources such as sensors, equipment logs, and operation records, and there are differences in format and semantics among these data sources.

[0007] Complexity of the concept set: The concept set involved in the field of intelligent factories for stainless steel products is huge and complex, and traditional ontology construction methods are difficult to handle effectively.

[0008] Identification of core concepts: In the huge concept set, identifying the most core concepts is the key to constructing an effective ontology, but this process is often time-consuming and error-prone.

[0009] Low similarity between concepts: The similarity between concepts in the field of intelligent factories for stainless steel products is relatively low, which increases the difficulty of concept integration and reasoning.

[0010] Uncomplicated hierarchical relationship: The hierarchical relationship of concepts in the field of intelligent factories for stainless steel products is relatively simple, but traditional ontology construction methods are often too complex and not suitable for the characteristics of this field. Summary of the Invention

[0011] To solve the above technical problems existing in the prior art, the present invention proposes a method for constructing an ontology-based intelligent factory model for stainless steel products, and its specific technical solutions are as follows: A method for constructing an ontology-based intelligent factory model for stainless steel products, comprising: Collecting heterogeneous data sources of the intelligent factory for stainless steel products and performing preprocessing; Extracting concepts in the field of intelligent factories for stainless steel products from the preprocessed data, constructing a backbone concept ontology, and a plurality of branch ontologies associated with the backbone concept ontology; Constructing an ontology model based on the backbone concept ontology and the branch ontologies; Iteratively optimizing the ontology model.

[0012] Furthermore, through expert knowledge, literature review, and data analysis, identifying the most core concepts in the field of intelligent factories for stainless steel products, constructing a backbone concept ontology, adopting modular design, each backbone concept ontology is independently encapsulated into a microservice architecture to support on-demand expansion, and the backbone concept ontology includes an intelligent logistics system, an intelligent production system, and an intelligent management system.

[0013] Furthermore, based on the backbone concept ontology, constructing a plurality of branch ontologies associated with the bottom-layer core concepts of the backbone concept ontology. The branch ontology of the intelligent logistics system includes an AGV automatic guidance system, a product sales system, a raw material procurement system, an intelligent stereoscopic warehouse, and a distribution system; the branch ontology of the intelligent production system includes an alarm device, an intelligent work station, an intelligent coding system, a detection device, an automated processing device, an equipment operation and maintenance system, and an anti-error system; the branch ontology of the intelligent management system includes a safety lighting system, a data monitoring system, an intelligent control system, a production execution management system, a production monitoring management system, a production planning management system, an energy consumption management system, an order management system, an operation and maintenance management system, and a panel system.

[0014] Furthermore, based on the above-mentioned backbone concept ontology and branch ontology, use Protege software to build an ontology model that includes all relevant concepts and entities in the field of intelligent factories for stainless steel products, and introduce a digital twin interface branch ontology to realize the real-time mapping between the physical factory and the virtual model; at the same time, the Pellet reasoner and SWRL rule engine are integrated in the Protege, logical rules are defined, and the ontology model is constrained and verified using SHACL.

[0015] Furthermore, apply graph theory and machine learning techniques to analyze the relationships between concepts and entities in the ontology model, identify key semantic associations and potential reasoning paths; use the owl:equivalentProperty and owl:sameAs properties of OWL to establish cross-ontology equivalent relationships and perform dynamic associations.

[0016] Furthermore, build a dynamic concept recognition model based on graph neural networks to monitor the changes in the production environment of the intelligent factory for stainless steel products in real time, automatically update the core concept library, and derive new conclusions or information from the existing knowledge using logical rules and algorithms.

[0017] Furthermore, adopt a distributed data acquisition framework to perform real-time or batch data acquisition on the heterogeneous data sources of the intelligent factory for stainless steel products, and the heterogeneous data sources include sensor time series data, device logs, ERP system structured data, and operator text records.

[0018] Furthermore, preprocess the collected data, including data cleaning, format unification, and semantic annotation. Among them, a hybrid cleaning strategy combining a rule engine and machine learning is introduced in the data cleaning stage, and ISO 8000 international general industrial standards are adopted for format and semantic unification.

[0019] Furthermore, extract the core terms of the stainless steel product industry from the historical document data through natural language processing technology to form a domain dictionary of concepts.

[0020] Furthermore, eliminate semantic ambiguities across data sources through an ontology alignment tool to unify the entity representations in heterogeneous data sources.

[0021] The present invention has the following advantages: Efficient data integration: Through the construction of the backbone concept ontology and branch ontology, information from multiple data sources can be efficiently integrated; the ontology model is integrated into the information system of the intelligent factory for stainless steel products to achieve data consistency and sharing, and provide intelligent decision-making support.

[0022] Simplified concept management: By identifying and defining core concepts, the management of the vast and complex set of concepts in the field of intelligent factories for stainless steel products is simplified; key concepts are extracted from requirements, and the relationships between these concepts are defined to form a structured ontology model; ontology description languages such as OWL are used to achieve a standardized representation of the ontology, ensuring the interoperability and scalability of the ontology.

[0023] Enhanced semantic reasoning ability: Based on the semantic reasoning process of the ontology model and relational network analysis, new conclusions or information can be derived from existing knowledge, enhancing the decision-making support ability of the ontology model for intelligent factories of stainless steel products.

[0024] Strong practicality: The ontology model of the present invention can adapt to the characteristics of low similarity between concepts and uncomplicated hierarchical relationships in the field of intelligent factories for stainless steel products, having strong adaptability. At the same time, by deeply analyzing the requirements of intelligent factories for stainless steel products, the goals and scope of ontology construction are determined to ensure the practicality and pertinence of the ontology. Description of the Drawings

[0025] Figure 1 is the conventional workflow diagram of a stainless steel product factory; Figure 2 is the flowchart for constructing the ontology model for the intelligent factory industry of stainless steel products in this embodiment; Figure 3 is the entity relationship restriction diagram for the intelligent factory industry of stainless steel products in this embodiment; Figure 4 is the overall ontology model diagram for the intelligent factory industry of stainless steel products in this embodiment. Detailed Implementation Manner

[0026] In order to make the purpose, technical solutions, and technical effects of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings of the specification and embodiments.

[0027] The present invention discloses a method for constructing an ontology-based intelligent factory model for stainless steel products, aiming to construct a unified knowledge model through ontology technology to achieve information integration and intelligent decision-making support for each link of the intelligent factory for stainless steel products, which is mainly achieved through the following aspects: Definition of the backbone concept ontology: Heterogeneous data sources of the intelligent factory for stainless steel products are collected, and one or more backbone concept ontologies are defined, which are used as the core of the intelligent factory ontology model for integrating and managing the most basic core concepts.

[0028] Construction of branch ontologies: Based on the backbone concept ontology, multiple branch ontologies are constructed, and the branch ontologies are associated with the most basic core concepts of the backbone concept ontology for processing data and information in specific fields.

[0029] Construction of the ontology model: Using ontology-based knowledge graph technology, construct an ontology model that includes all relevant concepts and entities in the field of intelligent factories for stainless steel products. This model can clearly display the semantic relationships between concepts.

[0030] Relationship network analysis: Through relationship network analysis technology, identify and analyze the complex relationships between concepts and entities in the field of intelligent factories for stainless steel products, providing a basis for semantic reasoning.

[0031] Semantic reasoning process: Based on the semantic associations and logical rules of the ontology model, develop a semantic reasoning process that can derive new conclusions or information from existing knowledge.

[0032] Optimization of the ontology construction method: Analyze the problems and difficulties faced by existing ontology construction methods, find the reasons for the problems, and optimize the construction process on the basis of the original construction method to construct an intelligent factory ontology model that does not have the above problems or solves some of the problems.

[0033] More specifically, as Figure 2 shown, it includes the following steps: The first step: For the heterogeneous data sources of the intelligent factory for stainless steel products, adopt a distributed data acquisition framework, such as Apache NiFi, for real-time or batch scraping. The heterogeneous data sources include sensor time-series data, device logs, ERP system structured data, operator text records, etc.

[0034] The second step: Preprocess various data sources from the intelligent factory for stainless steel products, including steps such as data cleaning, unification of format and semantics, and semantic annotation, to ensure the quality and consistency of the data. In the data cleaning stage, introduce a hybrid cleaning strategy that combines a rule engine and machine learning. Define domain-specific cleaning rules, such as: removing sensor outliers and fixing incorrect timestamp formats. Data standardization adopts international general industrial standards such as ISO 8000 to ensure the unity of format and semantics.

[0035] The third step: Through expert knowledge, literature review, and data analysis, identify the most core concepts in the field of intelligent factories for stainless steel products and construct the backbone concept ontology. This step is the basis of the entire ontology model. It is necessary to find out the classes contained in the intelligent factory industry for stainless steel products, and then find object and data attributes for classification to ensure the accuracy and comprehensiveness of the core concepts. As Figure 3 shown, in this embodiment, three backbone concept ontologies are selected through classification, namely the intelligent logistics system, the intelligent production system, and the intelligent management system.

[0036] Step 4: Extract the core terms of the stainless steel products industry from historical documents through natural language processing technology to form a domain dictionary containing concepts such as "stamping equipment" and "vacuum heat treatment process". The natural language processing technology can adopt the BERT pre-trained model.

[0037] Step 5: Eliminate semantic ambiguities across data sources through an ontology alignment tool, such as unifying "AGV cart" and "automated guided vehicle" into the same entity. The ontology alignment tool is, for example, LogMap.

[0038] Step 6: Based on the backbone concept ontology, construct a branch ontology associated with a specific domain. The branch ontology contains more specific sub-concepts and entities related to the backbone concept. Continuing to refer to Figure 3 , for the backbone ontology of the intelligent logistics system, the branch ontologies include the AGV automated guidance system, product sales system, raw material procurement system, intelligent stereoscopic warehouse, and distribution system; for the intelligent production system, the branch ontologies include alarm devices, intelligent workstations, intelligent coding systems, detection devices, automated processing devices, equipment operation and maintenance systems, and error-proofing systems. Among them, the automated processing devices include stamping equipment, shearing equipment, spot welding machines, vacuum heat treatment furnaces, and surface treatment equipment; for the intelligent management system, the branch ontologies include safety lighting systems, data monitoring systems, intelligent control systems, production execution management systems, production monitoring management systems, production planning management systems, energy consumption management systems, order management systems, operation and maintenance management systems, and panel systems.

[0039] Step 7: Use the Protege software to construct an ontology model containing all relevant concepts and entities in the field of intelligent factories for stainless steel products, as Figure 4 shown. Refine the "error-proofing system" in the intelligent production system into "process error-proofing" and "equipment error-proofing" sub-branches. The "process error-proofing" branch is associated with process parameter thresholds, such as the punching force range, and the "equipment error-proofing" branch integrates equipment self-checking protocols, such as vibration sensor anomaly detection. Introduce a "digital twin interface" branch ontology into the intelligent management system to achieve real-time mapping between the physical factory and the virtual model and support simulation optimization. Through continuous iteration and optimization, ensure the accuracy and integrity of the ontology model.

[0040] Step 8: Integrate the Pellet inference engine and the SWRL rule engine in Protege to support complex logical reasoning. For example, define the rule: Copy Device(?d) ∧ hasStatus(?d, "fault") → triggerMaintenanceRequest(?d), and use SHACL (Shapes Constraint Language) to perform constraint verification on the ontology model to ensure data integrity. For example, it is stipulated that the "vacuum heat treatment furnace" must be associated with temperature sensor data.

[0041] Step 9: Apply graph theory and machine learning techniques to deeply analyze the relationships between concepts and entities in the ontology model, and identify key semantic associations and potential reasoning paths. The backbone concept ontology adopts a modular design, and each backbone concept ontology is independently encapsulated into a microservice architecture to support on-demand expansion. For example, the "intelligent production system" ontology accesses edge computing nodes through APIs and receives device status data in real time.

[0042] Step 10: Use the owl:equivalentProperty and owl:sameAs properties of OWL to establish cross-ontology equivalent relationships. For example, dynamically associate the inventory quantity property of the "intelligent stereoscopic warehouse" with the material requirement property of the "production plan management system".

[0043] Step 11: Build an automated test environment based on Robot Framework to simulate factory scenarios such as a surge in orders and equipment failures, and verify the inference response speed and accuracy of the ontology model. Evaluate the logical branch coverage of the ontology model through a coverage analysis tool to ensure no redundancy or omission. The coverage analysis tool is, for example, JaCoCo.

[0044] Step 12: Develop a semantic reasoning process based on the results of the ontology model and relationship network analysis. Build a dynamic concept recognition model based on a graph neural network to monitor changes in the production environment in real time, such as the addition of new equipment models, and automatically update the core concept library. This process will use logical rules and algorithms to derive new conclusions or information from existing knowledge.

[0045] Step 13: Verify the effectiveness and accuracy of the ontology model through actual cases and experiments. Continuously iterate and optimize the ontology model according to feedback and results to adapt to the changes and developments in the intelligent factory industry for stainless steel products.

[0046] In summary, the present invention deeply studies ontology-based knowledge graphs and relationship network analysis, and proficiently masters the semantic reasoning process of using semantic associations and logical rules based on knowledge graphs to derive new conclusions or information. Analyze the problems and difficulties faced by existing ontology construction methods, find the reasons for the problems, and thus build an ontology model for the intelligent factory industry for stainless steel products that does not have the above problems or solves some of the problems on the basis of the original construction method. As a result, this model can efficiently manage and integrate information from multiple data sources, and at the same time solve the problems of a large and complex concept set, a small number of core concepts, low similarity between concepts, and not complex hierarchical relationships in the field of intelligent factories for stainless steel products.

[0047] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the implementation process of the present invention has been described in detail above, those skilled in the art can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for constructing an intelligent factory model of stainless steel products based on ontology, characterized in that, Including: Collecting and preprocessing heterogeneous data sources of the intelligent factory for stainless steel products; Extracting concepts in the field of the intelligent factory for stainless steel products from the preprocessed data, constructing a backbone concept ontology, and multiple branch ontologies associated with the backbone concept ontology; Constructing an ontology model based on the backbone concept ontology and the branch ontologies; Iteratively optimizing the ontology model.

2. The method for constructing an intelligent factory model of stainless steel products according to claim 1, characterized in that Identifying the most core concepts in the field of the intelligent factory for stainless steel products through expert knowledge, literature review, and data analysis, constructing a backbone concept ontology, which adopts a modular design, and each backbone concept ontology is independently encapsulated as a microservice architecture, supporting on-demand expansion. The backbone concept ontology includes an intelligent logistics system, an intelligent production system, and an intelligent management system.

3. The method for constructing an intelligent factory model of stainless steel products according to claim 2, wherein Based on the backbone concept ontology, constructing multiple branch ontologies associated with the bottom-layer core concepts of the backbone concept ontology. The branch ontology of the intelligent logistics system includes an AGV automatic guiding system, a product sales system, a raw material procurement system, an intelligent stereoscopic warehouse, and a distribution system; the branch ontology of the intelligent production system includes an alarm device, an intelligent workstation, an intelligent coding system, a detection device, an automated processing device, an equipment operation and maintenance system, and an anti-error system; the branch ontology of the intelligent management system includes a safety lighting system, a data monitoring system, an intelligent control system, a production execution management system, a production monitoring management system, a production planning management system, an energy consumption management system, an order management system, an operation and maintenance management system, and a panel system.

4. The method for constructing an intelligent factory model of stainless steel products according to claim 3, characterized in that, Based on the above-mentioned backbone concept ontology and branch ontologies, using Protege software to construct an ontology model containing all relevant concepts and entities in the field of the intelligent factory for stainless steel products, and introducing a digital twin interface branch ontology to realize real-time mapping between the physical factory and the virtual model; meanwhile, a Pellet reasoner and an SWRL rule engine are integrated in Protege to define logical rules and use SHACL to perform constraint verification on the ontology model.

5. The method for constructing an intelligent factory model of stainless steel products according to claim 4, wherein, Applying graph theory and machine learning techniques to analyze the relationships between concepts and entities in the ontology model, identifying key semantic associations and potential reasoning paths; using the owl:equivalentProperty and owl:sameAs properties of OWL to establish cross-ontology equivalent relationships and perform dynamic associations.

6. The method for constructing an intelligent factory model of stainless steel products according to claim 5, wherein Constructing a dynamic concept recognition model based on a graph neural network to monitor the changes in the production environment of the intelligent factory for stainless steel products in real time, automatically updating the core concept library, and deriving new conclusions or information from the existing knowledge using logical rules and algorithms.

7. The method for constructing an intelligent factory model of stainless steel products according to claim 1, characterized in that, Adopting a distributed data collection framework to perform real-time or batch data collection on the heterogeneous data sources of the intelligent factory for stainless steel products, and the heterogeneous data sources include sensor time-series data, equipment logs, ERP system structured data, and operator text records.

8. The method for constructing an intelligent factory model of stainless steel products according to claim 1, wherein, Preprocessing the collected data, including data cleaning, format unification, and semantic annotation. Among them, a hybrid cleaning strategy combining a rule engine and machine learning is introduced in the data cleaning stage, and ISO 8000 international general industrial standards are adopted for format and semantic unification.

9. The method for constructing an intelligent factory model of stainless steel products according to claim 1, wherein, Extract the core terms in the historical document data of the stainless steel products industry through natural language processing technology to form a domain dictionary of concepts.

10. The method for constructing an intelligent factory model of stainless steel products according to claim 1, wherein Eliminate semantic ambiguities across data sources through an ontology alignment tool to unify the entity representations in heterogeneous data sources.

Images