An intelligent verification system and method for industrial safety measures based on knowledge graphs and big data.

Abstract

This invention discloses an intelligent verification system and method for industrial safety measures based on knowledge graphs and big data. It includes a multi-source heterogeneous data standardization and structuring module, an industrial safety measures domain knowledge graph construction module, an AI-based intelligent verification engine for safety measures, and an intelligent auxiliary interaction module. The multi-source heterogeneous data standardization and structuring module collects historical work orders, operation tickets, and equipment ledger data, and uses natural language processing technology to extract key entity information such as work tasks, work locations, involved equipment, isolation points, and isolation measures. The industrial safety measures domain knowledge graph construction module is connected to the multi-source heterogeneous data standardization and structuring module. This intelligent verification system and method for industrial safety measures based on knowledge graphs and big data constructs a knowledge graph that integrates historical data and domain knowledge, and uses an AI engine for semantic reasoning to achieve automatic verification and intelligent assistance of safety measures in newly created documents.

Description

Technical Field

[0001] This invention relates to the field of industrial safety production management and intelligent operation and maintenance technology, specifically to an intelligent verification system and method for industrial safety measures based on knowledge graphs and big data. Background Technology

[0002] In complex industrial sectors such as power, chemical, and energy, work permits and operation permits, collectively known as "two permits," are core systems for ensuring on-site operational safety. Their core lies in the formulation and verification of correct safety measures to ensure reliable isolation of operational energy. However, existing methods for formulating and verifying these measures have significant flaws. First, they rely excessively on human experience. The issuance and verification process heavily depends on the operator's and issuer's personal experience and memory. Faced with complex equipment topologies, intertwined system connections, and changing on-site conditions, human intervention is highly prone to overlooking critical isolation points (such as bypass valves or secondary power supplies) or selecting incorrect isolation methods, creating potential safety hazards. Second, the value of massive amounts of historical data remains untapped. Industrial sites have accumulated a vast amount of historical two permit data, but currently... The data is only used for archiving and has not been transformed into structured knowledge that can guide current operations. Valuable operational experience is lost with personnel turnover and cannot be effectively passed on. Furthermore, the automated verification capability is insufficient. Existing five-prevention systems or electrical interlocking devices are mostly based on simple hard logic rules and lack the ability to understand the complex semantic relationship between work tasks, equipment status, and safety measures. They cannot judge the completeness and logical rationality of safety measures, especially when facing atypical operations or special working conditions. Finally, the audit efficiency is low. During periods of intensive tasks such as major overhauls and technical upgrades, auditors need to manually check each safety measure, which is time-consuming, labor-intensive, and difficult to standardize, becoming a bottleneck affecting maintenance efficiency. To address this, we propose an intelligent industrial safety measure verification system and method based on knowledge graphs and big data. Summary of the Invention

[0003] The technical problem to be solved by this invention is to overcome the existing defects and provide an intelligent verification system and method for industrial security measures based on knowledge graphs and big data. By constructing a knowledge graph that integrates historical data and domain knowledge, and using an AI engine for semantic reasoning, the system can automatically verify and intelligently assist the security measures in newly created invoices, which can effectively solve the problems in the background technology.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent verification system and method for industrial safety measures based on knowledge graphs and big data, including a multi-source heterogeneous data standardization and structuring module, an industrial safety measures knowledge graph construction module, an AI-based intelligent verification engine for safety measures, and an intelligent auxiliary interaction module; Multi-source heterogeneous data standardization and structuring module: used to collect historical work orders, operation tickets and equipment ledger data, and use natural language processing technology to extract key entity information such as work tasks, work locations, equipment involved, isolation points and isolation measures; The Industrial Safety Measures Knowledge Graph Construction Module connects with the Multi-Source Heterogeneous Data Standardization and Structure Module. It is used to define and construct an ontology model with equipment, operation, hazard source, isolation point and safety measures as the core concepts based on the extracted entity information, and establish semantic relationship between entities to form an industrial safety measures knowledge graph. The industrial safety knowledge graph also integrates a reasoning rule base based on industry safety regulations and expert experience. The AI-based safety measure intelligent verification engine connects to the knowledge graph construction module in the industrial safety measure field. It is used to receive newly created work tasks and proposed safety measure information, analyze the semantics of the work through context awareness, and perform retrieval, reasoning and comparison in the knowledge graph to perform integrity and rationality verification of the safety measures. Intelligent Assisted Interaction Module: Connects to the AI-based security measure intelligent verification engine to provide users with security measure missing warnings, redundant item prompts, and security measure recommendations based on the verification results. By constructing a knowledge graph that integrates historical data and domain knowledge, and using the AI ​​engine for semantic reasoning, it realizes automatic verification and intelligent assistance of security measures in newly created invoices.

[0005] Furthermore, the multi-source heterogeneous data includes unstructured historical work orders and operation tickets text data, structured equipment ledger database information, and process flow documents, realizing the diversity of data sources and ensuring that the data foundation for knowledge graph construction includes both experiential documents and objective equipment information, thereby enhancing the comprehensiveness and accuracy of knowledge.

[0006] Furthermore, the entity attributes extracted from unstructured text by the natural language processing technology include at least: isolation point name, isolation point type, isolation measure type, and isolation method. By limiting the specific attributes extracted by NLP, the accuracy and practicality of information extraction from unstructured text are ensured, laying the foundation for building a high-quality knowledge graph.

[0007] Furthermore, the relationships between entities in the industrial safety measures knowledge graph include: operation-involved-equipment, equipment-associated isolation point-isolation point, operation-must-take-safety measures, and topological connections between equipment. By clarifying the core semantic relationships in the knowledge graph, the system can understand the complex relationships between operations, equipment, and safety measures, which is a prerequisite for realizing semantic intelligent verification.

[0008] Furthermore, the integrity verification performed by the verification engine specifically involves: retrieving a set of standard safety measures for similar historical operations in the knowledge graph, comparing the current proposed safety measures with these, and identifying the absence of key isolation points. The rationality verification specifically involves: analyzing whether the isolation points in the proposed safety measures can effectively cut off the energy source based on the device topology and logical relationships in the knowledge graph, and detecting logical contradictions or excessive defenses. It can discover oversights and contradictions that are difficult for humans to detect from two dimensions: the completeness of historical experience and the correctness of on-site logic, thus greatly improving the depth of verification.

[0009] Furthermore, it also includes a safety isolation and interlocking system interface module, which is used to convert the verified and finalized safety measures into a standard instruction format and send it to the on-site safety isolation and interlocking system as the logical basis for on-site operation and interlocking. By adding an interface with the on-site interlocking system, seamless information transmission from plan formulation to on-site execution is achieved, forming a complete safety chain of online intelligent verification and offline reliable interlocking, thereby improving the system's practicality.

[0010] Furthermore, the safety recommendations provided by the intelligent auxiliary interaction module are based on knowledge graph path analysis, outputting a set of isolation points that appear most frequently and have the most comprehensive coverage for the current task. This allows for the rapid delivery of historical best practices to users, significantly reducing the difficulty of invoicing, improving efficiency, and promoting standardized operations.

[0011] Furthermore, the knowledge graph construction module supports incremental learning, which can continuously incorporate newly added historical invoice data into the graph construction and update process, realize dynamic optimization of verification rules, enable the system to adapt to changes such as equipment updates and process transformations, ensure the continuous effectiveness of verification capabilities, and have long-term vitality.

[0012] An intelligent verification method for industrial safety measures based on knowledge graphs and big data includes the following steps: Step 1: Receive the safety measures information proposed by the user for the newly created job task.

[0013] Step 2: Parse the semantics of the task and locate the associated equipment and historical task entities in the industrial safety measures knowledge graph.

[0014] Step 3: Perform integrity verification: Retrieve the set of historical standard safety measures corresponding to the task from the knowledge graph, compare the user-proposed safety measures with the set of standard safety measures, and identify the missing key isolation points.

[0015] Step 4: Perform a rationality check: Based on the device topology relationships and pre-set reasoning rules in the knowledge graph, analyze whether the isolation point settings in the proposed security measures can effectively cut off the energy source, and detect logical contradictions or redundant measures.

[0016] Step 5: Based on the verification results of Step 3 and Step 4, generate intelligent auxiliary suggestions that include safety measure missing warnings, redundancy prompts, and safety measure recommendations.

[0017] Step Six: Provide the intelligent assistance suggestions to the user and receive the user's feedback on the security measures based on the suggestions.

[0018] Step 7: Output the final confirmed safety measures plan to guide on-site operations or distribute it to the safety isolation and interlocking system.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: The intelligent verification system and method for industrial safety measures based on knowledge graphs and big data have the following advantages: 1. Through big data analysis and knowledge graph reasoning, hidden logical errors and security omissions can be discovered, thus forming a dual guarantee of human and technical defenses, effectively preventing malicious misoperations.

[0020] 2. It can transform historical data into a visual and usable knowledge graph, making tacit experience explicit, reducing reliance on the experience of individual experts, and thus helping enterprises to accumulate knowledge and train new employees.

[0021] 3. Intelligent verification can achieve millisecond-level response and provide auxiliary recommendations, which greatly shortens the time of invoice issuance and review, and its benefits are particularly significant during periods of heavy workload.

[0022] 4. The system can continuously learn and evolve as new data is generated, enabling the verification rules to automatically adapt to new equipment and processes and maintain long-term effectiveness. Attached Figure Description

[0023] Figure 1 This is a block diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart of the intelligent verification method of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Please see Figure 1-2 This embodiment provides a technical solution: an intelligent verification system and method for industrial safety measures based on knowledge graphs and big data, including a multi-source heterogeneous data standardization and structuring module, an industrial safety measures knowledge graph construction module, an AI-based intelligent verification engine for safety measures, and an intelligent auxiliary interaction module; A multi-source heterogeneous data standardization and structuring module is used to collect historical work orders, operation tickets, and equipment ledger data. Natural language processing (NLP) technology is used to extract key entity information such as work tasks, work locations, involved equipment, isolation points, and isolation measures. This multi-source heterogeneous data includes unstructured historical work order and operation ticket text data, structured equipment ledger database information, and process flow documents. This ensures the diversity of data sources and guarantees that the data foundation for knowledge graph construction includes both experiential documents and objective equipment information, enhancing the comprehensiveness and accuracy of knowledge. The entity attributes extracted from unstructured text using NLP technology include at least the following: The isolation point name, isolation point type, isolation measure type, and isolation method, by limiting the specific attributes extracted by NLP, ensure the accuracy and practicality of information extraction from unstructured text, laying the foundation for building a high-quality knowledge graph. The standardization and structuring module of multi-source heterogeneous data is the foundation of the entire system. During implementation, it is necessary to establish data interfaces with production management systems (such as two-ticket systems, ERP) and equipment management systems. The NLP model can use pre-trained models (such as BERT) for domain fine-tuning to train entity recognition (NER) tasks, accurately extracting fields such as isolation objects and isolation measures from the ticket. For structured data such as equipment ledgers, direct mapping and association are performed. The industrial safety measures knowledge graph construction module connects with the multi-source heterogeneous data standardization and structuring module. Based on extracted entity information, it defines and constructs an ontology model with equipment, operations, hazards, isolation points, and safety measures as core concepts, and establishes semantic relationships between entities to form an industrial safety measures knowledge graph. This knowledge graph also integrates a reasoning rule base based on industry safety regulations and expert experience. The relationships between entities in the industrial safety measures knowledge graph include: operations-involved-equipment, equipment-associated isolation points-isolation points, operations-must-take-safety measures, and topological connections between equipment. Equipment entity attributes may include ID, name, type, and system. Isolation point attributes may include type (e.g., valve, switch), status, and associated hazard type. Relationship extraction can be achieved through rule matching (…). The system combines co-occurrence analysis and relational classification models. The reasoning rule base is stored in IF-THEN format. By clarifying the core semantic relationships in the knowledge graph, the system can understand the complex relationships between operations, equipment, and safety measures. This is a prerequisite for semantic intelligent verification. The knowledge graph construction module supports incremental learning and can continuously incorporate newly added historical ticket data into the graph construction and update process, realizing dynamic optimization of verification rules. The system establishes a data buffer and a new data annotation process. Daily newly added ticket data that has been verified as safe and valid on-site is automatically or semi-automatically sent to the data standardization module after review, triggering incremental updates of the graph. This keeps the knowledge base up-to-date and enables the system to adapt to changes such as equipment updates and process modifications, ensuring that the verification capability remains effective and has long-term viability. The AI-based intelligent safety measure verification engine connects to a knowledge graph construction module in the industrial safety measure field. It receives newly created work tasks and proposed safety measure information, analyzes the semantics of the work through context awareness, and performs retrieval, reasoning, and comparison within the knowledge graph. It then performs integrity and rationality checks on the safety measures. The integrity check specifically involves: retrieving a set of standard safety measures for similar historical work from the knowledge graph, comparing the currently proposed safety measures with these sets, and identifying missing key isolation points. The rationality check involves: analyzing the equipment topology and logical relationships in the knowledge graph, determining whether the isolation points in the proposed safety measures can effectively cut off the energy source, and detecting logical contradictions or excessive defense. The integrity check process involves the intelligent safety measure verification engine parsing the user-inputted #2 boiler induced draft fan maintenance task. In the graph, locate the #2 boiler induced draft fan equipment node, and find all similar maintenance tasks and their safety measures along the equipment-related-historical operation path. Generate a set of standard safety measures, compare the user's proposed safety measures with it, calculate the matching degree, and the rationality verification process: The safety measure intelligent verification engine calls the topological relationship in the graph such as induced draft fan-driven by motor, motor-power supply from switch cabinet K1, etc. If the user's safety measures only stop the power supply of the induced draft fan body, but do not stop the power supply of the motor heater, the safety measure intelligent verification engine will judge that the energy source is not completely isolated according to the topology and issue a warning. If the user adds irrelevant operations such as closing adjacent water pump valves, it is marked as redundant. It can discover omissions and contradictions that are difficult to be detected by humans from two dimensions: the completeness of historical experience and the correctness of on-site logic, which greatly improves the verification depth. Intelligent Assisted Interaction Module: Connected to the AI-based intelligent safety measure verification engine, this module provides users with safety measure missing warnings, redundant item prompts, and safety measure recommendations based on the verification results. The safety measure recommendations provided by the intelligent assisted interaction module are based on knowledge graph path analysis, outputting the set of isolation points with the highest frequency and most comprehensive coverage for the current task. The recommendation algorithm is not only based on frequency but can also incorporate weighted scoring, such as considering the experience level of the safety measure developer and the zero-accident record in the history of the safety measure's execution. This allows for the recommendation of optimal practices rather than just the most common practices, enabling the rapid delivery of historical best practices to users, significantly reducing the difficulty of invoicing, improving efficiency, and promoting operational standardization. By constructing a knowledge graph that integrates historical data and domain knowledge and using an AI engine for semantic reasoning, the module achieves automatic verification and intelligent assistance for safety measures in newly created invoices. It also includes a safety isolation and interlocking system interface module, which is used to convert the verified and finalized safety measures into a standard instruction format and send it to the on-site safety isolation and interlocking system as the logical basis for on-site operation and interlocking. The safety isolation and interlocking system interface module encapsulates the list of isolation points (e.g., switch cabinet K1-disconnect-tag) in the final safety measures into an operation instruction sequence according to the protocol agreed upon by the safety isolation and interlocking system (e.g., IEC 61850, OPC UA) and pushes it to the interlocking system, providing a direct basis for implementing physical or logical interlocking on-site. By adding an interface with the on-site interlocking system, seamless information transmission from plan formulation to on-site execution is achieved, forming a complete safety chain of online intelligent verification and offline reliable interlocking, improving the system's practicality.

[0026] An intelligent verification method for industrial safety measures based on knowledge graphs and big data includes the following steps: Step 1: Receive the safety measures information proposed by the user for the newly created job task; Step 2: Parse the semantics of the task and locate the associated equipment and historical task entities in the industrial safety measures knowledge graph; Step 3: Perform integrity verification: Retrieve the set of historical standard safety measures corresponding to the task from the knowledge graph, compare the user-proposed safety measures with the set of standard safety measures, and identify the missing key isolation points; Step 4: Perform a rationality check: Based on the device topology relationship and pre-set reasoning rules in the knowledge graph, analyze whether the isolation point setting in the proposed security measures can effectively cut off the energy source, and detect logical contradictions or redundant measures; Step 5: Based on the verification results of Step 3 and Step 4, generate intelligent auxiliary suggestions that include safety measure missing warnings, redundancy prompts, and safety measure recommendations; Step Six: Provide the intelligent assistance suggestions to the user and receive the user's feedback and confirmation on the security measures based on the suggestions; Step 7: Output the final confirmed safety measures plan to guide on-site operations or distribute it to the safety isolation and interlocking system.

[0027] The working principle of the industrial safety intelligent verification system and method based on knowledge graphs and big data provided by this invention is as follows: When using the industrial safety intelligent verification system based on knowledge graphs and big data, a knowledge construction phase is first carried out, which can be conducted offline or online: The system continuously extracts data from various connected databases, including historical ticket databases and equipment asset databases. Through data standardization and structuring modules, NLP technology is used to transform unstructured ticket text into structured entity relation tables. This structured data is sent to the industrial safety knowledge graph construction module. The constructed module, based on predefined domain ontology (equipment, operation, safety measures, etc.), uses graph database technology (such as Neo4j) to build a huge knowledge network rich in semantic relationships. Expert experience and safety procedures are injected into the graph in the form of reasoning rules, making the entire graph the intelligent brain of the system. Then, the intelligent verification phase is carried out: When the operator creates a new work ticket in the ticketing system and fills in the operation task (such as #2 pot), the system performs the intelligent verification phase. After overhauling the furnace induced draft fan and formulating preliminary safety measures, the system initiates an online verification process. Step 1: The ticket information is sent to the intelligent safety measure verification engine. Step 2: The intelligent safety measure verification engine performs context awareness, analyzes task keywords, and locates relevant nodes in the knowledge graph. Step 3 (Completeness Verification): The intelligent safety measure verification engine searches for safety measure patterns from similar historical operations along the graph's relationship path, forming a standard safety measure set, which is then compared with the user's proposed safety measures. Step 4 (Reasonableness Verification): Based on the equipment topology and logical rules in the graph, the intelligent safety measure verification engine analyzes the logical correctness of the proposed safety measures. Step 5: The verification results (missing items, contradictory items, redundant items) and the recommended safety measure list generated based on the graph are sent to the intelligent auxiliary interaction module. The process flow is as follows: The interaction module clearly displays auxiliary suggestions in the sidebar or pop-up window of the user's ticketing interface. For example, a red warning indicates a missing item: motor heater power supply isolation; a yellow prompt asks for review: is closing the circulating water pump inlet valve necessary? Step 7: The user completes the safety measures according to the prompts and finally confirms and submits them. The final execution linkage stage is then carried out: The final approved safety measures plan is automatically converted into standard instructions through the safety isolation and interlocking system interface module and sent to the on-site five-prevention system or mechanical / electronic interlocking device. On-site staff perform isolation operations according to the prompts on the ticket and the interlocking system, realizing closed-loop management from intelligent verification to safe execution.

[0028] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. An intelligent verification system for industrial safety measures based on knowledge graphs and big data, characterized in that: It includes a multi-source heterogeneous data standardization and structuring module, an industrial safety measures knowledge graph construction module, an AI-based safety measures intelligent verification engine, and an intelligent auxiliary interaction module; Multi-source heterogeneous data standardization and structuring module: used to collect historical work orders, operation tickets and equipment ledger data, and use natural language processing technology to extract key entity information such as work tasks, work locations, equipment involved, isolation points and isolation measures; The knowledge graph construction module for industrial safety measures is connected to the multi-source heterogeneous data standardization and structuring module. It is used to define and construct an ontology model with equipment, operation, hazard source, isolation point and safety measures as the core concepts based on the extracted entity information, and establish semantic relationship between entities to form an industrial safety measures knowledge graph. The industrial safety knowledge graph also integrates a reasoning rule base based on industry safety regulations and expert experience. The AI-based safety measure intelligent verification engine connects to the knowledge graph construction module in the industrial safety measure field. It is used to receive newly created work tasks and proposed safety measure information, analyze the semantics of the work through context awareness, and perform retrieval, reasoning and comparison in the knowledge graph to perform integrity and rationality verification of the safety measures. Intelligent Assisted Interaction Module: Connects to the AI-based safety measure intelligent verification engine to provide users with safety measure missing warnings, redundant item prompts, and safety measure recommendations based on the verification results.

2. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: The multi-source heterogeneous data includes unstructured historical work orders and operation tickets text data, structured equipment ledger database information, and process flow documents.

3. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 2, characterized in that: The natural language processing technology used to extract entity attributes from unstructured text includes at least: isolation point name, isolation point type, isolation measure type, and isolation method.

4. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: The relationships between entities in the industrial safety measures knowledge graph include: operation-involved-equipment, equipment-associated isolation point-isolation point, operation-must-take-safety measures, and topological connections between equipment.

5. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: The integrity verification performed by the verification engine specifically involves: retrieving a set of standard safety measures for similar historical operations in the knowledge graph, comparing the current proposed safety measures with these, and identifying the absence of key isolation points. The rationality verification specifically involves: analyzing whether the isolation points in the proposed safety measures can effectively cut off the energy source based on the device topology and logical relationships in the knowledge graph, and detecting logical contradictions or excessive defense.

6. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: It also includes a safety isolation and interlocking system interface module, which is used to convert the verified and finalized safety measures into a standard instruction format and send it to the on-site safety isolation and interlocking system as the logical basis for on-site operation and interlocking.

7. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: The safety recommendations provided by the intelligent auxiliary interaction module are based on knowledge graph path analysis, outputting a set of isolation points that appear most frequently and have the most comprehensive coverage for the current task.

8. The intelligent verification system for industrial safety measures based on knowledge graphs and big data according to claim 1, characterized in that: The knowledge graph construction module supports incremental learning, which can continuously incorporate newly added historical invoice data into the graph construction and update process, thereby achieving dynamic optimization of verification rules.

9. A method for intelligent verification of industrial safety measures based on knowledge graphs and big data, characterized in that: The industrial safety intelligent verification system based on knowledge graphs and big data, as described in any one of claims 1-8, includes the following steps: Step 1: Receive the safety measures information proposed by the user for the newly created job task; Step 2: Parse the semantics of the task and locate the associated equipment and historical task entities in the industrial safety measures knowledge graph; Step 3: Perform integrity verification: Retrieve the set of historical standard safety measures corresponding to the task from the knowledge graph, compare the user-proposed safety measures with the set of standard safety measures, and identify the missing key isolation points; Step 4: Perform a rationality check: Based on the device topology relationship and pre-set reasoning rules in the knowledge graph, analyze whether the isolation point setting in the proposed security measures can effectively cut off the energy source, and detect logical contradictions or redundant measures; Step 5: Based on the verification results of Step 3 and Step 4, generate intelligent auxiliary suggestions that include safety measure missing warnings, redundancy prompts, and safety measure recommendations; Step Six: Provide the intelligent assistance suggestions to the user and receive the user's feedback and confirmation on the security measures based on the suggestions; Step 7: Output the final confirmed safety measures plan to guide on-site operations or distribute it to the safety isolation and interlocking system.

Images