Industrial risk identification rule automatic evolution method, system, device and program product

Abstract

This invention discloses an automatic evolution method, system, device, and program product for industrial risk identification rules. The method includes: acquiring industrial risk data; inputting the industrial risk data into an industrial risk identification model based on a dynamic risk identification rule base for risk identification, and outputting the risk identification result; mapping the risk event identification result to an industrial safety knowledge graph for semantic reasoning; performing consistency verification on the risk event identification result and the semantic reasoning result, and determining a mismatch event if a consistency mismatch exists; performing causal chain backtracking analysis on the mismatch event in the industrial safety knowledge graph to obtain the triplet feature set leading to the mismatch; and updating the rules of the dynamic risk identification rule base based on the triplet feature set. This invention realizes the evolution and updating of industrial risk identification rules, constructs a dual-optimization closed loop of identification rules and models, effectively improves the accuracy of industrial risk identification, and reduces the false negative rate.

Description

Technical Field

[0001] This invention relates to the field of deep learning technology, and in particular to an automatic evolution method, system, device and program product for industrial risk identification rules. Background Technology

[0002] Traditional industrial safety identification systems typically rely on manually set static rule bases. Once deployed, these rules lack adaptability and cannot effectively cope with complex and ever-changing industrial operating scenarios. Especially when new risk events occur, static rules cannot respond in a timely manner, resulting in high rates of missed or false alarms. Furthermore, rule updates require manual intervention and downtime maintenance, severely impacting production continuity and safety.

[0003] Therefore, there is an urgent need to invent an automatic evolution method for industrial risk identification rules to solve the problems of poor adaptability of existing static identification rule bases, inability to respond to new risk events in a timely manner, and the need for manual intervention for rule updates. Summary of the Invention

[0004] In view of this, embodiments of the present invention provide an automatic evolution method, system, device and program product for industrial risk identification rules, which at least partially solves the problems existing in the prior art.

[0005] Other features and advantages of the invention will become apparent from the following detailed description, or may be learned in part by practice of the invention.

[0006] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions:

[0007] According to a first aspect of the present invention, an automatic evolution method for industrial risk identification rules is provided, the method comprising:

[0008] Obtain industrial risk data;

[0009] The industrial risk data is input into a preset industrial risk identification model based on a dynamic risk identification rule base to identify risk events and output the risk event identification results.

[0010] The risk event identification results are mapped to a preset industrial safety knowledge graph for semantic reasoning to obtain semantic reasoning results;

[0011] The consistency of the risk event identification result and the semantic reasoning result is checked. If there is a consistency mismatch, it is determined to be a mismatch event.

[0012] A causal chain backtracking analysis was performed on the mismatch event in the industrial safety knowledge graph to obtain the triplet feature set that led to the mismatch.

[0013] Based on the triplet feature set, the dynamic risk identification rule base is updated to obtain the updated dynamic risk identification rule base.

[0014] Furthermore, the method also includes:

[0015] The updated dynamic risk identification rule base is used to update the industrial risk identification model and / or the industrial safety knowledge graph.

[0016] Furthermore, the industrial risk data includes industrial video data and sensor data;

[0017] The dynamic risk identification rule base supports online hot updates, rollbacks, and namespace-isolated deployment.

[0018] Furthermore, the industrial risk identification model is a deep learning model built on a dynamic risk identification rule base;

[0019] The risk event identification results include the risk event category and confidence level.

[0020] Furthermore, the construction process of the industrial safety knowledge graph includes:

[0021] Entity nodes are extracted from multi-source data, including personnel nodes, equipment nodes, environment nodes, and / or behavior nodes;

[0022] Identify the semantic relationships between each entity node, store the entity nodes and the semantic relationships between entity nodes in a graph database, and obtain the industrial safety knowledge graph.

[0023] Furthermore, based on the triplet feature set, the dynamic risk identification rule base is updated, including:

[0024] Based on the cost minimization algorithm, the risk identification rules in the dynamic risk identification rule base are adaptively optimized according to the triplet feature set.

[0025] According to a second aspect of the present invention, an automatic evolution system for industrial risk identification rules is provided, the system comprising:

[0026] The data acquisition module is used to acquire industrial risk data;

[0027] The model recognition module is used to input the industrial risk data into a preset industrial risk recognition model based on a dynamic risk recognition rule base to identify risk events and output the risk event recognition results.

[0028] The graph reasoning module is used to map the risk event identification results to a preset industrial safety knowledge graph for semantic reasoning, and obtain semantic reasoning results.

[0029] The result verification module is used to verify the consistency between the risk event identification result and the semantic reasoning result. If there is a consistency mismatch, it is determined to be a mismatch event.

[0030] The mismatch backtracking module is used to perform causal chain backtracking analysis on the mismatch event in the industrial safety knowledge graph, and to obtain the triplet feature set that leads to the mismatch.

[0031] The rule evolution module is used to update the dynamic risk identification rule base according to the triplet feature set, so as to obtain the updated dynamic risk identification rule base.

[0032] Furthermore, the system also includes:

[0033] The feedback optimization module is used to update the industrial risk identification model and / or the industrial safety knowledge graph using the updated dynamic risk identification rule base.

[0034] According to a third aspect of the present invention, an automatic evolution device for industrial risk identification rules is provided, the device comprising: a processor and a memory;

[0035] The memory is used to store one or more program instructions;

[0036] The processor is configured to run one or more program instructions to perform the steps of an automatic evolution method for industrial risk identification rules as described in any of the preceding claims.

[0037] According to a fourth aspect of the present invention, a computer program product is provided, the computer program product comprising a computing program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions that, when executed by a computer, cause the computer to implement the steps of an automatic evolution method for industrial risk identification rules as described in any of the preceding claims.

[0038] This invention provides an automatic evolution method, system, device, and program product for industrial risk identification rules. The method includes: acquiring industrial risk data; inputting the industrial risk data into an industrial risk identification model based on a dynamic risk identification rule base for risk identification, and outputting risk event identification results; mapping the risk event identification results to an industrial safety knowledge graph for semantic reasoning; performing consistency verification on the risk event identification results and semantic reasoning results, and determining a mismatch event if a mismatch exists; performing causal chain backtracking analysis on the mismatch event in the industrial safety knowledge graph to obtain the triplet feature set leading to the mismatch; and updating the dynamic risk identification rule base according to the triplet feature set. This invention realizes the evolution and updating of industrial risk identification rules, constructs a dual-optimization closed loop of identification rules and models, effectively improves the accuracy of industrial risk identification, and reduces the false negative rate. Attached Figure Description

[0039] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0040] Figure 1 A flowchart illustrating an automatic evolution method for industrial risk identification rules provided in an embodiment of the present invention;

[0041] Figure 2 This is a schematic diagram of the structure of an automatic evolution system for industrial risk identification rules provided in an embodiment of the present invention;

[0042] Figure 3 This is a schematic diagram of the structure of the industrial safety knowledge graph provided in an embodiment of the present invention;

[0043] Figure 4 This is a schematic diagram of the structure of the dynamic risk identification rule base provided in an embodiment of the present invention;

[0044] Figure 5 This is a schematic diagram of the structure of the risk identification rule and model dual optimization closed loop provided in an embodiment of the present invention;

[0045] Figure 6 This is a schematic diagram illustrating a practical application of the present invention for safety monitoring in chemical plants. Detailed Implementation

[0046] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0047] Furthermore, the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product, or apparatus.

[0048] Figure 1 The figure shows a flowchart of an automatic evolution method for industrial risk identification rules according to an embodiment of the present invention.

[0049] like Figure 1 As shown, the automatic evolution method for industrial risk identification rules according to an embodiment of the present invention may include steps S100, S200, S300, S400, S500 and S600.

[0050] In step S100, industrial risk data is acquired.

[0051] Specifically, the aforementioned industrial risk data includes industrial video data and sensor data.

[0052] Next, in step S200, the industrial risk data is input into a preset industrial risk identification model based on a dynamic risk identification rule base to identify risk events and output the risk event identification results.

[0053] Specifically, the aforementioned industrial risk identification model is a deep learning model (YOLOv8, Transformer) built on a dynamic risk identification rule base.

[0054] By analyzing real-time video frames using an industrial risk identification model, the output risk event identification results include risk event category and confidence level.

[0055] In step S300, the risk event identification results are mapped to a preset industrial safety knowledge graph for semantic reasoning to obtain semantic reasoning results.

[0056] Specifically, the construction process of the aforementioned industrial safety knowledge graph includes:

[0057] Entity nodes are extracted from multi-source data, including personnel nodes, equipment nodes, environment nodes, and / or behavior nodes; semantic relationships between various entity nodes are identified, and the entity nodes and their semantic relationships are stored in a graph database to obtain an industrial safety knowledge graph.

[0058] The graph databases mentioned above are Neo4j or JanusGraph, and the inference engines for the industrial safety knowledge graphs mentioned above are SWRL or SPARQL.

[0059] Next, in step S400, the consistency of the risk event identification result and the semantic reasoning result is checked. If there is a consistency mismatch, it is determined to be a mismatch event.

[0060] In step S500, a causal chain backtracking analysis is performed on the mismatch event in the industrial safety knowledge graph to obtain the triplet feature set that leads to the mismatch.

[0061] Finally, in step S600, the dynamic risk identification rule base is updated according to the triplet feature set to obtain the updated dynamic risk identification rule base.

[0062] Specifically, the above steps include:

[0063] The above triplet feature set is mapped to a preset rule template. Based on the cost minimization algorithm, the threshold of the risk identification rule is adaptively calculated. The calculation formula is as follows: .

[0064] The optimized risk identification rules are written into a dynamic risk identification rule base, which supports online hot updates, rollbacks, and namespace-isolated deployments.

[0065] Preferably, the above method further includes: updating the industrial risk identification model and / or industrial safety knowledge graph using the updated dynamic risk identification rule base.

[0066] The model update process includes: adding the scenario samples corresponding to the updated risk identification rules in the risk identification rule base to the training set of the above-mentioned industrial risk identification model, and updating the deep learning model through batch incremental learning to improve the adaptability of the industrial risk identification model.

[0067] In addition, embodiments of the present invention also provide an automatic evolution system for industrial risk identification rules, the system comprising:

[0068] The data acquisition module is used to acquire industrial risk data. The data acquisition module includes a video acquisition unit and a sensor data acquisition unit.

[0069] The model recognition module is used to input industrial risk data into a preset industrial risk recognition model based on a dynamic risk recognition rule base to identify risk events and output the risk event recognition results.

[0070] The graph reasoning module is used to map the risk event identification results to a preset industrial safety knowledge graph for semantic reasoning, and obtain the semantic reasoning results.

[0071] Figure 3 The diagram illustrates the structure of the industrial safety knowledge graph provided in this embodiment of the invention. The industrial safety knowledge graph includes a node extraction unit, a relation extraction unit, and a graph storage unit. The node extraction unit is used to extract entity nodes (personnel P, equipment E, environment Env, behavior A) from multi-source data. The relation extraction unit is used to identify semantic relationships between entities. The graph storage unit is used to store nodes and relationships in a graph database.

[0072] The result verification module is used to verify the consistency between the risk event identification result and the semantic reasoning result. If there is a consistency mismatch, it is determined to be a mismatch event.

[0073] The mismatch backtracking module is used to perform causal chain backtracking analysis on mismatch events in the industrial safety knowledge graph, and to obtain the triple feature set that led to the mismatch.

[0074] The rule evolution module is used to update the dynamic risk identification rule base based on the triple feature set, so as to obtain the updated dynamic risk identification rule base.

[0075] The aforementioned dynamic risk identification rule base has rule version control, permission management, and rollback functions. The dynamic risk identification rule base includes a rule version control unit, a rule storage unit (supporting namespace isolation, hot update modules, and permission management modules), and a rule storage unit. Figure 4 A schematic diagram of the structure of the dynamic risk identification rule base provided in an embodiment of the present invention is shown.

[0076] The feedback optimization module is used to update the industrial risk identification model and / or industrial safety knowledge graph using the updated dynamic risk identification rule base.

[0077] Figure 5 The diagram illustrates the structure of the risk identification rule and model dual optimization closed loop provided in an embodiment of the present invention. The new rule feedback module is used to add the samples generated by the new rule to the model training set, the model retraining module is used to update the model weights based on incremental learning, and the optimization result verification module is used to verify the consistency between the updated model and the rule.

[0078] Figure 6 A schematic diagram illustrating a practical application of the present invention for safety monitoring in chemical plants is shown in an embodiment of the invention.

[0079] In practical applications of safety monitoring in chemical plants, industrial data comes from thermal imaging cameras and pressure sensors. When temperature anomalies and pressure fluctuations simultaneously trigger mismatch, the system automatically generates multi-condition composite rules and updates them to the dynamic risk identification rule library.

[0080] In the practical application scenario of high temperature and high pressure monitoring in chemical plants, the embodiments of the present invention can shorten the alarm response time by 65% ​​and reduce the false alarm rate by 19%.

[0081] The embodiments of the present invention can also be used in warehouse forklift safety identification scenarios. In warehouse forklift safety identification scenarios, compared with the prior art before the improvement, the safety identification performance improved by the automatic evolution method of industrial risk identification rules provided by the embodiments of the present invention is shown in Table 1.

[0082] Table 1 Comparison of Safety Identification Performance of Warehouse Forklifts

[0083]

[0084] The embodiments of the present invention can also be used for risk early warning in underground mines. By combining gas detection and personnel positioning, automatic evacuation rules can be generated. The generated new rules take effect across the entire network within 90 seconds through a dynamic risk identification rule base.

[0085] In addition, this embodiment of the invention also provides an automatic evolution device for industrial risk identification rules, the device comprising: a processor and a memory; the memory for storing one or more program instructions; the processor for running one or more program instructions to execute the steps of an automatic evolution method for industrial risk identification rules as described above.

[0086] In addition, embodiments of the present invention also provide a computer program product, which includes computer program instructions that, when executed by a processor, implement the steps of an automatic evolution method for industrial risk identification rules as described above.

[0087] This invention provides an automatic evolution method, system, device, and program product for industrial risk identification rules. It constructs an industrial safety knowledge graph encompassing multi-dimensional semantic nodes and relationships, including personnel (P), equipment (E), environment (Env), and behavior (A). A deep learning model is used to identify risk events. The identification results are mapped to knowledge graph nodes, triggering semantic reasoning. When there is a mismatch between the reasoning conclusion and the model output, causal chain error attribution analysis is performed to extract the triplet feature set causing the mismatch. The triplet feature set is mapped to executable new rules or threshold adjustment strategies, updated in real-time to a versioned dynamic risk identification rule base, and the rules are driven to take effect online in subsequent identifications. Simultaneously, effective rules are fed back to the model training set for retraining, achieving a dual optimization closed loop for both the model and rules.

[0088] Compared with existing static rule base solutions, the present invention can complete the deployment of new rules within ≤2 minutes. Based on the embodiments of the present invention, the cross-scenario risk identification accuracy is improved by ≥15%, the false negative rate is reduced by ≥17%, and the manual maintenance cost is reduced by ≥40%. The embodiments of the present invention significantly enhance the adaptability and interpretability of industrial risk identification.

[0089] In this embodiment of the invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in this embodiment of the invention. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in this embodiment of the invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods. The storage medium can be memory, for example, volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM).The storage media described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable types of memory. Those skilled in the art will recognize that the functions described in the above examples can be implemented using a combination of hardware and software. When applied software, the corresponding functions can be stored in a computer-readable medium or transmitted as one or more instructions or code on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers. Although the invention has been described in detail above with general description and specific embodiments, modifications or improvements can be made to it, which will be apparent to those skilled in the art. Therefore, such modifications or improvements made without departing from the spirit of the invention are all within the scope of protection claimed by the invention.

[0090] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent changes, or alterations made by those skilled in the art using the disclosed technical content shall fall within the protection scope of the present invention.

Claims

1. An industrial risk identification rule automatic evolution method, characterized in that, The method includes: Obtain industrial risk data; The industrial risk data is input into a preset industrial risk identification model based on a dynamic risk identification rule base to identify risk events and output the risk event identification results. The risk event identification results are mapped to a preset industrial safety knowledge graph for semantic reasoning to obtain semantic reasoning results; The consistency of the risk event identification result and the semantic reasoning result is checked. If there is a consistency mismatch, it is determined to be a mismatch event. A causal chain backtracking analysis was performed on the mismatch event in the industrial safety knowledge graph to obtain the triplet feature set that led to the mismatch. Based on the triplet feature set, the dynamic risk identification rule base is updated to obtain the updated dynamic risk identification rule base.

2. The method of claim 1, wherein, The method further includes: The updated dynamic risk identification rule base is used to update the industrial risk identification model and / or the industrial safety knowledge graph.

3. The method of claim 1, wherein, The industrial risk data includes industrial video data and sensor data; The dynamic risk identification rule base supports online hot updates, rollbacks, and namespace-isolated deployment.

4. The method of claim 1, wherein, The industrial risk identification model is a deep learning model built on a dynamic risk identification rule base; The risk event identification results include the risk event category and confidence level.

5. The method of claim 1, wherein, The process of constructing the industrial safety knowledge graph includes: Entity nodes are extracted from multi-source data, including personnel nodes, equipment nodes, environment nodes, and / or behavior nodes; Identify the semantic relationships between each entity node, store the entity nodes and the semantic relationships between entity nodes in a graph database, and obtain the industrial safety knowledge graph.

6. The method of claim 1, wherein, Based on the triplet feature set, the dynamic risk identification rule base is updated, including: Based on the cost minimization algorithm, the risk identification rules in the dynamic risk identification rule base are adaptively optimized according to the triplet feature set.

7. An industrial risk identification rule automatic evolution system, characterized in that, The system includes: The data acquisition module is used to acquire industrial risk data; The model recognition module is used to input the industrial risk data into a preset industrial risk recognition model based on a dynamic risk recognition rule base to identify risk events and output the risk event recognition results. The graph reasoning module is used to map the risk event identification results to a preset industrial safety knowledge graph for semantic reasoning, and obtain semantic reasoning results. The result verification module is used to verify the consistency between the risk event identification result and the semantic reasoning result. If there is a consistency mismatch, it is determined to be a mismatch event. The mismatch backtracking module is used to perform causal chain backtracking analysis on the mismatch event in the industrial safety knowledge graph, and to obtain the triplet feature set that leads to the mismatch. The rule evolution module is used to update the dynamic risk identification rule base according to the triplet feature set, so as to obtain the updated dynamic risk identification rule base.

8. The system of claim 7, wherein, The system also includes: The feedback optimization module is used to update the industrial risk identification model and / or the industrial safety knowledge graph using the updated dynamic risk identification rule base.

9. An industrial risk identification rule automatic evolution device characterized by comprising: The device includes: a processor and a memory; The memory is configured to store one or more program instructions; The processor is configured to execute the one or more program instructions to implement the method for automatically evolving an industrial risk identification rule according to any one of claims 1 to 6.

10. A computer program product, characterised in that, The computer program product comprises computer program instructions, which, when executed by a processor, implement the method for automatically evolving an industrial risk identification rule according to any one of claims 1 to 6.

Images