Accident analysis method, system and equipment based on large model knowledge enhancement and medium

Abstract

The invention relates to an accident analysis method, system and equipment based on large model knowledge enhancement and a medium, and the method comprises the steps: inputting obtained safety accident text data into a pre-constructed large language model, and extracting a place and a production link of a safety accident; utilizing a large language model to extract entity and personnel operation behaviors in the place and the production link; performing mixed knowledge retrieval according to the extracted entities to obtain an entity knowledge set: performing classification and reordering on the obtained entity knowledge set according to pre-established knowledge standards, places, production links and personnel operation behaviors to obtain a reordered entity knowledge list; and forming prompt words according to the safety accident text, the place, the production link, the reordered entity knowledge list and the safety accident reason request, and inputting the prompt words into a large language model to generate a safety accident reason analysis text. The method can be widely applied to the technical field of knowledge enhancement and reordering in computer natural language processing, in particular to the field of industrial safety production.

Description

Technical Field

[0001] This invention belongs to the field of knowledge enhancement and reordering technology in computer natural language processing, specifically relating to the field of industrial safety production, and particularly to an accident analysis method, system, equipment and medium based on large model knowledge enhancement. Background Technology

[0002] Industrial safety is one of the important factors in ensuring normal production, and the ability to quickly analyze the causes of sudden abnormal situations is crucial for taking effective emergency response measures.

[0003] While existing large language models possess strong language generation capabilities, they suffer from the risk of "illusion" in analyzing specific results within a particular domain and lack the ability to dynamically assess the safety status of key entities within the context. This leads to two typical defects in the accident causes generated by these models: first, it is difficult to accurately locate key causal factors; second, even if key causes are identified, their importance weights cannot be correctly assessed. These limitations severely restrict the practical application value of large language models in the field of industrial safety: on the one hand, misjudgment or omission of core causes may delay the optimal response time; on the other hand, improper weight allocation can cause subsequent safety measures to deviate from the true risk management focus, making it difficult for this advanced technology to play its due decision support role in key aspects such as preventive maintenance and emergency response. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to provide an accident analysis method, system, device, and medium based on large-scale model knowledge enhancement, in order to overcome the shortcomings of existing large-scale language models in industrial safety accident analysis, solve the two core problems of inaccurate key cause identification and improper importance assessment, and enable large-scale language model technology to play its due role in decision support in the field of industrial safety.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides an accident analysis method based on large model knowledge enhancement, comprising the following steps: The acquired text data of safety incidents is input into a pre-built large language model to extract the location and production process of the safety incidents. Based on the extracted location and production process, the large language model is used to extract the entity and personnel operation behaviors within that location and production process; Based on the extracted entities, a hybrid knowledge retrieval is performed to obtain an entity knowledge set: The obtained entity knowledge set is classified according to pre-established knowledge standards, locations, production processes, and personnel operation behaviors. The classified entity knowledge is then reordered to obtain a reordered entity knowledge list. Based on the safety incident text, location and production process, reordered entity knowledge list, and safety incident cause request, prompt words are generated and input into a large language model to generate safety incident cause analysis text.

[0006] Furthermore, the process of inputting the acquired safety accident text data into a pre-constructed large language model to extract the location and production process of the safety accident includes: Construct clues for extraction location and production process, represented as follows: {First-Prompt}={Accident}+{First-Query} Here, {First-Prompt} represents the prompt words indicating the location and production process; {Accident} represents the text of the safety incident; and {First-Query} represents the request to indicate the location and production process of the safety incident. Input the extracted location and production process prompts into the large language model, and output the location and production process of the safety accident.

[0007] Furthermore, based on the extracted location and production process, the extraction of entity and personnel operational behaviors within that location and production process using a large language model includes: Based on the extraction location and production process, construct prompt words for the extraction entity and personnel operation behavior respectively; The prompts for extracting entities are represented as follows: {Second-Prompt}={Accident}+{Routine}+{Second-Query} The prompts for extracting personnel's operational behavior are represented as follows: {Third-Prompt}={Accident}+{Third-Query} Where {Second-Prompt} and {Third-Prompt} represent prompts for extracting entity and personnel operation behaviors, respectively; {Accident} represents the safety incident text; {Routine} represents the location and production process; {Second-Query} represents a request to extract entities within the location and production process from the safety incident text; and {Third-Query} represents a request to extract personnel operation behaviors from the safety incident text. Input the constructed prompts for extracting entities and personnel actions into the large language model to obtain a preset number of entities and personnel actions.

[0008] Furthermore, the step of performing hybrid knowledge retrieval based on the extracted entities to obtain an entity knowledge set includes: The search engine takes entities as input for requests and crawls descriptive and functional entity knowledge sets of those entities. If the entity knowledge crawling fails in the previous step, construct prompt words based on the entities and input them into the large language model to obtain the entity knowledge set.

[0009] Furthermore, the entity knowledge set obtained is classified according to pre-established knowledge standards, locations, production processes, and personnel operational behaviors, and the classified entity knowledge is reordered to obtain a reordered entity knowledge list, including: The obtained entity knowledge set is classified based on pre-established knowledge standards and personnel operational behaviors; The categorized entity knowledge is reordered to generate a reordered entity knowledge list.

[0010] Furthermore, the classification of the obtained entity knowledge set based on pre-established knowledge standards and personnel operational behaviors includes: Establish knowledge standards to classify entity knowledge sets into three categories: relevant and valid knowledge, irrelevant knowledge, and relevant knowledge. Based on the established knowledge standards, locations, production processes, and personnel operational behaviors, entity knowledge classification prompts are constructed. The entity knowledge classification prompts are as follows: {Fourth-Prompt}={Knowledge-Norm}+{ENT1-Knowledge1}+{ENT2-Knowledge2}+{ENT3-Knowledge3}+…+{ENT K -Knowledge K}+{Routine}+{Action}+{Fourth-Query} Where {Fourth-Prompt} represents entity knowledge classification prompts; {Knowledge-Norm} represents knowledge standards; {Routine} represents location and production process; {ENT1-Knowledge1}, ..., {ENT K -Knowledge K {} represents entity knowledge; {Action} represents human action; {Fourth-Query} represents a request to classify according to knowledge standards; Input entity knowledge classification prompts into a large language model and output the classified entity knowledge.

[0011] Furthermore, the step of reordering the categorized entity knowledge to generate a reordered entity knowledge list includes: The prompt words for constructing entity knowledge reordering are represented as follows: {Fifth-Prompt}={Classified-Knowledge}+{Fifth-Query} Where {Fifth-Prompt} represents the prompt word for entity knowledge reordering; {Classified-Knowledge} represents the entity knowledge after classification; and {Fifth-Query} indicates the knowledge sorting request to sort relevant valid knowledge first, relevant knowledge last, and irrelevant knowledge removed. Input the constructed entity knowledge reordering prompts into the large language model, and output the reordered entity knowledge list.

[0012] Secondly, the present invention provides an accident analysis system based on large model knowledge enhancement, comprising: The location and process flow extraction module is used to input the acquired safety accident text data into a pre-built large language model to extract the location and production process of the safety accident. The entity and task operation extraction module is used to extract the entity and personnel operation behaviors within the extracted location and production process using a large language model. The hybrid knowledge retrieval module is used to perform hybrid knowledge retrieval based on the extracted entities to obtain a set of entity knowledge. The hybrid knowledge reordering module is used to classify the obtained entity knowledge set according to pre-established knowledge standards, location, production process and personnel operation behavior, and reorder the classified entity knowledge to obtain a reordered entity knowledge list. The output module is used to generate prompt words based on the safety accident text, location and production process, reordered entity knowledge list and safety accident cause request, and input large language model to generate safety accident cause analysis text.

[0013] Thirdly, the present invention provides a computer-readable storage medium for storing one or more programs, the one or more programs including instructions that, when executed by a computing device, cause the computing device to perform the incident analysis method based on large model knowledge enhancement.

[0014] Fourthly, the present invention provides a computing device comprising: one or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs including instructions for executing the incident analysis method based on large model knowledge enhancement.

[0015] The present invention has the following advantages due to the adoption of the above technical solutions: This invention introduces a hybrid knowledge retrieval component and a hybrid knowledge reordering component, enabling large language models to combine internal knowledge with knowledge retrieved by search engines. Compared with the method of automatically generating general large language models, this invention can effectively reduce model illusions and improve the ability of large language models to mine and infer the causes of safety accidents.

[0016] Therefore, this invention can be widely applied in the fields of knowledge enhancement and reordering technology in computer natural language processing. Attached Figure Description

[0017] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings: Figure 1 This is a flowchart of the accident analysis method based on large model knowledge enhancement provided in the embodiments of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0019] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0020] In some embodiments of the present invention, an accident analysis method based on large model knowledge enhancement is provided, comprising a location and production process extraction component, an entity and personnel operation behavior extraction component, a hybrid knowledge retrieval component, a hybrid knowledge reordering component, and a structured prompt word component. The present invention reduces the risk of language model illusion and enhances the ability of language models to uncover the causes of accidents by providing filtered entity knowledge that is strongly correlated with safety accidents.

[0021] Correspondingly, in other embodiments of the present invention, an accident analysis system, device, and medium based on large model knowledge enhancement are provided.

[0022] Example 1 like Figure 1 As shown, this invention provides an accident analysis method based on large model knowledge enhancement, which includes the following steps: 1) Input the acquired safety accident text data into a pre-built large language model to extract the location and production process of the safety accident; 2) Based on the extracted location and production process, use a large language model to extract the entity and personnel operation behaviors within that location and production process; 3) Perform hybrid knowledge retrieval based on the extracted entities to obtain an entity knowledge set: 4) Classify the obtained entity knowledge set according to the pre-established knowledge standards, locations, production processes, and personnel operation behaviors, and reorder the classified entity knowledge to obtain a reordered entity knowledge list. 5) Based on the safety accident text, location and production process, the reordered entity knowledge list and the safety accident cause request, construct prompt words, input them into the large language model to generate the safety accident cause analysis text.

[0023] Furthermore, step 1) above includes the following steps: 1.1) Construct prompts for extraction locations and production processes.

[0024] In this embodiment, the constructed prompt for extracting the location and production process includes a safety incident text and a request ("Indicate the location and production process of the safety incident"). Assuming the safety incident text is denoted as {Accident}, the prompt for extracting the location and production process is denoted as {First-Prompt}, and the request is denoted as {First-Query}, then the prompt for extracting the location and production process is represented as follows: {First-Prompt}={Accident}+{First-Query} 1.2) Input the constructed prompts for the extraction location and production process into the large language model, and output the location and production process of the safety accident.

[0025] Furthermore, step 2) above includes the following steps: 2.1) Based on the extraction location and production process, construct prompt words for the extraction entity and personnel operation behavior respectively.

[0026] In this embodiment, the constructed entity extraction prompt includes the security incident text, location and production process, and request ("Please extract the entities within the location and production process in the security incident text"). Assuming the entity extraction prompt is denoted as {Second-Prompt}, the request as {Second-Query}, and the location and production process as {Routine}, then the entity extraction prompt is represented as: {Second-Prompt}={Accident}+{Routine}+{Second-Query} The constructed prompt for retrieving personnel operation behavior includes a safety incident text and a request ("Please extract the personnel's operations from the safety incident text"). Assuming the request is denoted as {Third-Query}, the prompt for retrieving personnel operation behavior is represented as follows: {Third-Prompt}={Accident}+{Third-Query} 2.2) Input the constructed prompts for extracting entities and personnel operation behaviors into the large language model to obtain a preset number of entities and personnel operation behaviors.

[0027] In this embodiment, a constant K is set to represent the entity extracted by this invention, called Entity, or ENT. After the prompt words for extracting the entity are input into the large language model, the large language model outputs K entities, denoted as: {ENT1, ENT2, ..., ENT...} K After inputting the prompts for the extracted personnel's actions into the large language model, the large language model outputs the personnel's actions, denoted as {Action}.

[0028] Furthermore, step 3) above includes the following steps: 3.1) Use entities as input to the search engine to crawl the descriptive and functional entity knowledge sets of the entities; 3.2) If crawling entity knowledge fails in step 3.1), construct prompt words based on entities and input them into the large language model to obtain the entity knowledge set.

[0029] In this embodiment, the entity knowledge set is denoted as: {ENT1-Knowledge1}, {ENT2-Knowledge2}, {ENT3-Knowledge3}, ..., {ENT K -Knowledge K}

[0030] Furthermore, step 4) above includes the following steps: 4.1) Classify the obtained entity knowledge set based on pre-established knowledge standards and personnel operation behaviors.

[0031] 4.2) Reorder the categorized entity knowledge to generate a reordered entity knowledge list.

[0032] Furthermore, step 4.1 above includes the following steps: 4.1.1) Establish knowledge standards.

[0033] In this embodiment, the knowledge standard is denoted as {Knowledge-Norm}, which is used to classify entity knowledge into the following three categories: ① Relevant valid knowledge: applicable locations and production processes, and specific knowledge pointing out human operational errors; ② Irrelevant knowledge: Entity knowledge that is not applicable to the location and production process; ③ Relevant knowledge: The remaining knowledge after classifying relevant and irrelevant knowledge.

[0034] 4.1.2) Based on the established knowledge standards, locations, production processes, and personnel operation behaviors, construct entity knowledge classification prompt words.

[0035] In this embodiment, the constructed entity knowledge classification prompt is denoted as {Fourth-Prompt}, which includes knowledge standards, entity knowledge, location and production process, personnel operation behavior, and request ("Please classify according to knowledge standards"), represented as: {Fourth-Prompt}={ Knowledge-Norm}+{ENT1-Knowledge1}+{ENT2-Knowledge2}+{ENT3-Knowledge3}+…+{ENT K -Knowledge K}+{Routine}+{Action}+{Fourth-Query}.

[0036] 4.1.3) Input the entity knowledge classification prompts into the large language model and output the classified entity knowledge.

[0037] In this embodiment, the classified entity knowledge is denoted as: {Classified-Knowledge}.

[0038] Furthermore, step 4.2 above includes the following steps: 4.2.1) Construct prompt words for entity knowledge reordering.

[0039] In this embodiment, the constructed entity knowledge reordering prompt is denoted as {Fifth-Prompt}, which includes the classified entity knowledge {Classified-Knowledge} and the request ("Please sort the knowledge according to the method of prioritizing relevant and valid knowledge, relevance to relevant knowledge, and removal of irrelevant knowledge"), denoted as {Fifth-Query}, and represented as follows: {Fifth-Prompt}={Classified-Knowledge}+{Fifth-Query} 4.2.2) Input the constructed entity knowledge reordering prompts into the large language model and output the reordered entity knowledge list.

[0040] In this embodiment, the reordered entity knowledge is denoted as: {Reordered-Knowledge}.

[0041] Furthermore, step 5) above includes the following steps: 5.1) Construct prompt words for safety incident analysis text.

[0042] In this embodiment, the prompt word for the constructed safety incident analysis text is denoted as {Sixth-Prompt}, which includes the safety incident text, location and production process, reordered entity knowledge, and request ("Please analyze the cause of the incident as comprehensively as possible based on the knowledge"), denoted as {Sixth-Query}. The prompt word is then represented as: {Sixth-Prompt}={Accident}+{Rountine}+{Reordered-Knowledge}+{Sixth-Query} 5.2) Input the prompt words of the safety accident analysis text into the large language model and output the safety accident cause analysis text.

[0043] In this embodiment, the text analyzing the cause of a safety incident is denoted as: {Accident-Analysis}.

[0044] Example 2 This embodiment statistically illustrates the advantages of the accident analysis method based on large model knowledge enhancement proposed in this invention over the method of automatically generating safety accident reports using general large language models.

[0045] The test module used 200 security incident texts, and the scoring was done using BertScore. BertScore is a semantic evaluation metric based on a pre-trained language model. It calculates the similarity between the generated text and the reference text in the BERT embedding space, and comprehensively measures the quality of text generation from three dimensions: precision, recall, and F1 score. It can more accurately reflect the performance of the generated text in terms of semantic fidelity, content completeness, and overall consistency.

[0046] Table 1. Comparison of basic prompt words between this invention and the general large language model.

[0047] As shown in Table 1, adding basic formatting requirements, such as "Please analyze the causes of the safety incident as comprehensively as possible," to the basic prompts of the method of this invention significantly improves the overall report quality score compared to reports generated using only the same general large language model. This indicates that the method of this invention can reduce model illusion while improving the ability of the large language model to discover and infer the causes of safety incidents.

[0048] Example 3 The above-described embodiment 1 provides an accident analysis method based on large model knowledge enhancement. Correspondingly, this embodiment provides an accident analysis system based on large model knowledge enhancement. The system provided in this embodiment can implement the accident analysis method based on large model knowledge enhancement of embodiment 1. The system can be implemented through software, hardware, or a combination of both. For example, the system may include integrated or separate functional modules or units to execute the corresponding steps in the methods of embodiment 1. Since the system in this embodiment is basically similar to the method embodiment, the description process in this embodiment is relatively simple. For relevant details, please refer to the description of embodiment 1. The system embodiment provided in this embodiment is merely illustrative.

[0049] The accident analysis system based on large model knowledge enhancement provided in this embodiment includes: The location and production process extraction module is used to input the acquired safety accident text data into a pre-built large language model to extract the location and production process of the safety accident. The entity and task operation extraction module is used to extract the entity and personnel operation behaviors within the extracted location and production process using a large language model. The hybrid knowledge retrieval module is used to perform hybrid knowledge retrieval based on the extracted entities to obtain a set of entity knowledge. The hybrid knowledge reordering module is used to classify the obtained entity knowledge set according to pre-established knowledge standards, location, production process and personnel operation behavior, and reorder the classified entity knowledge to obtain a reordered entity knowledge list. The output module is used to generate prompt words based on the safety accident text, location and production process, reordered entity knowledge list and safety accident cause request, and input large language model to generate safety accident cause analysis text.

[0050] Example 4 This embodiment provides a processing device corresponding to the accident analysis method based on large model knowledge enhancement provided in Embodiment 1. The processing device can be a client-side processing device, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Embodiment 1.

[0051] The processing device includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus to enable communication between them. The memory stores a computer program that can run on the processor. When the processor runs the computer program, it executes the accident analysis method based on large model knowledge enhancement provided in Embodiment 1.

[0052] Preferably, the memory may be high-speed random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage device.

[0053] Preferably, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and there is no limitation herein.

[0054] Example 5 The incident analysis method based on large model knowledge enhancement in Embodiment 1 can be specifically implemented as a computer program product. The computer program product may include a computer-readable storage medium on which computer-readable program instructions for executing the incident analysis method based on large model knowledge enhancement described in Embodiment 1 are loaded.

[0055] A computer-readable storage medium can be a tangible device that holds and stores instructions for use by an instruction execution device. A computer-readable storage medium can be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof.

[0056] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. An accident analysis method based on large-scale model knowledge enhancement, characterized in that, Includes the following steps: The acquired text data of safety incidents is input into a pre-built large language model to extract the location and production process of the safety incidents. Based on the extracted location and production process, the large language model is used to extract the entity and personnel operation behaviors within that location and production process; Based on the extracted entities, a hybrid knowledge retrieval is performed to obtain an entity knowledge set: The obtained entity knowledge set is classified according to pre-established knowledge standards, locations, production processes, and personnel operation behaviors. The classified entity knowledge is then reordered to obtain a reordered entity knowledge list. Based on the safety incident text, location and production process, reordered entity knowledge list, and safety incident cause request, prompt words are generated and input into a large language model to generate safety incident cause analysis text.

2. The accident analysis method based on large model knowledge enhancement as described in claim 1, characterized in that, The process of inputting the acquired safety accident text data into a pre-constructed large language model to extract the location and production process of the safety accident includes: Construct clues for extraction location and production process, represented as follows: {First-Prompt}={Accident}+{First-Query} Here, {First-Prompt} represents the prompt words indicating the location and production process; {Accident} represents the text of the safety incident; and {First-Query} represents the request to indicate the location and production process of the safety incident. Input the extracted location and production process prompts into the large language model, and output the location and production process of the safety accident.

3. The accident analysis method based on large model knowledge enhancement as described in claim 1, characterized in that, The step of extracting entity and personnel operational behaviors within the extracted location and production process using a large language model includes: Based on the extraction location and production process, construct prompt words for the extraction entity and personnel operation behavior respectively; The prompts for extracting entities are represented as follows: {Second-Prompt}={Accident}+{Routine}+{Second-Query} The prompts for extracting personnel's operational behavior are represented as follows: {Third-Prompt}={Accident}+{Third-Query} Where {Second-Prompt} and {Third-Prompt} represent prompts for extracting entity and personnel operation behaviors, respectively; {Accident} represents the safety incident text; {Routine} represents the location and production process; {Second-Query} represents a request to extract entities within the location and production process from the safety incident text; and {Third-Query} represents a request to extract personnel operation behaviors from the safety incident text. Input the constructed prompts for extracting entities and personnel actions into the large language model to obtain a preset number of entities and personnel actions.

4. The accident analysis method based on large model knowledge enhancement as described in claim 1, characterized in that, The process of performing hybrid knowledge retrieval based on the extracted entities to obtain an entity knowledge set includes: Use entities as input to the search engine to crawl descriptive and functional entity knowledge of the entities; If the entity knowledge crawling fails in the previous step, construct prompt words based on the entities and input them into the large language model to obtain the entity knowledge set.

5. The accident analysis method based on large model knowledge enhancement as described in claim 1, characterized in that, The entity knowledge set is classified according to pre-established knowledge standards, locations, production processes, and personnel operational behaviors. The classified entity knowledge is then reordered to obtain a reordered entity knowledge list, including: The obtained entity knowledge set is classified based on pre-established knowledge standards and personnel operational behaviors; The categorized entity knowledge is reordered to generate a reordered entity knowledge list.

6. The accident analysis method based on large model knowledge enhancement as described in claim 5, characterized in that, The classification of the entity knowledge set obtained based on pre-established knowledge standards and personnel operational behaviors includes: Establish knowledge standards to classify entity knowledge into three categories: relevant and valid knowledge, irrelevant knowledge, and relevant knowledge. Based on the established knowledge standards, locations, production processes, and personnel operational behaviors, entity knowledge classification prompts are constructed. Among them, the entity knowledge classification prompts are: {Fourth-Prompt}={Knowledge-Norm}+{ENT1-Knowledge1}+{ENT2-Knowledge2}+{ENT3-Knowledge3}+…+{ENT K -Knowledge K }+{Routine}+{Action}+{Fourth-Query} Where {Fourth-Prompt} represents entity knowledge classification prompts; {Knowledge-Norm} represents knowledge standards; {Routine} represents location and production process; {ENT1-Knowledge1}, ..., {ENT K -Knowledge K {} represents entity knowledge; {Action} represents human action; {Fourth-Query} represents a request to classify according to knowledge standards; Input entity knowledge classification prompts into a large language model and output the classified entity knowledge.

7. The accident analysis method based on large model knowledge enhancement as described in claim 1, characterized in that, The process of reordering the categorized entity knowledge to generate a reordered entity knowledge list includes: The prompt words for constructing entity knowledge reordering are represented as follows: {Fifth-Prompt}={Classified-Knowledge}+{Fifth-Query} Where {Fifth-Prompt} represents the prompt word for entity knowledge reordering; {Classified-Knowledge} represents the entity knowledge after classification; and {Fifth-Query} indicates the knowledge sorting request to sort relevant valid knowledge first, relevant knowledge last, and irrelevant knowledge removed. Input the constructed entity knowledge reordering prompts into the large language model, and output the reordered entity knowledge list.

8. An accident analysis system based on large-scale model knowledge enhancement, characterized in that, include: The location and production process extraction module is used to input the acquired safety accident text data into a pre-built large language model to extract the location and production process of the safety accident. The entity and task operation extraction module is used to extract the entity and personnel operation behaviors within the extracted location and production process using a large language model. The hybrid knowledge retrieval module is used to perform hybrid knowledge retrieval based on the extracted entities to obtain a set of entity knowledge. The hybrid knowledge reordering module is used to classify the obtained entity knowledge set according to pre-established knowledge standards, location, production process and personnel operation behavior, and reorder the classified entity knowledge to obtain a reordered entity knowledge list. The output module is used to generate prompt words based on the safety accident text, location and production process, reordered entity knowledge list and safety accident cause request, and input large language model to generate safety accident cause analysis text.

9. A computer-readable storage medium for storing one or more programs, characterized in that, The one or more programs include instructions that, when executed by a computing device, cause the computing device to perform any of the methods described in claims 1 to 7.

10. A computing device, characterized in that, include: One or more processors and a memory, wherein the memory stores one or more programs and is configured to be executed by the one or more processors, the one or more programs including instructions for performing any of the methods described in claims 1 to 7.

Images