Coal mine safety early warning system and method based on large model

Abstract

The invention discloses a coal mine safety early warning system and method based on a large model, and relates to the technical field of coal mine safety early warning, and the system comprises a coal mine safety early warning special large model, a safety monitoring module, a data processing module, and a safety early warning tool. The coal mine safety early warning special large model is constructed based on target data and related knowledge. A large model technology is applied to coal mine safety early warning business, and a coal mine safety early warning large model is constructed through methods of model training, knowledge base construction, knowledge base evaluation and the like on the basis of powerful semantic understanding, logical reasoning and decision planning capabilities of a general large model. A special large model capable of recognizing various safety risk accidents of a coal mine, judging potential risks and hidden dangers and automatically generating early warning information is obtained, the reminding quality and efficiency of safety risk early warning services are effectively improved, safety and efficiency of coal mine production are guaranteed, and the multi-dimensional and expert-perspective safety early warning analysis and reminding functions are achieved.

Description

Technical Field

[0001] This invention relates to the field of coal mine safety early warning technology, specifically to a coal mine safety early warning system and method based on a large model. Background Technology

[0002] Traditional coal mine safety early warning systems mainly consist of a sensor network, a data transmission network, a monitoring and control center, a data analysis and processing center, and an early warning and decision support system. The system collects various data in real time during coal mine production, such as gas concentration, temperature, humidity, wind speed, and equipment status, through the sensor network and transmits the data to the monitoring and control center. At the monitoring and control center, the data is analyzed and processed to assess the safety status of the coal mine and generate early warning information. This early warning information can be sent to relevant personnel through various means so that timely measures can be taken to prevent risks.

[0003] The data analysis and processing mainly compares the received data from each module with the data type and threshold. If the data exceeds a certain range, an early warning message is generated. However, it cannot analyze the security issues of data types such as images and videos, nor can it analyze and warn of potential or indirect security issues. It cannot achieve multi-dimensional, expert-perspective security early warning analysis functions. Therefore, this application proposes a coal mine safety early warning system and method based on a large model to solve the above problems. Summary of the Invention

[0004] (a) Technical problems to be solved

[0005] In view of the shortcomings of the prior art, the present invention provides a coal mine safety early warning system and method based on a large model, which solves the technical problems mentioned in the background above.

[0006] (II) Technical Solution

[0007] To achieve the above objectives, the technical solution adopted by the present invention is: a coal mine safety early warning system based on a large model, which includes a proprietary large model for coal mine safety early warning, a safety monitoring module, a data processing module, and safety early warning tools;

[0008] The proprietary large-scale model for coal mine safety early warning is constructed based on target data and relevant knowledge.

[0009] The safety monitoring module is used to acquire and process various data from the coal mine.

[0010] The data processing module is used to clean and preprocess the data to be processed, complete the data format conversion, classify and splice and assemble the data, complete the construction of prompt words and input them into the large model;

[0011] The safety early warning tool is used to notify coal mine safety management personnel of early warning information.

[0012] A method for early warning of coal mine safety based on a large model, the method comprising the following steps:

[0013] S1: Construct a proprietary large-scale model for coal mine safety early warning based on target data and related knowledge, so that it has professional identification, processing and early warning capabilities in the field of coal mine safety early warning, on the basis of having the knowledge understanding ability, logical reasoning ability and decision planning ability of general large-scale models;

[0014] S2: Collect data from various safety monitoring systems in coal mines, such as data values, images, and videos, label them as data to be processed and store them in the database. The data processing module cleans and preprocesses the data to be processed in sequence to ensure data quality and output the data to be tested.

[0015] S3: Based on the data to be detected output in step S2, the data processing module completes the data format conversion, classifies and splices the data, completes the construction of prompt words, and converts it into input data suitable for large model analysis;

[0016] S4: Based on the data to be input assembled in step S3, input the data into the proprietary large-scale model for coal mine safety early warning and wait for the large-scale model to analyze and process it.

[0017] S5: After analyzing the large model, it outputs the detection results and automatically calls and executes various security warning tools based on the results;

[0018] S6: After receiving early warning information through safety early warning tools, coal mine safety management personnel shall immediately take corresponding measures to ensure that the risk of safety accidents is eliminated in a timely manner.

[0019] Preferably, in step S1, the specific process of constructing a proprietary large-scale model for coal mine safety early warning based on target data and relevant knowledge mainly includes:

[0020] (1) Knowledge base analysis: Determine the target purpose, subject scope and content scope of the knowledge base, mainly including data early warning, fault diagnosis and risk analysis of various monitoring and control systems in the field of coal mine safety; the user group and characteristics of the knowledge base mainly include coal mine safety supervisors and business personnel who need to see safety early warning information in a timely and comprehensive manner;

[0021] (2) Target data preparation: Collect data related to the coal mine safety early warning knowledge base, including parameter values, on-site images, monitoring videos, etc., and clean and preprocess the collected data to ensure data quality; then convert the data into a format suitable for large model training.

[0022] (3) Model training: Select a suitable large model, such as BERT or GPT-3, and train it to understand and generate content related to the coal mine safety early warning knowledge base; evaluate the performance of the model and fine-tune the model based on the evaluation results;

[0023] (4) Knowledge base construction: Using the trained large model, extract knowledge from the prepared data, such as national standards, local standards, industry norms, enterprise system regulations, and various safety early warning and handling records, logs, expert historical experience, etc., related to coal mine safety. Organize this knowledge into a structured knowledge base and store it in the vector database, and then verify and improve it.

[0024] (5) Knowledge base assessment: Assess the accuracy, completeness, consistency, usability and accessibility of the knowledge base to ensure that it has the ability to understand and analyze most coal mine safety issues, and make improvements based on the assessment results;

[0025] (6) Model Deployment: Deploy the knowledge base to the production environment, provide access and query interfaces, and complete the construction of a proprietary large model for coal mine safety early warning.

[0026] Preferably, in step S2, the data collected from various safety monitoring and control systems in the coal mine mainly includes various data from the coal mine production process. According to the subsystem classification, these include: safety monitoring, personnel positioning, vehicle positioning, tunneling and anchoring integration, digital industrial video, fully mechanized mining, power supply, drainage, compressed air, ventilation, etc., and are respectively labeled with corresponding data tags. For example, the data tag for the safety monitoring system is "Safety_Monitoring_System", and the tag for data to be processed is "wait_process". The data processing module will clean and preprocess the above-mentioned data to be processed in sequence to ensure data quality. Among them, subsystems and equipment with higher priority will be preprocessed first.

[0027] Preferably, in step S3, the construction steps of the prompt word project mainly include:

[0028] (1) Design reasonable introductory terms

[0029] Based on data type, system function, equipment attributes, etc., analyze and design reasonable introductory terms, such as "Please carefully review the following real-time safety monitoring data from the XX coal mine area, use your safety early warning analysis capabilities to provide us with a detailed safety assessment report, and promptly issue early warnings of safety risks";

[0030] For example, “Suppose you are an experienced coal mine safety expert. Now, based on the following monitoring data, please predict and point out any potential safety risks and issue early warnings about risks that may affect safe production.”

[0031] (2) Data description and input format specifications

[0032] Data descriptions and input format specifications are defined according to data type, system function, and device attributes. Below is a simple example:

[0033] Data source and type: such as "These data cover multiple key monitoring points of XX Coal Mine, including but not limited to: methane concentration (CH4), carbon monoxide concentration (CO), temperature, humidity, wind speed, equipment operating status (such as ventilation fans, gas extraction pumps), etc."

[0034] Input format example: "Data will be provided in JSON format. Each data item contains a timestamp, monitoring point name, monitoring indicator, and corresponding value. For example: {"timestamp":"2023-04-01T12:00:00","location":"main tunnel","metrics":[{"name":"CH4","value":0.5},{"name":"Temperature","value":25}]}."

[0035] (3) Task specification

[0036] Based on data type, system function, device attributes, etc., the security risks requiring early warning are specified into specific tasks. The following is a simple example:

[0037] Safety early warning analysis task: For example, “For each monitoring point, please analyze whether the methane concentration exceeds the standard (generally the warning line is 1.0%), whether the temperature is abnormal (considering the normal temperature range of the coal mine working face), and whether any equipment is in an abnormal operating state. For any potential risks found, please clearly indicate the risk type (such as methane accumulation, fire risk, equipment failure), hazard level (such as low, medium, high), and recommended emergency measures and early warning tool interfaces.”

[0038] (4) Expected output request word

[0039] Based on data type, system function, device attributes, etc., the format of the expected output words is standardized. The following is a simple example:

[0040] Expected output format: "The output should be in structured text or JSON format, containing the safety assessment results for each monitoring point. For example: [{"location":"Main tunnel","risks":[{"type":"Methane accumulation","level":"Medium","measures":"Increase ventilation, strengthen monitoring"}],"overall_safety_level":"Good"}]."

[0041] If the output includes warning information, then the corresponding output should include risk type, hazard level, handling method, and warning tool interface. For example: [{"location":"Main roadway","risks":[{"type":"Methane accumulation","level":"High","measures":"Immediate on-site handling is required, and the main handling measures are xxxxxx",alert":"Alarm has been triggered",alert_interface":

[0042] "http: / / 1.85.1.195:8080 / alertxxx",}]}].

[0043] Preferably, in step S4, the order in which the input data is processed by the large model is based on the order in which the data is assembled in S3. In particular, for some important coal mine monitoring systems and equipment, high-priority detection can be set independently to ensure that important risks can be identified and warned in a timely manner.

[0044] Preferably, in step S5, various security warning tools include, but are not limited to, computers, mobile phones, voice broadcasts, alarm linkage, video calls, etc., and various analysis tools are called by the large model in the form of HTTP interfaces when a security warning is generated.

[0045] (III) Beneficial Effects

[0046] The beneficial effects of this invention are as follows:

[0047] This coal mine safety early warning system and method, based on a large model, applies large model technology to coal mine safety early warning operations. Building upon the powerful semantic understanding, logical reasoning, and decision-making capabilities of general large models, it constructs a large coal mine safety early warning model through model training, knowledge base construction, and knowledge base evaluation. This results in a proprietary large model capable of identifying various safety risks and accidents in coal mines, judging potential risks and hazards, and automatically generating early warning information. This effectively improves the quality and efficiency of safety risk early warning operations, ensures the safety and efficiency of coal mine production, and enables multi-dimensional, expert-perspective safety early warning analysis and alert functions. Attached Figure Description

[0048] Figure 1 This is an overall flowchart of the present invention;

[0049] Figure 2 This is a flowchart illustrating the workflow of the proprietary large-scale model for coal mine safety early warning in this invention.

[0050] Figure 3 This is a flowchart illustrating the process of constructing the prompt word project for this invention. Detailed Implementation

[0051] 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.

[0052] Example

[0053] A coal mine safety early warning system based on a large model, comprising a proprietary large model for coal mine safety early warning, a safety monitoring module, a data processing module, and safety early warning tools;

[0054] The proprietary large-scale model for coal mine safety early warning is constructed based on target data and relevant knowledge.

[0055] The safety monitoring module is used to acquire and process various data from the coal mine.

[0056] The data processing module is used to clean and preprocess the data to be processed, complete the data format conversion, classify and splice and assemble the data, complete the construction of prompt words and input them into the large model;

[0057] The safety early warning tool is used to notify coal mine safety management personnel of early warning information.

[0058] A method for early warning of coal mine safety based on a large model, the method comprising the following steps:

[0059] S1: Construct a proprietary large-scale model for coal mine safety early warning based on target data and related knowledge, so that it has professional identification, processing and early warning capabilities in the field of coal mine safety early warning, on the basis of having the knowledge understanding ability, logical reasoning ability and decision planning ability of general large-scale models;

[0060] S2: Collect data from various safety monitoring systems in coal mines, such as data values, images, and videos, label them as data to be processed and store them in the database. The data processing module cleans and preprocesses the data to be processed in sequence to ensure data quality and output the data to be tested.

[0061] S3: Based on the data to be detected output in step S2, the data processing module completes the data format conversion, classifies and splices the data, completes the construction of prompt words, and converts it into input data suitable for large model analysis;

[0062] S4: Based on the data to be input assembled in step S3, input the data into the proprietary large-scale model for coal mine safety early warning and wait for the large-scale model to analyze and process it.

[0063] S5: After analyzing the large model, it outputs the detection results and automatically calls and executes various security warning tools based on the results;

[0064] S6: After receiving early warning information through safety early warning tools, coal mine safety management personnel shall immediately take corresponding measures to ensure that the risk of safety accidents is eliminated in a timely manner.

[0065] In step S1, the specific process of constructing a proprietary large-scale model for coal mine safety early warning based on target data and relevant knowledge mainly includes:

[0066] (1) Knowledge base analysis: Determine the target purpose, subject scope and content scope of the knowledge base, mainly including data early warning, fault diagnosis and risk analysis of various monitoring and control systems in the field of coal mine safety; the user group and characteristics of the knowledge base mainly include coal mine safety supervisors and business personnel who need to see safety early warning information in a timely and comprehensive manner;

[0067] (2) Target data preparation: Collect data related to the coal mine safety early warning knowledge base, including parameter values, on-site images, monitoring videos, etc., and clean and preprocess the collected data to ensure data quality; then convert the data into a format suitable for large model training.

[0068] (3) Model training: Select a suitable large model, such as BERT or GPT-3, and train it to understand and generate content related to the coal mine safety early warning knowledge base; evaluate the performance of the model and fine-tune the model based on the evaluation results;

[0069] (4) Knowledge base construction: Using the trained large model, extract knowledge from the prepared data, such as national standards, local standards, industry norms, enterprise system regulations, and various safety early warning and handling records, logs, expert historical experience, etc., related to coal mine safety. Organize this knowledge into a structured knowledge base and store it in the vector database, and then verify and improve it.

[0070] (5) Knowledge base assessment: Assess the accuracy, completeness, consistency, usability and accessibility of the knowledge base to ensure that it has the ability to understand and analyze most coal mine safety issues, and make improvements based on the assessment results;

[0071] (6) Model Deployment: Deploy the knowledge base to the production environment, provide access and query interfaces, and complete the construction of a proprietary large model for coal mine safety early warning.

[0072] In step S2, the data collected from various safety monitoring and control systems in the coal mine mainly includes various data from the coal mine production process. According to the subsystem classification, these include: safety monitoring, personnel positioning, vehicle positioning, tunneling and anchoring integrated system, digital industrial video, fully mechanized mining, power supply, drainage, compressed air, ventilation, etc., and are labeled with corresponding data tags. For example, the data tag for the safety monitoring system is "Safety_Monitoring_System", and the tag for data to be processed is "wait_process". The data processing module will clean and preprocess the above data to be processed in sequence to ensure data quality. Among them, subsystems and equipment with higher priority will be preprocessed first.

[0073] In step S3, the main steps in constructing the prompt word project include:

[0074] (1) Design reasonable introductory terms

[0075] Based on data type, system function, equipment attributes, etc., analyze and design reasonable introductory terms, such as "Please carefully review the following real-time safety monitoring data from the XX coal mine area, use your safety early warning analysis capabilities to provide us with a detailed safety assessment report, and promptly issue early warnings of safety risks";

[0076] For example, “Suppose you are an experienced coal mine safety expert. Now, based on the following monitoring data, please predict and point out any potential safety risks and provide early warnings of risks that may affect safe production.”

[0077] (2) Data description and input format specifications

[0078] Data descriptions and input format specifications are defined according to data type, system function, and device attributes. Below is a simple example:

[0079] Data source and type: such as "These data cover multiple key monitoring points of XX Coal Mine, including but not limited to: methane concentration (CH4), carbon monoxide concentration (CO), temperature, humidity, wind speed, equipment operating status (such as ventilation fans, gas extraction pumps), etc."

[0080] Input format example: Data will be provided in JSON format. Each data item contains a timestamp, monitoring point name, monitoring indicator, and corresponding value. For example: {"timestamp":"2023-04-01T12:00:00","location":"main roadway","metrics":[{"name":"CH4","value":0.5},{"name":"Temperature","value":25}]}.

[0081] (3) Task specification

[0082] Based on data type, system function, device attributes, etc., the security risks requiring early warning are specified into specific tasks. The following is a simple example:

[0083] Safety early warning analysis task: For example, "For each monitoring point, please analyze whether its methane concentration exceeds the standard (generally the warning line is 1.0%), whether the temperature is abnormal (considering the normal temperature range of the coal mine working face), and whether any equipment is in an abnormal operating state. For any potential risks found, please clearly indicate the risk type (such as methane accumulation, fire risk, equipment failure), hazard level (such as low, medium, high), and recommended emergency measures and early warning tool interfaces."

[0084] (4) Expected output request word

[0085] Based on data type, system function, device attributes, etc., the format of the expected output words is standardized. The following is a simple example:

[0086] Expected output format: "The output should be in structured text or JSON format, containing the safety assessment results for each monitoring point. For example: [{"location":"Main tunnel","risks":[{"type":"Methane accumulation","level":"Medium","measures":"Increase ventilation and strengthen monitoring"}],"overall_safety_level":"Good"}]."

[0087] If the output includes warning information, then the corresponding output should include risk type, hazard level, handling method, and warning tool interface. For example: [{"location":"Main roadway","risks":[{"type":"Methane accumulation","level":"High","measures":"Immediate on-site handling is required, and the main handling measures are xxxxxx",alert":"Alarm has been triggered",alert_interface":

[0088] "http: / / 1.85.1.195:8080 / alertxxx",}]}]".

[0089] In step S4, the order in which the input data is processed by the large model follows the chronological order in which the data was assembled in S3. In particular, for some important coal mine monitoring systems and equipment, high-priority detection can be set independently to ensure that important risks can be identified and warned in a timely manner.

[0090] In step S5, various security warning tools include, but are not limited to, computers, mobile phones, voice broadcasts, alarm linkage, video calls, etc. All kinds of analysis tools are called by the large model in the form of HTTP interfaces when a security warning is generated.

[0091] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A coal mine safety early warning system based on a large model, characterized by: The system includes a large model for coal mine safety early warning, a safety monitoring module, a data processing module, and a safety early warning tool; The coal mine safety early warning proprietary large model is constructed based on target data and relevant knowledge; The safety monitoring module is used to obtain data to be processed from various data of the coal mine; The data processing module is used to clean and preprocess the data to be processed, complete data format conversion, classify and splice the data, assemble the data, complete the prompt word engineering construction and input it into the large model; The safety warning tool is used to notify the coal mine safety management personnel of the warning information.

2. The early warning method of a coal mine safety early warning system based on a large model according to claim 1, characterized in that: The method comprises the following steps: S1: Construct a coal mine safety early warning proprietary big model based on target data and related knowledge, so that it has the professional identification, processing and early warning capabilities in the field of coal mine safety early warning on the basis of the knowledge understanding, logical reasoning and decision-making planning capabilities of the general big model; S2: Collect various safety monitoring system data of coal mines, such as data values, images, videos, etc., mark them as data to be processed and store them in the database. The data processing module cleans and preprocesses the data to be processed in turn to ensure data quality and output the data to be tested; S3: Based on the data to be detected outputted in step S2, the data processing module completes data format conversion, classifies and splices the data, assembles the data, completes the construction of the prompt word project, and converts the data into input data suitable for large model analysis; S4: Based on the data to be input assembled in step S3, the data is input into the coal mine safety early warning dedicated large model, and waits for the large model to analyze and process; S5: After the large model analyzes, it outputs the detection results and automatically calls and executes various safety warning tools based on the results; S6: After receiving the warning information through the safety warning tool, the coal mine safety management personnel will immediately take corresponding measures to ensure that the risk of safety accidents is eliminated in a timely manner.

3. The early warning method of a coal mine safety early warning system based on a large model according to claim 2 is characterized by: In step S1, the specific process of constructing a large-scale coal mine safety early warning proprietary model based on target data and relevant knowledge mainly includes: (1) Knowledge base analysis: Determine the target purpose, subject scope and content scope of the knowledge base, mainly including data warning, fault judgment, risk analysis, etc. of various monitoring and control systems in the field of coal mine safety; the user groups and characteristics of the knowledge base, mainly including coal mine safety supervisors and business personnel, who need to see safety warning information in a timely and comprehensive manner; (2) Target data preparation: Collect data related to the topic of the coal mine safety warning knowledge base, including parameter values, on-site images, monitoring videos, etc., and clean and preprocess the collected data to ensure data quality; then convert the data into a format suitable for large model training. (3) Model training: Select a suitable large model, such as BERT, GPT-3, etc., and train it to enable it to understand and generate content related to the topic of the coal mine safety warning knowledge base; evaluate the performance of the model and fine-tune the model based on the evaluation results; (4) Knowledge base construction: Use the trained large model to extract knowledge from the prepared data, such as national standards, local standards, industry specifications, enterprise regulations related to coal mine safety, and various safety warning and disposal records, logs, and expert historical experience. Organize this knowledge into a structured knowledge base and store it in the vector database for verification and improvement. (5) Knowledge base evaluation: Evaluate the accuracy, completeness, consistency, usability, and accessibility of the knowledge base to ensure that it has the ability to understand and analyze most coal mine safety issues, and make improvements based on the evaluation results; (6) Model deployment: Deploy the knowledge base to the production environment, provide access and query interfaces, and complete the construction of a large model for coal mine safety warning.

4. The early warning method of a coal mine safety early warning system based on a large model according to claim 2 is characterized by: In the step S2, the data collected from various safety monitoring systems of coal mines mainly include various data in the production process of coal mines, which are classified according to subsystems, including: safety monitoring, personnel positioning, vehicle positioning, integrated digging and anchoring, digital industrial video, comprehensive mining, power supply, drainage, compressed air, ventilation, etc., and are marked with corresponding data labels respectively. The data processing module will clean and pre-process the above-mentioned data to be processed in turn to ensure data quality. Among them, subsystems and equipment with higher priorities will be pre-processed first.

5. The early warning method of a coal mine safety early warning system based on a large model according to claim 2 is characterized by: In step S3, the steps of constructing the prompt word project mainly include: (1) Design reasonable introduction words Analyze and design reasonable introduction words according to data types, system functions, device attributes, etc.; (2) Data description and input format specifications According to data type, system function, device attributes, etc., data description and input format specification are carried out respectively; (3) Task Specification According to data types, system functions, equipment attributes, etc., tasks are specified for security risks that require early warning; (4) Expected output request word According to data type, system function, device attribute, etc., the format of the expected output result words is standardized.

6. The early warning method of a coal mine safety early warning system based on a large model according to claim 2 is characterized by: In step S4, the order of the input data for large-model analysis and processing is the chronological order of the data assembly in S3. In particular, for some important coal mine monitoring systems and equipment, high-priority detection can be set independently to ensure that important risks can be identified and warned in a timely manner.

7. The early warning method of a coal mine safety early warning system based on a large model according to claim 2 is characterized by: In step S5, various security warning tools include but are not limited to computers, mobile phones, voice broadcasts, alarm linkage, video calls, etc. Various analysis tools are called by the big model in the form of http interfaces when a security warning is generated.

Images