Mining area supervision method and system based on intelligent agent, and electronic equipment

Abstract

The invention relates to the field of deep application of an artificial intelligence technology, in particular to a mining area supervision method and system based on an intelligent agent and electronic equipment. The method comprises the following steps: in response to a received potential safety hazard analysis task for a mining area, calling an intelligent research and judgment agent to obtain mining area scene potential hazard information, and calling an intelligent text analysis agent to obtain mining area data potential hazard information; in response to the received law enforcement scheme compiling task, calling a law enforcement scheme compiling agent, and generating a law enforcement scheme corresponding to the mining area scene hidden danger information and the mining area data hidden danger information; in response to the received on-site law enforcement task for the mining area, calling a law enforcement assistant agent to generate supervision prompt information according to the law enforcement scheme, and obtaining an on-site law enforcement result corresponding to the supervision prompt information; and in response to the received law enforcement document generation task, calling the law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result. By adopting the scheme, the intelligent level of supervision and supervision can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of deep application of artificial intelligence technology, and in particular to an agent-based mine area supervision and inspection method, system and electronic device. Background Art

[0002] Supervision and inspection is an important means to ensure the safe production of coal mines. Limited by the limited number of supervision and inspection personnel, the experience of law enforcement officers has limitations, and the lack of intelligent law enforcement means. At present, there are still problems in law enforcement, such as the supervision and law enforcement plan is not accurate enough, mainly relying on the law enforcement department and regional law enforcement personnel to conduct investigations, spot checks and surprise inspections; the material comparison and law enforcement process mainly rely on manpower, resulting in missed inspections, inability to detect problems, large resource consumption, high labor costs and low efficiency; the supervision and law enforcement system cannot empower on-site law enforcement, and the law enforcement results cannot be fairly obtained. Summary of the Invention

[0003] The present disclosure aims to at least partly solve one of the technical problems in the related art.

[0004] To this end, the first object of the present disclosure is to propose an agent-based mine area supervision and inspection method to improve the intelligent level of supervision and inspection.

[0005] The second object of the present disclosure is to propose an agent-based mine area supervision and inspection system.

[0006] The third object of the present disclosure is to propose an electronic device.

[0007] The fourth object of the present disclosure is to propose a computer-readable storage medium.

[0008] The fifth object of the present disclosure is to propose a computer program product.

[0009] To achieve the above object, the first aspect embodiment of the present disclosure proposes an agent-based mine area supervision and inspection method, including: In response to receiving a safety hazard analysis task for a mine area, calling an intelligent judgment agent to obtain mine area scene hazard information, and calling an intelligent text analysis agent to obtain mine area material hazard information; In response to receiving a law enforcement plan compilation task, calling a law enforcement plan arrangement agent to generate a law enforcement plan corresponding to the mine area scene hazard information and the mine area material hazard information; In response to receiving an on-site law enforcement task for the mine area, calling a law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtaining an on-site law enforcement result corresponding to the supervision prompt information; In response to receiving a law enforcement document generation task, calling a law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

[0010] Optionally, the method further includes: Obtaining supervision and monitoring information input by law enforcement officers; Decomposing the supervision and monitoring information into tasks to determine the supervision and monitoring tasks corresponding to the supervision and monitoring information, where the supervision and monitoring tasks are any one of the potential hazard analysis tasks, the law enforcement plan compilation tasks, the on-site law enforcement tasks, and the law enforcement document generation tasks.

[0011] Optionally, the potential hazard information of the mining area scene includes personnel active warning information, and obtaining the potential hazard information of the mining area scene includes: Obtaining mining area monitoring images, locating the human figures in the mining area monitoring images, and constructing a skeleton model of the human figures; Associating and analyzing the movement trajectories of the skeleton models in consecutive frames of the mining area monitoring images to obtain the spatial movement rules and action frequencies of the skeleton models; Matching the spatial movement rules and the dynamic behaviors corresponding to the action frequencies based on temporal action features, and if the dynamic behavior is a personnel active warning dynamic behavior, generating the personnel active warning information corresponding to the dynamic behavior.

[0012] Optionally, the potential hazard information of the mining area scene includes abnormal personnel behavior trajectory information, and obtaining the potential hazard information of the mining area materials includes: Obtaining mining area monitoring images and detecting the personnel movement trajectories in the mining area monitoring images; Obtaining positioning card data, comparing the personnel movement trajectories with the positioning card data, and if the personnel movement trajectories do not match the positioning card data, generating abnormal personnel behavior trajectory information corresponding to the personnel movement trajectories and the positioning card data.

[0013] Optionally, the potential hazard information of the mining area scene includes abnormal environmental parameter information, and obtaining the potential hazard information of the mining area materials includes: Obtaining gas monitoring data; Predicting the temporal characteristics of gas concentration corresponding to the gas monitoring data; Matching the gas monitoring data with an abnormal pattern library to obtain abnormal samples corresponding to the gas monitoring data; Calculating the similarity between the temporal characteristics of gas concentration and the abnormal samples based on the weighted dynamic time warping algorithm; Determining the abnormal environmental parameter information corresponding to the gas monitoring data according to the similarity.

[0014] Optionally, the calling of the intelligent text analysis agent to obtain the potential hazard information of the mining area materials includes: Obtain the multi-source heterogeneous data uploaded by the mining area, and perform feature extraction on the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data; Extract the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base; Determine the potential hazard information of the mining area materials corresponding to the multi-source heterogeneous data according to the sub-graph.

[0015] Optionally, before extracting the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base, the method further includes: Obtain the regulation data in the coal mine safety production knowledge base; Extract the entities and associated production features in the regulation data, and map the associated production features to the relationships between different entities in the initial knowledge graph to obtain the knowledge graph corresponding to the coal mine safety production knowledge base.

[0016] Optionally, the performing feature extraction on the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data includes: Perform structured processing on the multi-source heterogeneous data to obtain structured multi-source heterogeneous data; Control the business rule engine to perform feature extraction on the structured multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data.

[0017] Optionally, the extracting the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base includes: Perform word space similarity matching between the feature data and the knowledge graph to obtain the sub-graph corresponding to the feature data.

[0018] To achieve the above object, an embodiment of the second aspect of the present disclosure proposes an intelligent agent-based mining area supervision and inspection system, including: A potential hazard analysis unit, configured to, in response to receiving a safety potential hazard analysis task for a mining area, call an intelligent judgment intelligent agent to obtain the potential hazard information of the mining area scene, and call an intelligent text analysis intelligent agent to obtain the potential hazard information of the mining area materials; A plan preparation unit, configured to, in response to receiving a law enforcement plan preparation task, call a law enforcement plan arrangement intelligent agent to generate a law enforcement plan corresponding to the potential hazard information of the mining area scene and the potential hazard information of the mining area materials; A law enforcement assistance unit, configured to, in response to receiving a on-site law enforcement task for the mining area, call a law enforcement assistant intelligent agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; A document generation unit, configured to, in response to receiving a law enforcement document generation task, call a law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

[0019] To achieve the above object, an embodiment of the third aspect of the present disclosure provides an electronic device, including: A memory, configured to store executable program code; A processor, configured to call and run the executable program code from the memory, so that the electronic device executes the method shown in any one of the foregoing first aspects.

[0020] To achieve the above object, an embodiment of the fourth aspect of the present disclosure provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the computer program is executed, the method shown in any one of the foregoing first aspects is implemented.

[0021] To achieve the above object, an embodiment of the fifth aspect of the present disclosure provides a computer program product, including a computer program, and when the computer program is executed by a processor, the method shown in any one of the foregoing first aspects is implemented.

[0022] In summary, the method, system, and electronic device provided by the present disclosure can sink the safety hazard inspection to the mining area data layer by applying an agent. Hidden major safety hazards gradually emerge through multi-modal and multi-source data analysis, solving the problem of the last mile of on-site law enforcement, which is conducive to urging the mining party to actively create safety and prevent major and particularly serious mining accidents. Secondly, the application of the agent can solve the problem of precise and differentiated law enforcement, significantly reducing the frequency of law enforcement running mines, which is helpful for improving the production efficiency of the mining area and effectively assisting law enforcement officers to improve law enforcement efficiency. In addition, by applying the agent to empower the supervision and law enforcement business process, the intelligence of supervision and inspection can be improved based on artificial intelligence technology means, the intelligence level of on-site law enforcement can be improved, the law enforcement efficiency can be improved, and the usability, operability, and reliability of the traditional supervision and inspection system can be improved.

[0023] The additional aspects and advantages of the present disclosure will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The above and / or additional aspects and advantages of the present disclosure will become obvious and easy to understand from the following description of the embodiments in conjunction with the drawings, where: Figure 1 is a schematic flowchart of a method for mining area supervision and inspection based on an agent provided by an embodiment of the present disclosure; Figure 2 is a schematic flowchart of a method for mining area supervision and inspection based on an agent provided by an embodiment of the present disclosure; Figure 3 Schematic flowchart of an agent - based mining area supervision and inspection method provided by an embodiment of the present disclosure; Figure 4 Schematic flowchart of a method for obtaining potential hazard information of mining area data provided by an embodiment of the present disclosure; Figure 5 Schematic structural diagram of an agent - based mining area supervision and inspection system provided by an embodiment of the present disclosure. Detailed implementation manners

[0025] The following details the embodiments of the present disclosure. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements with the same or similar functions throughout. The embodiments described by referring to the accompanying drawings are exemplary and are intended to explain the present disclosure, rather than to be construed as a limitation of the present disclosure.

[0026] It should be noted that the current supervision and law - enforcement process mainly has the following problems: Problem 1: There is a lack of intelligent inspection means for the multi - element basic data uploaded by the mining area; Problem 2: There is no plan for law - enforcement inspections or spot checks, and it is impossible to accurately implement differential inspections; Problem 3: The law - enforcement results lack fairness. The compilation of a series of documents is time - consuming and prone to human errors.

[0027] In the context of the continuous iteration of artificial intelligence technology and the increasing perfection of the intelligent construction of the mining area, the application of artificial intelligence (AI) technology is advancing in depth from basic scenarios to core business fields. Its development shows significant characteristics of leaping from single - perception to multi - dimensional interaction and upgrading from simple recognition to complex decision - making. This technological evolution trend provides a new path for building accurate supervision and inspection based on intelligent algorithms, which can effectively improve the scientific nature and execution efficiency of supervision work, and strengthen the objective fairness of the supervision and law - enforcement process through a data - driven decision - making mechanism.

[0028] The following details the present disclosure with specific embodiments.

[0029] In the first embodiment, as Figure 1 shown, Figure 1 Schematic flowchart of an agent - based mining area supervision and inspection method provided by an embodiment of the present disclosure. This method can be implemented depending on a computer program and can run on a system for agent - based mining area supervision and inspection. This computer program can be integrated into an application or run as an independent tool - type application.

[0030] Among them, the agent - based mining area supervision and inspection method can be executed by an electronic device.

[0031] Exemplarily, the agent-based mining area supervision and inspection method includes the following steps: S101, in response to receiving a safety hazard analysis task for a mining area, call the intelligent judgment agent to obtain mining area scene hazard information, and call the intelligent text analysis agent to obtain mining area data hazard information; According to some embodiments, the safety hazard analysis task is used to indicate the analysis of safety hazards in the mining area.

[0032] In some embodiments, the intelligent judgment agent refers to an agent used to obtain mining area scene hazard information. The mining area scene hazard information refers to hazard information related to the mining area scene.

[0033] In some embodiments, the intelligent text analysis agent refers to an agent used to obtain mining area data hazard information. The mining area data hazard information refers to hazard information related to the production data of the mining area.

[0034] S102, in response to receiving a law enforcement plan compilation task, call the law enforcement plan arrangement agent to generate a law enforcement plan corresponding to the mining area scene hazard information and the mining area data hazard information; According to some embodiments, the law enforcement plan compilation task is used to indicate the compilation of a law enforcement plan.

[0035] In some embodiments, the law enforcement plan arrangement agent refers to an agent used to compile a law enforcement plan. The law enforcement plan arrangement agent can generate a law enforcement plan corresponding to the mining area scene hazard information and the mining area data hazard information.

[0036] S103, in response to receiving a on-site law enforcement task for a mining area, call the law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; According to some embodiments, the on-site law enforcement task is used to indicate that law enforcement officers are conducting on-site law enforcement in the mining area and need law enforcement assistance.

[0037] In some embodiments, the law enforcement assistant agent refers to an agent used for assisting law enforcement. The law enforcement assistant agent can generate supervision prompt information in real time for law enforcement officers to refer to during the process of on-site law enforcement in the mining area by law enforcement officers.

[0038] In some embodiments, the on-site law enforcement result refers to the law enforcement result obtained after law enforcement officers complete on-site law enforcement in the mining area.

[0039] S104, in response to receiving a law enforcement document generation task, call the law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

[0040] According to some embodiments, the law enforcement document generation task is used to indicate the generation of intelligent documents.

[0041] In some embodiments, the law enforcement document generation agent refers to an agent that generates law enforcement documents. This law enforcement document generation agent can be used to generate intelligent documents corresponding to on-site law enforcement results.

[0042] In summary, in the method provided in this embodiment, by applying agents, the safety hazard inspection can be sunk to the mining area data layer, and hidden major safety hazards gradually emerge through multi-modal and multi-source data analysis, solving the problem of the last mile of on-site law enforcement, which is conducive to urging the mining party to actively create a safe environment and prevent major and particularly serious mining accidents; secondly, the application of agents can solve the problem of precise and differentiated law enforcement, significantly reducing the frequency of law enforcement evasion, which is helpful for improving the production efficiency of the mining area and effectively assisting law enforcement officers to improve law enforcement efficiency; in addition, by applying agents to empower the supervision and law enforcement business process, the intelligence of supervision and inspection can be improved based on artificial intelligence technology means, the intelligence level of on-site law enforcement can be improved, law enforcement efficiency can be increased, and the effectiveness, usability and reliability of traditional supervision and inspection systems can be improved.

[0043] Another embodiment of the present disclosure provides a mining area supervision and inspection method based on agents. This method can be executed by an electronic device.

[0044] Exemplarily, the mining area supervision and inspection method based on agents may include the following steps: S201, obtain the supervision and monitoring information input by law enforcement officers; According to some embodiments, the supervision and monitoring information can be text information or language information. All functions can be called through voice / natural language at one entrance for unified display.

[0045] In some embodiments, Figure 2 is a schematic flow diagram of a mining area supervision and inspection method based on agents provided in an embodiment of the present disclosure. As Figure 2 shown, the supervision and monitoring information can be, for example, the tasks and / or problems input by law enforcement officers.

[0046] S202, perform task decomposition on the supervision and monitoring information to determine the supervision and monitoring tasks corresponding to the supervision and monitoring information; According to some embodiments, as Figure 2 shown, the task decomposition device can perform task decomposition on the supervision and monitoring information to determine the supervision and monitoring tasks corresponding to the supervision and monitoring information.

[0047] In some embodiments, the task decomposition device can decompose the tasks and / or problems input by law enforcement officers according to semantic understanding, and divide them into individual tasks, namely Tasks.

[0048] In some embodiments, the task disassembler can be served by a large model (such as deepseek), for example.

[0049] According to some embodiments, the supervision and monitoring task is any one of a potential safety hazard analysis task, a law enforcement plan compilation task Task3, a on-site law enforcement task Task4, and a law enforcement document generation task Task5; among them, the potential safety hazard analysis task includes a one-key potential safety hazard analysis task Task1 and a mine area basic data analysis and early warning task Task2, as Figure 2 shown.

[0050] In some embodiments, before law enforcement, law enforcement officers can initiate a potential safety hazard analysis task through language information and / or text information, thereby automatically triggering a one-key potential safety hazard analysis task Task1 and a mine area basic data analysis and early warning task Task2.

[0051] According to some embodiments, as Figure 2 shown, the supervision and monitoring tasks correspond one-to-one with the agents in the Agent Center. Among them, the intelligent judgment agent is used to execute Task1, the text generation agent is used to execute Task2, the law enforcement plan arrangement agent is used to execute Task3, the law enforcement assistant agent is used to execute Task4, and the intelligent text analysis agent is used to execute Task5.

[0052] In some embodiments, as Figure 2 shown, the output results of each agent in the Agent Center, that is, the key information, can be transmitted to other tasks through the global parameter transmission agent to achieve the overall goal.

[0053] In some embodiments, as Figure 2 shown, the output of each agent in the Agent Center can be summarized and output for law enforcement officers to evaluate and give feedback, assisting in the iterative upgrade of the agents.

[0054] In some embodiments, as Figure 2 shown, an enhanced retrieval agent (RAG Agent) can also be set up to retrieve the knowledge base, solve the problems of knowledge limitation and hallucination, improve the processing ability of complex tasks, and optimize the output quality and professionalism.

[0055] In some embodiments, as Figure 2As shown, a prompt word transmitter Agent can also be set up to transmit a detailed description of the task, related information, and the output result of the previous task to each agent in the form of a prompt, ensuring that the agent receives accurate and complete input information, thereby guiding the agent to perform the task according to the expected goal and improving the accuracy and efficiency of task processing.

[0056] It should be noted that by adopting the "planning-execution-tool-verification" architecture, the task decomposer is used as planning, each task is used as execution, the Agent Center is used as a packaged tool, and the summary output is used as verification. Therefore, artificial intelligence technology can be used to improve the intelligence of supervision and law enforcement, solve the problem of the last mile of on-site law enforcement, and effectively assist law enforcement personnel to improve law enforcement efficiency.

[0057] S203, in response to receiving a safety hazard analysis task for a mining area, calling an intelligent analysis agent to obtain scene hazard information of the mining area; According to some embodiments, the intelligent analysis and judgment agent can conduct a one-click inspection of various monitoring systems and law enforcement-specific scenario algorithms for target law enforcement units, and finally summarize all safety hazards identified.

[0058] For example, the intelligent judgment agent can conduct remote intelligent judgment of mining risks by combining vision and data analysis. In this case, the hidden danger information of mining scenes includes but is not limited to personnel active warning information, personnel behavior trajectory abnormal information, and environmental parameter abnormal information. Therefore, the three core scene problems of personnel active warning, environmental parameter abnormality, and personnel behavior abnormality can be solved to mine law enforcement clues and evidence.

[0059] Among them, the intelligent analysis and judgment agent can use multimodal data fusion analysis based on deep learning to obtain personnel active warning information, abnormal personnel behavior trajectory information, and abnormal environmental parameter information.

[0060] In some embodiments, for active warning information from personnel, the intelligent judgment agent can obtain mining area monitoring images, locate the human figure in the mining area monitoring images, and construct a skeleton model of the human figure; associate and analyze the motion trajectory of the skeleton model of consecutive frames in the mining area monitoring images to obtain the spatial movement law and action frequency of the skeleton model; match the spatial movement law and the dynamic behavior corresponding to the action frequency based on the temporal action characteristics, and if the dynamic behavior is a dynamic behavior of active warning from personnel, then generate active warning information from personnel corresponding to the dynamic behavior. Therefore, by extracting motion features through human posture estimation and combining the temporal action mode to capture dynamic laws, accurate behavior recognition can be achieved, thereby improving the accuracy of obtaining active warning information from personnel.

[0061] Among them, the dynamic behavior of personnel taking the initiative to give a warning can be, for example, "waving to give a warning". In this case, regarding the spatial movement pattern, it mainly lies in identifying the spatial movement pattern of the hand and elbow in consecutive frames (such as the change in the angle of the elbow joint, the arc of the wrist trajectory, etc.). Therefore, it is possible to put people first, focus on the initiative of personnel to give warnings, and for the underground environment with complex lighting and task occlusion, based on human pose estimation and temporal feature modeling, identify the waving of personnel, and achieve high-robustness and low-latency dynamic behavior determination.

[0062] Among them, a deep learning model can be used to locate the portrait in the mining area monitoring image and extract the key point coordinates of joints such as the shoulder, elbow, and wrist in real time to construct a skeleton model of the portrait.

[0063] In some embodiments, for the abnormal information of the personnel behavior trajectory, the intelligent judgment agent can obtain the mining area monitoring image and detect the personnel movement trajectory in the mining area monitoring image; obtain the positioning card data, and compare the personnel movement trajectory with the positioning card data. If the personnel movement trajectory does not match the positioning card data, the abnormal information of the personnel behavior trajectory corresponding to the personnel movement trajectory and the positioning card data is generated. Therefore, by combining the personnel movement trajectory and the positioning card data to determine the abnormal information of the personnel behavior trajectory, the accuracy of obtaining the abnormal information of the personnel behavior trajectory can be improved.

[0064] Among them, when detecting the personnel movement trajectory in the mining area monitoring image, the personnel movement trajectory in a specific area can be monitored, and the specific area can be adjusted according to the actual application scenario. The specific area can be delimited, for example, by using virtual electronic fence technology. Therefore, the safety and effectiveness of area control can be ensured.

[0065] Among them, by fusing multi-source perception data and a logical discrimination mechanism, the personnel movement trajectory and the positioning card data can be associated and matched to identify typical abnormal events such as abnormal card carrying, long-term stay, abnormal handover of leaders, and abnormal trajectories of key inspection personnel. For example, if abnormal behaviors such as a person entering the mine without carrying a card or a person staying in the mine without leaving are found, the alarm mechanism can be triggered immediately, so as to achieve timely identification and response to illegal behaviors.

[0066] In some embodiments, for environmental parameter anomaly information, the intelligent judgment agent can obtain gas monitoring data; predict the temporal characteristics of gas concentration corresponding to the gas monitoring data; match the gas monitoring data with the anomaly pattern library to obtain the anomaly samples corresponding to the gas monitoring data; based on the Weighted Dynamic Time Warping (WDTW) algorithm, calculate the similarity between the temporal characteristics of gas concentration and the anomaly samples; and determine the environmental parameter anomaly information corresponding to the gas monitoring data according to the similarity. Therefore, by forming a closed-loop gas monitoring - analysis - warning process, the real-time and reliability of obtaining environmental parameter anomaly information can be ensured; secondly, by introducing an anomaly matching mechanism based on the anomaly pattern library and WDTW, the interpretability and traceability of the warning output can be enhanced.

[0067] Among them, Figure 3 is a schematic flowchart of a method for mine area supervision and inspection based on an intelligent agent provided by an embodiment of the present disclosure. As Figure 3 shown, after gas monitoring data is obtained through data collection, the gas monitoring data can be preprocessed to obtain the preprocessed gas monitoring data; then, the preprocessed gas monitoring data can be respectively input into the prediction model and the anomaly model library to obtain the temporal characteristics of gas concentration and the anomaly samples corresponding to the preprocessed gas monitoring data; then, warning judgment can be carried out according to the temporal characteristics of gas concentration and the anomaly samples. When carrying out warning judgment, based on the weighted dynamic time warping algorithm and combined with position weight adjustment, the similarity between the temporal characteristics of gas concentration and the anomaly samples is calculated; finally, anomaly recognition is carried out according to the similarity, the most likely anomaly category and evolution trend are identified, and the result is output.

[0068] Among them, the anomaly pattern library can be constructed through public accident datasets and simulation anomaly injection. The anomaly pattern library includes but is not limited to a methane anomaly pattern library, a carbon monoxide anomaly pattern library, etc.

[0069] Among them, by applying the weighted dynamic time warping (WDTW) algorithm to real-time gas time series data and combining with the position weight strategy, the matching accuracy can be improved. Specifically, WDTW has good time consistency feature modeling ability. To further highlight the importance of the anomaly occurrence area, a time position weight and a local gradient change weight based on the Gaussian distribution are introduced, so that when calculating the similarity between the gas concentration sequence and the known anomaly samples, more attention is paid to the mutation area and key nodes. This not only ensures the flexible matching of asynchronous time series but also improves the sensitivity to anomaly types, is applicable to gas time series data collected at high frequencies in real-time scenarios, and can effectively assist anomaly traceability and type determination.

[0070] For example, for carbon monoxide monitoring data, based on data processing, a deep learning model can be used to mine the temporal features of methane concentration for prediction. Through statistical analysis, a dynamic threshold can be constructed to achieve hierarchical early warning. The pattern matching algorithm can be used to search and compare in the carbon monoxide abnormal pattern library to identify the abnormal type and give early warning. Further, the sliding window statistical method can be adopted to analyze the distribution of historical carbon monoxide concentration data, extract the 85% and 95% quantiles as the dynamic thresholds for the first-level and second-level early warnings, and combine the fixed alarm thresholds in the relevant coal mine safety regulations to construct a three-level early warning standard, and construct a three-level early warning model based on dynamic classification. On this basis, an abnormal pattern matching mechanism is introduced, and the constructed carbon monoxide abnormal pattern library is used to compare the similarity of real-time data; among them, the improved dynamic time warping (DTW) method is used for the comparison method to enhance the recognition ability of non-linear temporal changes, so as to achieve accurate recognition and hierarchical early warning of different types of abnormalities (such as drilling operations, roof caving, etc.).

[0071] For example, for methane monitoring data, the data smoothing technology can be used to process the original data. Relying on the hybrid model of Long Short-Term Memory (LSTM)-Attention mechanism, the temporal dependence relationship of carbon monoxide concentration can be grasped for prediction. Through the dynamic confidence interval and the duration of the concentration rising trend, hierarchical early warning is set. With the help of the improved pattern matching algorithm, the abnormal category can be identified in the methane abnormal pattern library and early warning can be given. Further, the model inputs a fixed-length historical methane concentration sequence and outputs the predicted values of future multi-step concentrations; based on the prediction results, a dynamic confidence interval is established to quantify the prediction uncertainty, and combined with the duration of the continuous rise of the methane concentration, it is used as the trigger condition for hierarchical early warning, so as to realize the timely identification and grading of abnormal changes.

[0072] Among them, the LSTM unit can effectively capture the long-term and short-term dynamic changes in the concentration sequence through its gating mechanism, and solve the problem of gradient disappearance of the traditional recurrent neural network; the attention mechanism endows the model with weighted attention to the key time steps, especially the information before and after the abnormal mutation of the concentration, which can improve the accuracy and robustness of the prediction.

[0073] It should be noted that for the abnormal information of environmental parameters, based on the gas monitoring data, combined with the time series prediction model, dynamic threshold analysis and abnormal pattern recognition technology, the accurate prediction of the gas / carbon monoxide concentration trend and multi-level early warning judgment can be realized.

[0074] S204, call the intelligent text analysis agent to obtain the hidden danger information of the mining area materials; According to some embodiments, it is possible to obtain the multi-source heterogeneous data uploaded by the mining area, extract features from the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data; extract the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base; and determine the potential hazard information of the mining area materials corresponding to the multi-source heterogeneous data according to the sub-graph.

[0075] In some embodiments, Figure 4 is a schematic flowchart of a process for obtaining potential hazard information of mining area materials provided by an embodiment of the present disclosure. As Figure 4 shown, the intelligent text analysis agent can be based on the coal mine safety knowledge enhancement framework of Graph RAG, based on the coal mine safety production knowledge base, combine major hazard identification criteria and law enforcement cases to construct a business rule engine, and through the construction of a collaborative mechanism between the domain knowledge graph and the large language model, realize the process from semantic retrieval to causal reasoning, so as to realize the one-key analysis and early warning of the risk of the multi-source heterogeneous basic material content uploaded by the mining area, and solve the problems of weak knowledge relevance and insufficient logical reasoning ability of traditional RAG technology.

[0076] Among them, the data in the multi-source heterogeneous data includes but is not limited to tabular data, picture data, text data, other data, etc. After obtaining the multi-source heterogeneous data, the multi-source heterogeneous data can be processed, and the processed multi-source heterogeneous data can be stored in the material report data pool. When processing the multi-source heterogeneous data, the multi-source heterogeneous data can be structured to obtain structured multi-source heterogeneous data. Therefore, the basic material data of the mining area can be normatively reviewed to ensure the effectiveness and consistency of the data. Secondly, the structured multi-source heterogeneous data can be labeled and classified for alignment, and finally stored in the material report data pool.

[0077] Among them, when checking the data in the material report data pool, the data to be checked can be put into the set of mining area materials to be checked, and then, the extraction and / or inspection of the set of mining area materials to be checked (the set of data to be checked) is carried out.

[0078] Among them, after obtaining the structured multi-source heterogeneous data, the structured multi-source heterogeneous data can be subjected to index extraction, extracting data such as multi-type file data, key features, and associated features in tables, word, pdf, pictures, etc.; operations such as associated feature table operation and maintenance, document attribution analysis, and document legality verification can also be carried out.

[0079] For example, the business rule engine can be controlled to extract features from structured heterogeneous data to obtain feature data corresponding to the heterogeneous data. The business rule engine refers to the business rules / workflows solidified according to the major hidden danger judgment criteria and the experience of law enforcement officers. For example, the judgment of over-capacity production mainly depends on the annual approved production capacity and monthly output, and thus it can be inferred whether there is an illegal act of over-capacity production. Another example is the abnormal extraction of work safety expenses, etc.

[0080] Among them, the business rule engine can obtain structured heterogeneous data from the set of data of the mining area to be inspected, and then extract features from it.

[0081] In some embodiments, before extracting the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine work safety knowledge base, it is also necessary to obtain the regulatory data in the coal mine work safety knowledge base; extract the entities and associated production features in the regulatory data, and map the associated production features to the relationships between different entities in the initial knowledge graph to obtain the knowledge graph corresponding to the coal mine work safety knowledge base. Among them, the associated production feature refers to the regulatory feature.

[0082] Among them, the coal mine work safety knowledge base refers to the database related to the coal mine work safety industry. The coal mine work safety knowledge base is an "intelligent toolbox" tailored by the supervision and inspection unit, which covers core materials such as national laws and regulations, industry standards, risk and hidden danger investigation lists, historical accident cases, law enforcement cases, mining area ledger reports, map materials, emergency plans and expert experience, etc., and can transform scattered safety information into a "ready-to-use" practical guide.

[0083] For example, the source documents in the regulatory data can be automatically structured through large model prompt engineering, and the regulatory features in the source documents can be automatically extracted, and then mapped to the entities and the relationships between entities in the knowledge graph, so as to realize the automatic construction of the knowledge graph.

[0084] In some embodiments, when extracting the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine work safety knowledge base, the feature data can be matched with the knowledge graph in terms of word space similarity to obtain the sub-graph corresponding to the feature data, so that the judgment logic and basis for illegal determination can be obtained.

[0085] Among them, the retriever can be configured according to the extracted entities and associated production features, and word space similarity matching is performed based on knowledge graph embedding technology to return the associated sub-graph, and then the existence judgment of hidden dangers and the feedback of illegal basis are combined with the business rule engine.

[0086] S205, in response to receiving the law enforcement plan compilation task, call the law enforcement plan orchestration agent to generate a law enforcement plan corresponding to the hidden danger information of the mining area scene and the hidden danger information of the mining area data; According to some embodiments, the law enforcement plan orchestration agent can, based on the semantic understanding ability and planning ability of the applied large model, and according to the results of safety hazard inspections (including mine area scene hazard information and mine area data hazard information), automatically generate corresponding law enforcement plans in combination with the law enforcement list.

[0087] In some embodiments, the law enforcement plan orchestration agent can also support law enforcement officers to online add, delete, modify, and check the generated law enforcement plans. Therefore, the problems of precise and differentiated law enforcement can be solved, and the inspection frequency can be reduced.

[0088] S206, in response to receiving a on-site law enforcement task for the mine area, call the law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; According to some embodiments, the law enforcement assistant agent can apply large model distillation technology to deploy the large model to the underground terminal mobile phone through a lightweight deep neural network engine (Mobile Neural Network, MNN), and can be used offline for underground weak network / no network scenarios. After connecting to the network, the data is automatically synchronized to solve the problem of operation interruption caused by the network.

[0089] In some embodiments, when the law enforcement officer arrives at the location corresponding to the law enforcement plan in the mine area, the intelligent assistant agent can automatically prompt the law enforcement officer through the earphone about the content, methods, and steps that need to be focused on for inspecting the hazards at this location. Therefore, the law enforcement quality can be improved, and the problem of missed inspections can be solved.

[0090] In some embodiments, for various types of knowledge such as hazard standards and rules and regulations that law enforcement officers have doubts about or need to verify, the law enforcement assistant agent can provide an offline colloquial intelligent question-and-answer service, and provide citation sources to provide on-site knowledge support to law enforcement officers at any time.

[0091] According to some embodiments, the intelligent assistant agent can also support diverse evidence collection methods such as taking pictures, videos, and texts, associate with the specific content and progress of the law enforcement plan, form a traceable electronic evidence chain, meet the law enforcement evidence collection requirements in complex scenarios, and thus generate corresponding on-site law enforcement results.

[0092] S207, in response to receiving a law enforcement document generation task, call the law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

[0093] According to some embodiments, the law enforcement document generation agent can, based on the on-site law enforcement result and historical fair law enforcement cases, apply the logical reasoning and summarization ability of the large model to combine structured case information and document templates, and automatically complete the generation of a series of document contents such as on-site transcripts, law enforcement documents, case handling reports, and case-filing decision letters.

[0094] In summary, for the method provided in this embodiment, before law enforcement, law enforcement officers can initiate a one-key analysis task for potential safety hazards, automatically call the intelligent judgment and intelligent text analysis agents to obtain the potential hazard information of the corresponding mining area. After that, initiate a law enforcement plan compilation task, automatically call the plan arrangement agent, and formulate a corresponding law enforcement plan according to the above inspection content, while supporting online editing. Then, when the law enforcement officers arrive at the scene, they can initiate a on-site law enforcement task, which can automatically call the law enforcement assistant agent. It will automatically remind what should be inspected, the inspection steps, the inspection basis, and on-site evidence collection according to the corresponding location; at the same time, real-time problems and solutions can be understood through intelligent Q&A. After law enforcement is completed, a law enforcement document generation task can be initiated, automatically call the document generation agent, and automatically generate various documents such as on-site transcripts, law enforcement documents, and penalty decisions according to the on-site law enforcement and evidence collection situations, and evaluate the assistance process of the law enforcement agents. Therefore, by following the design concept of "data-driven, knowledge-enhanced, intelligent decision-making, and closed-loop response", constructing law enforcement agents, the overall technical route is based on the expert knowledge base in the field of coal mine safety production, applying dynamic knowledge graphs, large language models, and multi-agent collaboration technologies, integrating multi-modal perception, knowledge fusion, intelligent judgment, and automatic response, which can empower the intelligence of on-site law enforcement in supervision and inspection, and provide full-process empowerment for the intelligent law enforcement process of mine safety supervision and inspection law enforcement officers.

[0095] To implement the above embodiment, the present disclosure also proposes an intelligent agent-based mining area supervision and inspection system.

[0096] Exemplarily, Figure 5 is a schematic structural diagram of an intelligent agent-based mining area supervision and inspection system provided by an embodiment of the present disclosure. As Figure 5 shown, the intelligent agent-based mining area supervision and inspection system 500 includes: A potential hazard analysis unit 501, configured to, in response to receiving a potential safety hazard analysis task for a mining area, call an intelligent judgment agent to obtain potential hazard information of the mining area scene, and call an intelligent text analysis agent to obtain potential hazard information of the mining area materials; A plan compilation unit 502, configured to, in response to receiving a law enforcement plan compilation task, call a law enforcement plan arrangement agent to generate a law enforcement plan corresponding to the potential hazard information of the mining area scene and the potential hazard information of the mining area materials; A law enforcement assistance unit 503, configured to, in response to receiving a on-site law enforcement task for a mining area, call a law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; A document generation unit 504, configured to, in response to receiving a law enforcement document generation task, call a law enforcement document generation agent to generate intelligent documents corresponding to the on-site law enforcement result.

[0097] Optionally, the agent-based mining area supervision and inspection system 500 further includes a task acquisition unit for: Obtain the supervision and monitoring information input by law enforcement officers; Decompose the supervision and monitoring information into tasks to determine the supervision and monitoring tasks corresponding to the supervision and monitoring information, where the supervision and monitoring tasks are any one of safety hazard analysis tasks, law enforcement plan compilation tasks, on-site law enforcement tasks, and law enforcement document generation tasks.

[0098] Optionally, the mining area scene hazard information includes personnel active warning information. When the hazard analysis unit 501 is used to obtain the mining area scene hazard information, it specifically is used for: Obtain mining area monitoring images, locate the human figures in the mining area monitoring images, and construct a skeleton model of the human figures; Associate and analyze the movement trajectories of the skeleton models in consecutive frames of the mining area monitoring images to obtain the spatial movement rules and action frequencies of the skeleton models; Based on the temporal action features, match the spatial movement rules and the dynamic behaviors corresponding to the action frequencies. If the dynamic behavior is a personnel active warning dynamic behavior, generate the personnel active warning information corresponding to the dynamic behavior.

[0099] Optionally, the mining area scene hazard information includes abnormal personnel behavior trajectory information. When the hazard analysis unit 501 is used to obtain the mining area material hazard information, it specifically is used for: Obtain mining area monitoring images and detect the personnel movement trajectories in the mining area monitoring images; Obtain positioning card data, and compare the personnel movement trajectories with the positioning card data. If the personnel movement trajectories do not match the positioning card data, generate the abnormal personnel behavior trajectory information corresponding to the personnel movement trajectories and the positioning card data.

[0100] Optionally, the mining area scene hazard information includes abnormal environmental parameter information. When the hazard analysis unit 501 is used to obtain the mining area material hazard information, it specifically is used for: Obtain gas monitoring data; Predict the gas concentration temporal characteristics corresponding to the gas monitoring data; Match the gas monitoring data with the abnormal pattern library to obtain the abnormal samples corresponding to the gas monitoring data; Based on the weighted dynamic time warping algorithm, calculate the similarity between the gas concentration temporal characteristics and the abnormal samples; According to the similarity, determine the abnormal environmental parameter information corresponding to the gas monitoring data.

[0101] Optionally, when the hazard analysis unit 501 is used to call the intelligent text analysis agent to obtain the mining area material hazard information, it specifically is used for: Obtain the multi-source heterogeneous data uploaded by the mining area, and perform feature extraction on the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data; Extract the subgraph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base; Determine the potential hazard information of the mining area materials corresponding to the multi-source heterogeneous data according to the subgraph.

[0102] Optionally, before extracting the subgraph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base, the potential hazard analysis unit 501 is further configured to: Obtain the regulation data in the coal mine safety production knowledge base; Extract the entities and associated production features in the regulation data, and map the associated production features to the relationships between different entities in the initial knowledge graph to obtain the knowledge graph corresponding to the coal mine safety production knowledge base.

[0103] Optionally, when the potential hazard analysis unit 501 is configured to perform feature extraction on the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data, it is specifically configured to: Perform structured processing on the multi-source heterogeneous data to obtain structured multi-source heterogeneous data; Control the business rule engine to perform feature extraction on the structured multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data.

[0104] Optionally, when the potential hazard analysis unit 501 is configured to extract the subgraph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base, it is specifically configured to: Perform word space similarity matching between the feature data and the knowledge graph to obtain the subgraph corresponding to the feature data.

[0105] It should be noted that the foregoing explanation of the embodiment of the mining area supervision and inspection method based on the intelligent agent also applies to the mining area supervision and inspection system based on the intelligent agent of this embodiment, and will not be elaborated here.

[0106] In summary, the system provided by the embodiments of the present disclosure can sink the safety hazard inspection to the mining area data layer by applying the intelligent agent. Hidden major safety hazards gradually emerge through multi-modal multi-source data analysis, solving the problem of the last mile of on-site law enforcement, which is conducive to urging the mining party to actively create safety and prevent major and particularly serious accidents in mines; secondly, the application of the intelligent agent can solve the problem of precise and differentiated law enforcement, significantly reducing the frequency of law enforcement personnel running to mines, which is helpful for improving the production efficiency of the mining area and effectively assisting law enforcement personnel to improve law enforcement efficiency; in addition, by applying the intelligent agent to empower the supervision and law enforcement business process, the supervision and inspection intelligence can be improved based on artificial intelligence technology means, the on-site law enforcement intelligence level can be improved, the law enforcement efficiency can be improved, and the effectiveness, usability and reliability of the traditional supervision and inspection system can be improved.

[0107] To implement the above embodiments, the present disclosure also provides an electronic device, including: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided in the foregoing embodiments.

[0108] Among them, the electronic device includes but is not limited to edge devices, law enforcement devices, etc., and the edge devices include but are not limited to mobile phones, computers, etc. Deploying the method provided in the foregoing embodiments to the electronic device can achieve lightweight edge deployment of the large model.

[0109] To implement the above embodiments, the present disclosure also provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the method provided in the foregoing embodiments when executed by a processor.

[0110] To implement the above embodiments, the present disclosure also provides a computer program product including a computer program, which implements the method provided in the foregoing embodiments when executed by a processor.

[0111] The collection, storage, use, processing, transmission, provision, and disclosure of the user's personal information involved in the present disclosure all comply with the provisions of relevant laws and regulations and do not violate public order and good customs.

[0112] It should be noted that personal information from users should be collected for legal and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after obtaining the informed consent of the user, including but not limited to notifying the user to read the user agreement / user notice and sign an agreement / authorization including authorizing relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others with access to the personal information data comply with their privacy policies and procedures.

[0113] The present disclosure anticipates providing embodiments that allow users to selectively block the use or access of personal information data. That is, the present disclosure anticipates providing hardware and / or software to prevent or block access to such personal information data. Once personal information data is no longer needed, the risk can be minimized by restricting data collection and deleting the data. In addition, when applicable, personal identifiers are removed from such personal information to protect the privacy of the user.

[0114] In the descriptions of the foregoing embodiments, the descriptions with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present disclosure. In this specification, the schematic representations of the above terms are not necessarily directed to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0115] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In the description of the present disclosure, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0116] Any process or method description in a flowchart or otherwise described herein can be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiments of the present disclosure includes additional implementations, where the functions may be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present disclosure pertain.

[0117] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definitional sequence list of executable instructions for implementing logical functions, and can be embodied specifically in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM) or flash memory, an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which a program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or other appropriate processing as necessary, and then storing it in a computer memory.

[0118] It should be understood that various parts of the present disclosure can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0119] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of implementing the above embodiments can be completed by a program instructing relevant hardware, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0120] In addition, each functional unit in various embodiments of the present disclosure may be integrated into one processing module, may exist physically alone for each unit, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0121] The above-mentioned storage medium may be a read-only memory, a magnetic disk, an optical disc, etc. Although the embodiments of the present disclosure have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present disclosure. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present disclosure.

Claims

1. An agent-based mining area supervision and inspection method, characterized in that, Including: In response to receiving a safety hazard analysis task for a mining area, call the intelligent judgment agent to obtain hidden danger information of the mining area scene, and call the intelligent text analysis agent to obtain hidden danger information of the mining area materials; In response to receiving a law enforcement plan compilation task, call the law enforcement plan arrangement agent to generate a law enforcement plan corresponding to the hidden danger information of the mining area scene and the hidden danger information of the mining area materials; In response to receiving a on-site law enforcement task for the mining area, call the law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; In response to receiving a law enforcement document generation task, call the law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

2. The method according to claim 1, wherein The method further includes: Obtain the supervision and monitoring information input by law enforcement personnel; Decompose the supervision and monitoring information into tasks to determine the supervision and monitoring tasks corresponding to the supervision and monitoring information, where the supervision and monitoring tasks are any one of the safety hazard analysis task, the law enforcement plan compilation task, the on-site law enforcement task, and the law enforcement document generation task.

3. The method according to claim 1, wherein The hidden danger information of the mining area scene includes personnel active warning information, and obtaining the hidden danger information of the mining area scene includes: Obtain mining area monitoring images, locate the human figures in the mining area monitoring images, and construct a skeleton model of the human figures; Associate and analyze the movement trajectories of the skeleton models in consecutive frames of the mining area monitoring images to obtain the spatial movement rules and action frequencies of the skeleton models; Match the spatial movement rules and the dynamic behaviors corresponding to the action frequencies based on temporal action features. If the dynamic behavior is a personnel active warning dynamic behavior, generate the personnel active warning information corresponding to the dynamic behavior.

4. The method according to claim 1, characterized in that The hidden danger information of the mining area scene includes abnormal information on personnel behavior trajectories, and obtaining the hidden danger information of the mining area materials includes: Obtain mining area monitoring images and detect the personnel movement trajectories in the mining area monitoring images; Obtain positioning card data, and compare the personnel movement trajectories with the positioning card data. If the personnel movement trajectories do not match the positioning card data, generate abnormal information on the personnel behavior trajectories corresponding to the personnel movement trajectories and the positioning card data.

5. The method according to claim 1, characterized in that, The hidden danger information of the mining area scene includes abnormal information on environmental parameters, and obtaining the hidden danger information of the mining area materials includes: Obtain gas monitoring data; Predict the gas concentration time series characteristics corresponding to the gas monitoring data; Match the gas monitoring data with the abnormal pattern library to obtain the abnormal samples corresponding to the gas monitoring data; Based on the weighted dynamic time warping algorithm, calculate the similarity between the gas concentration time series characteristics and the abnormal samples; According to the similarity, determine the abnormal information on environmental parameters corresponding to the gas monitoring data.

6. The method according to claim 1, wherein The calling the intelligent text analysis agent to obtain the hidden danger information of the mining area materials includes: Obtain the multi-source heterogeneous data uploaded by the mining area, and perform feature extraction on the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data; Extract the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base; Determine the hidden danger information of the mining area materials corresponding to the multi-source heterogeneous data according to the sub-graph.

7. The method according to claim 6, wherein Before extracting the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base, the method further includes: Obtain the regulation data in the coal mine safety production knowledge base; Extract the entities and associated production features in the regulation data, and map the associated production features to the relationships between different entities in the initial knowledge graph to obtain the knowledge graph corresponding to the coal mine safety production knowledge base.

8. The method according to claim 6, characterized in that, The feature extraction of the multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data includes: Perform structured processing on the multi-source heterogeneous data to obtain structured multi-source heterogeneous data; Control the business rule engine to perform feature extraction on the structured multi-source heterogeneous data to obtain the feature data corresponding to the multi-source heterogeneous data.

9. The method according to claim 6, characterized in that The extraction of the sub-graph corresponding to the feature data from the knowledge graph corresponding to the coal mine safety production knowledge base includes: Perform word space similarity matching between the feature data and the knowledge graph to obtain the sub-graph corresponding to the feature data.

10. An agent-based mining area supervision and inspection system, characterized in that, Includes: A hidden danger analysis unit, configured to, in response to receiving a safety hidden danger analysis task for a mining area, call an intelligent judgment agent to obtain the hidden danger information of the mining area scene, and call an intelligent text analysis agent to obtain the hidden danger information of the mining area materials; A plan compilation unit, configured to, in response to receiving a law enforcement plan compilation task, call a law enforcement plan arrangement agent to generate a law enforcement plan corresponding to the hidden danger information of the mining area scene and the hidden danger information of the mining area materials; A law enforcement assistance unit, configured to, in response to receiving a on-site law enforcement task for the mining area, call a law enforcement assistant agent to generate supervision prompt information according to the law enforcement plan, and obtain the on-site law enforcement result corresponding to the supervision prompt information; A document generation unit, configured to, in response to receiving a law enforcement document generation task, call a law enforcement document generation agent to generate an intelligent document corresponding to the on-site law enforcement result.

11. An electronic device, characterized in that, The electronic device includes: A memory, configured to store executable program code; A processor, configured to call and run the executable program code from the memory, so that the electronic device executes the method according to any one of claims 1 to 9.

Images